UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Morbidity and mortality related to tuberculosis (tb) in British Columbia (BC), Canada Moniruzzaman, Akm 2010

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata

Download

Media
24-ubc_2010_fall_moniruzzaman_akm.pdf [ 1.74MB ]
Metadata
JSON: 24-1.0071046.json
JSON-LD: 24-1.0071046-ld.json
RDF/XML (Pretty): 24-1.0071046-rdf.xml
RDF/JSON: 24-1.0071046-rdf.json
Turtle: 24-1.0071046-turtle.txt
N-Triples: 24-1.0071046-rdf-ntriples.txt
Original Record: 24-1.0071046-source.json
Full Text
24-1.0071046-fulltext.txt
Citation
24-1.0071046.ris

Full Text

MORBIDITY AND MORTALITY RELATED TO TUBERCULOSIS (TB) IN BRITISH COLUMBIA (BC), CANADA  by  AKM MONIRUZZAMAN MBBS, UNIVERSITY OF DHAKA, 1994 MBA, UNIVERSITY OF DHAKA, 1999 M.Sc., UNIVERSITY OF BRITISH COLUMBIA, 2004  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES (Health Care and Epidemiology)  THE UNIVERSITY OF BRITISH COLUMBIA (VANCOUVER)  July 2010  © Akm Moniruzzaman, 2010  ABSTRACT BACKGROUND: The epidemiology of tuberculosis (TB) related recurrence and mortality is not well characterized for British Columbia (BC) or Canada. The objectives of this thesis were: 1) to estimate the incidence of recurrence of TB and identify predictors associated with TB recurrence; to investigate the relative contribution of exogenous re‐infection as a mechanism of recurrence; to characterize mortality among TB patients and to identify potentially modifiable risk factors for patients whose deaths were attributable to TB. METHODS: This study was conducted using population‐based data maintained by the centralized provincial TB service (TB Control Division at BC Center for Disease Control). All TB cases recorded with this Division from 1990 to 2006 were reviewed. TB patients who developed recurrence and who died during the observation period were identified. Cox regression was performed to identify risk factors associated with recurrence and mortality. RESULTS: During the study period (1990 to 2006), over 5400 TB patients were registered with the provincial TB control program. The incidence of recurrence was 370 per 100,000 person‐ years (pys). Several factors such as foreign‐birth, incomplete treatment, poor‐compliance to treatment, place of initial diagnosis, HIV and drug abuse were significantly associated with TB recurrence. The relative contribution of re‐infection was 8% and the incidence of reinfection was 19 per 100,000 PYs. There was an excess mortality among TB patients compared to the general BC population. 1069 TB patients died during 1990 to 2006 (29 per 1000 PYs). The cumulative mortality at the first 6 and 12 months were 8% and 10% respectively. Increasing age, Aboriginal ethnicity, miliary TB, HIV/AIDS, alcoholism, substance abuse, etc was significantly associated with all‐cause mortality in multivariable analyses. Among these deaths, 109 (109/5408=2%) deaths were primarily caused by TB and another 177 (177/5408=3.3%) deaths were partially contributed by TB (one of the causes of death). Miliary TB, far advanced PTB and Aboriginal ethnicity were the strongest predictors of mortality related to TB. CONCLUSIONS: This study identified several important risk factors in a population‐based TB cohort. Effective interventions targeting these high‐risk populations are urgently required in order to prevent recurrence and mortality related to TB.  ii  TABLE OF CONTENTS ABSTRACT .........................................................................................................................................ii TABLE OF CONTENTS....................................................................................................................... iii LIST OF TABLES ............................................................................................................................... vii LIST OF FIGURES ............................................................................................................................... x LIST OF ABBREVIATIONS ................................................................................................................. xi ACKNOWLEDGEMENTS .................................................................................................................. xii DEDICATION .................................................................................................................................. xiii CO‐AUTHORSHIP STATEMENT ...................................................................................................... xiv CHAPTER 1: BACKGROUND, STUDY JUSTIFICATION, AND OBJECTIVES .......................................... 1 1.1 OVERVIEW OF THE DISSERTATION ....................................................................................... 2 1.2 BACKGROUND ....................................................................................................................... 2 1.2.1 Global Burden of Tuberculosis ....................................................................................... 2 1.2.2 Tuberculosis in Canada .................................................................................................. 3 1. 2.3 Recurrent TB ................................................................................................................. 4 1.2.4 Mechanism of Recurrence: Relapse vs. Reinfection...................................................... 5 1.2.5: Tuberculosis Mortality .................................................................................................. 7 1.3 RATIONALE AND JUSTIFICATION ........................................................................................... 8 1.4 OBJECTIVES ........................................................................................................................... 9 1.5 HYPOTHESES ......................................................................................................................... 9 1.6 REFERENCES ........................................................................................................................ 11 CHAPTER 2: A POPULATION BASED STUDY OF RECURRENT TUBERCULOSIS IN BRITISH COLUMBIA, CANADA..................................................................................................................... 20 2.1. INTRODUCTION .................................................................................................................. 21 2.2 METHODS ............................................................................................................................ 22 2.2.1 Setting and Source of Information .............................................................................. 22 2.2.2 Study Design ................................................................................................................ 23 2.2.3 Study Population and Relevant Definition................................................................... 23  iii  2.2.4 Follow‐up Time ............................................................................................................ 24 2.2.5 Variables Description ................................................................................................... 25 2.2.6 Statistical Analysis ........................................................................................................ 35 2.3 RESULTS............................................................................................................................... 36 2.3.1 Prevalence of Recurrence ............................................................................................ 36 2.3.2 Incidence of Recurrence .............................................................................................. 37 2.4 DISCUSSION ......................................................................................................................... 49 2.5 CONCLUSIONS ..................................................................................................................... 58 2.6 REFERENCES ........................................................................................................................ 60 CHAPTER 3: A POPULATION BASED STUDY OF TUBERCULOSIS RECURRENCE‐RELAPSE VERSUS REINFECTION ................................................................................................................................. 67 3.1. INTRODUCTION .................................................................................................................. 68 3.2 METHODS ............................................................................................................................ 69 3.2.1 Recurrent Cases ........................................................................................................... 72 3.2.2 Follow‐up Time ............................................................................................................ 72 3.2.3 Mycobacteriology ........................................................................................................ 72 3.2.4 Culture and DNA Extraction ......................................................................................... 73 3.2.5 DNA Fingerprinting ...................................................................................................... 73 3.2.6 Variables Description ................................................................................................... 74 3.2.7 Statistical Analysis ........................................................................................................ 74 3.3 RESULTS............................................................................................................................... 75 3.3.1 Description of Study Cohort......................................................................................... 75 3.3.2 Recurrence ................................................................................................................... 77 3.3.3 Culture Positive Recurrent Cases ................................................................................. 79 3.3.4 Reinfection and Relapse .............................................................................................. 89 3.4 DISCUSSION ......................................................................................................................... 91 3.5 CONCLUSIONS ..................................................................................................................... 94 3.6 REFERENCES ........................................................................................................................ 95  iv  CHAPTER 4: A SYSTEMATIC REVIEW ON RISK FACTORS OF MORTALITY AMONG TB PATIENTS 101 4.1 INTRODUCTION ................................................................................................................. 102 4.2 METHODS .......................................................................................................................... 103 4.2.1 Data Extraction and Calculation of Death Rate/Proportion ...................................... 104 4.3 RESULTS............................................................................................................................. 105 4.3.1 Risk Factors for Mortality among TB Patients ........................................................... 113 4.4 DISCUSSION ....................................................................................................................... 123 4.5 CONCLUSIONS ................................................................................................................... 129 4.6 REFERENCES ...................................................................................................................... 130 CHAPTER 5: AN ALL CAUSE MORTALITY STUDY AMONG PATIENTS WITH TUBERCULOSIS IN BRITISH COLUMBIA ..................................................................................................................... 138 5.1 INTRODUCTION ................................................................................................................. 139 5.2 METHODS .......................................................................................................................... 140 5.2.1 Variables Description ................................................................................................. 140 5.2.2 Statistical Analysis ...................................................................................................... 141 5.3 RESULTS............................................................................................................................. 145 5.3.1 Description of Study Cohort and Mortality ............................................................... 145 5.3.2 Comparisons of Mortality between TB Cohort and General Population .................. 152 5.3.3 Follow‐up and Risk Factors for All‐cause Mortality among TB Patients .................... 153 5.4 DISCUSSION ....................................................................................................................... 162 5.5 CONCLUSIONS ................................................................................................................... 172 5.5 REFERENCES ...................................................................................................................... 174 CHAPTER 6: A TB‐RELATED MORTALITY STUDY IN BRITISH COLUMBIA, CANADA FROM 1990‐ 2006 ............................................................................................................................................ 182 6.1 INTRODUCTION ................................................................................................................. 183 6.2 METHODS .......................................................................................................................... 184 6.2.1 Cause of Death ........................................................................................................... 185 6.2.2 Variables Description ................................................................................................. 187 6.2.3 Statistical Analysis ...................................................................................................... 188 v  6.3 RESULTS............................................................................................................................. 189 6.4 DISCUSSION ....................................................................................................................... 205 6.4.1 Older Age ................................................................................................................... 207 6.4.2 Failure of Diagnosis/Missed Diagnosis ...................................................................... 208 6.4.3 Miliary TB ................................................................................................................... 209 6.4.4 Aboriginal People ....................................................................................................... 210 6.4.5 Caucasian ................................................................................................................... 211 6.4.6 Far Advanced PTB and CNS/Meningeal TB ................................................................ 211 6.4.7 Drug Resistance.......................................................................................................... 212 6.4.8 HIV .............................................................................................................................. 212 6.4.9 Co‐morbidities ........................................................................................................... 213 6.5 CONCLUSIONS ................................................................................................................... 216 6.6 REFERENCES ...................................................................................................................... 217 CHAPTER 7: SUMMARY, CONTRIBUTIONS, POLICY IMPLICATIONS & RECOMMENDATIONS, FUTURE RESERACH AND CONCLUSIONS ..................................................................................... 225 7.1 SUMMARY OF OBJECTIVES ............................................................................................... 226 7.2 SUMMARY OF FINDINGS ................................................................................................... 226 7.3 UNIQUE CONTRIBUTIONS TO RESEARCH.......................................................................... 229 7.4 LIMITATIONS ..................................................................................................................... 230 7.5 POLICY IMPLICATIONS AND RECOMMENDATIONS FOR TB CONTROL PROGRAM ........... 231 7.6 FUTURE RESEARCH............................................................................................................ 237 7.7 CONCLUSIONS ................................................................................................................... 238 7.8 REFERENCES ...................................................................................................................... 240 APPENDIX A ................................................................................................................................. 244 APPENDIX B: HUMAN ETHICS APPROVAL CERTIFICATE.............................................................. 260  vi  LIST OF TABLES Table 2.1: Characteristics of study cohort (N=4464) .................................................................... 39 Table 2.2: Occurrence of incident recurrent cases (n=108) over the follow‐up period (interval between treatment completion time and date of recurrence).................................................... 40 Table 2.3: Occurrence of recurrence (incidence) among foreign‐born individuals (n=87) since their arrival in Canada ................................................................................................................... 42 Table 2.4: Comparisons of socio‐demographic characteristics between TB patients who developed a recurrence (n=108) and who did not (n=4356) ....................................................... 43 Table 2.5: Comparisons of treatment and related factors between TB patients who developed a recurrence (n=108) and who did not (n=4356) ............................................................................ 44 Table 2.6: Comparisons of co‐morbidity and other clinical characteristics between TB patients who developed a recurrence (n=108) and who did not (n=4356) ............................................... 45 Table 2.7: Multivariable cox regression analysis of risk factors for recurrence ........................... 47 Table 2.8: Incidence of recurrence across several risk factors among TB patients who successfully completed treatment during the primary episode (n=3529) ................................... 48 Table 3.1: Socio‐demographic characteristics of study cohort .................................................... 76 Table 3.2: Treatment and other related characteristics of study cohort ..................................... 78 Table 3.3: Co‐morbidity and behavioral characteristics of TB patients in BC between 1995 and 2006 .............................................................................................................................................. 79 Table 3.4: Incidence of recurrence, relapse and reinfection ........................................................ 81 Table 3.5: Characteristics of 24 culture positive recurrent cases with available DNA fingerprint82 Table 3.6: Characteristics of 24 culture positive recurrent cases with paired isolates ................ 84 Table 3.7: Treatment (Rx) characteristics of 24 culture positive recurrent cases during their initial episode ................................................................................................................................ 86 Table 3.8: Characteristics of specimens from 24 culture positive recurrent cases during their initial and recurrent episode ........................................................................................................ 88 Table 4.1: Mortality among TB patients from selected studies (n=66) ..................................... 107 Table 4.2: Mortality according to the burden of TB disease (low, intermediate and high) ....... 108  vii  Table 4.3: Socio‐demographic risk factors for dying in patients with Tuberculosis (TB) ........... 115 Table 4.4: Co‐morbidities associated with mortality in patients with Tuberculosis (TB) ........... 117 Table 4.5: Health‐care delivery and treatment‐ related risk factors of mortality ...................... 119 Table 4.6: Risk factors associated with diagnosis and the type of disease ................................ 121 Table 5.1: Socio‐demographic characteristics of study cohort (n=5408) ................................... 146 Table 5.2: Comparisons of crude death rates (CDR) and age standardized mortality ratios (ASMR) in British Columbia from 1990 to 2006 (TB cohort vs. general population) ................. 147 Table 5.3: Number of deaths among TB patients in BC over time (1990‐2006) by birth cohort 150 Table 5.4: Comparisons of socio‐demographic characteristics between TB patients who died during follow up period (n=953) and who survived (n=4068) .................................................... 156 Table 5.5: Comparisons of diagnosis type and related factors between TB patients who died during follow up period (n=953) and who survived (n=4068) .................................................... 157 Table 5.6: Comparisons of co‐morbidities between TB patients who died during follow up period (n=953) and those who survived (n=4068) ..................................................................... 158 Table 5.7: Multivariable cox regression analysis for risk factors of mortality among TB patients ..................................................................................................................................................... 159 Table 5.8: Univariate and multivariable logistic regression analysis for risk factors of mortality for patients who died during treatment (n=309) and who did not (n=4904) ............................ 161 Table 6.1: Socio‐demographic characteristics of study cohort (n=5408) ................................... 190 Table 6.2: Characteristics of TB patients who died during 1990 to 2006 in British Columbia (N=1069) ..................................................................................................................................... 191 Table 6.3: Number of cause‐specific deaths by calendar year (1990 to 2006) .......................... 193 Table 6.4: Occurrence of TB‐related death during the treatment period .................................. 196 Table 6.5: Socio‐demographic and diagnosis‐related characteristics and prevalence of associated illness across several types of death (n=1069) and survivors (n=4339) ................... 197 Table 6.6: Univariate multinomial logistic regression analysis for socio‐demographic risk factors of TB‐related mortality ............................................................................................................... 200 Table 6.7: Univariate multinomial logistic regression analysis for diagnosis type and related risk factors of TB‐related mortality ................................................................................................... 201 viii  Table 6.8: Univariate multinomial logistic regression analysis for co‐morbid risk factors of TB‐ related mortality ......................................................................................................................... 202 Table 6.9: Multivariable multinomial logistic regression analysis for risk factors of TB‐related mortality among TB patients ...................................................................................................... 203 Table 6.10: Comparisons of socio‐demographic and clinical characteristics between TB patients who were diagnosed postmortem (137) and who were diagnosed alive (n=5235) .................. 204 Table 6.11: Relationship between HIV and cause of death ........................................................ 205 Table A.1: Socio‐demographic and clinical characteristics of study cohort (n=5403)................ 245 Table A.2: Comparisons of socio‐demographic characteristics between TB patients who had a history of recurrence (n=490) and who had not (n=4913) ......................................................... 247 Table A.3: Comparisons of treatment and related behaviors between TB patients who had a history of recurrence (n=490) and who had not (n=4913) ......................................................... 248 Table A.4: Comparisons of co‐morbidity and other clinical characteristics between TB patients who developed a recurrence (n=108) and who did not (n=4356) ............................................. 249 Table A.5: Multivariable cox regression analysis to evaluate the effect of co‐morbid conditions on TB recurrence......................................................................................................................... 250 Table A.6: Risk factors for recurrence among patients with more than 6 months of inactivity 251 Table A.7: Distribution of treatment incompletion and non‐compliance to treatment across several risk factors ...................................................................................................................... 252 Table A.8: Lists of variables that were included in multivariable cox regression model for risk factors of all‐cause mortality among TB patients ....................................................................... 253 Table A.9 Comparisons of co‐morbidity between TB patients who died during follow up period (n=953) and who survived (n=4068) ........................................................................................... 256 Table A.10: Multivariable cox regression analysis for risk factors of mortality among TB patients ..................................................................................................................................................... 257 Table A.11: Comparisons of risk factors among canadian‐born non‐aboriginals, aboriginals, and foreign‐born people (n=5191) .................................................................................................... 258 Table A.12: Univariate multinomial logistic regression analysis for co‐morbid risk factors of TB‐ related mortality ......................................................................................................................... 259 ix  LIST OF FIGURES Figure 2.1: Study population for incidence of recurrence (n=4464) ............................................ 26 Figure 2.2: Trend of recurrent (prevalent) cases from 1990 to 2006 in British Columbia, Canada ....................................................................................................................................................... 38 Figure 2.3: Trend of incident recurrent cases in British Columbia from 1990 to 2006 ................ 41 Figure 3.1: Study population ......................................................................................................... 71 Figure 3.2: MIRU‐VNTR results of 24 recurrent TB cases with paired isolated ............................ 90 Figure 4.1: Mortality among TB patients according to background incidence of disease ......... 111 Figure 4.2: Mortality among HIV positive and HIV negative TB patients by background incidence of disease .................................................................................................................................... 112 Figure 5.1: Study population for all‐cause mortality among TB patients (n=5408) ................... 143 Figure 5.2: Trend of crude death rates for TB cohort in British Columbia from 1990 to 2006 .. 148 Figure 5.3: Trend of age standardized mortality ratios (ASMR) for TB cohort in British Columbia from 1990 to 2006 ...................................................................................................................... 149 Figure 5.4: Trend of deaths among TB patients in British Columbia from 1990 to 2006 by birth cohort .......................................................................................................................................... 151 Figure 5.5: Histogram of age at death for TB patients who died between 1990 and 2006 (n=1069) ...................................................................................................................................... 153 Figure 5.6: Survival curve among aboriginal (bottom line), canadian‐born non‐aboriginal and foreign‐born (top line) TB patients ............................................................................................. 155 Figure 6.1: Study population for TB‐related mortality in British Columbia (n=5408) ................ 186 Figure 6.2: Trend of cause‐specific death percentage for TB cohort in British Columbia from 1990 to 2006 ............................................................................................................................... 194 Figure 6.3: TB‐related death rate (per 100, 000 BC population) in BC from 1990‐2006............ 195  x  LIST OF ABBREVIATIONS -  AHR: Adjusted Hazard Ratio  -  AIDS: Acquired Immune Deficiency Syndrome  -  AOR: Adjusted Odds Ratio  -  BC: British Columbia.  -  BCCDC: British Columbia Centre for Disease Control  -  CB: Canadian‐born  -  CI: Confidence Interval  -  DTB: Division of TB Control of British Columbia  -  EMB: Ethambutol  -  EPTB: Extra‐pulmonary Tuberculosis  -  FB: Foreign‐born  -  HIV: Human Immunodeficiency Virus  -  INH: Isoniazid  -  MDR: Multi‐drug Resistance  -  MIRU‐VNTR: Mycobacterial Interspersed Repetitive Unit‐Variable Number  Tandem Repeats -  MTB: M. tuberculosis  -  PTB: Pulmonary Tuberculosis  -  RFLM: Restricted Fragment Length Polymorphism  -  PZA: Pyrazinamide  -  RM: Rifampin  -  SM: Streptomycin  -  TB: Tuberculosis  -  UHR: Unadjusted Hazard Ratio  -  UOR: Unadjusted Odds Ratio  -  WHO: World Health Organization  xi  ACKNOWLEDGEMENTS I offer sincere gratitude to my co‐supervisors, Dr. Arminee Kazanjian and Dr. J Mark FitzGerald, who shared so generously of their time, knowledge and general enthusiasm for research. I also wish to thank my committee members Dr. Kevin Elwood and Dr. Hubert Wong for their invaluable mentorship and support. They consistently provided me with sound advice, great learning opportunities and support in all forms. They really made the completion of this degree a great learning experience. I feel very privileged to have had such an amazing thesis committee and will do my best to follow the example they have set.  I also offer sincere thanks to Ms. Fay Hutton and other staff at the Division of TB Control in British Columbia, for their hard work in preparing and maintaining a comprehensive TB database. I would like to thank the staff at the Provincial TB laboratory for preparing specimens for the molecular epidemiology study. I would also like to thank Canadian Institutes of Health Research (CIHR) and National Reference Mycobacterium Laboratory in Winnipeg for their generous support. I would also like to thank the School of Population and Public Health for providing a fantastic learning environment filled with opportunities. My special thanks to Drs. Patricia Spittal and Chris Richardson for their continued support.  Special thanks to my wife, my family and my mother for their endless support throughout my Ph. D. I also extend my gratitude to my peers and colleagues: Margo Pearce, Sheetal Patel, Elisa Lloyd‐Smith, Hasanat Alamgir, Margot Kuo and Darlene Taylor.  xii  DEDICATION  To the memory of my father: Md. Khalilur Rahman  xiii  CO­AUTHORSHIP STATEMENT This statement describes the contributions of PhD candidate to the manuscripts presented in this thesis. The thesis was conceived, instrumented, written and disseminated by PhD candidate. The co‐authors of the manuscripts that were identified in this dissertation made contributions only if it deemed appropriate with committee or collegial duties. Dr. Arminee Kazanjian and Dr. J. Mark FitzGerald were thesis co‐supervisors. Dr. Hubert Wong and Dr. Kevin Elwood were other committee members. Each of the co‐authors reviewed all the manuscripts and provided feedback. However, the student was responsible for preparing the drafts as well as revisions of all manuscripts as suggested by co‐authors.  xiv  CHAPTER 1: BACKGROUND, STUDY JUSTIFICATION, AND OBJECTIVES  1  1.1 OVERVIEW OF THE DISSERTATION The PhD dissertation incorporates a manuscript‐based format as outlined by the Faculty of Graduate Studies at the University of British Columbia. There are seven chapters in this thesis. The first chapter is the introductory chapter, which discusses background, study justification and the objectives of this thesis. Each chapter from two to six is representative of an individual manuscript resulted from thesis related research. The primary focus of Chapter 2 is the risk factors for recurrent tuberculosis (TB), whether Chapter 3 is an estimation of re‐infection and relapse as a mechanism of recurrent TB. Chapter 4 is a systematic review of published literature on TB‐related mortality. Chapter 4 and 5 discuss all‐cause mortality and cause‐specific mortality among TB patients respectively. Chapter 7 is the concluding chapter that summarizes the key findings from Chapters two to six, and outlines implications for future research and intervention efforts.  1.2 BACKGROUND 1.2.1 Global Burden of Tuberculosis In 1990, the World Health Organization (WHO) conducted a special study to estimate the global burden of tuberculosis (TB). This report showed that approximately one third of the global population was infected with Mycobacterium tuberculosis (MTB) (1). Since the publication of the WHO report, the epidemiology of TB has changed in response to a number of factors including the increased prevalence of human immunodeficiency virus (HIV), the emergence of drug resistance disease and changes in TB management practices (2). Currently, there are more new infections than ever, which translate a newly infected person with TB bacillus every second (3). Fortunately, but not every case of latent TB infection progresses to active disease, only 5‐ 10% of all infected cases will develop active illness during their life time (3,4).  2  The most recent estimate published in 2009 predicted a total of 9.4 million new TB cases globally in 2008 (5), a substantial increase of TB cases than that in last decades (6.6 million and 8.3 million in 1990 and 2000 respectively) (6). The burden of this disease has been greatest in Southeast Asia and sub‐Saharan Africa in terms of both the absolute number of cases and incidence. For example, about 55% of these incident cases occurred in Asian countries and 31% in African countries. According to geographical region, the sub‐Saharan countries have, overall, the highest incidence of TB (350 per 100, 000), almost two‐times higher than that of Southeast Asia region, which accounted for the highest number of new TB cases (3). Until 2005, the global incidence of TB had been increasing by about 1.1% annually despite the increased global efforts to control this disease (7). Since then, although the global incidence of TB has been falling, but the annual rate of decline has been very slow (less than 1%).  1.2.2 Tuberculosis in Canada According to Health Canada, approximately 1600 new TB cases (e.g., 1547 new and re‐ treatment cases in 2007) occur each year in Canada (8‐10). Although the overall TB incidence in Canada has shown a decreasing trend during the last two decades, the burden of TB recurrence did not decline proportionally. Alberta, British Columbia (BC), Ontario and Quebec‐ the four most populous provinces of Canada – are where approximately 85% of the total TB burden of TB occurs , with BC having the highest incidence rate of TB among these provinces (8,9). In 2007, the TB incidence rate for BC was 6.3 per 100, 000 (national rate was 4.7) compared to a rate of 5.1 (per 100, 000) in Ontario, 3.2 in Alberta and 3.0 in Quebec (8). There is also an over‐ representation of TB among Canadian‐born aboriginals. In 2007, Canadian‐born aboriginal persons accounted for 20% of total reported cases, while Canadian‐born non‐Aboriginal persons only made up 11% of total cases despite making up a much higher proportion of the population (8).  In Canada, like many developed countries where a large number of immigrants are admitted each year, the majority of TB cases occur among the foreign‐born. In Canada, approximately, 3  80% of immigrants originate from a country where the burden of TB is very high (9). The proportion of all TB cases occurring in the foreign‐born has increased significantly over time ‐ from 18% in 1970 to 66% in 2007 (8,10). In BC, 75% of all TB cases were among foreign‐born people (although immigrants accounted for only a quarter of total BC population according to 2006 census) and the incidence of TB among foreign‐born people was 19.2 per 100,000 (1.6 among non‐Aboriginal Canadians) in 2004 (11).  To address the large number of cases among the foreign‐born, TB control programs in the industrialized nations need to actively support the development and implementation of TB control initiatives in developing countries (12‐14). A recent article indicates that investment in immigrants’ countries of origin, where the incidence of diseases is particularly high, may be a cost effective means of reducing the incidence of TB among future immigrants migrating to countries with a low burden of TB (14). The Canadian Lung Association has also recommended a similar strategy.  1. 2.3 Recurrent TB TB recurrence (the term retreatment is also frequently used as a measure of recurrence globally) is a second episode of TB that takes place after the first occurrence has been defined as cured or inactive (15). In low incidence countries (e.g., Canada) attempting to eliminate TB, prevention of recurrence is critical for control strategies because it helps to reduce the transmission of TB by decreasing the number of contagious cases. From a monitoring perspective, the recurrence/retreatment rate is useful for assessing the quality of TB control programs due to its potential association with treatment coverage and completion rates (16). Several factors such as age, male gender, Aboriginal ethnicity, incomplete TB treatment, drug resistant status, silicosis, alcoholism, HIV co‐infection, and extent of disease (e.g., advanced stage, cavitations) have been found to be associated with recurrence (17‐26). However, the majority of studies, focusing on recurrent disease, have not been conducted in population‐ based cohorts but rather in an institutional or clinical setting, which limits the application of 4  study findings to the general population. Moreover, population based studies, which more accurately reflect burden of illness, can be of great use to health policy makers at the national and international level. Additionally, information on TB recurrence from the Canadian setting is limited. A case‐control study was conducted in the province of Manitoba during the 1980s, which was limited by a relatively small number of cases, study methodology and duration of follow‐up (23). The study also did not provide an estimate of incidence of recurrence. The estimation of incidence and identification of risk factors for recurrence using population based data represents a potentially useful means of improving TB control programs by targeting the high‐risk population.  1.2.4 Mechanism of Recurrence: Relapse vs. Reinfection Recurrence may be due to either endogenous reactivation (relapse) or exogenous re‐infection. If both the recurrent episode and primary episode of TB are caused by the same strain of TB bacillus, then it is considered to be a relapse. However, if a different strain is found for both the initial and recurrent episode, re‐infection is considered the likely mechanism. Although complex, it is important to know whether the recurrence is linked to ongoing community transmission (i.e., re‐infection) or has occurred because of inadequate treatment of the initial episode (i.e., relapse). Re‐infection is thus an important indicator of community transmission. A precise estimate of the extent of re‐infection existing in specific communities is vital when attempting to prioritize control efforts (e.g., effective detection and management of contagious cases) to decrease the number of cases with new infection and re‐infection.  It is noteworthy to mention that the relative contribution of reinfection to recurrence has direct implications on the development of an improved vaccine against TB (15,27). If the infection from the first episode does not provide protective immunity against a new strain of Mycobacterium tuberculosis, the development of new TB vaccines based on existing knowledge of host defenses becomes more challenging (15,27,28). Moreover, if re‐infection is a frequent cause of recurrence, then the evaluation of TB treatment regimens used in clinical trials 5  becomes problematic because it can be difficult to determine their effectiveness (i.e., successfully treated patients may unknowingly become reinfected during the study) (29). This type of information is vital to the development of efficient and effective TB management programs.  The relative contributions of relapse and reinfection as a cause of recurrence has been debated for many decades, since both types of recurrences are not clinically distinct (15,27). Studies published in the last few decades described relapse exclusively as the mechanism for recurrence (29‐31). During this period, TB researchers were limited in resources or access to techniques that could differentiate relapse from re‐infection. However, with the advent of molecular epidemiology techniques, it is now possible to make quantitative assessment of the contribution of reinfection to recurrence.  To date, few studies have been conducted to explicitly examine the contribution of reinfection to TB recurrence. The relative contribution of re‐infection varies widely (0‐100%) across the studies that have been conducted (29,31‐44). The majority of these studies have not been population based and reported only the proportion of recurrence due to reinfection (no incidence estimate of reinfection was available). A recent systematic review (15) on reinfection raised concerns about the quality of many of the included studies. The authors of this review (15) strongly recommended that future studies assessing this question should utilize a more rigorous methodology to allow for an improved understanding of the possible mechanisms of recurrence. Although studies have been conducted in several countries (mostly in the developed world), no such information is available in Canada. A study (43) on reinfection was recently conducted across North America among patients participating in a clinical drug trial, no specific estimates were provided for Canadian patients. In addition because this was a clinical trial, with very selective recruitment of patients, the study does not provide a true population based estimate.  6  1.2.5: Tuberculosis Mortality TB‐related mortality continues to pose a significant public health problem locally and internationally. Globally, TB accounts for 6% of all deaths and more than one fourth of all preventable adult deaths (45‐47). An estimated two million people die of TB each year (1.8 million deaths in 2008 including 0.5 million HIV co‐infected deaths) with a case fatality rate (CFR) of up to 53% (overall global CFR is 23%) %) (2,5,6). The number of deaths caused by TB bacillus was higher than that of any other infectious agent in the pre‐HIV era (1,48). Currently TB is the second commonest infectious cause of death due to a single agent (HIV ranks first). In the modern era, all TB related deaths are potentially preventable. This raises an important question, why do so many patients diagnosed with TB die when for the majority of cases effective therapy is available. It is commonly acknowledged that further epidemiological research is required on the determinants of TB mortality.  Although studies conducted in this area have identified several factors (e.g., older age, HIV infection, multi‐drug resistance, co‐morbid conditions, and immune‐suppression status) as being associated with TB‐related mortality (46,49‐55), the epidemiological risk factors for mortality in TB patients requires further exploration. In addition, there is a scarcity of longitudinal, population‐based studies. Most studies have been conducted on hospitalized patients or in selected populations (46,50,51,56‐63)  Studies based on narrowly defined populations are prone to have limited external validity, which often affects their application to the general population. Research focused on examining the impact of preventable risk factors on TB related mortality in a general population setting is clearly required to address these gaps. In addition, there is limited Canadian research evaluating TB mortality (64‐66). A study (64) evaluating TB‐related mortality was conducted in BC during the 1980s but this study was conducted in the pre‐HIV era. Since then, the epidemiology of TB has changed substantially both globally and in Canada. In addition, the prior BC study was limited by in terms of the number of cases, analytical techniques, and study  7  duration. Recently, concern regarding TB mortality has emerged as a problem in many industrialized countries of world (e.g., the USA) due to an increase in the number of TB related deaths (56,67,68). This situation reinforces the need for an investigation of the risk factors for TB related mortality in a Canadian setting.  1.3 RATIONALE AND JUSTIFICATION Despite to increased efforts focusing on TB control, this condition continues to pose a significant public health problem throughout the world. Canada, a country with a low incidence of TB aims to eliminate TB. Prevention of infection and re‐infection are important components of this elimination strategy. There is significant mortality among TB patients despite available treatment strategies, which, at least for drug sensitive disease, is effective. Despite this there is a limited body of research on TB recurrence and mortality in Canadian setting. Data from studies conducted in other countries might not be relevant to Canada due to fundamental differences in health care systems and population demographics. In addition, there are only a limited number of population‐based studies. Studies based on selective groups of individuals are prone to have limited external validity, which often affects their application to the general population.  Moreover, there is a scarcity of systematic reviews on TB‐related morality. In addition, the majority of mortality studies among TB patients evaluated the risk of dying from TB while patients were on treatment. Only a limited number of studies investigated specific causes of death among TB patients and identified risk factors among TB patients whose deaths were truly attributable to TB (64,69,70). These studies demonstrated that a substantial proportion of deaths while on treatment were not directly related to TB (the underlying cause), rather the deaths were attributed to other illness/causes. Evaluating the performance of TB control program based deaths during treatment might be misleading. Interventions based on such findings have limited applicability in preventing deaths from other causes. Therefore, it is widely acknowledged that more epidemiological research is required on the rates, patterns and determinants of TB recurrence and mortality. 8  1.4 OBJECTIVES This thesis explores the following objectives: 1. To estimate the incidence of TB recurrence and to identify socio‐demographic and clinical risk factors for recurrent TB in all TB patients in British Columbia 2. To estimate the incidence of re‐infection (exogenous re‐infection) and relapse (endogenous reactivation) as the cause of recurrence using molecular epidemiology techniques (DNA finger printing) in patients with culture proven TB 3. Conduct a systematic review of the literature on potential risk factors for TB‐related mortality 4. To characterize the mortality among TB patients, to investigate the trend of deaths compared to general population of BC and identify the risk factors associated with mortality among TB patients 5. To identify potentially modifiable or preventable risk factors for patients whose deaths were attributable to TB in all TB patients in British Columbia  1.5 HYPOTHESES   Socio‐demographic and clinical factors including foreign birth, drug‐resistance status, HIV co‐infection, non‐adherence to TB treatment, and radiographic evidence of cavitation will be positively associated with tuberculosis recurrence.    The relative contribution of re‐infection (exogenous reinfection) will be lower compared to relapse (endogenous reactivation) among culture proven TB cases and occurrence of re‐infection (rate of reinfection) parallels the overall TB incidence in the general population.    TB patients experience excess mortality compared to the general population    Specific modifiable risk factors such as failure of diagnosis/missed diagnosis, alcoholism, drug abuse and HIV co‐infection will be positively correlated with TB‐related mortality.  9  Findings from this thesis, described in the following chapters, will improve our understanding of the epidemiology of TB recurrence and mortality and provide a basis for health policy makers and service planners to implement effective interventions targeting high risk patients.  10  1.6 REFERENCES  (1) Kochi A. The global tuberculosis situation and the new control strategy of the World Health Organization. Tubercle 1991 Mar;72(1):1‐6. (2) Dye C, Scheele S, Dolin P, Pathania V, Raviglione MC. Consensus statement. Global burden of tuberculosis: estimated incidence, prevalence, and mortality by country. WHO Global Surveillance and Monitoring Project. JAMA 1999 Aug 18;282(7):677‐686. (3) World Health Organization. Fact Sheets on Tuberculosis. 2009; Available at: http://www.who.int/tb/publications/factsheets/en/index.html. Accessed April/02, 2010. (4) World Health Organization. Global Tuberculosis Control 2009: Epidemiology Strategy Financing. 2009; Available at: http://www.who.int/tb/publications/global_report/2009/pdf/full_report.pdf. Accessed April/02, 2010. (5) World Health Organization. Tuberculosis Facts 2009 Update. 2009; Available at: http://www.who.int/tb/publications/2009/factsheet_tb_2009update_dec09.pdf. Accessed April/02, 2010. (6) Dolin PJ, Raviglione MC, Kochi A. Global tuberculosis incidence and mortality during 1990‐ 2000. Bull.World Health Organ. 1994;72(2):213‐220. (7) Maher D, Raviglione M. Global epidemiology of tuberculosis. Clin.Chest Med. 2005 v; Jun;26(2):167‐182. (8) Public Health Agency of Canada. Tuberculosis in Canada 2007, Pre‐Release. 2007; Available at: http://www.phac‐aspc.gc.ca/publicat/2008/tbcanpre07/pdf/tbpre2007‐eng.pdf. Accessed March/31, 2010.  11  (9) Public Health Agency of Canada and Canadian Lung Association. Canadian Tuberculosis Standards, 6th Edition. 2007; Available at: http://www.phac‐aspc.gc.ca/tbpc‐ latb/pubs/pdf/tbstand07_e.pdf. Accessed March/31, 2010. (10) Health Canada & Public Health Agency of Canada. Tuberculosis in Canada 2000 & Tuberculosis among the Foreign‐born in Canada. Canada Communicable Disease Report 2003;29(2):9. (11) BC Center for Disease Control. TB Annual Report 2004. 2004; Available at: http://www.bccdc.ca/NR/rdonlyres/F80BA3E8‐4E46‐4AFC‐9B05‐ CE3643C2947E/0/2004_TB_Annual.pdf. Accessed March/31, 2010. (12) Maher D, Raviglione M. Global epidemiology of tuberculosis. Clin.Chest Med. 2005 v; Jun;26(2):167‐182. (13) Fanning A, Billo N, Tannenbaum T, Phypers M, Little C, Graham B, et al. Stop TB‐Halte a la Tuberculose‐Canada: engaging industrialised nations in the challenge to meet global targets. International Journal of Tuberculosis & Lung Disease 2004 Jan;8(1):147‐150. (14) Schwartzman K, Oxlade O, Barr RG, Grimard F, Acosta I, Baez J, et al. Domestic returns from investment in the control of tuberculosis in other countries. N.Engl.J.Med. 2005 Sep 8;353(10):1008‐1020. (15) Lambert ML, Hasker E, Van Deun A, Roberfroid D, Boelaert M, Van der Stuyft P. Recurrence in tuberculosis: relapse or reinfection? The Lancet Infectious Diseases 2003 May;3(5):282‐287. (16) Dobler CC, Crawford AB, Jelfs PJ, Gilbert GL, Marks GB. Recurrence of tuberculosis in a low‐ incidence setting. European Respiratory Journal 2009 Jan;33(1):160‐167. (17) Chan‐Yeung M, Noertjojo K, Chan SL, Tam CM. Sex differences in tuberculosis in Hong Kong. International Journal of Tuberculosis & Lung Disease 2002 Jan;6(1):11‐18.  12  (18) Cox H, Kebede Y, Allamuratova S, Ismailov G, Davletmuratova Z, Byrnes G, et al. Tuberculosis recurrence and mortality after successful treatment: impact of drug resistance. PLoS Medicine / Public Library of Science 2006 Oct;3(10):e384. (19) Mallory KF, Churchyard GJ, Kleinschmidt I, De Cock KM, Corbett EL. The impact of HIV infection on recurrence of tuberculosis in South African gold miners. International Journal of Tuberculosis & Lung Disease 2000 May;4(5):455‐462. (20) A controlled clinical comparison of 6 and 8 months of antituberculosis chemotherapy in the treatment of patients with silicotuberculosis in Hong Kong. Hong Kong Chest Service/tuberculosis Research Centre, Madras/British Medical Research Council. Am.Rev.Respir.Dis. 1991 Feb;143(2):262‐267. (21) Johnson JL, Okwera A, Nsubuga P, Nakibali JG, Whalen CC, Hom D, et al. Efficacy of an unsupervised 8‐month rifampicin‐containing regimen for the treatment of pulmonary tuberculosis in HIV‐infected adults. Uganda‐Case Western Reserve University Research Collaboration. International Journal of Tuberculosis & Lung Disease 2000 Nov;4(11):1032‐1040. (22) Cowie RL. Silicotuberculosis: long‐term outcome after short‐course chemotherapy. Tubercle & Lung Disease 1995 Feb;76(1):39‐42. (23) Johnson IL, Thomson M, Manfreda J, Hershfield ES. Risk factors for reactivation of tuberculosis in Manitoba. CMAJ Canadian Medical Association Journal 1985 Dec 15;133(12):1221‐1224. (24) Snider DE,Jr. Reactivation of tuberculosis in Oklahoma: 1970‐1973. Chest 1975 Jul;68(1):36‐ 40. (25) Segarra F, Sherman DS. Relapses in pulmonary tuberculosis. Dis.Chest 1967 Jan;51(1):59‐ 63.  13  (26) Mitchison DA, Nunn AJ. Influence of initial drug resistance on the response to short‐course chemotherapy of pulmonary tuberculosis. Am.Rev.Respir.Dis. 1986 Mar;133(3):423‐430. (27) Fine PE, Small PM. Exogenous reinfection in tuberculosis. N.Engl.J.Med. 1999 Oct 14;341(16):1226‐1227. (28) Repique CJ, Li A, Collins FM, Morris SL. DNA immunization in a mouse model of latent tuberculosis: effect of DNA vaccination on reactivation of disease and on reinfection with a secondary challenge. Infection & Immunity 2002 Jul;70(7):3318‐3323. (29) van Rie A, Warren R, Richardson M, Victor TC, Gie RP, Enarson DA, et al. Exogenous reinfection as a cause of recurrent tuberculosis after curative treatment. N.Engl.J.Med. 1999 Oct 14;341(16):1174‐1179. (30) Chiang CY, Riley LW. Exogenous reinfection in tuberculosis. The Lancet Infectious Diseases 2005 Oct;5(10):629‐636. (31) Garcia de Viedma D, Marin M, Hernangomez S, Diaz M, Ruiz Serrano MJ, Alcala L, et al. Tuberculosis recurrences: reinfection plays a role in a population whose clinical/epidemiological characteristics do not favor reinfection. Arch.Intern.Med. 2002 Sep 9;162(16):1873‐1879. (32) Shen G, Xue Z, Shen X, Sun B, Gui X, Shen M, et al. The study recurrent tuberculosis and exogenous reinfection, Shanghai, China. Emerging Infectious Diseases 2006 Nov;12(11):1776‐ 1778. (33) Das S, Chan SL, Allen BW, Mitchison DA, Lowrie DB. Application of DNA fingerprinting with IS986 to sequential mycobacterial isolates obtained from pulmonary tuberculosis patients in Hong Kong before, during and after short‐course chemotherapy. Tubercle & Lung Disease 1993 Feb;74(1):47‐51. (34) Das S, Paramasivan CN, Lowrie DB, Prabhakar R, Narayanan PR. IS6110 restriction fragment length polymorphism typing of clinical isolates of Mycobacterium tuberculosis from patients 14  with pulmonary tuberculosis in Madras, south India. Tubercle & Lung Disease 1995 Dec;76(6):550‐554. (35) Godfrey‐Faussett P, Githui W, Batchelor B, Brindle R, Paul J, Hawken M, et al. Recurrence of HIV‐related tuberculosis in an endemic area may be due to relapse or reinfection. Tubercle & Lung Disease 1994 Jun;75(3):199‐202. (36) el‐Sadr WM, Perlman DC, Matts JP, Nelson ET, Cohn DL, Salomon N, et al. Evaluation of an intensive intermittent‐induction regimen and duration of short‐course treatment for human immunodeficiency virus‐related pulmonary tuberculosis. Terry Beirn Community Programs for Clinical Research on AIDS (CPCRA) and the AIDS Clinical Trials Group (ACTG). Clinical Infectious Diseases 1998 May;26(5):1148‐1158. (37) Sahadevan R, Narayanan S, Paramasivan CN, Prabhakar R, Narayanan PR. Restriction fragment length polymorphism typing of clinical isolates of Mycobacterium tuberculosis from patients with pulmonary tuberculosis in Madras, India, by use of direct‐repeat probe. J.Clin.Microbiol. 1995 Nov;33(11):3037‐3039. (38) Vernon A, Burman W, Benator D, Khan A, Bozeman L. Acquired rifamycin monoresistance in patients with HIV‐related tuberculosis treated with once‐weekly rifapentine and isoniazid. Tuberculosis Trials Consortium. Lancet 1999 May 29;353(9167):1843‐1847. (39) Lourenco MC, Grinsztejn B, Fandinho‐Montes FC, da Silva MG, Saad MH, Fonseca LS. Genotypic patterns of multiple isolates of M. tuberculosis from tuberculous HIV patients. Tropical Medicine & International Health 2000 Jul;5(7):488‐494. (40) Caminero JA, Pena MJ, Campos‐Herrero MI, Rodriguez JC, Afonso O, Martin C, et al. Exogenous reinfection with tuberculosis on a European island with a moderate incidence of disease. American Journal of Respiratory & Critical Care Medicine 2001 Mar;163(3 Pt 1):717‐ 720.  15  (41) Hawken M, Nunn P, Gathua S, Brindle R, Godfrey‐Faussett P, Githui W, et al. Increased recurrence of tuberculosis in HIV‐1‐infected patients in Kenya. Lancet 1993 Aug 7;342(8867):332‐337. (42) Heldal E, Docker H, Caugant DA, Tverdal A. Pulmonary tuberculosis in Norwegian patients. The role of reactivation, re‐infection and primary infection assessed by previous mass screening data and restriction fragment length polymorphism analysis. International Journal of Tuberculosis & Lung Disease 2000 Apr;4(4):300‐307. (43) Jasmer RM, Bozeman L, Schwartzman K, Cave MD, Saukkonen JJ, Metchock B, et al. Recurrent tuberculosis in the United States and Canada: relapse or reinfection?. American Journal of Respiratory & Critical Care Medicine 2004 Dec 15;170(12):1360‐1366. (44) Bandera A, Gori A, Catozzi L, Degli Esposti A, Marchetti G, Molteni C, et al. Molecular epidemiology study of exogenous reinfection in an area with a low incidence of tuberculosis. J.Clin.Microbiol. 2001 Jun;39(6):2213‐2218. (45) Raviglione MC, Snider DE,Jr, Kochi A. Global epidemiology of tuberculosis. Morbidity and mortality of a worldwide epidemic. JAMA 1995 Jan 18;273(3):220‐226. (46) Abos‐Hernandez R, Olle‐Goig JE. Patients hospitalised in Bolivia with pulmonary tuberculosis: risk factors for dying. International Journal of Tuberculosis & Lung Disease 2002 Jun;6(6):470‐474. (47) Olle‐Goig JE. Patients with tuberculosis in Bolivia: why do they die?. Pan American Journal of Public Health 2000 Sep;8(3):151‐155. (48) Kochi A. The global tuberculosis situation and the new control strategy of the World Health Organization. 1991. Bull.World Health Organ. 2001;79(1):71‐75.  16  (49) Gustafson P, Gomes VF, Vieira CS, Samb B, Naucler A, Aaby P, et al. Clinical predictors for death in HIV‐positive and HIV‐negative tuberculosis patients in Guinea‐Bissau. Infection 2007 Apr;35(2):69‐80. (50) Connolly C, Davies GR, Wilkinson D. Impact of the human immunodeficiency virus epidemic on mortality among adults with tuberculosis in rural South Africa, 1991‐1995. International Journal of Tuberculosis and Lung Disease 1998;2(11):919‐925. (51) Rao VK, Iademarco EP, Fraser VJ, Kollef MH, Rao VK, Iademarco EP, et al. The impact of comorbidity on mortality following in‐hospital diagnosis of tuberculosis. Chest 1998 Nov;114(5):1244‐52. (52) Borgdorff MW, Veen J, Kalisvaart NA, Nagelkerke N. Mortality among tuberculosis patients in the netherlands in the period 1993‐1995. European Respiratory Journal 1998 Apr;11(4):816‐ 820. (53) Harries AD, Nyangulu DS, Kang'ombe C, Ndalama D, Glynn JR, Banda H, et al. Treatment outcome of an unselected cohort of tuberculosis patients in relation to human immunodeficiency virus serostatus in Zomba Hospital, Malawi. Transactions of the Royal Society of Tropical Medicine & Hygiene 1998 May‐Jun;92(3):343‐347. (54) Glynn JR, Warndorff DK, Fine PE, Munthali MM, Sichone W, Ponnighaus JM. Measurement and determinants of tuberculosis outcome in Karonga District, Malawi. Bull.World Health Organ. 1998;76(3):295‐305. (55) van den Broek J, Mfinanga S, Moshiro C, O'Brien R, Mugomela A, Lefi M. Impact of human immunodeficiency virus infection on the outcome of treatment and survival of tuberculosis patients in Mwanza, Tanzania. International Journal of Tuberculosis & Lung Disease 1998 Jul;2(7):547‐552.  17  (56) Pablos‐Mendez A, Sterling TR, Frieden TR. The relationship between delayed or incomplete treatment and all‐cause mortality in patients with tuberculosis. JAMA 1996 Oct 16;276(15):1223‐1228. (57) Zahar JR, Azoulay E, Klement E, De Lassence A, Lucet JC, Regnier B, et al. Delayed treatment contributes to mortality in ICU patients with severe active pulmonary tuberculosis and acute respiratory failure. Intensive Care Med. 2001 Mar;27(3):513‐520. (58) Mathur P, Sacks L, Auten G, Sall R, Levy C, Gordin F. Delayed diagnosis of pulmonary tuberculosis in city hospitals. Arch.Intern.Med. 1994 Feb 14;154(3):306‐310. (59) Katz I, Rosenthal T, Michaeli D. Undiagnosed tuberculosis in hospitalized patients. Chest 1985 Jun;87(6):770‐774. (60) Sacks LV, Pendle S, Sacks LV, Pendle S. Factors related to in‐hospital deaths in patients with tuberculosis. Archives of Internal Medicine 1998 Sep 28;158(17):1916‐22. (61) Penner C, Roberts D, Kunimoto D, Manfreda J, Long R. Tuberculosis as a primary cause of respiratory failure requiring mechanical ventilation. American Journal of Respiratory & Critical Care Medicine 1995 Mar;151(3 Pt 1):867‐872. (62) Zafran N, Heldal E, Pavlovic S, Vuckovic D, Boe J, Zafran N, et al. Why do our patients die of active tuberculosis in the era of effective therapy? Tubercle & Lung Disease 1994 Oct;75(5):329‐ 33. (63) Greenaway C, Menzies D, Fanning A, Grewal R, Yuan L, FitzGerald JM. Delay in diagnosis among hospitalized patients with active tuberculosis ‐ Predictors and outcomes. American Journal of Respiratory and Critical Care Medicine 2002 01 Apr;165(7):927‐933. (64) Xie HJ, Enarson DA, Chao CW, Allen EA, Grzybowski S, Xie HJ, et al. Deaths in tuberculosis patients in British Columbia, 1980‐1984. Tubercle & Lung Disease 1992 Apr;73(2):77‐82.  18  (65) Khan K, Campbell A, Wallington T, Gardam M. The impact of physician training and experience on the survival of patients with active tuberculosis. CMAJ Canadian Medical Association Journal 2006 September 26;175(7):749‐753. (66) Enarson DA, Grzybowski S, Dorken E. Failure of diagnosis as a factor in tuberculosis mortality. Can.Med.Assoc.J. 1978 Jun 24;118(12):1520‐1522. (67) Kourbatova EV, K. LM,Jr, Romero J, Kraft C, del Rio C, Blumberg HM, et al. Risk factors for mortality among patients with extrapulmonary tuberculosis at an academic inner‐city hospital in the US. European Journal of Epidemiology 2006;21(9):715‐21. (68) de Meer G, van Geuns HA. Rising case fatality of bacteriologically proven pulmonary tuberculosis in The Netherlands. Tubercle & Lung Disease 1992 Apr;73(2):83‐86. (69) Bustamante‐Montes LP, Escobar‐Mesa A, Borja‐Aburto VH, Gomez‐Munoz A, Becerra‐ Posada F, Bustamante‐Montes LP, et al. Predictors of death from pulmonary tuberculosis: the case of Veracruz, Mexico. International Journal of Tuberculosis & Lung Disease 2000 Mar;4(3):208‐15. (70) Walpola HC, Siskind V, Patel AM, Konstantinos A, Derhy P. Tuberculosis‐related deaths in Queensland, Australia, 1989‐1998: characteristics and risk factors. International Journal of Tuberculosis & Lung Disease 2003 Aug;7(8):742‐750.  19  CHAPTER 2: A POPULATION BASED STUDY OF RECURRENT TUBERCULOSIS IN BRITISH COLUMBIA, CANADA1  1  ‐ “A version of this chapter will be submitted for publication. Moniruzzaman, A. Kazanjian, A. Wong, H. Elwood, R.K. and FitzGerald, J. M. (2010) Recurrence of Tuberculosis: a population‐based study.”  20  2.1. INTRODUCTION Globally, in terms of morbidity and mortality, tuberculosis (TB) is one of the leading infectious diseases, which caused the highest number of deaths by a single pathogenic agent in the pre‐ HIV era (1,2) and now ranks second to HIV. Recurrent TB remains a significant public health problem for TB control programs globally, especially in resource‐poor countries. Recurrent cases have higher mortality rates (3‐6) and also continue to transmit infection in the community. A recent South African study reported an elevated risk of new TB (another episode) among patients with a history of previous TB (7). In addition, management and treatment becomes are more challenging in a patient with recurrent TB. Retreatment frequently involves more expensive drugs and a longer duration, and thus provides an extra burden to resource poor countries. In low incidence countries (e.g. Canada), attempting to eliminate TB, the prevention of recurrence is critical for control strategies because it helps to reduce the transmission of TB by decreasing the number of contagious cases. From a quality assurance perspective, the recurrence rate is useful for assessing the effectiveness of a TB control program due to its potential association with treatment coverage and completion rates (8‐11) . Recurrence is widely used as an outcome measure in studies evaluating either efficacy or effectiveness of a TB treatment regimen (12‐17). An improved understanding of recurrence is valuable for future development of treatment regimens.  Several factors such as age, male gender, Aboriginal ethnicity, incomplete TB treatment, drug resistant status, silicosis, alcoholism, HIV co‐infection, and extent of disease (e.g., advanced stage, cavitations) have been found to be associated with the recurrence of TB (10,13,18‐25). However, wide discrepancies are observed across studies in terms of methodology and criteria that are used to define reactivated cases (8,23,26‐28) and the majority of studies, evaluating recurrence, have been conducted on specific populations in institutional or clinical settings, which limit the application of study findings to the general population. For example, findings from clinical trials differ from population‐based studies as clinical trials usually follow rigid selection criteria and a vigorous patient follow up and monitoring system. These criteria are not being applicable in real‐life settings. 21  Additionally, few studies have investigated risk factors of recurrence in the presence of multiple factors including demographic and clinical characteristics, individual prognostic factors and treatment related behaviors (26). Information on TB recurrence from the Canadian setting is also limited. To our knowledge, the study published by Johnson et al. in 1985 is the only Canadian study and this study investigated the recurrence of TB in the province of Manitoba (22). However, it was a case‐control study that included only 58 reactivated cases and did not provide an estimate of incidence of recurrence. The objectives of this study are to estimate the incidence of, as well as the risk factors, for TB recurrence in the province of British Columbia using population based data.  2.2 METHODS 2.2.1 Setting and Source of Information This study was conducted in British Columbia, with a population of approximately 3.4 million. Like other parts of Canada, this province has many immigrants, mainly from Asian countries.  The provincial TB service program is administered by the Division of TB Control (DTB) at the British Columbia Centre for Disease Control (BCCDC), located in Vancouver (the largest city of this province). This center acts as a referral centre for the prevention, control and treatment of all TB‐related disease and infection occurring in BC. Since TB is a notifiable disease, this division is notified with regard to all new active cases of TB that are diagnosed in BC. In addition, the TB medications are dispensed by the Pharmacy division of BCCDC and myco‐bacteriology for all active cases is also coordinated from the provincial TB laboratory located at the BCCDC. These centralized activities work as an additional safeguard and ensure that the DTB is informed of all TB and TB‐related cases that occur in the province.  A centralized computer based system has been in place from the 1960’s but this system was upgraded and integrated into a more comprehensive database since 1990., The database 22  maintains a comprehensive electronic database with records for all subjects whose care is managed through the TB control program. This database contains several types of information including: socio‐demographic profile, medical risk factors, treatment and related information, mycobacteriology. Mycobacteriology related data is generated from the BCCDC TB laboratory and is entered into the TB control by a trained TB clerk. Socio‐demographic and risk factor information is obtained from several sources. If TB patients are seen in person in the BCCDC TB clinic, then an interview clerk usually collects this information and enters it into the TB control database. If the client is not seen in person (i.e. TB Control is consulted by private physicians or the patient is in the hospital), then usually it is the local Public Health Nurse who collects the socio‐demographic information for the patient in their area. A standardized data collection form is used for this purpose. The information is then sent (mailed, faxed) to TB Control and input by Medical Records Staff and Field Operations nursing personnel.  2.2.2 Study Design A population‐based retrospective cohort design was used for this study. This study used data from 1990 to 2006.  2.2.3 Study Population and Relevant Definition The study population included all TB cases recorded in the DTB case registry from 1990 to 2006. According to Health Canada, a tuberculosis patient with “documented evidence or history of previously active tuberculosis which became inactive” (29) is defined as a recurrent/relapse case. A TB patient who did not have history of prior TB was termed as a new case. In BC, the same criteria set by Health Canada are being used to determine new and recurrent cases. It has been recommended by Health Canada that all subjects who are diagnosed with TB in Canada need to be recorded under the headings new and relapsed cases. Prevalence and incidence of recurrence were investigated separately. A case whose recurrent episode was determined between 1990 and 2006 regardless of its primary episode was termed as a prevalent recurrent 23  case. However, a recurrent case whose both primary and 2nd episode occurred between 1990 and 2006 was defined as an incident case. Therefore, the number of eligible cases differed across these analyses. For prevalence, all recurrent cases reported between the observation periods were included in the analysis. A TB patient with an unknown status as a new or recurrent case was excluded from both the prevalence and incidence study. However, a restricted sample was used to estimate incidence of recurrence as well as factors associated with incidence of recurrence. In the incident study, recurrent cases whose primary and subsequent episodes of TB occurred between 1990 and 2006 were considered for this study. Any case having a history of a primary episode before 1990 was excluded from the analysis. In addition, new TB cases diagnosed in 2005 and 2006 were not considered in the incident study in order to allow a minimum of two years follow up time (since their primary diagnosis date) for the development of recurrence. The period of inactivity (difference between date of diagnosis of subsequent episode and the treatment end date of previous episode) differs across studies ranging from 0 to 12 months (8,23,27,28,30). The Canada TB reporting system recommends a 6 month period of inactivity to consider the case as a recurrent one. The primary analysis for risk factors were conducted among all incident recurrent patients regardless of period of inactivity (n=108). However, cases with less than 6 months of inactivity were identified and an additional analysis was done among these recurrent patients (with more than 6 months of inactivity).  2.2.4 Follow­up Time Follow up time was estimated from the differences between time 1 and time 0. Treatment end date of the primary episode was time zero and censoring date was time 1. Recurrent cases were censored at time of 1st recurrence (time 1). Patients without recurrence were censored (time 1) when death occurred, or they left the province or their follow‐up ended (Study end date, 31st December, 2006).  24  2.2.5 Variables Description  2.2.5.1 Outcome (Dependent) Variable – Recurrence The recurrence of TB is our primary concern and therefore it was considered as the outcome variable. For our prevalence study, prevalence of recurrence was the outcome while for our incident study; it was the incidence of recurrence. New cases were the comparison group for both scenarios. In the previous section, both recurrent and new cases were defined.  25  Figure 2.1: Study population for incidence of recurrence (n=4464)  5433 TB cases registered with TB Division between 1990 and 2006  Excluded (n=969)  Eligible for follow up (n= 4464)  1) TB patients with unknown status (n=5) 2) Primary episode before 1990 (n=383) 3) Recurrent cases with unknown primary episode (n=24) 3) Recently (in 2005 and 2006) diagnosed new cases (n=557)  Follow up interval: time1 – time0 Time 0: Treatment end date of 1st episode Time 1: Date of recurrence/Date of Death/Lost to follow up/Study end date Outcome: Recurrence of TB (event=108) Observation time: 29178 PYs  Incident cases with less than 6 months of inactivity (n=14)  Incident cases with more than 6 months of inactivity (n=94)  Primary analysis: Cox regression  26  2.2.5.2 Independent Variables  2.2.5.2.1 Socio‐demographic Variables Age  In the database, the exact age was not available; rather it was calculated from the differences between date of diagnosis and date of birth. For patients with multiple episodes, date of diagnosis at primary episode was used to calculate age. Age at diagnosis was analyzed as a continuous variable. Biological Gender  The data on gender in the primary database was almost complete and was coded as male, female and trans‐sexual. For cases with trans‐sexual status, current gender was used. The risk for females of recurrence was estimated by comparing females with males (reference category). Country of Origin/ Immigration Status Under this variable, data for several categories (Canadian born, foreign‐born Canadian citizen, landed immigrant, visitors, refugee, etc.) were included in the primary database. All categories were collapsed into three groups‐ Canadian born non‐Aboriginals, Canadian born Aboriginals and foreign‐born. The foreign‐born group includes landed immigrants, foreign‐born Canadian citizens, foreign‐born people living on a work permit, minister’s permit, or as a refugee, student, visitor, etc.  For several subjects, immigration status was missing or unknown.  Attempts were made to see if this information could be extracted from other variables such as birth country, arrival year and Aboriginal community. Subjects with missing or unknown status were reassessed for their immigration status according to these variables and adjustments were made accordingly. 27  Ethnicity  In the primary database, several categories were used to determine ethnicity. There were substantial numbers of patients in certain ethnic groups, but few in other groups. In order to optimize the number of variables during the analysis, these categories were collapsed into seven groups as follows:   Caucasians‐ used as reference category    Aboriginals ‐ includes both registered and non‐registered subjects, Inuit and Métis    Chinese    Southeast Asian‐ includes East Indian, Punjabi, East Asian, South Asian, and Southeast Asian,    Vietnamese    Filipino    Other, which includes the rest of the ethnicities  In the current study, Filipino and Vietnamese have been analyzed as a single group. Birth Country Regions  As there were exhaustive lists of countries of birth‐place for TB subjects in the database, it would be difficult to run the analysis for all countries. In light of this, all listed counties were collapsed into several geographical regions according to WHO criteria. The birth country regions are as follows:   Pan American Heath Region (PAHO) ‐ includes patients who were born in countries in North America, South America and Central America. PAHO was used as the reference category  28    African Health Region: includes mostly sub‐Saharan countries (except Sudan and Somalia) and other African countries (except Egypt, Libya, Morocco and Tunisia)    European Health Region ‐ includes all people born in Europe and Russia    Southeast Asia Region‐ includes Bangladesh, Bhutan, DPR Korea, India, Indonesia, Maldives, Myanmar, Nepal, Sri Lanka, Thailand, and Timor‐Leste    Eastern Mediterranean Health Region‐includes Afghanistan, Pakistan, middle‐east countries and few African countries ( Egypt, Libya, Morocco, Somalia, Sudan, and Tunisia)    Western Pacific Health Region‐ includes Australia, Cambodia, China, Fiji, Hong Kong, Japan, South Korea, Laos, Macau, Malaysia, Mongolia, New Zealand, Papua New Guinea, Philippines, Singapore, Vietnam  Few cases belonged to the Eastern Mediterranean and European regions. Therefore, both regions were collapsed into a single group and were analyzed accordingly.  Place of Diagnosis of Primary Episode  Several places were mentioned as the place of diagnosis for the primary episode in the TB databases. Therefore, all the places were collapsed into three groups‐ BC, other Canadian provinces, and outside Canada. Since there were few cases that belonged to the category to “Other Canadian Provinces”, it was combined with the group Outside of Canada and was analyzed as a single group (Outside of BC included patients who were diagnosed in other Canadian provinces as well as in overseas countries). Patients who were diagnosed initially at BC were used as the reference category.  Marital Status  There were several categories enlisted in the primary database. We collapsed the categories into the two following groups‐ 29    Single‐includes single, separated, divorced, and widowed    Family –includes common law groups and married groups  The category “family” was used as the reference group during analysis.  2.2.6.2.2 Diagnosis and Treatment Related Factors Type of Diseases  Several types of diagnosis were enlisted in the primary database under diseases type. All this information was grouped into three broad categories as follows:   Pulmonary TB (PTB)‐ TB that involves primarily lungs and also includes pleural TB    Extra‐pulmonary TB (EPTB): includes TB that involves extra‐pulmonary sites such as genitourinary organs, abdominal, meninges, bones or joints, lymph nodes etc. Patients with disseminated/miliary TB were considered as EPTB.    Both: patients who had both pulmonary and extra‐pulmonary involvement  Cavitary Disease  This variable was derived from diagnosis type of diseases. Patients who had been diagnosed as ‘Cavitations of Lung TB’ were classified as ‘yes’. A patient who did not have such a diagnosis was coded as ‘no’ and was used as the reference category.  30  Duration of Treatment  Treatment duration was calculated from treatment end date and treatment start date. Patients who did not receive any treatment due to post‐mortem diagnosis were assigned zero time and were not accounted for in the analysis.  Completion of Treatment  This variable was derived from the variable “Reason Treatment Ended”. Several categories for Reason Treatment ended were enlisted in the original database such as: completed satisfactorily, Incomplete, Incomplete due to death, and Incomplete due to departure from province. All these groups were classified into three groups:   Completed treatment/successful completion of treatment2: includes patients who received at least 6 months therapy and completed medically recommended treatment within acceptable timeframe and acceptable completion rate.    Incomplete treatment: includes patients who could not complete treatment due to individual causes (such as clients’ decision not to continue treatment, non‐compliance, drug reaction). In addition, patients whose treatment status had been coded as incomplete, due to loss to follow up, were included in this category.    Other: includes patients who could not complete treatment due to death or relocation out of province or could not receive treatment due to post‐mortem diagnosis or other reasons.  In the current analyses, the effect of incomplete treatment was assessed compared to complete treatment. A patient who was included in the “other treatment” category was not accounted for in this analysis.  2  ‐Successful completion of treatment mostly represents patients whose outcome was ‘cure’ (bacteriological) and whose outcome was ‘treatment completed’.  31  Compliance with Treatment  Information on compliance with treatment was coded in the TB databases as a percentage such as “less than 50%”, “50% to 80%”, and 100%. Patients who had 80% compliance or less were classified as non‐compliant/poor compliant while patients who had more than 80% compliance were classified as compliant.  Administration of Therapy  This variable was originally coded in the database as ‘Self‐administered’, ‘Supervised once daily’, ‘Supervised twice weekly’, ‘DOT’ and ‘Supervised thrice weekly’. All these groups were categorized into two groups:   Self‐administered therapy: includes ‘Self‐administered’    Supervised therapy: includes ‘Supervised once daily’, ‘Supervised twice weekly’, ‘DOT’ and ‘Supervised thrice weekly’  2.2.5.2.3 Mycobacteriology Results TB Culture  When a TB is suspected as diagnosis, smear and culture tests on available specimens are routinely performed in order to confirm the TB diagnosis. Based on culture results, all TB patients were categorized into two groups‐ culture positive cases and culture negative cases. Patients whose culture reports were not available or could not be done were treated as unknown. The effect of positive culture on recurrence was investigated and culture negative/unknown cases were used as the reference group.  32  Drug Sensitivity  Once culture tests are available, drug susceptibility tests were also routinely conducted on all positive cultures. In the BCCDC TB lab, susceptibility testing is performed for four first‐lines anti TB drugs: rifampin (RIF), isoniazid (INH), ethambutol (EM) and streptomycin (SM). If the subjects are resistant to isoniazid (INH) or rifampin (RM), then the susceptibility testing for pyrazinamide (PZA) is done. This based on the infrequent presence of PZA resistance. Since the test reports for PZA susceptibility were unavailable for most of the cases, the results of PZA were not considered to determine the status of susceptibility (i.e., whether it should be considered as sensitive or resistant cases). Based on the susceptibility results, three subgroups for resistance patterns and one group for sensitivity were created as follows:   Mono resistance: resistance to a single drug    Multidrug resistance (MDR): resistance to rifampin and isoniazid    Poly‐resistance: resistance to at least two drugs excluding MDR cases    Sensitive: sensitive to all four drugs  The laboratory report of drug susceptibility was originally coded as sensitive, resistant, borderline sensitive/resistant cases. In addition to clearly resistant groups, all borderline categories are also considered as resistant cases. In the current analysis, the poly and mono‐ resistance groups were collapsed into a single group and patients with sensitive isolates were used as the reference group.  2.2.5.2.4 Co‐morbid Conditions Under co‐morbid conditions, the variables that were analyzed were as follows: diabetes mellitus, malnutrition, alcoholism, drugs or substance abuse, HIV status, malignancy and use of immunosuppressive medications.  33  Overall, all these variables were classified as a dichotomous variable ‐yes or no. “Yes” indicates presence of co‐morbidity or illness while “No” indicates the absence of that particular disease. There was an additional two categories (uncertain and not asked) in the primary database, which were treated as unknown observations. Patients without a valid response (missing or unknown) to co‐morbid conditions and patients without co‐morbidity were used as the reference group.  The term “Alcoholism Yes” indicates a history or current intake of an excessive amount of alcohol (subjective judgment), while “No” indicates the absence of such an intake of alcohol. The variable “Substance abuse” was derived from drug abuse and methadone use and coded as yes and no. Yes includes both past and present use of either non‐injection or injection drugs (including methadone) while no means absence of such a habit. In the primary database, the information on HIV and AIDS was available separately. Both variables were combined and classified into a binary variable ‐ yes vs. no. Yes means sero‐positivity for HIV which includes AIDS defining illness. No means negative for HIV.  The variable “Malignancy” was derived from four variables (lymphoma, leukemia and other malignancy, chemotherapy for cancer) that were available in the current database. It was coded into yes (presence of any malignancy) and no (absence of such a malignancy. The variable “Chronic Renal Failure” was derived from two variables renal failure and kidney transplant, which was later coded as yes or no. Immunosuppressive medications is a combination variable of immune‐suppressive medications and steroid therapy. Patients who received immunosuppressive or steroid therapy or are being treated with these medications currently were coded as yes. Patients who did not have a history of such medications were classified as no.  34  2.2.6 Statistical Analysis An initial descriptive analysis for the study population was performed. Comparisons of continuous variables among groups (such as new vs. recurrent cases) were done using student t test and Wilcoxon rank sum test whenever appropriate. Categorical variables were compared using the chi‐square test and fisher’s exact test (if one of the expected cell values were less than 5). For the prevalence study, logistic regression analysis was conducted to identify factors associated with TB recurrence. Since the primary focus of this study was the incidence of recurrence, a detailed description of the analysis for incident study was presented here.  For the incident study, incidence was estimated per person‐year of follow up and expressed in terms of 100, 000 person‐years. Univariate and multivariable Cox regression were conducted to identify predictors of incidence of recurrence. Variables which were found significant (p value ≤0.05) in Univariate analysis were considered for multi‐variable analysis. The effect of universal confounders such as age and gender were adjusted for in the multivariable model regardless of their statistical significance in bi‐variate analysis. Clinical relevance and evidence from literature were also considered in choosing variables for the multiple regression analysis.  A series of multivariable model (Model 1 to Model 4) were conducted. Variables that were considered for multivariable modeling were as follows: age, gender, immigration status, birth‐ country regions, place of diagnosis, culture status, completion of treatment successfully, adherence to treatment, drug abuse, alcoholism and HIV/AIDS. The effect of “incomplete treatment” and “non‐adherence to treatment” was evaluated in a separate multivariable model in the presence of similar controlling variables. Immigration status and birth country regions were not considered together in model building process. Model 1 and Model 2 included variable “immigration status” whether Model 3 and Model 4 included “birth country regions” rather than immigration status. Similarly, Model 1 and Model 3 evaluated the effect of “Incomplete treatment” and Model 2 and Model 4 evaluated the effect of “non‐adherence to treatment”. Otherwise, variables that were considered in these four multivariable models were  35  the same. The strength and magnitude of the association between incidence of recurrence and the risk factors was expressed in terms of a hazard ratio with 95% confidence intervals. Variables were considered significant if the 95% confidence interval did not contain one. Two additional analyses were conducted for this study‐ one among patients who successfully completed their TB treatment during the primary episode and other analysis among patients who were identified as recurrent TB cases using rigid criteria (more than 6 months of inactivity since treatment end).  For socio‐demographic, diagnosis and treatment related factors, the categories “unknown/not known”, “not asked” and “uncertain” were treated as missing.  However, for co‐morbid  conditions, unknown/missing were combined with “no” response and treated as the reference group in the analysis. SPSS (Windows 17.0 version) and S‐plus (Windows 8.1 version) statistical packages were used to run these analyses.  2.3 RESULTS  2.3.1 Prevalence of Recurrence During the study period (1990‐2006), 5433 cases of TB were registered with the TB Control Division (DTB) at the British Columbia Center for Disease Control. Among them, thirty patients were not considered in the current analysis because of unknown outcome status (new or recurrent) or they were diagnosed as new cases before 1990. The characteristics of the study cohort (n=5403) are described in Table (Appendix A.1). The mean age of the study cohort at the time of diagnosis was 46 years while the median was 44 years with a range from 1 to 104 years. There was predominance of males (55%) in this TB cohort. Foreign‐born individuals accounted for 67% of the total TB cases, while Aboriginal people and non‐aboriginal Canadian‐ born accounted for 13% and 16% respectively.  36  Out of 5403 cases, 490 patients had at least one recurrent episode. Overall the recurrence rate was 9.1 % (95% CI: 8.3%, 9.9%). The prevalence was the highest in 1990 (12%) and the lowest in 1995 (6%). Although the prevalence of recurrence was decreasing over time, using the chi‐ square test for trend, it did not reach statistical significance (p=0.39). The trend of recurrence from 1990 to 2006 is presented in Figure 2.1. Comparisons of socio‐demographic characteristics and treatment‐related behaviors between prevalent recurrent cases and new cases were presented in the Appendix A (Table A.2 and Table A.3).  2.3.2 Incidence of Recurrence Out of 5433 registered cases, 4464 cases were eligible for this study. In total, 969 TB patients were excluded from the analysis. The reasons for exclusions were: primary episode before 1990, recent diagnosis as a new case, and unknown status of primary episode. A detailed description of the study cohort is presented in Figure 2.2. During the study period, 108 (2.4%) patients developed recurrent disease. The incidence was 370 cases per 100,000 person years. A few patients had more than one recurrence but our analysis was limited to the 1st recurrent diagnosis. According to calendar year, the most number of incident cases was 14 (5%) in the year of 2000 followed by 12 (4%) cases in 2002. The trend of the incidence of recurrence is shown in Figure 2.3. As expected no case was observed in 1990. The characteristic of the study cohort is presented in Table 2.1.  37  Figure 2.2: Trend of recurrent (prevalent) cases from 1990 to 2006 in British Columbia, Canada  The median age at the time of 1st diagnosis was 45 years with an inter‐quarterlies range from 30‐66 years. This study cohort was over‐represented by foreign‐born people who contributed two‐third of total cases while Canadian‐born non‐aboriginal contributed only one sixth of total cases. In terms of ethnicity, the largest group was Chinese (23%) followed by Caucasian (18%). In addition, 566 (13%) patients were of Aboriginal ancestry.  38  Table 2.1: Characteristics of study cohort (N=4464)  Variables Case Status New cases Incident recurrent cases Cumulative Incidence Overall At first year At 2nd year At 5th year Median follow up time (IQR) All cases Incident cases New cases Age at diagnosis (primary episode) Mean (SD) Median (IQR) Biological gender Male Female Birth place Canadian‐ born non‐Aboriginal Canadian‐ born Aboriginal Foreign‐born Unknown Birth Country regions PAHO African Region East. Mediterranean European Region Southeast Asia Region Western Pacific Region Unknown/missing Ethnicity Caucasian Aboriginals Chinese South‐East Asian Filipino Vietnamese Other Unknown/Missing Place of Diagnosis (1st episode) In BC Inside of Canada, outside of BC Outside of Canada Unknown/Missing  N (%) 4356 (98%) 108 (2%) Per 100, 000 PYs 370 854 1500 2400 Years 6.1 (2.1, 10.7) 2.1 (0.8, 4.5) 6.2 (2.2, 10.8) Years 47.4 (21.8) 45 (30‐66) 2405 (54) 2059 (46) 754 (17) 566 (13) 2943 (66) 201 (4) 1403 (31) 81 (2) 101 (2) 225 (5) 639 (14) 1762 (40) 253 (6) 803 (18) 566 (13) 1047 (23) 704 (16) 351 (8) 265 (6) 442 (10) 286 (6) 3909 (88) 5 (<1) 162 (4) 388 (8)  39  Variables  N (%)  Marital status Single Not single Other Unknown/Missing  1992 (22) 1321 (30) 221 (5) 1930 (43)  Table 2.2: Occurrence of incident recurrent cases (n=108) over the follow‐up period (interval between treatment completion time and date of recurrence)  Number of cases Percentage of total incident cases Cumulative Percentage of total incident cases  Within 6 months  Between 6 ‐12 months  Between 12‐24 months  Between 24‐36 months  Between 36 ‐48 months  Between 48 ‐60 months  After 60 months  14  18  21  17  9  4  25  13%  17%  19%  16%  8%  4%  23%  13%  30%  49%  65%  73%  77%  100%  The median follow up for the entire cohort was 6.1 years and was 6.2 years for new cases. The median time to recurrence after treatment completion was 2.1 (IQR: 0.8‐4.5) years. Table 2.2 presents the occurrence of recurrence during the follow‐up period. About 30% of recurrence cases occurred within 1 year and 60% within four years of treatment completion. Twenty‐five (23%) cases developed recurrence even after the five years of follow up. Table 2.3 presents time of recurrence among foreign‐born patients since their arrival in Canada. About 52% of recurrence occurred within four years of arrival among overseas patients (Table 2.3).  40  Figure 2.3: Trend of incident recurrent cases in British Columbia from 1990 to 2006  Table 2.4 presents unadjusted hazard ratios and the corresponding 95% confidence interval for the socio‐demographic determinants for the development of recurrence. Compared to Canadian‐born non‐aboriginal, foreign‐born people (UHR: 2.52) were three‐fold higher risk to have recurrence. As expected, the risk of recurrence was 4 times higher among patients born in African countries, twice as high in patients from Southeast Asia and also from the Western Pacific region compared to patients born in the Pan‐American region. In terms of ethnicity, Chinese (UHR: 2.84), Filipino/Vietnamese (UHR: 2.89) and Southeast Asian (UHR: 2.84) had significantly higher risk to have recurrence compared to Caucasian subjects. Aboriginal people had an elevated risk of having recurrence (UHR=1.75, 293 per 100, 000 PYs) compared to non‐ Aboriginal Canadian, but the 95% CI of the corresponding HR contained one. Patients whose  41  primary diagnosis took place outside of BC were fifteen‐times (UHR=14.82) more likely to develop recurrence compared to patients who were diagnosed locally.  Table 2.3: Occurrence of recurrence (incidence) among foreign‐born individuals (n=87) since their arrival in Canada  Primary Episode  Number of cases % of total incident cases Cumulative % of total incident cases  Before arrival  Same year  1 year Interval  2 year Interval  3 year Interval  4 year interval  5 year interval  After 5 years  30  12  5  4  0  3  1  32  35%  14%  6%  5%  0%  3%  1%  36%  49%  55%  60%  60%  63%  64%  100%  Recurrent Episode Number of cases  ‐  11  14  9  5  6  3  39  % of total incident cases  ‐  13%  16%  10%  6%  7%  3%  45%  Cumulative % of total incident cases  ‐  13%  29%  39%  45%  52%  55%  100%  42  Table 2.4: Comparisons of socio‐demographic characteristics between TB patients who developed a recurrence (n=108) and who did not (n=4356)  Variable  Incidence of recurrence Overall (all cases) Culture positive cases Cases who completed treatment successfully3 Age at diagnosis (primary episode) Mean (SD) Biological gender Male Female Birth place Canadian born non‐Aboriginal Canadian born Aboriginal Foreign born Birth Country regions PAHO African Region East. Mediterranean and Europe Southeast Asia Region Western Pacific Region Ethnicity Caucasian Aboriginals Chinese South‐East Asian Filipino/Vietnamese Other Ethnicity Unknown/missing Marital status Single Not single Place of initial diagnosis In BC Outside of Canada4  Number of Incident cases/Total person years  Incidence per 100, 000 person years  Unadjusted Hazard Ratio (95% CI)  108/29178 63/21064 73/27174  370 299 269  Years 43.3 (19.2)  Years 47.5 (21.9)  1.00 (0.99, 1.00)  53/14851 55/14328  357 384  Reference 1.08 (0.74, 1.58)  9/5197 12/4095 87/19363  173 293 449  Reference 1.75 (0.74, 4.14) 2.52 (1.27, 5.00)  22/9797 5/533 2/2276 20/4013 59/11898  225 939 88 498 496  Reference 4.0 (1.51, 10.56) 0.39 (0.09, 1.65) 2.07 (1.13, 3.79) 2.13 (1.30, 3.47)  9/5468 12/4095 31/7102 23/4611 21/4232 8/2698 4/971  165 293 436 499 496 296 412  Reference 1.80 (0.76, 4.26) 2.54 (1.21, 5.34) 2.84 (1.31, 6.31) 2.89 (1.32, 8.82) 1.68 (0.65, 4.35) ‐  16/8440 33/10836  190 305  Reference 0.65 (0.36, 1.17)  67/27402 41/1104  245 3715  Reference 14.82 (10.04, 21.87)  3  ‐Recurrence among patients who completed initial treatment successfully was termed as relapse in several studies 4 ‐Outside of Canada also included patients who were diagnosed outside of BC, but within Canada (only few cases belong to this category‐ outside of BC, but within Canada)  43  Table 2.5: Comparisons of treatment and related factors between TB patients who developed a recurrence (n=108) and who did not (n=4356)  Variable  Duration of Treatment Mean (SD) Median (IQR) Rx completion of primary episode Completed successfully5 Incomplete Rx compliance of primary episode Complaint Non‐compliant Major mode of Treatment at primary episode Supervised Self‐administered Culture results at primary episode Negative/no/unknown Positive Drug resistance status at primary episode Sensitive Mono/poly resistance MDR Disease type of initial episode Pulmonary Extra‐pulmonary Both Presence of Cavitations at initial episode Yes No  5  Number of Incident cases/Total person years  Incidence per 100, 000 person years  Unadjusted Hazard Ratio (95% CI)  Years 0.67 (0.39) 0.65 (0.50, 0.83)  Years 0.68 (0.31) 0.67 (0.50, 0.81)  73/27174 24/1621  269 1480  Reference 5.62 (3.54, 8.92)  65/27487 18/1139  236 1581  Reference 6.53 (3.88, 11.01)  18/5893 56/22265  305 252  Reference 0.78 (0.46, 1.33)  45/8114 63/21064  555 299  1.96 (1.36, 2.93) Reference  55/19104 7/1755 1/70  288 399 1420  Reference 1.40 (0.64, 3.08) 3.85 (0.53, 27.83)  65/19203 37/8661 6/1315  338 427 456  Reference 1.29 (0.86, 1.93) 1.30 (0.56, 2.99)  4/838 74/28213  477 262  1.26 (0.46, 3.47) Reference  0.48 (0.22, 1.08)  ‐This represents relapse, which was also used as a global measure of recurrence  44  Table 2.6: Comparisons of co‐morbidity and other clinical characteristics between TB patients who developed a recurrence (n=108) and who did not (n=4356)  Variable  Diabetes Mellitus Yes No/unknown Malnutrition Yes No/unknown Alcoholism Yes No/unknown Drug abuse Yes No/unknown Either HIV or AIDS Yes No/unknown Any Malignancy Yes No/unknown Immunosuppressive medication Yes No/unknown Chronic Renal failure Yes No/unknown  Number of Incident cases/Total person years  Incidence per 100, 000 person years  Unadjusted Hazard Ratio (95% CI)  5/1714 103/27464  292 375  0.75 (0.30, 1.83) Reference  2/239 106/28939  835 366  2.17 (0.54, 8.80) Reference  7/1390 101/27781  504 364  1.39 (0.64, 2.98) Reference  12/731 96/28434  1641 338  4.28 (2.35, 7.81) Reference  14/838 94/28285  1670 332  4.28 (2.47, 7.40) Reference  2/378 106/28804  529 368  1.37 (0.34, 5.53) Reference  4/479 104/28699  835 362  2.10 (0.77, 5.71) Reference  2/237 106/28941  845 366  2.24 (0.55, 9.08) Reference  Table 2.5 presents the hazard of recurrence from Univariate Cox regression model for treatment and related factors. The factors that had the largest effects on recurrence were incomplete treatment (UHR: 5.62), and poor compliance to treatment (UHR: 6.53). TB patients with negative or without culture (UHR=1.92) were two‐fold more likely to have recurrence compared to culture positive TB patients. However, the mean duration of treatment and major mode of treatment was similar among new and recurrent cases. In terms of diagnosis related factors, no significantly elevated risk of recurrence was observed among patients with EPTB and 45  patients with Cavitations. The impact of co‐morbid factors on the incidence of recurrence is presented in Table 2.6. The relative hazard associated with drug abuse (UHR: 4.28) and positive HIV status (UHR: 4.28) were significantly higher among recurrent cases.  The multivariable Cox regression showed that foreign‐born individuals, place of diagnosis, incomplete treatment, poor‐compliance with treatment, drug abuse and positive HIV status were consistently significant risk factors for recurrence (Table 2.7). Variables that did not reach statistical significance in multivariable models were age, gender, being culture positive and alcoholism. Table 2.8 presents findings from the analysis was restricted to patients who successfully completed their treatment during 1st episode. Variable that was significant in univariate Cox regression were foreign‐birth, born in African, Southeast Asia and Western Pacific regions, diagnosis of TB outside of BC, positive HIV status and substance abuse.  Since information about co‐morbidities was limited in terms of missing observations, a sensitivity analysis was conducted among patients with valid responses to the presence co‐ morbid conditions. Results of this analysis are presented in the appendix (Table A.4 and Table A.5). Additional Cox Regression analysis was conducted among a selective group of recurrent patients based on stricter definition criteria (who had a period of inactivity over 6 months). Results of this additional analysis are presented in the Appendix A (Table A.6). Comparisons of treatment completion and poor‐compliance rates across several risk factors are also presented in Appendix (Table A.7).  46  Table 2.7: Multivariable cox regression analysis of risk factors for recurrence  Variable  Age at primary episode (per year) Biological gender Male Female Birth place CB6 non‐Aboriginal Canadian‐born Aboriginal Foreign‐ born Birth Country regions Pan‐American Health Region African Region East. Mediterranean/Europe Southeast Asia Region Western Pacific Region Place of Diagnosis Outside of Canada7 In BC Culture positivity Positive Negative/unknown Rx completion of primary episode Completed successfully Incomplete Rx compliance of primary episode Compliant Non‐complaint Alcoholism Drug abuse Either HIV or AIDS  Model 1 Adjusted Hazard Ratio (95% CI)  Model 2 Adjusted Hazard Ratio (95% CI)  Model 3 Adjusted Hazard Ratio (95% CI)  Model 4 Adjusted Hazard Ratio (95% CI)  1.00 (0.99, 1.01)  1.00 (0.99, 1.01)  1.00 (1.00, 1.01)  1.00 (0.99, 1.01)  Reference 1.19 (0.79, 1.80)  Reference 1.34 (0.86, 2.08)  Reference 1.15 (0.76, 1.74)  Reference 1.26 (0.81, 1.97)  Reference 1.11 (0.46, 2.71) 2.62 (1.25, 5.46)  Reference 1.30 (0.54, 3.14) 2.63 (1.21, 5.72) Reference 2.57 (0.82, 8.07) 0.78 (0.17, 3.56) 3.20 (1.43, 7.16) 3.82 (1.92, 7.56)  Reference 3.85 (1.28, 11.59) 0.38 (0.05, 2.97) 2.73 (1.18, 6.33) 2.94 (1.45, 5.93)  9.98 (5.97, 16.69) Reference  5.07 (2.62, 9.84) Reference  9.24 (5.50, 15.52) Reference  4.59 (2.33, 9.03) Reference  1.25 (0.76, 2.05) Reference  1.35 (0.76, 2.40) Reference  1.18 (0.72, 1.95) Reference  1.29 (0.73, 2.32) Reference  Reference 4.29 (2.60, 7.08)  ‐  Reference 4.64 (2.78, 7.68)  ‐  Reference 3.98 (3.23, 7.10) 1.58 (0.64, 3.96) 2.50 (0.99, 6.29)8 4.97 (2.28, 10.87)  1.46 (0.62, 3.43) 2.68 (1.06, 6.76) 3.61 (1.69, 7.71)  1.84 (0.77, 4.41) 3.06 (1.13, 8.25) 4.16 (1.75, 9.87)  Reference 4.04 (2.24, 7.26) 1.76 (0.69, 4.44) 3.14 (1.18, 8.31) 4.63 (1.99, 10.82)  6  ‐CB: Canadian‐born ‐ Outside of Canada also included patients who were diagnosed outside of BC, but within Canada (only few cases belong to this category‐ outside of BC, but within Canada) 8 P value was 0.052 7  47  Table 2.8: Incidence of recurrence across several risk factors among TB patients who successfully completed treatment during the primary episode (n=3529)  Overall  Number of Incident cases/Total person years  Incidence per 100, 000 person years  73/27174  269  Unadjusted Hazard Ratio (95% CI)  1.00 (0.99, 1.01)  Age at diagnosis Male  36/13793  261  Reference  Female  37/13380  277  1.07 (0.68, 1.69)  Canadian‐born non‐Aboriginal  6/4737  127  Reference  Canadian‐born Aboriginal  6/3685  163  1.34 (0.43, 4.16)  Foreign‐born  61/18795  325  2.53 (1.09, 5.84)  Pan‐ American Health Region  12/8874  135  Reference  African Region  3/3502  597  4.22 (1.19, 14.97)  Eastern Mediterranean and European Region  1/12253  44  0.40 (0.04, 2.61)  Southeast Asia Region  12/123901  308  2.13 (0.96, 4.74)  Western Pacific Region  45/11583  389  2.78 (1.47, 5.25)  In BC diagnosis  50/25564  196  Reference  23/979  2348  11.98 (7.31, 19.64)  Positive culture  49/19765  248  0.70 (0.43, 1.15)  Negative/no/unknown culture  24/7409  324  Reference  PTB  42/18001  233  Reference  EPTB  29/7906  367  1.60 (1.00, 2.57)  PTB & EPTB  2/1267  158  0.55 (0.16, 2.67)  Diagnosis outside of Canada9  9  ‐ Outside of Canada also included patients who were diagnosed outside of BC, but within Canada (only few cases belong to this category‐ outside of BC, but within Canada)  48  Number of Incident cases/Total person years  Incidence per 100, 000 person years  Unadjusted Hazard Ratio (95% CI)  Cavitations yes  2/801  250  0.80 (0.19, 3.27)  Cavitations no  56/26306  213  Reference 1.53 (0.66, 3.55)  Rx duration (per year) 59/27037  218  Reference  2/394  507  2.20 (0.54, 9.01)  Supervised  15/5470  274  Reference  Self‐administered  38/21295  178  0.62 (0.34, 1.13)  Diabetes mellitus  4/1640  244  0.88 (0.32, 2.40)  Malnutrition yes  1/201  498  1.75 (0.24, 12.57)  Alcoholism yes  4/1191  336  1.26 (0.46, 3.45)  HIV/AIDS positive  4/601  666  3.95 (1.96, 7.94)  Drug abuse yes  9/707  1273  3.42 (1.48, 7.89)  Malignancy yes  2/370  540  2.10 (0.52, 8.59)  Compliant Non‐compliant  2.4 DISCUSSION Between 1990 and 2006, recurrent cases of active TB accounted for 9% of the total active TB cases reported in BC. In comparison to national data reported by Health Canada (6% to 12%), BC had a higher relapse proportion (% total BC cases) for most calendar years (31). A higher prevalence of recurrence continues to pose a concern to Canadian TB control programs because it is a source of ongoing transmission of infection for new cases in the community. 49  This is the first Canadian study which estimated the annual incidence of TB recurrence. The crude incidence was 370 cases per 100,000 person‐years. The cumulative incidence at the first year and 2nd year were 854 and 1500 per 100,000 pys respectively. This incidence is lower than has been reported in other comparable studies. For example, the North American TB trial consortium study 22 and 23 conducted by Jasmer and others reported a recurrence rate of 3800 per 100, 000 person‐years, which was ten times higher than our study (32). The incidence of recurrence was 550 per 100,000 person‐years in a Brazilian study (33). The incidence varies widely under study conditions due to differences in background prevalence of tuberculosis (high vs. low burden countries), selected population (population based vs. institutional setting), study design (clinical trial vs. observational studies) and criteria to define recurrence. Several studies (9,15,19,34,35) reported recurrence of TB (which sometimes was termed as a relapse) in patients whose treatment outcome was ‘cure’ and/or ‘treatment completed’. Our study demonstrated that the recurrence rate was 269 per 100, 000 pys in patients who completed their initial treatments successfully (which represents patients with cure and treatment completed). A recent Australian study restricted to culture positive cases reported that the crude annual incidence of recurrence was 71 per 100, 000 after completion of successful treatment (8). Their rate was much lower (compared to current study) which shares a similar profile with Canada in terms of TB incidence (In our study, the incidence among culture positive patients who completed treatment successfully was 248 per 100,000 pys).  As expected, the incidence of recurrence was much higher in subjects from countries with a high incidence of TB. For example, a recent study conducted in a high‐burden setting (Karakalpakstan, Uzbekistan) reported a high level of recurrence (36%) among patients who were treated in the framework of a DOTS program (19). A Vietnamese study reported a relapse rate of 9% among patients whose treatment outcome was documented as ‘cure’ (34). In another study from Vietnam, the investigators reported a 13 % relapse rate among defaulters or participants who had transferred out (36). The reported recurrence rate after treatment completion in a systematic review (37) based on controlled trials (2290 per 100,000 at first  50  year) and observational studies (7850 per 100,000 at first year) were much higher than in the current study.  A recent population based study conducted in South Carolina reported an overall recurrence rate of 2.9% over a period of 33 years (1970 to 2002), but did not provide any incidence estimates (38). It must be noted, a wide discrepancy exists in setting criteria to define recurrent cases across studies, which makes comparisons difficult. However, this American study (38) used a minimum 12 months inactivity period to define recurrent cases as recommended by Center for Disease Control (CDC), Atlanta. One possible explanation of the low incidence in this setting might be related to quality of heath delivery through a centralized system, which ensures uniform prescription of standard treatment regimens.  Our study showed almost a quarter of recurrent cases occurred five years after treatment completion. The time to recurrent diagnoses varies widely from across studies Clinical trials usually follow patients for a shorter period and in addition more difficult subjects are often excluded. In addition many trials now include DOTs as an integral part of the study protocol. Therefore, their reported time to recurrence is based on one or two years. Most observational studies have a longer follow up. Kopanoff and others found (30) that 30% of total recurrent cases occurred within five years of their prior TB diagnosis and 48% occurred after 11 years. Another American study conducted in the 1970s found the median interval between the original diagnosis and recurrence was 14 years (23). The Division of TB control in BC does active surveillance of patients post treatment completion from several months to several years depending on the severity of initial episode and associated medical conditions. Physician should keep in mind that recurrence could occur a long period after the initial episode.  Several factors consistently emerge in this study as having a strong impact on recurrence in multivariable setting: foreign‐birth, place of initial diagnosis, failure to complete initial TB therapy, poor compliance to such therapy, positive HIV status and substance abuse. This study demonstrated that foreign‐born people experience a significant, two‐fold higher risk of 51  recurrence compared to Canadian‐born people. Although foreign‐born people bear a disproportionate burden of TB in both USA and Canada, their increased risk of recurrence was not consistently found in American studies. One American study demonstrated that foreign‐ born people were more likely to have recurrence (38) while another American study conducted in New York City demonstrated contrasting findings that US‐born people were more likely to have recurrence (9).  This study identified a particular group of TB patients with a very high risk of recurrence (about 40% of total recurrence occurred in this group). Patients whose initial diagnosis took place outside of Canada had at least five times higher risk of recurrence compared to patients with a diagnosis in BC. This was more reflective of patients who were born outside of Canada with a prior history of TB. Vancouver (the largest city of BC) is a place of immigrants, especially from Asian countries (such as China, Vietnam, Hong Kong, Philippines and India). Resources and infrastructure for TB‐control programs are severely limited in these countries. Persons treated for TB in these countries may receive inadequate or incomplete treatment. This puts foreign‐ born persons at greater risk for disease recurrence. Moreover, in foreign‐born TB patients, about 45% of recurrence occurred within three years since their arrival in Canada. These findings have significant policy implications to local TB control program. This offers an important body of evidence to prioritize the development and implementation of effective screening and treatment protocols for this high risk group (foreign‐born patients with prior history of TB at their host country) since their arrival in Canada. Interventions such as aggressive monitoring, post‐arrival active surveillance and prophylaxis should be prioritized in the current TB control program.  This study demonstrated that patients with incomplete therapy were four to five fold more likely (1680 vs. 269 per 100,000 pys) to have recurrence compared to patients who had completed their initial treatment regimen successfully. The effect of incomplete treatment remained significant in all multivariable models. This result is expected and similar findings have been widely documented in previous studies (7,23,30,38). Incomplete therapy leads to 52  inadequate bacteriological cure, which eventually increases TB patients’ vulnerability for a new episode of active TB. A recent study conducted in Vietnam reported a 13% recurrence among patients who defaulted or transferred out. This rate could be much higher in an area burdened by high HIV infection. For examples, the study by Verver et al in South Africa reported a much higher recurrence (28%, 6500 per 100, 000 person‐years) among those who defaulted from treatment (7). The effects of incomplete therapy were well documented in studies conducted pre‐HIV era (23). Although many current studies investigate recurrence rates among patients who either completed therapy for their primary episode or were declared cured following initial treatment, treatment effects are also reported in several recent studies. For example, a recent American study reported that patients with sub‐optimal regimens were six time more likely to have recurrence compared to patients who completed therapy within the optimal timeframe (38). The study of Kopanoff and others demonstrated that 20% of total patients had not received any chemotherapy and another 20% had received inadequate chemotherapy (30). In a commentary, Bloom and Murray reported that about 35‐50% of patients from several American cities did not complete an appropriate TB regimen (39).  This study demonstrated that poor compliance was one of the strongest risk factors for recurrence. This finding has been consistently and widely documented in previous studies (9,30,38,40). Kopanoff and others demonstrated a high rate of non‐compliance among their recurrent cases (198/477) (30). Non‐compliance has been a long‐standing issue and an on‐going problem for TB control. Both older and more recent studies identified non‐compliance to initial TB treatment as the most threatening problem for recurrence (9,30,38,40). Non‐compliance to TB treatment has increased substantially due to an increased prevalence of drug abuse, alcoholism, HIV infection and homelessness (41). This study also demonstrated a significantly higher rate of poor adherence among those reported as having alcoholism and drug abuse (Table A.7 in the appendix).  Despite increased efforts, non‐completion of treatment and non‐adherence to treatment are reasons for concern in resource‐rich countries like Canada. Interventions such as enhanced 53  surveillance activities and aggressive monitoring targeting high‐risk patients (who default or did not comply the initial therapy) should be prioritized to the current TB control program. Moreover, efforts including involvement of well‐trained and committed outreach workers, offering of incentives (such as bus tickets and meal voucher), increased education for patients/family members and forceful confinement for defaulters are needed in order to improve adherence to treatment to prevent the recurrence of TB. Improved understanding and identifying patients with risk of non‐adherence at early stage could potentially improve the treatment outcome.  As expected, this study also observed a significantly higher recurrence rate among patients who had HIV infection (1680 per 100, 000 pys). In the era of HIV, the effect of HIV on increasing TB recurrence have been reported in many studies (42,43). However, the magnitude of effect differs across studies depending on the background incidence of TB and HIV and treatment regimens. The Kenyan study (42) reported a 34‐fold increased recurrence (16.7 vs. 0.5 per 100 PYs) among HIV positive patients. The Kinshasa study (43) demonstrated a relapse rate of 18.1 per 100 pys among HIV infected persons. Several studies also reported a relatively higher recurrence rate and non‐compliance among HIV infected persons compared to non‐infected patients, but the difference was not statistically different (14,16,17,44). This study did not demonstrate a high effect of HIV infection like Kenyan study. This might be due to treatment regimens containing rifampin as a potent sterilizing drug, which has been very effective in the management of HIV co‐infected persons. The Ugandan study conducted by Johnson et al among HIV co‐infected participants demonstrated that patients receiving thiacetazone containing regimens were at increased risk of recurrence. Studies conducted by Hawken et al. and Perriens et al. reported significantly higher relapse rates among patients whose regimens contained thiacetazone in their continuation phase (42,43). These results also reflect the fact that these are second line drugs  In this study, the incidence of recurrence (1670 per 100, 000 pys) among HIV co‐infected patients was higher than with other risk factors. Treatment incompletion and poor‐adherence 54  was significantly higher among those reported to be infected with HIV. Interventions targeting HIV infected persons should be employed so that these high‐risk populations complete the treatment regimens within an acceptable time‐frame and with good adherence.  This study demonstrated that the incidence of recurrence was very high among those reporting a history of substance abuse (1641 per 100, 000 person years) which is almost five times higher than the overall incidence. The effect of substance abuse has been consistently found in previous studies (9,28), in particular in studies conducted in North America (9,45). Patients who use injection drugs were more likely not to comply with recommended treatment regimens, which increase the likelihood of recurrence. This study also confirmed that the rate of treatment incompletion and poor‐compliance was significantly higher in this high risk of population. A systematic approach in identifying patients with substance abuse and refereeing them for appropriate interventions should be prioritized in the current TB control program.  Surprisingly, the present study identified that supervised therapy does not have a significant impact on the development of recurrence despite several studies demonstrating the effectiveness of supervised therapy in reducing relapse, emergence of drug resistance and mortality (17,41,46,47). The study by Alwood et al. reported significantly higher treatment completion among patients who received DOT compared to non‐DOT treatment (41). However, in the current study, the incidence of recurrence was slightly higher among patients who were administered supervised therapy. The rate of non‐compliance was similar in both groups (11% vs. 11%). Poor‐compliance and non‐completion of treatment were the most important predictors of recurrence. The purpose of supervised therapy is to improve the rate of adherence, and eventually increase the treatment completion rate. In the current study, the administration of supervised therapy did not lead to a higher compliance rate when compared to reported self‐administered therapy, and therefore failed to demonstrate a significant effect on recurrence. Administration of anti‐TB therapy was not random in this setting, rather than it was selective in nature. Patients that were deemed to have a high risk of recurrence often received supervised therapy. 55  Since the introduction of Directly Observed Therapy (DOTS), it has become the cornerstone of TB treatment strategies. Although the WHO strongly recommends using DOTS under program conditions, the effectiveness of DOTS was not consistently documented in previous reports. A number of clinical trials conducted in high incidence settings such as India, Pakistan, and Thailand demonstrated DOTS inability to achieve the expected treatment outcome (48‐50). The controlled clinical trial in India, which evaluated three oral short course regimes, showed a significantly lower relapse rate among the unsupervised treatment arm (48). This study concluded partially or completely unsupervised treatment regimen as a viable alternative in resource poor settings (48). However, a study in Uzbekistan (19) reported a very high rate of recurrence and mortality among patients who were successfully treated with standard chemotherapy under the DOTS program. A recent American study reported that recurrence was 3.3% (78/2344) among patients who received DOTS, and higher than their overall recurrence rate 2.3% (123/4571) observed in the same study (9). Another study conducted in San Francisco reported similar rates of relapse, treatment failure and acquired drug resistance among patients who received DOTS therapy and among those who received self‐administered therapy (32).  Although the effect of drug resistance, in particular multi‐drug resistance (MDR) has been significant in other studies (19), our study could not confirm this finding. Recurrence was higher among drug resistant cases compared to sensitive cases and much higher in MDR cases, but the differences were not statistically significant. Low number of MDR cases might be lead to this non‐significant association. However, other potential explanation would be better management of TB cases through this specialized center and in particular its use of a dedicated TB ward which provides treatment for drug resistant cases, thus further preventing the recurrence among drug resistant cases.  Presence of cavitations on chest radiograph has been frequently observed as a risk factor of relapse (10,51). This study could not demonstrate any significant association between the presence of cavitations and recurrence. Several African studies also reported no increased risk 56  of recurrence among patients with residual cavitary disease. Treatment with potent sterilizing TB drugs such as rifampin and pyrazinamide might increase the likelihood of persons with cavitations being more successfully treated and hence reduce the risk of recurrence. Another reason might be related to the use of potent sterilizing drugs that killed all potential organisms, which would have otherwise caused further recurrence (53).  A particular strength of this study is the use of population‐based data under program conditions. These data are more representative of real life settings when compared to data from clinical trials. This study followed patients for a maximum period of 17 years, one of the longest periods of time for observational studies on recurrence. TB is a notifiable disease in Canada and TB control activities in this province are centralized. It is unlikely that this provincial TB control program missed any potential subjects who developed TB while living in British Colombia. Because of information from real‐life settings and our large sample size (over five thousand patients), the findings of the current study are more robust and useful to health policy makers. The confidence intervals for most findings were relatively small, providing additional support for the significance of the findings.  This study has several limitations. Due to passive nature follow up, this study might not account for a few recurrent cases which have moved from BC, which might lead to an underestimation of incidence. A study conducted by Moran et al. 2004 in BC among TB contacts showed that migration of TB patients was low (54).  The criteria set by World Health Organization (WHO) to define treatment cure or completion could not be applied in this study. The inability to have treatment outcome information, which is not systematically recorded in the current database, might make the findings less comparable to other studies based on WHO criteria. Since bacteriological confirmation was not systematically recorded during the study period, we could not categorize patients into relapse, failure and default to be consistent with the WHO. However, treatment related information was available from 1990‐2006. The treatment completion also included physician judgment on 57  the clinical cure of patients. It is expected that the variable completion of treatment used in this study is more representative of both treatment cure and completion. Several studies reported no significant differences in recurrence rate among these two groups‐ cure and treatment completed (10). We do not think that difference in methodology have substantially affected our risk ratio estimates.  Another limitation of this research project was the percentage of missing observations. The percentage of incomplete information was low for socio‐demographic characteristics. However, the frequency of missing observations was much higher for co‐morbid conditions. We assumed that patients with incomplete responses were more representative of patients with “No” responses due to nature of collection of information followed in the TB control program. Therefore, the effect of co morbidity was evaluated in comparison to patients with valid responses and patients with unknown status. However, we also conducted separate multi‐ variable model restricting to patients with a valid response (yes or no) for each co‐morbid condition. Although the magnitude of association differed, the patterns of significant variables were similar (Table A.4 and Table A.5) in both scenarios.  2.5 CONCLUSIONS This study shows that the incidence of TB recurrence is low in BC. However, it is not as low as in a recent Australian study. The stable prevalence of recurrence is an issue of concern due to this being a source of ongoing increased risk of infection. This study identified several important risk factors for recurrent TB in a population‐based cohort. Treatment related factors such as incomplete treatment and poor‐adherence to treatment were the strongest predictors of recurrence. Development and implementation of effective interventions which work to enhance completion rates and compliance to TB treatment are urgently required in order to prevent recurrence. Besides treatment factors, foreign‐birth, place of diagnosis and HIV/AIDS consistently emerged as risk factors contributing to the burden of recurrence. This study also provides an empiric evidence to prioritize the development and implementation of effective 58  screening and treatment protocols for foreign‐born patients with prior history of TB at their host country. This very high risk group of patients should be monitored aggressively since their arrival in Canada.  Several behavioral factors such as substance abuse and alcoholism, which were associated with higher incompletion rates and poor adherence to treatment, were more frequent among patients with recurrence. These behavioral factors including poor adherence to treatment are subject to modification. Efforts should be made to improve the adherence rates of patients through involvement of well‐trained and committed outreach workers. In addition, interventions targeting these high‐risk populations would be valuable in the elimination of TB from Canada.  59  2.6 REFERENCES  (1) Kochi A. The global tuberculosis situation and the new control strategy of the World Health Organization. Tubercle 1991 Mar;72(1):1‐6. (2) Kochi A. The global tuberculosis situation and the new control strategy of the World Health Organization. 1991. Bull.World Health Organ. 2001;79(1):71‐75. (3) Lawn SD, Acheampong JW, Lawn SD, Acheampong JW. Pulmonary tuberculosis in adults: factors associated with mortality at a Ghanaian teaching hospital. West African Journal of Medicine 1999 Oct‐Dec;18(4):270‐4. (4) Gustafson P, Gomes VF, Vieira CS, Samb B, Naucler A, Aaby P, et al. Clinical predictors for death in HIV‐positive and HIV‐negative tuberculosis patients in Guinea‐Bissau. Infection 2007 Apr;35(2):69‐80. (5) Mathew TA, Ovsyanikova TN, Shin SS, Gelmanova I, Balbuena DA, Atwood S, et al. Causes of death during tuberculosis treatment in Tomsk Oblast, Russia. International Journal of Tuberculosis and Lung Disease 2006 Aug;10(8):857‐863. (6) Santha T, Garg R, Frieden TR, Chandrasekaran V, Subramani R, Gopi PG, et al. Risk factors associated with default, failure and death among tuberculosis patients treated in a DOTS programme in Tiruvallur District, South India, 2000. International Journal of Tuberculosis and Lung Disease 2002 Sep;6(9):780‐788. (7) Verver S, Warren RM, Beyers N, Richardson M, van der Spuy GD, Borgdorff MW, et al. Rate of reinfection tuberculosis after successful treatment is higher than rate of new tuberculosis. American Journal of Respiratory & Critical Care Medicine 2005 Jun 15;171(12):1430‐1435. (8) Dobler CC, Crawford AB, Jelfs PJ, Gilbert GL, Marks GB. Recurrence of tuberculosis in a low‐ incidence setting. European Respiratory Journal 2009 Jan;33(1):160‐167.  60  (9) Driver CR, Munsiff SS, Li J, Kundamal N, Osahan SS. Relapse in persons treated for drug‐ susceptible tuberculosis in a population with high coinfection with human immunodeficiency virus in New York City. Clinical Infectious Diseases 2001 Nov 15;33(10):1762‐1769. (10) Mallory KF, Churchyard GJ, Kleinschmidt I, De Cock KM, Corbett EL. The impact of HIV infection on recurrence of tuberculosis in South African gold miners. International Journal of Tuberculosis & Lung Disease 2000 May;4(5):455‐462. (11) Shamputa IC, Van Deun A, Salim MA, Hossain MA, Fissette K, de Rijk P, et al. Endogenous reactivation and true treatment failure as causes of recurrent tuberculosis in a high incidence setting with a low HIV infection. Tropical Medicine & International Health 2007 Jun;12(6):700‐ 708. (12) Jindani A, Nunn AJ, Enarson DA. Two 8‐month regimens of chemotherapy for treatment of newly diagnosed pulmonary tuberculosis: international multicentre randomised trial. Lancet 2004 Oct 2‐8;364(9441):1244‐1251. (13) Johnson JL, Okwera A, Nsubuga P, Nakibali JG, Whalen CC, Hom D, et al. Efficacy of an unsupervised 8‐month rifampicin‐containing regimen for the treatment of pulmonary tuberculosis in HIV‐infected adults. Uganda‐Case Western Reserve University Research Collaboration. International Journal of Tuberculosis & Lung Disease 2000 Nov;4(11):1032‐1040. (14) Sterling TR, Alwood K, Gachuhi R, Coggin W, Blazes D, Bishai WR, et al. Relapse rates after short‐course (6‐month) treatment of tuberculosis in HIV‐infected and uninfected persons. AIDS 1999 Oct 1;13(14):1899‐1904. (15) Thomas A, Gopi PG, Santha T, Chandrasekaran V, Subramani R, Selvakumar N, et al. Predictors of relapse among pulmonary tuberculosis patients treated in a DOTS programme in South India. International Journal of Tuberculosis & Lung Disease 2005 May;9(5):556‐561.  61  (16) Perriens JH, St Louis ME, Mukadi YB, Brown C, Prignot J, Pouthier F, et al. Pulmonary tuberculosis in HIV‐infected patients in Zaire. A controlled trial of treatment for either 6 or 12 months. N.Engl.J.Med. 1995 Mar 23;332(12):779‐784. (17) Chaisson RE, Clermont HC, Holt EA, Cantave M, Johnson MP, Atkinson J, et al. Six‐month supervised intermittent tuberculosis therapy in Haitian patients with and without HIV infection. American Journal of Respiratory & Critical Care Medicine 1996 Oct;154(4 Pt 1):1034‐1038. (18) Chan‐Yeung M, Noertjojo K, Chan SL, Tam CM. Sex differences in tuberculosis in Hong Kong. International Journal of Tuberculosis & Lung Disease 2002 Jan;6(1):11‐18. (19) Cox H, Kebede Y, Allamuratova S, Ismailov G, Davletmuratova Z, Byrnes G, et al. Tuberculosis recurrence and mortality after successful treatment: impact of drug resistance. PLoS Medicine / Public Library of Science 2006 Oct;3(10):e384. (20) A controlled clinical comparison of 6 and 8 months of antituberculosis chemotherapy in the treatment of patients with silicotuberculosis in Hong Kong. Hong Kong Chest Service/tuberculosis Research Centre, Madras/British Medical Research Council. Am.Rev.Respir.Dis. 1991 Feb;143(2):262‐267. (21) Cowie RL. Silicotuberculosis: long‐term outcome after short‐course chemotherapy. Tubercle & Lung Disease 1995 Feb;76(1):39‐42. (22) Johnson IL, Thomson M, Manfreda J, Hershfield ES. Risk factors for reactivation of tuberculosis in Manitoba. CMAJ Canadian Medical Association Journal 1985 Dec 15;133(12):1221‐1224. (23) Snider DE,Jr. Reactivation of tuberculosis in Oklahoma: 1970‐1973. Chest 1975 Jul;68(1):36‐ 40. (24) Segarra F, Sherman DS. Relapses in pulmonary tuberculosis. Dis.Chest 1967 Jan;51(1):59‐ 63. 62  (25) Mitchison DA, Nunn AJ. Influence of initial drug resistance on the response to short‐course chemotherapy of pulmonary tuberculosis. Am.Rev.Respir.Dis. 1986 Mar;133(3):423‐430. (26) Chang KC, Leung CC, Yew WW, Ho SC, Tam CM. A nested case‐control study on treatment‐ related risk factors for early relapse of tuberculosis. American Journal of Respiratory & Critical Care Medicine 2004 Nov 15;170(10):1124‐1130. (27) Dobler CC, Marks GB, Simpson SE, Crawford AB. Recurrence of tuberculosis at a Sydney chest clinic between 1994 and 2006: reactivation or reinfection?. Med.J.Aust. 2008 Feb 4;188(3):153‐155. (28) Cacho J, Perez Meixeira A, Cano I, Soria T, Ramos Martos A, Sanchez Concheiro M, et al. Recurrent tuberculosis from 1992 to 2004 in a metropolitan area. European Respiratory Journal 2007 Aug;30(2):333‐337. (29) Health Canada. Tuberculosis in Canada 1997. 1997; Available at: http://www.hc‐ sc.gc.ca/pphb‐dgspsp/publicat/tbcan97/pdf/tbcan97e.pdf. Accessed March/31, 2010. (30) Kopanoff DE, Snider DE,Jr, Johnson M. Recurrent tuberculosis: why do patients develop disease again? A United States Public Health Service cooperative survey. Am.J.Public Health 1988 Jan;78(1):30‐33. (31) Public Health Agency of Canada. Tuberculosis in Canada 2007, Pre‐Release. 2007; Available at: http://www.phac‐aspc.gc.ca/publicat/2008/tbcanpre07/pdf/tbpre2007‐eng.pdf. Accessed March/31, 2010. (32) Jasmer RM, Seaman CB, Gonzalez LC, Kawamura LM, Osmond DH, Daley CL. Tuberculosis treatment outcomes: directly observed therapy compared with self‐administered therapy. American Journal of Respiratory & Critical Care Medicine 2004 Sep 1;170(5):561‐566. (33) Ruffino‐Netto A. Recurrence of tuberculosis. Jornal Brasileiro De Pneumologia: Publicacao Oficial Da Sociedade Brasileira De Pneumologia E Tisilogia 2007 Oct;33(5). 63  (34) Vree M, Huong NT, Duong BD, Sy DN, Van LN, Hung NV, et al. Survival and relapse rate of tuberculosis patients who successfully completed treatment in Vietnam. Int.J.Tuberc.Lung Dis. 2007 Apr;11(4):392‐397. (35) Shen G, Xue Z, Shen X, Sun B, Gui X, Shen M, et al. The study recurrent tuberculosis and exogenous reinfection, Shanghai, China. Emerging Infectious Diseases 2006 Nov;12(11):1776‐ 1778. (36) Vree M, Huong NT, Duong BD, Sy DN, Van le N, Co NV, et al. Mortality and failure among tuberculosis patients who did not complete treatment in Vietnam: a cohort study. BMC Public Health 2007;7:134. (37) Panjabi R, Comstock GW, Golub JE. Recurrent tuberculosis and its risk factors: adequately treated patients are still at high risk. International Journal of Tuberculosis & Lung Disease 2007 Aug;11(8):828‐837. (38) Selassie AW, Pozsik C, Wilson D, Ferguson PL. Why pulmonary tuberculosis recurs: a population‐based epidemiological study. Ann.Epidemiol. 2005 Aug;15(7):519‐525. (39) Bloom BR, Murray CJ. Tuberculosis: commentary on a reemergent killer. Science 1992 Aug 21;257(5073):1055‐1064. (40) Burman WJ, Cohn DL, Rietmeijer CA, Judson FN, Sbarbaro JA, Reves RR. Noncompliance with directly observed therapy for tuberculosis. Epidemiology and effect on the outcome of treatment. Chest 1997 May;111(5):1168‐1173. (41) Alwood K, Keruly J, Moore‐Rice K, Stanton DL, Chaulk CP, Chaisson RE. Effectiveness of supervised, intermittent therapy for tuberculosis in HIV‐infected patients. AIDS 1994 Aug;8(8):1103‐1108.  64  (42) Hawken M, Nunn P, Gathua S, Brindle R, Godfrey‐Faussett P, Githui W, et al. Increased recurrence of tuberculosis in HIV‐1‐infected patients in Kenya. Lancet 1993 Aug 7;342(8867):332‐337. (43) Perriens JH, Colebunders RL, Karahunga C, Willame JC, Jeugmans J, Kaboto M, et al. Increased mortality and tuberculosis treatment failure rate among human immunodeficiency virus (HIV) seropositive compared with HIV seronegative patients with pulmonary tuberculosis treated with "standard" chemotherapy in Kinshasa, Zaire. Am.Rev.Respir.Dis. 1991 Oct;144(4):750‐755. (44) Kassim S, SassanMorokro M, Ackah A, Abouya LY, Digbeu H, Yesso G, et al. Two‐year follow‐up of persons with HIV‐1‐ and HIV‐2‐associated pulmonary tuberculosis treated with short‐course chemotherapy in West Africa. AIDS 1995 October;9(10):1185‐1192. (45) Benator D, Bhattacharya M, Bozeman L, Burman W, Cantazaro A, Chaisson R, et al. Rifapentine and isoniazid once a week versus rifampicin and isoniazid twice a week for treatment of drug‐susceptible pulmonary tuberculosis in HIV‐negative patients: a randomised clinical trial. Lancet 2002 Aug 17;360(9332):528‐534. (46) Chaulk CP, Moore‐Rice K, Rizzo R, Chaisson RE. Eleven years of community‐based directly observed therapy for tuberculosis. JAMA 1995 Sep 27;274(12):945‐951. (47) Weis SE, Slocum PC, Blais FX, King B, Nunn M, Matney GB, et al. The effect of directly observed therapy on the rates of drug resistance and relapse in tuberculosis. N.Engl.J.Med. 1994 Apr 28;330(17):1179‐1184. (48) A controlled clinical trial of oral short‐course regimens in the treatment of sputum‐positive pulmonary tuberculosis. Tuberculosis Research Centre. International Journal of Tuberculosis & Lung Disease 1997 Dec;1(6):509‐517.  65  (49) Walley JD, Khan MA, Newell JN, Khan MH. Effectiveness of the direct observation component of DOTS for tuberculosis: a randomised controlled trial in Pakistan. Lancet 2001 Mar 3;357(9257):664‐669. (50) Kamolratanakul P, Sawert H, Lertmaharit S, Kasetjaroen Y, Akksilp S, Tulaporn C, et al. Randomized controlled trial of directly observed treatment (DOT) for patients with pulmonary tuberculosis in Thailand. Transactions of the Royal Society of Tropical Medicine & Hygiene 1999 Sep‐Oct;93(5):552‐557. (51) Sonnenberg P, Murray J, Glynn JR, Shearer S, Kambashi B, Godfrey‐Faussett P. HIV‐1 and recurrence, relapse, and reinfection of tuberculosis after cure: a cohort study in South African mineworkers. Lancet 2001 Nov 17;358(9294):1687‐1693. (52) Johnson JL, Okwera A, Vjecha MJ, Byekwaso F, Nakibali J, Nyole S, et al. Risk factors for relapse in human immunodeficiency virus type 1 infected adults with pulmonary tuberculosis. International Journal of Tuberculosis & Lung Disease 1997 Oct;1(5):446‐453. (53) Aber VR, Nunn AJ. [Short term chemotherapy of tuberculosis. Factors affecting relapse following short term chemotherapy]. Bull.Int.Union Tuberc. 1978 Dec;53(4):276‐280. (54) A. Morán Mendoza. The value of the tuberculin skin test size in predicting the development of tuberculosis in contacts of active cases. University of British Columbia: University of British Columbia; 2004.  66  CHAPTER 3: A POPULATION BASED STUDY OF TUBERCULOSIS RECURRENCE­RELAPSE VERSUS REINFECTION10  10  ‐“A version of this chapter will be submitted for publication. Moniruzzaman, A. Kazanjian, A. Wong, H. Elwood, R.K. Rodrigues, M. Tang, P, Sharma, M and FitzGerald, J. M. (2010) Recurrence of Tuberculosis: Relapse vs. Reinfection.”  67  3.1. INTRODUCTION Recurrence refers to a new episode of tuberculosis (TB) after the prior episode has defined as cured or inactive. Recurrence may be due to either endogenous reactivation (relapse) or exogenous reinfection (reinfection). If both the recurrent episode and primary episode of TB are caused by the same strain of TB bacillus, then it is considered to be a relapse. However, if different strains are found for both episodes, reinfection is considered the likely mechanism. Although complex, it is important to know whether recurrence is linked to an ongoing community transmission (i.e., reinfection) or acquired through individual host mechanism (i.e., relapse). A patient living in a community that has a high prevalence of TB infection is at an increased risk for a new episode of active TB. Reinfection is thus an important indicator of community transmission. A precise estimate of the extent of reinfection existing in specific communities is vital when attempting to prioritize control efforts (e.g., effective detection and management of contagious cases) to decrease the number of cases with new infection and reinfection.  It is noteworthy to mention that the relative contribution of reinfection to recurrence has direct implications on policy‐making and effective allocation of resources for TB control and prevention, such as the development of an improved vaccine against TB (1,2). If the infection from the first episode does not provide protective immunity against a new strain of Mycobacterium tuberculosis, the development of new TB vaccines based on existing knowledge of host defenses becomes challenging (1‐3). Moreover, if reinfection is a frequent cause of recurrence, then the evaluation of TB treatment regimens used in clinical trials becomes problematic because it can be difficult to determine their effectiveness (i.e., successfully treated patients may unknowingly become re‐infected during or after the study during the follow up period) (4). This type of information is vital to the development of efficient and effective TB management programs.  The relative contribution of relapse and re‐infection as a cause of recurrence has been an issue of fundamental debate for many decades because both types of recurrence are not clinically 68  distinct (1,2). Studies published in the last few decades described relapse exclusively as the mechanism for recurrence (4‐6). During this period, TB researchers were limited in resources or access to techniques that could differentiate relapse from re‐infection. However, with the advent of molecular epidemiology techniques, it is now possible to make quantitative assessment of the contribution of re‐infection to recurrence.  To date, few studies have been conducted to explicitly examine the contribution of re‐infection to TB recurrence. The relative contribution of re‐infection varies widely (0‐100%) across the studies that have been conducted (4,6‐19). The majority of these studies have been conducted on selected populations and reported only the proportion of recurrence due to re‐infection (no incidence estimate of re‐infection was available).  A recent systematic review (1) on re‐infection raised concern about the quality of supporting evidence presented in many studies. The authors of this review strongly recommended that future studies in this field utilize a more rigorous methodology to allow for an improved understanding of the possible mechanisms of recurrence. Although studies have been conducted in several countries (mostly in the developed world), similar information is not available in Canada. A study (18) on re‐infection was recently conducted across North America among patients participating in a clinical drug trial, no specific estimates were provided for Canadian patients. In addition, no formal investigation of re‐infection has been conducted in British Columbia. The objective of this study is to investigate the relative frequency of relapse vs. re‐infection as a mechanism of recurrence and to estimate the incidence of relapse and re‐ infection separately.  3.2 METHODS The Provincial TB service (TB Control Division) at the British Columbia Centre for Disease Control (BCCDC) acts as a referral centre for the prevention, control and treatment of TB and latent TB infection in British Columbia (BC). TB control activities in BC are centralized; therefore, 69  this center maintains a central registry that includes information on all TB patients diagnosed in this province. In addition, all mycobacteriology tests are coordinated by this center. The BCCDC TB laboratory, which performs all culture and drug susceptibility tests, also serves as the reference lab for this province. Since 1995, TB laboratory preserves all positive isolates that had drug susceptibility testing. Therefore, this study was restricted to TB patients who were diagnosed from 1995 to 2006. Using the myco‐bacteriological records, all patients with positive culture for MTB complex from 1995 to 2006 were identified and included in the study. Patients who were diagnosed as new cases in 2005 and 2006 were excluded from the analysis. In addition, recurrent cases that had a primary episode before 1995 were not accounted in this study.  70  Figure 3.1: Study population 3621 TB Cases between 1995 and 2006 Excluded  Recently (in 2005 and 2006) diagnosed new cases (n=557)  Total TB cases between 95 and 04 (3061)  Excluded  Cases with Culture negative or no culture (n=639, 21% of total cases)  Study Cohort for Reinfection/Relapse: Culture Positive Cases (n=2425)  Successfully treated (n=1897, 78%)  Incomplete treatment (n=133, 6%)  35 recurrent cases (2%)  10 recurrent cases (7%)  18 culture positive recurrent cases (0.84%)  8 culture positive recurrent cases (6%) (Paired Isolates available for all cases)  Paired isolates available for 16 cases  Total available paired isolates =24  Treatment other (n=395, 16%)  Identical MIRU pattern = 22 Not identical MIRU pattern=2  71  3.2.1 Recurrent Cases According to the Canadian Tuberculosis Reporting guidelines, all subjects who are diagnosed with TB in Canada need to be reported to Heath Canada under the headings new or relapsed (recurrent) cases. They define a recurrent case as a TB patient with “documented evidence or history of previously active tuberculosis which became inactive” (20). A TB patient who did not have history of prior TB is termed as a new case. The BC TB control program follows the same criteria set by Canadian Tuberculosis Reporting System to determine new and recurrent cases. A recurrent case whose both primary and second episode occurred between 1995 and 2006 was only included in the current analysis. Recurrent cases that had positive culture for both initial and recurrent episodes were identified and isolates from their positive cultures were sent for DNA fingerprinting.  3.2.2 Follow­up Time For each TB patient, person‐time was estimated from the interval between time 1 and time 0. Treatment end date of the primary episode was defined as time zero. Time 1 was defined at the time when patients were censored. Recurrent cases were censored at the day of diagnosis of recurrence. Patients without recurrence were censored when 1) death occurred, or 2) they left the Province or 3) follow‐up ended (Study end date, 31st December, 2006). Patients who were diagnosed post‐mortem (date of diagnosis and date of death was the same day) had contributed zero follow up time.  3.2.3 Mycobacteriology Mycobacterial culture (which is considered as the gold standard for a TB diagnosis) is routinely performed for available specimens of all suspected TB cases in order to confirm the diagnosis. Once culture of body fluids is reported positive for MTB complex, drug susceptibility testing to first‐line antibiotics is performed on at least one isolate from each culture positive patient. In 72  BC, routine susceptibility testing is performed for four drugs: rifampin, isoniazid, ethambutol and streptomycin. The susceptibility testing for pyrazinamide is only carried out when subjects are resistant to isoniazid or rifampin. In the BCCDC TB laboratory, the BACTEC radiometric system is used to determine susceptibility using standardized techniques. The drug concentrations used to identify resistant status were as follows: isoniazid: 0.1 µg/ml, rifampin: 2.0 µg/ml, ethambutol: 4.0 µg/ml and streptomycin: 2.0 µg/ml. This BACTEC method has been used by the BCCDC laboratory since 1990 and has been demonstrated to be 100% sensitive in identifying resistance to first‐line drugs (21‐23).  3.2.4 Culture and DNA Extraction Using mycobacteriology data, recurrent TB patients who were culture positive during initial and recurrent episode were identified. Clinical specimens from these culture positive patients were processed using standard methods. All this processing was performed at the BCCDC TB laboratory. After maintaining all the required protocols, these isolates were sent to the National Reference Mycobacterium Laboratory in Winnipeg for DNA fingerprinting.  3.2.5 DNA Fingerprinting As a fingerprinting technique, mycobacterial interspersed repetitive unit‐ variable number tandem repeats (MIRU‐VNTR) was applied in order to genotype the strain from each paired isolates. Although the Restricted Fragment Length Polymorphism (RFLP) method is the current gold standard for the DNA fingerprinting, this method is subject to several limitations (24‐26). RFLP not only involves large amount of DNA, but also requires extended technician time to complete the test (25,26). On the other hand, MIRU which is as discriminatory as Restricted Fragment Length Polymorphism (RFLP), has several advantages over RFLP such as being convenient and less labor intensive (27‐30). Moreover, this genotyping test has ability to discriminate isolates with low copy numbers of IS6110 (30). Recently, the Center for Diseases Control (CDC) has adopted the MIRU‐VNTR method as the first‐line genotyping technique (31). 73  Therefore, this method has been used in several recent molecular epidemiology studies (7,32). If MIRU‐VNTR of the both paired isolates had similar patterns, the patient was documented as a case of relapse. If the MIRU patterns were different for both isolates, the patient was classified as a case of re‐infection. In addition to the DNA fingerprinting method, clinical characteristics of each patient during the initial and recurrent episodes were evaluated to validate the identification of relapse or re‐infection as a cause of recurrence. MIRU‐VNTR was performed in National Reference Mycobacterium Laboratory in Winnipeg, Canada.  3.2.6 Variables Description A detailed description of variables has been discussed in the previous chapter (Chapter 2). Only several important variables will again be highlighted in this section. Age was calculated at the time of diagnosis of the initial episode. Successful treatment was defined as patients who received at least 6 months of TB therapy and were deemed to have clinical, radiological and bacteriological improvement. On the other hand, incomplete treatment included patients who could not complete treatment due to individual causes (such as clients’ decision not to continue treatment, non‐compliance, lost to follow up). Patients who could not complete treatment due to death or relocation out of province or due to the diagnosis post‐mortem or other reasons were categorized as other. Patients who had been administered therapy supervised once daily, supervised twice weekly and supervised thrice weekly were collapsed into supervised therapy. Based on drug susceptibility testing results, patient were categorized into mono resistant (resistance to a single drug), MDR (resistance to at least isoniazid and rifampin) and poly resistant (resistance to multiple drugs, not MDR).  3.2.7 Statistical Analysis An initial descriptive analysis for the study population was performed. Incidence of recurrence was estimated per person‐year of follow up and expressed in terms of 100, 000 person‐years.  74  3.3 RESULTS 3.3.1 Description of Study Cohort Overall, 3621 TB cases were registered with the provincial TB control program during the study period (1995‐ 2006). Among them, 557 new cases were excluded since they were diagnosed in 2005 and 2006. Out of 3064 eligible cases, 2425 (79%) cases were culture positive during their initial episode. Although the characteristics of 3064 cases were presented in this study, the primary analysis (estimation of re‐infection) was restricted to culture positive cases. Table 3.1, Table 3.2 and Table 3.3 describe the characteristics of culture positive cases compared to all cases, culture positive cases with successful therapy and cases with negative or no culture in respect to socio‐demographic characteristics, treatment related factors and co‐morbidities.  The mean age of culture positive cases at the time of diagnosis was 50 years. There were more male (53%) than female subjects. Foreign‐born individuals accounted for three quarters of the total culture positive TB cases, while Aboriginal people and non‐aboriginal Canadian‐born accounted for 12% and 14% respectively. In terms of birth country, almost half of patients originated from the Western Pacific region. The distribution of age, gender and other socio‐ demographic characteristics were similar across the four patient groups (Table 3.1) except the cases with negative or without culture were significantly older compared to the other groups. In addition, culture negative/unknown patients differ substantially from other groups in terms of location for the diagnosis of the primary episode. Outside of BC as a place of initial diagnosis was mentioned for 13% cases among patients with culture negative or unknown status (1‐3% for other three groups).  75  Table 3.1: Socio‐demographic characteristics of study cohort Variable  Age at diagnosis of primary episode Mean (SD) Median (IQR) Gender Male Female Birth place Canadian‐ born non‐Aboriginal Canadian‐ born Aboriginal Foreign‐born Ethnicity Caucasian Aboriginal Chinese South‐East Asian Filipino Vietnamese Other Birth Country based on WHO Regions PAHO African Region Eastern Mediterranean and European Region Southeast Asia Region Western Pacific Region Place of initial diagnosis Within BC Outside of BC Unknown/Missing  Total cases (3064)  Culture positive cases (2425)  Culture negative/ no (639)  N (%)  Culture positive cases with successful treatment (1897) N (%)  N (%) Years 48 (21.5) 45.5 (31‐66)  Years 50.2 (21.0) 47.3 (33‐68)  Years 48.9 (20.6) 46 .1 (32‐66)  Years 57.5 (21.8) 61.8 (38‐77)  1618 (53) 1446 (47)  1296 (53) 1129 (47)  971 (51) 926 (49)  322 (50) 317 (50)  468 (16) 321 (11) 2919 (73)  333 (14) 268 (12) 1701 (74)  261 (14) 192 (10) 1414 (76)  135 (22) 53 (9) 429 (69)  426 (14) 321 (11) 733 (25) 524 (18) 270 (9) 192 (6) 499 (17)  336 (14) 268 (12) 588 (25) 399 (17) 205 (9) 148 (6) 397 (17)  271 (15) 191 (10) 504 (27) 340 (18) 176 (9) 124 (7) 265 (14)  90 (15) 53 (9) 145 (23) 125 (20) 65 (10) 44 (7) 102 (16)  848 (29) 60 (2) 205 (7)  648 (28) 45 (2) 164 (7)  486 (26) 40 (2) 137 (7)  200 (33) 15 (2) 41 (7)  487 (17) 1293 (45)  387 (17) 1034 (46)  325 (18) 874 (47)  100 (16) 259 (42)  2580 (84) 103 (3) 381 (13)  2094 (86) 20 (1) 311 (13)  1636 (86) 15 (1) 246 (13)  486 (76) 83 (13) 70 (11)  N (%)  Among the culture positive cases, 2422 (79%) had a history of successfully completing therapy for their initial episode whereas 133 (6%) had a history of incomplete therapy. The mean duration of treatment of the initial episode for culture positive cases was 0.7 years (8 months). Self‐administered therapy was administered to the majority of patients (70%) whereas supervised therapy was administered to 13% of patients. Among patients whose drug 76  susceptibility test results were available, 91% were sensitive, 1% had multi‐drug resistant (MDR) and the remaining 8% had either mono or poly resistance. The frequency of treatment and related factors, among the four groups of patients, were comparable (Table 3.2). The percentage of missing observations for treatment completion (11 to 16%) and administration of therapy (13 to 17%) did not differ substantially across the four groups. The median follow up period for culture positive cases was 4.4 years whereas it was higher for cases with negative culture or unknown culture status (5.4 years).  Overall, 7% patients were HIV positive and 5% patients had a history of substance abuse. The prevalence of most co‐morbidity including HIV was lower among culture/no cases compared to all other groups (Table 3.3). However, the distribution of these risk factors was similar among all cases, culture positive cases and culture positive cases with successful treatment.  3.3.2 Recurrence Out of 3064 cases, 71 (2.3%) patients had another episode of active TB during the study period. Table 3.4 describes the incidence of recurrence, relapse and re‐infection among several groups. Overall, the incidence of recurrence in this study cohort was 475 per 100, 000 person‐years. Among initially culture positive cases, the total number of recurrent cases was 45 (1.9%). However, the proportion of recurrent patients (26 cases) among cases with initially negative or no culture results was high (4.1%) compared to culture positive cases. The incidence was 718/100,000 among culture negative/no culture cases and 397/100000 among culture positive cases. The incidence among culture negative/no cases was the highest compared to other groups and almost two times higher than that of culture positive cases. Out of 26 recurrent cases that occurred among patients with culture negative/unknown status, 21 (81%) had a history of initial diagnosis outside of Canada.  77  Table 3.2: Treatment and other related characteristics of study cohort  Variable  Disease type Pulmonary Extra‐pulmonary Both Duration of Treatment in years Mean (SD) Median (IQR) Rx compliance with primary episode Yes No Unknown Rx completion of primary episode Completed successfully Incomplete Other/unknown Main mode of Treatment Supervised Self‐administered Unknown Drug resistance status Sensitive Mono/poly resistance MDR  Total cases (3064)  Culture positive cases (2425)  Culture negative/no (639)  N (%)  Culture positive cases with successful treatment (1897) N (%)  N (%) 2041 (67) 853 (28) 170 (5)  1631 (67) 634 (26) 160 (7)  1280 (68) 496 (26) 121 (6)  410 (64) 219 (34) 10 (2)  0.67 (0.33) 0.63 (0.49‐0.80)  0.68 (0.33) 0.65 (0.50‐0.82)  0.75 (0.29) 0.72 (0.52‐ 0.85)  0.63 (0.30) 0.55 (0.49‐0.74)  2540 (83) 305 (10) 219 (7)  2007 (83) 245 (10) 173 (7)  1851 (98) 27 (1) 19 (1)  533 (83) 60 (9) 46 (7)  2422 (79) 176 (6) 466 (15)  1897 (78) 133 (6) 395(16)  ‐  525 (82) 43 (7) 71 (11)  365 (12) 2171 (71) 528 (17)  319 (13) 1689 (70) 417 (17)  259 (14) 1394 (73) 244 (13)  46 (7) 482 (75) 111 (17)  2197 (91) 196 (8) 19 (1)  2197 (91) 196 (8) 19 (1)  1724 (91) 151 (8) 12 (1)  ‐  N (%)  As expected, the incidence was lower (328/100,000) among culture positive patients who completed their initial TB therapy successfully. The median time to recurrence for all recurrent cases since their treatment completion was 1.8 years. However, the median time to recurrence was lower for culture negative/no cases (1.3 years) compared to culture positive (1.5 years) cases.  78  Table 3.3: Co‐morbidity and behavioral characteristics of TB patients in BC between 1995 and 2006  Variable  Total cases (3064)  Culture positive cases (2425)  Culture negative/no (639)  N (%)  Culture positive cases with successful treatment (1897) N (%)  N (%) Diabetes Mellitus  192 (6)  175 (7)  139 (7)  17 (6)  Malnutrition  40 (1)  38 (2)  20 (1)  2 (<1)  Alcoholism Drug abuse  147 (5) 140 (5)  133 (6) 132 (5)  86 (5) 72 (4)  14 (2) 8 (1)  Either HIV or AIDS Any Malignancy Renal failure Immunosuppressive medication  185 (6) 67 (2) 45 (2) 69 (2)  167 (7) 56 (2) 42 (2) 59 (2)  105 (6) 30 (2) 20 (1) 46 (2)  18 (3) 11 (2) 3 (1) 10 (2)  N (%)  3.3.3 Culture Positive Recurrent Cases Out of 45 recurrent cases that occurred among culture positive cases, 26 cases were culture positive at the time they presented with the recurrent episode. The incidence of culture positive recurrent cases was 230 per 100,000 person‐years. The incidence was much lower (169 per 100, 000) among culture positive patients with a history of successful completion of their initial TB treatment. This group had the lowest incidence of recurrence compared to all other groups.  Among 26 culture positive recurrent cases, 24 cases had paired isolates available for DNA fingerprinting. The remaining two culture positive cases were initially diagnosed outside of BC. Therefore, their specimens were not available for DNA fingerprinting. In terms of clinical specimen types available for DNA fingerprinting, 14 (58%) were sputum samples during the initial episode and 16 (67%) during the recurrent episode. The median interval between two 79  clinical lab specimens collection dates was 27 months with a minimum of 10 months and a maximum period of 78 months. The median period between the last positive sample related to the initial episode and, the sample associated with the recurrent episode was 26 months (minimum and maximum: 9‐68 months).  The socio‐demographic, treatment and diagnosis related characteristics and co‐morbid conditions of the 24 recurrent cases are presented in Table 3.6, Table 3.7 and Table 3.8. The mean age at the time of recurrence was 59 years, which was much higher than that of all recurrent cases. This recurrent cohort was over‐represented by Canadian‐born aboriginal (21%, overall ‐ 8%) and Canadian‐born non‐aboriginal people (21%, overall‐ 5%). The majority of patients (83%) were sensitive to first‐line drugs during their initial episode. Three patients (13%) were either mono or poly resistant and one case (4%) was MDR. However, four out of 20 sensitive cases developed resistance when presenting with a recurrent episode (three was mono‐resistant and one was MDR).  Out of these 24 recurrent cases, 16 cases occurred among patients who had history of successful treatment for their initial episode and the remaining 8 cases occurred among those who did not complete their initial TB regimen. The median time of recurrence (since treatment completion was 18 (IQR: 7‐38) months. Six patients developed recurrence within six months of treatment completion ‐ four among those who completed treatment successfully (Table 3.7, Case # 3, 8, 14 and 17) and two among those who did not complete their initial treatment successfully (Table 3.7, Case # 9 and 10). For Case # 9 and 10, the interval between the last positive sample from the initial episode and the recurrent sample was 9 (treatment for 7 months) and 11 (treatment for 10 months) months respectively (Table 3.8).  80  Table 3.4: Incidence of recurrence, relapse and reinfection  Variable  Total cases (3064)  Culture positive cases (2425)  Culture positive cases with successful Rx (1897)  Culture negative/no/un known (639)  Overall, number of recurrent cases  71/3064=2.3%  45/2425=1.9%  35/1897=1.8%  26/649=4.1%  Incidence per 100, 000 person years  475  397  328  718  Number of culture positive recurrent cases Incidence per 100, 000 person years  26/2425=1.1%  18/1897=1%  229  169  Relapse  22/2425=0.9%  16/1897=0.8%  Incidence per 100, 000 person years  194  150  Re‐infection  2/2425=0.08%  2/1897=0.11%  Incidence per 100, 000 person years  18  19  4.4 (1.7‐7.7)  5.4 (2.9‐8.3)  Median Follow up time (IQR) in years  4.6 (1.8‐8.0)  5.9 (2.6‐8.7)  81  Table 3.5: Characteristics of 24 culture positive recurrent cases with available DNA fingerprint Variable  Age at recurrence Mean (SD) Median (IQR) Years post initial episode Mean (SD) Median (IQR) Gender Male Female Birth place CB non‐Aboriginal CB Aboriginal Foreign‐born Birth Country by WHO region PAHO African Region Eastern Mediterranean and European Region Southeast Asia Region Western Pacific Region Place of initial diagnosis Within BC Outside of BC Marital status Married Single Disease type of initial episode Pulmonary Extra‐pulmonary Both Duration of Treatment Mean (SD) Median (IQR) Rx completion of 1st episode Complete Incomplete Other/unknown Rx compliance of 1st episode Yes No Unknown Major mode of Treatment of  (N=71)  Recurrence among Culture positive cases (N=45)  Recurrence among culture positive cases with successful treatment (N=35)  Culture positive recurrent cases with available DNA fingerprint (N=24)  Years 47.9 (20.2) 40.1 (30‐62) Years 2.5 (2.3) 1.8 (0.8‐3.6)  Years 51.6 (20.6) 50.0 (36‐69) Years 2.2 (2.0) 1.5 (0.6‐3.3)  Years 50 (21.2) 39.5 (33‐67) Years 2.1 (2.1) 1.5 (0.6‐3.2)  Years 58.9 (19.4) 56.7 (39‐78) Years 1.7 (1.5) 1.3 (0.5‐3.1)  32 (45) 39 (55)  20 (44) 25 (56)  16 (46) 19 (54)  13 (54) 11 (46)  5 (7) 6 (8) 60 (85)  5 (11) 5 (11) 35 (78)  2 (6) 3 (9) 30 (86)  5 (21) 5 (21) 14 (58)  11 (16) 4 (6) 1 (1)  10 (22) 2 (4) 1 (2)  5 (14) 2 (6) 1 (3)  10 (42) 2 (8) 0 (0)  13 (18) 42 (59)  7 (16) 25 (56)  5 (14) 22 (63)  5 (21) 7 (29)  45 (63) 26 (37)  40 (89) 5 (11)  31 (89) 4 (11)  24 (100) 0 (0)  17 (70) 7 (29)  11 (65) 6 (35)  11 (73) 4 (27)  3 (37) 5 (63)  43 (61) 24 (34) 4 (6) Years 0.70 (0.44) 0.66 (0.50‐0.89)  23 (51) 19 (42) 3 (7) Years 0.74 (0.51) 0.67 (0.50‐ 0.90)  19 (54) 15 (43) 1 (3) Years 0.82 (0.50) 0.74 (0.53‐ 0.90)  14 (58) 7 (29) 3 (13) Years 0.79 (0.66) 0.65 (0.51‐0.90)  51 (72) 16 (22) 4 (6)  35 (78) 10 (22) 0 (0)  ‐  16 (67) 8 (33) 0 (0)  46 (65) 10 (14) 15 (21)  35 (78) 7 (16) 3 (7)  31 (89) 1 (3) 3 (8)  19 (80) 5 (21) 0 (0)  Total recurrent case  82  Variable  Recurrence among Culture positive cases (N=45)  Recurrence among culture positive cases with successful treatment (N=35)  Culture positive recurrent cases with available DNA fingerprint (N=24)  7 (16) 33 (74) 5 (11)  6 (17) 24 (69) 5 (14)  6 (25) 18 (75) 0 (0)  37 (82) 7 (16) 1 (2)  30 (86) 4 (11) 1 )3)  2 (3)  1 (2)  0 (0)  20 (83) 3 (13) 1 (4) 1 (4)  9 (13) 10 (14) 2 (3)  7 (16) 8 (18) 2 (4)  5 (14) 6 (17) 2 (6)  7 (29) 8 (33) 2 (8)  Total recurrent case (N=71)  initial episode Supervised Self‐administered Unknown Drug resistance status of initial episode Sensitive Mono/poly resistance MDR Alcoholism Drug abuse Either HIV or AIDS Malignancy  8 (11) 39 (55) 24 (34) 37 (82) 7 (16) 1 (2)  83  Table 3.6: Characteristics of 24 culture positive recurrent cases with paired isolates11  Case #  Outcome  Sex  Ethnicity  Age  HIV/ AIDS  Diagnosis Year of initial episode  Diagnosis type at initial episode  Diagnosis Year of recurrent episode  Diagnosis type at recurrent episode  1  Relapse  F  CB Aboriginal  57  No  1995  1996  2  Relapse  M  FB  84  No  1996  3  Relapse  M  FB  34  No  1997  4  Re‐ infection  M  CB Aboriginal  32  Yes  1997  Gastro‐ Intestinal TB Minimal, Pulmonary TB Minimal Pulmonary TB Gastro‐ Intestinal TB  5  Relapse  F  CB non‐ Aboriginal  75  No  1998  2000  6  Relapse  M  FB  76  No  2000  Pulmonary TB, Unspecified Pulmonary Nodule  Gastro‐ Intestinal Pulmonary TB, Unspecified Lymph Node TB, Peripheral Bone Joint TB, Unspecified Pulmonary TB  7  Relapse  F  FB  33  Yes  1999  Miliary TB  2003  8  Relapse  M  FB  89  No  1999  Pleurisy TB  2000  9  Relapse  M  CB non‐ Aboriginal  39  Yes  2000  Miliary TB  2000  10  Relapse  M  CB non‐ Aboriginal  46  Yes  1999  2001  11  M  Yes  2000  M  CB non‐ Aboriginal FB  50  12  Re‐ infection Relapse  Pulmonary TB, Epididymis TB Miliary TB  55  No  2000  2002  13  Relapse  F  CB Aboriginal  51  No  2000  14  Relapse  F  FB  38  Yes  2000  Pulmonary TB Lymph Node TB, Peripheral Miliary TB, Unspecified  11  2000  2001  2001  2002  2002  2002  2001  Pulmonary TB, Unspecified Genitourinar y TB, Unspecified Pulmonary TB, Unspecified Lymph Node TB, Peripheral Pulmonary TB  Pulmonary TB, Pulmonary TB Joint TB  Miliary (Lung) TB  ‐ CB: Canadian‐born, FB: Foreign‐born  84  Case #  Outcome  Sex  Ethnicity  Age  HIV/ AIDS  Diagnosis Year of initial episode  Diagnosis type at initial episode  Diagnosis Year of recurrent episode  Diagnosis type at recurrent episode  15  Relapse  F  FB  65  No  2001  Pleurisy TB  2002  Pleurisy TB  16  Relapse  F  FB  83  No  2001  2003  17  Relapse  M  CB non‐ Aboriginal  36  Yes  2002  Pulmonary TB Pulmonary TB  18  Relapse  F  FB  80  No  2003  19  Relapse  F  CB Aboriginal  28  No  2002  20  Relapse  M  FB  55  No  2002  21  Relapse  M  FB  67  No  2000  Pulmonary TB Intra‐thoracic Lymph Node TB Pulmonary TB Pulmonary TB, Unspecified Pulmonary TB Pleurisy TB  22  Relapse  M  FB  70  No  1999  23  Relapse  F  FB  76  No  24  Relapse  F  CB Aboriginal  35  Yes  Pulmonary TB Cavitary Lung TB  2004  2004 2005  Pulmonary TB Infiltrative Lung TB Cavitary Lung TB  2005  2002  Pleurisy TB  2006  Pulmonary TB, Unspecified Miliary TB  2002  Pulmonary TB, Unspecified  2006  Pleurisy TB  2005 2006  85  Table 3.7: Treatment (Rx) characteristics of 24 culture positive recurrent cases during their initial episode Case #  Rx Duration  1  1  2  Rx completeness  Time of recurrence (since Rx end) in months  14  Self Administered  6  Incomplete, non‐ compliant Successful Rx  39  3  41  Successful Rx  4  4  15  Successful Rx  33  5  <1  29  6 7  6 13  Incomplete, Drug reaction Successful Rx Successful Rx  8  6  Successful Rx  4  9  7  2  10  10  11  6  Incomplete, lost to follow up Incomplete, non‐ compliant Successful Rx  Self Administered Self Administered Supervised Twice Weekly Self Administered DOT/ATLANTA Self Administered Self Administered Self Administered  12  7  Successful Rx  12  13  16  8  14  8  Incomplete, non‐ compliant Successful Rx  15  9  Successful Rx  9  16  11  Successful Rx  6  17  11  Successful Rx  5  18 39  Culture results12  Rx Administration  0‐3 months ‘+’  3‐6 months No cul  6‐9 moths No cul  9‐12 months No cul  ‘+’ & ‘‐’  ‘+’ & ‘‐’  No cul  No cul  ‘+’ & ‘‐’  No cul  No cul  ‘+’  ‘+’  No cul  No cul  No cul  ‘+’ & ‘‐’  No cul  No cul  No cul  ‘+’ & ‘‐’ ‘+’ & ‘‐’  ‘+’ & ‘‐’ No cul  ‘‐’ No cul  No cul No cul  ‘+’ & ‘‐’  No cul  ‘‐’  ‘+’  ‘+’  No cul  No cul  ‘+’ & ‘‐’  4  Supervised Twice Weekly  ‘+’  ‘+’  No cul  No cul  23  Supervised Twice Weekly Self Administered Self Administered  ‘+’  ‘‐’  ‘‐’  ‘‐’  ‘+’ & ‘‐’  ‘‐’  No cul  No cul  ‘+’  No cul  No cul  No cul  Supervised Twice Weekly Self Administered Self Administered Self  ‘+’ & ‘‐’  ‘‐’  ‘+’ & ‘‐’  ‘‐’  ‘+’ & ‘‐’  ‘‐’  No cul  No cul  ‘+’  ‘‐’  No cul  No cul  ‘+’ & ‘‐’  No cul  No cul  No cul  1  12  : ‘+’ means positive culture; ‘‐‘ means negative culture; No cul means culture was not available; ‘+’ & ‘‐‘ means both positive and negative culture was available (culture was done multiple times)  86  Case #  Rx Duration  Rx completeness  Time of recurrence (since Rx end) in months  Culture results12  Rx Administration  Administered 18  6  Successful Rx  9  19  12  Successful Rx  16  20  4  22  21  8  Incomplete, Drug reaction Successful Rx  22  11  66  23  1  24  10  Incomplete, non‐ compliant Incomplete, non‐ compliant Successful Rx  50  Self Administered Self Administered Self Administered Self Administered Self Administered  ‘+’ & ‘‐’  No cul  No cul  No cul  ‘+’ & ‘‐’  ‘‐’  No cul  No cul  ‘+’  ‘‐’  No cul  No cul  ‘+’  No cul  No cul  No cul  ‘+’  ‘+’ & ‘‐’  ‘+’ & ‘‐’  ‘+’  41  Self Administered  ‘+’ & ‘‐’  No cul  No cul  No cul  47  Supervised Daily  ‘+’ & ‘‐’  ‘‐’  ‘‐’  No cul  87  Table 3.8: Characteristics of specimens from 24 culture positive recurrent cases during their initial and recurrent episode Case #  Specimen type of initial episode13  Drug Sensitivity of initial episode  Specimen type of recurrent episode  Drug Sensitivity of recurrent episode  Interval between two lab0 and lab1 in months  1  Rectal lesion, MTB  Sensitive  Sensitive  15  15  2  Sputum, MTB  Mono‐resistant  Ascitic fluid, MTB Sputum, MTB  Mono  54  48  3  Sputum, MTB  Neck asp, MTB  MDR  46  34  4  48  48  Sensitive  Bone Marrow, MTB Sputum, MTB  Sensitive  5  Abdominal lymph node, MTB Sputum, MTB  Poly‐resistant, later MDR Sensitive  Sensitive  29  29  6  Sputum, MTB  Sensitive  Sputum, MTB  Sensitive  22  18  7  Blood, M bovis  Sensitive  Urine, M bovis  49  51  8  Pleural Fluid, MTB  Sensitive  Sputum, MTB  Mono‐ resistant Sensitive  10  10  9  Sputum, MTB, SIRE‐S  Sensitive  9  Urine, MTB  Sensitive  Mono‐ resistant Sensitive  10  10  Clavicle Aspiration, MTB Sputum, MTB  16  11  11  Sputum, MTB  Sensitive  Sputum, MTB  Sensitive  31  29  12  Sensitive  Sputum, MTB  Sensitive  18  17  13  Bronchial Washings, MTB Neck mass, MTB  Sensitive  Sensitive  25  24  14  Sputum, MTB  Sensitive  MDR  19  18  15  Pleural Tissue, MTB  Sensitive  Sensitive  18  18  16  Sputum, MTB  Sensitive  Synovial fluid, MTB Neck Swab, MTB Back swab (sinus), MTB Sputum, MTB  Sensitive  17  16  17  Sputum, MTB  Sensitive  Sputum, MTB  37  37  18  Sputum, MTB  Sensitive  Sputum, MTB  Mono‐ resistant Sensitive  15  14  19  Sputum, MTB  Sensitive  Sputum, MTB  Sensitive  28  26  20  Sputum, MTB  Poly‐resistant  Washing lul, MTB  Poly‐resistant  27  27  13  Interval between last positive culture of initial episode and lab1 date in months  ‐MTB: Mycobacterium tuberculosis; M bovis: Mycobacterium bovis; MDR: multi‐drug resistance;  88  Case #  Specimen type of initial episode13  Drug Sensitivity of initial episode  Specimen type of recurrent episode  Drug Sensitivity of recurrent episode  Interval between two lab0 and lab1 in months  21  Sensitive  Sputum, MTB  Sensitive  59  59  22  Bronchial washing, MTB Lung tissue, MTB  Mono‐resistant  Sputum, MTB  78  68  23  Sputum, MTB  Sensitive  Sputum, MTB  Mono‐ resistant Sensitive  44  43  24  Sputum, MTB  Sensitive  Sputum, MTB  Sensitive  56  56  Interval between last positive culture of initial episode and lab1 date in months  3.3.4 Reinfection and Relapse Out of 24 patients with paired isolates, 22 showed identical MIRU‐VNTR patterns. Two patients (Case 4 and 11) had distinct MIRU patterns. For Case # 4, the sample that yielded Mycobacterium tuberculosis during the recurrent episode was from the bone marrow. On the other hand, Case # 11 had a sputum sample that yielded Mycobacterium tuberculosis during the recurrent episode. Both patients had a history of successful treatment for the initial episode and had a sensitive organism during both episodes. Case # 4 was treated for 15 months with a supervised regimen whether Case # 11 was treated by with a supervised regimen for six months. Case # 4 did not have any negative cultures following the initial positive culture.  89  Figure 3.2: MIRU‐VNTR results of 24 recurrent TB cases with paired isolated  MIRU LOCUS 40*  MIRU LOCUS 39  MIRU LOCUS 31*  MIRU LOCUS 27`  MIRU LOCUS 26*  MIRU LOCUS 24  MIRU LOCUS 23*  MIRU LOCUS 20`  MIRU LOCUS 16  MIRU LOCUS 10*  MIRU LOCUS 4  100  MIRU LOCUS 2`  80  MIRU-VNTR  60  MIRU-VNTR  40  Categorical  Key  Specimen No.  MIRU pattern  0800737  Y11/A3TS35  225325153321  0800761  X11/A0TS126  225325153323  0800731  Y5/A0TS088  225315153223  0800755  X5/97TS410  225315153223  0800754  X4/97TS176  244323153323  0800739  Y13/A2TS270  233325173324  0800763  X13/A0TS222  233325173324  0800730  Y4/A1TS094  232325153324  0800735  Y9/A0TS302  232325153324  0800743  Y17/A5TS297  232325153324  0800759  X9/99TS382  232325153324  0800767  X17/A3TS009  232325153324  0800733  Y7/A3TS213  232324253322  0800757  X7/99TS006  232324253322  0800740  Y14/A2TS190  223426153324  0800764  X14/A0TS328  223426153324  0800736  Y10/A0TS375  223425152322  0800760  X10/99TS262  223425152322  0800746  Y20/A5TS043  225425173524  0800770  X20/A2TS338  225425173524  0800728  Y2/A0TS281  227425173422  0800752  X2/96TS142  227425173422  0800745  Y19/A5TS007  223325163533  0800769  X19/A2TS202  223325163533  0800734  Y8/A0TS020  222325163533  0800758  X8/99TS086  222325163533  0800749  Y23/A6TS192  225425163533  0800773  X23/A2TS300  225425163533  0800748  Y22/A6TS042  223225173533  0800772  X22/99TS160  223225173533  0800732  Y6/A2TS124  223325173533  0800756  X6/A0TS183  223325173533  0800738  Y12/A1TS415  223325173513  0800762  X12/A0TS221  223325173513  0800742  Y16/A3TS047  2213251(10)3433  0800766  X16/A1TS200  2213251(10)3433  0800744  Y18/A4TS201  225424173633  0800768  X18/A3TS110  225424173633  0800750  Y24/A6TS412  228225113222  0800774  X24/A2TS048  228225113222  0800727  Y1/96TS223  228223113221  0800751  X1/95TS093  228223113221  0800747  Y21/A5TS083  254327223532  0800771  X21/A0TS075  254327223532  0800741  Y15/A2TS288  254326223432  0800765  X15/A1TS040  254326223432  0800729  Y3/A2TS343  254326223333  0800753  X3/97TS387  254326223333  90  On the other hand, Case # 11 had consecutive negative cultures following the positive culture and had negative culture at the time of treatment completion. Both these cases were HIV positive. After investigating the clinical characteristics, both Case # 4 and Case # 11 was considered as confirmed cases of re‐infection. The relative contribution of re‐infection was: overall 2/24=8.5%, among successfully treated patients 2/16=12.5% and among patients with incomplete therapy 0/8=0%. The incidence of re‐infection among culture positive cases was 18 per 100, 000 person years. When restricted to culture positive patients who had a history of successful treatment, the incidence of re‐infection 19 per 100, 000 person years. The incidence of relapse among culture positive cases and culture positive cases with successful treatment were 194 and 150 per 100, 000 person‐years respectively (Table 3.4).  3.4 DISCUSSION The current study demonstrated that in a population‐based cohort of TB patients, the incidence of recurrence was low in a setting with low prevalence of TB. The estimated incidence of recurrence was 397 among culture confirmed TB cases. When restricted to culture positive recurrent cases, the incidence was 230 per 100, 000 person‐years. This incidence was much lower (169/100,000) among culture positive cases who were successfully treated with their initial episode. The incidence of recurrence varies widely across studies, mainly due to differences in background incidence of TB, methodology, prevalence of HIV infection and the quality of the local TB control program. A recent Australian study reported an incidence of 71 per 100, 000 person‐years in a setting where background incidence of TB was 6.5 per 100, 000 populations (33). This Australian study estimated culture positive incidence of recurrence among culture‐confirmed TB cases who completed treatment for their initial active disease. The current study was conducted in a population which shares similar characteristics to the Australian study in terms of epidemiology (such as a low incidence and over‐representation by foreign‐born individuals) and the delivery of care (such as free treatment). Our incidence of recurrence was twice as high compared to the Australian study. However, the incidence observed in the current study is lower than has been reported in other comparable studies. For 91  example, the North American TB trial consortium study 22 and 23 conducted by Jasmer et al. 2004, reported an annual incidence of recurrence rate of 3800 per 100, 000 population which was much higher than our study (18). The incidence of recurrence was 550 per 100,000 patient years in a Brazilian study (34). One potential explanation for the low recurrence rate might be related to the quality of health care delivery through a centralized TB control system, which ensures uniform prescription of standard treatment regimens and appropriate treatment support for complicated cases.  The current study estimated the relative contribution of re‐infection compared to recurrence. Out of 24 culture‐positive recurrent cases whose paired isolates were evaluated for DNA fingerprinting, only two cases showed distinct MIRU patterns. After a thorough investigation of clinical characteristics, both the cases were considered as a confirmed case of re‐infection. The relative contribution of re‐infection was 8.5 % (2/24) among culture positive cases. When limited to culture positive patients with history of successful treatment, the relative contribution of re‐infection was 12.5% (2/16). The results of this study are similar to what was found in the North American study (18). This controlled clinical trial reported three (4%) cases of re‐infection among 75 recurrent patients, whose paired isolates underwent RFLP genotyping (18). The Australian study (33), which was conducted under program conditions like the present study, reported four re‐infection cases (27%) among 15 recurrent patients.  A wide discrepancy was observed in the reported proportion of re‐infection in previous studies. These differences are attributable to variations in local rates of TB, the prevalence of HIV, study methodology and small sample sizes. Studies conducted in low to intermediate incidence settings (TB incidence < 50 per 100, 000 population) reported from zero to 100% recurrence due to re‐infection (6,11,13‐15,18,19,35‐39). Studies that reported either 100% or 0% reinfection rate had small number of cases (11,13,38). For examples, a USA study (13) who had 7 paired isolates did not observe any reinfection case. Several studies conducted in the USA and European countries that had more than ten paired isolates reported the proportion of reinfection ranging from 4 to 33%. For examples, the study conducted in Houston, Texas (39) 92  found 21% (8 out of 38 cases) were due to re‐infection while the New York study (35) observed a 29% re‐infection (5 out of 17 cases) among the recurrent cases.  Previous studies showed that re‐infection had become an increasingly frequent cause of recurrence in high incidence settings. The study conducted in Cape Town, South Africa, which has a remarkably high rate of TB, showed that re‐infection accounted for 77% of the total recurrent cases (40). Another South African study (41) conducted among gold miners demonstrated a lower (36%) contribution of re‐infection to recurrence (9 out of 33 recurrent cases). However, a lower contribution of re‐infection was observed in several studies based in regions with a high prevalence of TB. For instance, a study conducted in Ho Chi Minh City, Vietnam (42) did not observe any re‐infection case among 39 recurrent cases.  The present study estimated incidence of re‐infection, which was 19 per 100,000 person‐years among culture‐confirmed patients with a history of successful treatment for their initial episode. Reinfection rates usually reflect the incidence of TB infections in the community. Our study demonstrated that the incidence of reinfection was approximately three‐fold higher than the local TB incidence. In the current population (in BC), the TB incidence (in 2004) is 7.4 per 100, 000 (43) and the incidence of culture‐confirmed cases was 5.6 per 100, 000 (culture proven cases represent about 80% of total TB cases in BC) (43).  Few studies that investigated reinfection have reported the incidence of re‐infection. The incidence of reinfection in the Australian study was 18.9 cases per 100,000 (33), which was almost identical compared to the current study. Their reported reinfection rate was four‐fold higher compared to local (New South Wales, Australia) TB incidence (Dobler et al. 2009). The North American study (18) also reported incidence of reinfection, which was 150 per 100, 000 person‐years. This rate was eight times higher than that of the present study. The incidence of reinfection in South African study was remarkably high (2200 per 100, 000 person‐years) ‐ four times higher than local TB infection rate (40). Findings of these studies showed that patients with history of active TB are at increased risk of reinfection than that of new TB. 93  This study has several limitations. Recurrent culture‐positive patients who had less than 6 months of inactivity were included in this study. Inclusion of these cases might overestimate the rate of relapse. However, several studies on recurrence did not consider any period of inactivity. This might be a concern to patients who could not complete their therapy of initial episode. Both lab specimens were taken at least 10 months apart. In addition, the minimum interval between last positive culture of initial episode and recurrent sample was 9 months. Given this, we do not think this have distorted relapse estimates substantially. The likelihood of reinfection by the same strain and mixed infection could be other potential limitation of this study. However, given only two cases with non‐identical MIRU patterns, we do not think, reinfection estimates was affected substantially due to these.  This study could not investigate the risk factors for reinfection due to very small number of cases. Both cases who had non‐matching MIRU patterns during recurrent disease were HIV positive. The study of Sonnenberg et al. 2001 demonstrated HIV infection as an independent risk factor for reinfection (41).  3.5 CONCLUSIONS In conclusion, recurrence of TB among culture‐confirmed patients with successful treatment is low in an area with a low prevalence of disease. Re‐infection is also an infrequent cause of recurrence in this setting. Rate of tuberculosis reinfection after successful treatment is almost three‐fold higher than the local TB incidence.  94  3.6 REFERENCES  (1) Lambert ML, Hasker E, Van Deun A, Roberfroid D, Boelaert M, Van der Stuyft P. Recurrence in tuberculosis: relapse or reinfection? The Lancet Infectious Diseases 2003 May;3(5):282‐287. (2) Fine PE, Small PM. Exogenous reinfection in tuberculosis. N.Engl.J.Med. 1999 Oct 14;341(16):1226‐1227. (3) Repique CJ, Li A, Collins FM, Morris SL. DNA immunization in a mouse model of latent tuberculosis: effect of DNA vaccination on reactivation of disease and on reinfection with a secondary challenge. Infection & Immunity 2002 Jul;70(7):3318‐3323. (4) van Rie A, Warren R, Richardson M, Victor TC, Gie RP, Enarson DA, et al. Exogenous reinfection as a cause of recurrent tuberculosis after curative treatment. N.Engl.J.Med. 1999 Oct 14;341(16):1174‐1179. (5) Chiang CY, Riley LW. Exogenous reinfection in tuberculosis. The Lancet Infectious Diseases 2005 Oct;5(10):629‐636. (6) Garcia de Viedma D, Marin M, Hernangomez S, Diaz M, Ruiz Serrano MJ, Alcala L, et al. Tuberculosis recurrences: reinfection plays a role in a population whose clinical/epidemiological characteristics do not favor reinfection. Arch.Intern.Med. 2002 Sep 9;162(16):1873‐1879. (7) Shen G, Xue Z, Shen X, Sun B, Gui X, Shen M, et al. The study recurrent tuberculosis and exogenous reinfection, Shanghai, China. Emerging Infectious Diseases 2006 Nov;12(11):1776‐ 1778. (8) Das S, Chan SL, Allen BW, Mitchison DA, Lowrie DB. Application of DNA fingerprinting with IS986 to sequential mycobacterial isolates obtained from pulmonary tuberculosis patients in Hong Kong before, during and after short‐course chemotherapy. Tubercle & Lung Disease 1993 Feb;74(1):47‐51.  95  (9) Das S, Paramasivan CN, Lowrie DB, Prabhakar R, Narayanan PR. IS6110 restriction fragment length polymorphism typing of clinical isolates of Mycobacterium tuberculosis from patients with pulmonary tuberculosis in Madras, south India. Tubercle & Lung Disease 1995 Dec;76(6):550‐554. (10) Godfrey‐Faussett P, Githui W, Batchelor B, Brindle R, Paul J, Hawken M, et al. Recurrence of HIV‐related tuberculosis in an endemic area may be due to relapse or reinfection. Tubercle & Lung Disease 1994 Jun;75(3):199‐202. (11) el‐Sadr WM, Perlman DC, Matts JP, Nelson ET, Cohn DL, Salomon N, et al. Evaluation of an intensive intermittent‐induction regimen and duration of short‐course treatment for human immunodeficiency virus‐related pulmonary tuberculosis. Terry Beirn Community Programs for Clinical Research on AIDS (CPCRA) and the AIDS Clinical Trials Group (ACTG). Clinical Infectious Diseases 1998 May;26(5):1148‐1158. (12) Sahadevan R, Narayanan S, Paramasivan CN, Prabhakar R, Narayanan PR. Restriction fragment length polymorphism typing of clinical isolates of Mycobacterium tuberculosis from patients with pulmonary tuberculosis in Madras, India, by use of direct‐repeat probe. J.Clin.Microbiol. 1995 Nov;33(11):3037‐3039. (13) Vernon A, Burman W, Benator D, Khan A, Bozeman L. Acquired rifamycin monoresistance in patients with HIV‐related tuberculosis treated with once‐weekly rifapentine and isoniazid. Tuberculosis Trials Consortium. Lancet 1999 May 29;353(9167):1843‐1847. (14) Lourenco MC, Grinsztejn B, Fandinho‐Montes FC, da Silva MG, Saad MH, Fonseca LS. Genotypic patterns of multiple isolates of M. tuberculosis from tuberculous HIV patients. Tropical Medicine & International Health 2000 Jul;5(7):488‐494. (15) Caminero JA, Pena MJ, Campos‐Herrero MI, Rodriguez JC, Afonso O, Martin C, et al. Exogenous reinfection with tuberculosis on a European island with a moderate incidence of disease. American Journal of Respiratory & Critical Care Medicine 2001 Mar;163(3 Pt 1):717‐ 720. 96  (16) Hawken M, Nunn P, Gathua S, Brindle R, Godfrey‐Faussett P, Githui W, et al. Increased recurrence of tuberculosis in HIV‐1‐infected patients in Kenya. Lancet 1993 Aug 7;342(8867):332‐337. (17) Heldal E, Docker H, Caugant DA, Tverdal A. Pulmonary tuberculosis in Norwegian patients. The role of reactivation, re‐infection and primary infection assessed by previous mass screening data and restriction fragment length polymorphism analysis. International Journal of Tuberculosis & Lung Disease 2000 Apr;4(4):300‐307. (18) Jasmer RM, Bozeman L, Schwartzman K, Cave MD, Saukkonen JJ, Metchock B, et al. Recurrent tuberculosis in the United States and Canada: relapse or reinfection?. American Journal of Respiratory & Critical Care Medicine 2004 Dec 15;170(12):1360‐1366. (19) Bandera A, Gori A, Catozzi L, Degli Esposti A, Marchetti G, Molteni C, et al. Molecular epidemiology study of exogenous reinfection in an area with a low incidence of tuberculosis. J.Clin.Microbiol. 2001 Jun;39(6):2213‐2218. (20) Health Canada. Tuberculosis in Canada 1997. 1997; Available at: http://www.hc‐ sc.gc.ca/pphb‐dgspsp/publicat/tbcan97/pdf/tbcan97e.pdf. Accessed March/31, 2010. (21) Long R. Drug‐resistant tuberculosis. CMAJ Canadian Medical Association Journal 2000 Aug 22;163(4):425‐428. (22) Long R, Manfreda J, Mendella L, Wolfe J, Parker S, Hershfield E. Antituberculous drug resistance in Manitoba from 1980 to 1989. CMAJ Canadian Medical Association Journal 1993 May 1;148(9):1489‐1495. (23) Scalcini M, Carre G, Jean‐Baptiste M, Hershfield E, Parker S, Wolfe J, et al. Antituberculous drug resistance in central Haiti. Am.Rev.Respir.Dis. 1990 Sep;142(3):508‐511.  97  (24) van Embden JD, Cave MD, Crawford JT, Dale JW, Eisenach KD, Gicquel B, et al. Strain identification of Mycobacterium tuberculosis by DNA fingerprinting: recommendations for a standardized methodology. J.Clin.Microbiol. 1993 Feb;31(2):406‐409. (25) McNabb SJ, Braden CR, Navin TR. DNA fngerprinting of Mycobacterium tuberculosis: lessons learned and implications for the future. Emerging Infectious Diseases 2002 Nov;8(11):1314‐1319. (26) Braden CR, Crawford JT, Schable BA. Quality assessment of Mycobacterium tuberculosis genotyping in a large laboratory network. Emerging Infectious Diseases 2002 Nov;8(11):1210‐ 1215. (27) Mazars E, Lesjean S, Banuls AL, Gilbert M, Vincent V, Gicquel B, et al. High‐resolution minisatellite‐based typing as a portable approach to global analysis of Mycobacterium tuberculosis molecular epidemiology. Proc.Natl.Acad.Sci.U.S.A. 2001 Feb 13;98(4):1901‐1906. (28) Savine E, Warren RM, van der Spuy GD, Beyers N, van Helden PD, Locht C, et al. Stability of variable‐number tandem repeats of mycobacterial interspersed repetitive units from 12 loci in serial isolates of Mycobacterium tuberculosis. J.Clin.Microbiol. 2002 Dec;40(12):4561‐4566. (29) Supply P, Lesjean S, Savine E, Kremer K, van Soolingen D, Locht C. Automated high‐ throughput genotyping for study of global epidemiology of Mycobacterium tuberculosis based on mycobacterial interspersed repetitive units. J.Clin.Microbiol. 2001 Oct;39(10):3563‐3571. (30) Cowan LS, Mosher L, Diem L, Massey JP, Crawford JT. Variable‐number tandem repeat typing of Mycobacterium tuberculosis isolates with low copy numbers of IS6110 by using mycobacterial interspersed repetitive units. J.Clin.Microbiol. 2002 May;40(5):1592‐1602. (31) Centers for Disease Control and Prevention. Notice to Readers: New CDC Program for Rapid Genotyping of Mycobacterium tuberculosis Isolates . MMWR 2005;54(02):47.  98  (32) Shamputa IC, Van Deun A, Salim MA, Hossain MA, Fissette K, de Rijk P, et al. Endogenous reactivation and true treatment failure as causes of recurrent tuberculosis in a high incidence setting with a low HIV infection. Tropical Medicine & International Health 2007 Jun;12(6):700‐ 708. (33) Dobler CC, Crawford AB, Jelfs PJ, Gilbert GL, Marks GB. Recurrence of tuberculosis in a low‐ incidence setting. European Respiratory Journal 2009 Jan;33(1):160‐167. (34) Ruffino‐Netto A. Recurrence of tuberculosis. Jornal Brasileiro De Pneumologia: Publicacao Oficial Da Sociedade Brasileira De Pneumologia E Tisilogia 2007 Oct;33(5). (35) Small PM, Shafer RW, Hopewell PC, Singh SP, Murphy MJ, Desmond E, et al. Exogenous reinfection with multidrug‐resistant Mycobacterium tuberculosis in patients with advanced HIV infection. N.Engl.J.Med. 1993 Apr 22;328(16):1137‐1144. (36) Chaves F, Dronda F, Alonso‐Sanz M, Noriega AR. Evidence of exogenous reinfection and mixed infection with more than one strain of Mycobacterium tuberculosis among Spanish HIV‐ infected inmates. AIDS 1999 Apr 1;13(5):615‐620. (37) Sudre P, Pfyffer GE, Bodmer T, Prod'hom G, Furrer H, Bassetti S, et al. Molecular epidemiology of tuberculosis among HIV‐infected persons in Switzerland: a countrywide 9‐year cohort study. Swiss HIV Cohort Study. Infection 1999 Nov‐Dec;27(6):323‐330. (38) de Boer AS, van Soolingen D. Recurrent tuberculosis due to exogenous reinfection. N.Engl.J.Med. 2000 Apr 6;342(14):1050‐1051. (39) El Sahly HM, Wright JA, Soini H, Bui TT, Williams‐Bouyer N, Escalante P, et al. Recurrent tuberculosis in Houston, Texas: a population‐based study. International Journal of Tuberculosis & Lung Disease 2004 Mar;8(3):333‐340.  99  (40) Verver S, Warren RM, Beyers N, Richardson M, van der Spuy GD, Borgdorff MW, et al. Rate of reinfection tuberculosis after successful treatment is higher than rate of new tuberculosis. American Journal of Respiratory & Critical Care Medicine 2005 Jun 15;171(12):1430‐1435. (41) Sonnenberg P, Murray J, Glynn JR, Shearer S, Kambashi B, Godfrey‐Faussett P. HIV‐1 and recurrence, relapse, and reinfection of tuberculosis after cure: a cohort study in South African mineworkers. Lancet 2001 Nov 17;358(9294):1687‐1693. (42) Lan NT, Lien HT, Tung le B, Borgdorff MW, Kremer K, van Soolingen D. Mycobacterium tuberculosis Beijing genotype and risk for treatment failure and relapse, Vietnam. Emerging Infectious Diseases 2003 Dec;9(12):1633‐1635. (43) BC Center for Disease Control. TB Annual Report 2004. 2004; Available at: http://www.bccdc.ca/NR/rdonlyres/F80BA3E8‐4E46‐4AFC‐9B05‐ CE3643C2947E/0/2004_TB_Annual.pdf. Accessed March/31, 2010.  100  CHAPTER 4: A SYSTEMATIC REVIEW ON RISK FACTORS OF MORTALITY AMONG TB PATIENTS14  14  ‐ “A version of this chapter will be submitted for publication. Moniruzzaman, A. Kazanjian, A. Wong, H. Elwood, R.K. Chowdhury, M.M. and FitzGerald, J. M. (2010) A systematic review on TB‐related mortality.”  101  4.1 INTRODUCTION Worldwide, TB remains one of the leading infectious diseases in terms of morbidity and mortality (1,2). There are approximately 9.4 million new active cases of TB diagnosed annually and 1.8 million deaths associated with TB (including 0.5 million deaths among HIV‐confected cases) in 2008 (3). The burden of illness from TB is so enormous that the World Health Organization (WHO) declared it a global emergency in 1993 (1). Historically improvements in social conditions were associated with a decline in TB‐related mortality and this decline was accentuated with the introduction of chemotherapy. This decline has been more prominent in industrialized nations. The global TB mortality rate (excluding HIV) was 30 (per 100,000) in 1990, which has fallen to 21 in 2008 (4). One of the United Nations (UN) Millennium Development Goals was to reduce TB‐related mortality by 50% (compared to rate of 1990) by 2015 (2). According to current decline rate, the likelihood of achieving of such a goal is remote.  Moreover, during the last two decades, there has been a resurgence of TB in many countries, partly due to the HIV epidemic, the deterioration of public health care infrastructure and the emergence of drug resistant disease. As a result, TB still remains a major global cause of death (second commonest infectious cause of deaths after HIV), especially in countries with a high burden of disease (such as sub‐Saran Africa and several Asian countries). In the modern era, all TB‐related deaths are potentially preventable. This raises an important question, why are we seeing so many TB‐related deaths despite the availability of effective chemotherapy? If appropriate interventions are not effectively implemented, many patients with this treatable and preventable disease will continue to have avoidable morbidity and mortality.  Mortality among TB patents varies widely, due in part to differences in background burden of disease, geographical locations, methodology used in studies evaluating this issue and the quality of the health care delivery system. In order to prevent deaths among TB patients, identification of factors associated with mortality needs to be determined. Several factors such as older age, male gender, incomplete TB treatment, drug resistant status, malnutrition, missed diagnosis, alcoholism, HIV co‐infection, and extent of disease (e.g., advanced stage, cavitations) 102  have been found to be associated with mortality among TB patients (5‐14). However, the risk factors for mortality, based on published literature, especially in the era of HIV are still not well documented in a systematic way despite a substantial literature on TB mortality. To our knowledge, there is no systematic review of risk factors for mortality related to TB. Thus the assessment of the risk factors for mortality with a systematic review of the published literature would provide a robust estimate of TB‐related mortality and also improve our understanding of TB related mortality. The objective of this study is to make a quantitative assessment of mortality among TB patients and to identify potential risk factors for TB‐related mortality through a systematic review of the published literature.  4.2 METHODS We systematically searched the Medline and Embase databases for relevant studies using key words “Tuberculosis”, “Mortality” and “Risk factors”. We limited our search to the English literature and studies publishing until December 2007. Bibliographies of selected articles were also reviewed to identify the relevant studies on mortality among TB patients. We initially reviewed the titles and abstracts of the citations that were identified through the search strategy. Review of the abstracts was done independently by two reviewers. Differences between two observers related to data abstraction were resolved though discussions. The final selection of studies was based on a priori criteria of the patient population, study design and focus of study. The following inclusion criteria were used: studies reported mortality among TB patients and investigated the factors associated with mortality. Cases series and ecological studies that used the general population as the comparison group were excluded. We did not include mortality studies that were conducted on specific groups of TB patients such as TB patients with meningitis, TB patients with HIV and TB patients with end stage renal disease.  103  4.2.1 Data Extraction and Calculation of Death Rate/Proportion We developed a standard abstraction sheet to abstract information from the selected articles. We calculated the overall death rate among TB patients from the total number of deaths (numerator) and total number of TB cases (denominator) as reported by the selected studies. The follow‐up period varied widely across studies. The majority of studies reported death that occurred during the treatment phase. However, several studies reported deaths that occurred among TB patients during hospitalization and a number of studies also provided death rates that included a period of longer follow‐up. Therefore, we estimated death rates during hospitalization as well as during treatment where available. We classified death during hospitalization: death from any cause during hospitalization from any cause or within 1 month of treatment initiation. We defined death during treatment: death from any cause during the treatment period or between 6 months and 1 year of treatment initiation. If available, we extracted data on deaths during treatment or within 1 year from studies that followed patients for a longer period of time.  A specific cause of death was not reported in the majority of studies. A number of studies also reported whether death was related to TB in addition to overall mortality. We defined TB‐ caused death if TB was the underlying cause of death. If available, we distinguished TB‐caused death from all‐cause mortality and estimated overall TB‐caused death (% TB subjects dying from TB) using total number of TB‐caused death as numerator and total number of TB cases as denominator. In addition, the proportions of death due to TB out of all reported deaths for each study (whenever available) were calculated.  We also estimated the death rate according to HIV status (among HIV positive patients and HIV negative patients) when available and according to the background burden of disease. We categorized all the selected studies into three groups based on the location of the studies–low, intermediate and high burden countries. We defined low burden settings where the TB incidence was 20 or less (per 100, 000), intermediate burden from 21 to 100 (per 100, 000) and  104  high burden over 100 (per 100, 000) respectively. In order to classify into burden of disease, we used the local TB incidence if it was reported in the study. In studies that did not provide a local TB incidence rate, we used national TB incidence as reported by the WHO for the first year of the study period. The incidence of TB for all countries has been available on the WHO website since 1990. We used TB incidence for the year of 1990 (from the WHO) for studies without reported local TB incidence that were conducted before 1990. We used five‐point summary statistics (Minimum, 1st Quartile, Median/2nd Quartile, 3rd Quartile and Maximum) to characterize the death rates extracted from the studies.  Information on risk factors for mortality varied widely across studies. We considered a risk factor if it was reported in a study and the corresponding 95% CI of the estimate of risk factor did not include one. Several studies did not report a measure of association for a risk factor, but conducted a hypothesis test to evaluate the effect of factors on mortality among TB patients. For these studies, we used a cut off p value of 0.05 to be considered as a significant risk factor. In the case of discrepancies with regard to significant risk factors between the univariate and multivariate model, we considered risk factors to be significant if it was reported in a multivariate model. In studies that did not provide an adjusted estimate, and the risk factor was reported in a univariate model, as being significant we accepted this.  4.3 RESULTS The search of electronic databases yielded 564 citations for initial screening (Medline databases yielded 276 citations and Embase yielded another 288 citations). After the review of abstract and titles, 77 articles were selected for full review. Out of these studies, only 39 studies were finally selected for this systematic review. Bibliographic search of these 77 articles generated another 22 articles that met our inclusion criteria. Review of these articles identified another 5 articles that met our inclusion criteria for abstraction. Thus, in total, 66 articles were included in this systematic review.  105  Table 4.1 presents overall death (proportion of TB patients dying from any cause) among TB patients. Out of 66 studies, the total number of deaths and total number of TB cases was available in 60 studies. Five of six studies (15‐20) that did not provide proportion of TB subjects dying from TB (total number of deaths was reported but no information on total number of TB cases) were case‐control studies. The remaining study reported only cause‐specific deaths (death due to TB), but no information was available on total mortality (all‐cause). In total these studies reported a total of 12659 deaths (all‐cause) among 81791 TB patients that yielded a global crude death of 15.5%. The lowest mortality was 2.2% (21) and the highest was 65.2% (22). The median death (% of total TB patients dying from any cause) was 12.8% (IQR: 7.1%, 22.2%).  When we limited mortality (all‐cause) that occurred during treatment, 50 studies had appropriate data, in these studies the global death was slightly lower (14.5%,). The median death (% of total TB patients dying) was 10.3% (IQR: 5.9%, 20.5%). Overall, the lowest reported mortality (proportion of deaths of total TB subjects) during treatment was 2.2% and the highest mortality was 39.7% (23). Seven studies reported death (all‐cause) among TB patients during hospitalization, which yielded a global death of 7.8% (median: 8.4%, IQR: 7.4%, 12.1%) – much lower than the mortality that occurred during treatment. The lowest mortality (% of total TB patients dying) during hospitalization was 4.9% (24) and the highest was 16.1% (25).  106  Table 4.1: Mortality among TB patients from selected studies (n=66)  Number Mortality % (Total of studies deaths/Total TB cases)  Death Rate (Minimum 1Q, Median, 3Q, Maximum)  CDR 15(overall)  60  12659/81791=15.5%  2.2%, 7.1%, 12.8%, 22.2%, 65.2%  CDR (during treatment)  50  10743/73994=14.5%  2.2%, 5.9%, 10.3%, 20.5%, 39.7%  CDR (during hospitalization)  7  388/4955=7.8%  4.9%, 7.4%, 8.4%, 12.1%, 16.1%  CDR (TB‐caused death)  13  567/37240=1.5%  0.1%, 1.0%, 3.2%, 6.1%, 15.6%  TB‐caused death (% of total deaths)  15  534/1538=34.7%  1.4%, 25%, 35.3%, 42.0%, 75.3%  CDR in HIV positive patients  35  2231/7365=30.3%  0%, 13.7%, 28.8%, 38.9%, 70.1%  CDR in HIV negative patients  30  771/11760=6.6%  0.4%, 3.3%, 8.1%, 14.8%, 48.6%  15  CDR (Crude Death Rate) was estimated in percentage (total deaths/total TB cases)  107  Table 4.2: Mortality according to the burden of TB disease (low, intermediate and high)16  Low burden (16) Intermediate burden High burden (41) (9) CDR (overall)  CDR during Rx  CDR (TB‐ caused death)  TB‐caused death (% of total deaths)  HIV positive patients  Number of studies  15  6  39  Point estimate  1406/14988=9.4%  725/9262=7.8%  10528/57541=18.3%  Death rate (Min, 1Q, Median, 3Q, Max) Number of studies  4.9%, 8.7%, 11.7%, 14.6%, 44.1%  3.6%, 5.4%, 10.0%, 14.6%, 17.8%  2.2%, 7.0%, 14.7%, 24.3%, 65.2%  10  6  34  Point estimate  923/10612=8.7%  563/9154=6.2%  9257/54678=16.9%  Death rate (Min, 1Q, Median, 3Q, Max) Number of studies  6.8%, 8.1%, 9.1% , 18.0%, 36.1%  3.0%, 3.5%, 6.6%, 10.7%, 13.6%  2.2%, 5.7%, 11.7%, 23.1%, 39.7%  6  3  4  Point estimate  178/8257=2.2%  298/25021=1.2%  91/3962=2.8%  Death rate (Min, 1Q, Median, 3Q, Max) Number of studies  0.1%, 0.7%, 3.2% , 7.9%, 15.6%  0.6%, 6.8%, 7.5%  1.1%, 1.9%, 4.3%, 4.6%  7  2  6  Point estimate  214/874=24.5%  165/255=64.7%  155/409=37.9%  Death rate (Min, 1Q, Median, 3Q, Max) Number of studies  1.4%, 13.6%, 28.9%, 41.7%, 43.2%  42.0%, 75.3%  25.0%, 29.4%, 33.6%, 48.7%, 73.0%  7  4  24  Point estimate  128/553=23.2%  132/352=37.5%  1971/6460=30.5%  Death rate (Min, 1Q, Median, 3Q, Max)  6.0%, 8.8%, 21.0%, 37.5%, 43.9%  0.0%, 31.6%, 34.8%, 38.9%  6.1%, 13.8%, 31.2%, 28.8% , 41.9%, 70.1%  16  ‐ Original mortality rates were reported in categories that included five or less number studies. No five point summary was estimated for these scenarios.  108  Low burden (16) Intermediate burden High burden (41) (9) HIV negative patients  Number of studies  3  3  24  Point estimate  69/1142=6.0%  136/1180=11.5%  566/9438=6.0%  Death rate (Min, 1Q, Median, 3Q, Max)  5.6%, 8.9%, 11.5%  6.8%, 8.8%, 18.3%  0.4%, 2.2%, 6.9%, 15.2%, 48.6%  Approximately 20 percent of studies (13 studies) provided data regarding death due to TB and total number of TB cases. These studies reported a total of 567 TB‐related deaths among 37,240 TB patients, which yielded a global TB‐related death proportion (% of patients dying from TB), of 1.5% (median: 3.2%, IQR: 1%, 6.1%). The lowest TB‐related death (% of TB subjects dying from TB) was 0.1% (11) the highest was 15.6% (26). The proportion of deaths related to TB of all deaths was 34.7% (median: 35.3%, IQR: 25%, 42%) as reported in 20 studies.  Thirty five studies included data on subjects co‐infected with HIV. In total, these studies consisted of a total of 2231 deaths among 7365 HIV‐TB co‐infected patients. The global mortality among HIV positive TB patients was 30.3% (median: 28.8%, IQR: 13.7%, 38.9%). The lowest mortality among HIV co‐infected patients was 0/7= 0% (10) and the highest was 429/612=70.1% (22). Four out of 35 studies (22,27‐29) reported a mortality of over 50%, among HIV co‐infected TB patients (56.9%, 58.3%, 59.2% and 70.1% respectively). In contrast, five studies (9,10,30‐32) had mortality lower than 10% among HIV positive patients. As expected, the global mortality among TB patients who were HIV negative was much lower‐ 6.6% (median: 8.1%, IQR: 3.3%, 14.85). The highest mortality among HIV negative TB patients was 48.6% (22). Two (31,33) of 30 studies reported mortality among HIV negative patients lower than 1% (1/280=0.4% and 1/185=0.5% respectively). Based on background burden of diseases, most of the studies (41 studies, 62%) selected in this review were conducted in a high‐burden setting. Sixteen (24%) studies were conducted in a low 109  burden setting and 9 (14%) studies in an intermediate setting. Table 4.2 and Figure 4.1 present mortality data in these three settings (low, intermediate and high burden countries). The overall global mortality was almost two‐times higher (median: 8.7% vs. 14.7%) in a high burden setting compared to low burden countries. However, countries with intermediate burden of diseases had a relatively lower mortality (Median: 10.0%). The similar patterns were observed for mortality during treatment (Median: 9.1%, 6.6% and 11.7% respectively in low, intermediate and high burden countries). In low‐burden settings, the highest overall mortality and mortality during treatment in TB patients was 44.1% (34) and 36.1% (26) respectively. In the high‐burden category, five studies (7,21,31,35,36) reported overall mortality (% of total TB patients dying) lower than 5%.  110  Figure 4.1: Mortality among TB patients according to background incidence of disease  111  Figure 4.2: Mortality among HIV positive and HIV negative TB patients by background incidence of disease  112  As expected, TB‐related deaths (proportion of TB subjects dying specifically from TB) were highest in high‐burden setting (2.2%, 1.2% and 2.8% respectively in low, intermediate and high burden settings). Countries with an intermediate burden of disease had the lowest TB‐related death, which was strongly influenced by one study from Turkey (16) which included the largest number of TB cases (133/22,651=0.6%). The proportion of TB‐related deaths out of all reported deaths was the lowest in low‐burden setting (24.5%). However, this proportion was the highest in countries with an intermediate TB incidence (64.7%).  Figure 4.2 presents comparisons of death between HIV‐positive and HIV‐negative individuals in low, intermediate and high burden settings. Among HIV co‐infected TB patients, the deaths (% TB patients dying) was (23.2%, median: 21%) the lowest in low‐burden settings. However, the proportion of subjects dying from TB in intermediate burden settings (37.5%) was higher than that was found in high‐burden settings (30.5%). The median death (% of TB patients dying from TB) among HIV‐confected persons in high‐burden setting was 31.2% (IQR: 13.8%, 41.9%). A similar pattern was found among HIV negative TB patients (6.0%, 11.5% and 6.0% in low, intermediate and high burden settings respectively). In a high‐burden setting, the median death among HIV‐negative persons was 6.9% (IQR: 2.2%, 15.2%).  4.3.1 Risk Factors for Mortality among TB Patients Table 4.3 presents studies that reported socio‐demographic characteristics as risk factors for mortality among TB patients. Among them, older age or increasing age was widely cited as a significant risk factor for mortality. The relationship between gender and mortality was not consistent across studies. There was a significantly elevated risk of mortality among males in two studies (6,35) while female gender was significantly associated with mortality in two other studies (18,25). Unemployment has been demonstrated as a significant risk factor in three studies (11,14,37). Homelessness (14) and years of education such as no schooling (35) or schooling for less than 8 years (38) were reported, in some studies, as being associated with an 113  increased risk of death. Local‐born people including those of French (37) and Swiss‐origin (39) were shown to have an increased risk of death in some studies. In one study, mortality was significantly higher among Caucasians (40) while in another study non‐white people were at greater risk of dying (41). However, a number of other studies documented no increased risk of mortality among several racial groups including: African American (34), Hispanic (34), White (34), Asian (34) and Black Caribbean (40). Increased mortality among Indigenous population was found in one study (41). In some studies living in a rural setting (17), distance from hospital (5) and diagnosis in hospital (38) were found to be associated with an increased risk of mortality.  Table 4.4 presents a list of co‐morbid conditions and behavioral factors associated with mortality among TB patients as reported in the studies included in this review. Among these co‐ morbidities, HIV infection was consistently documented as a strong predictor of risk in many studies. Several HIV specific factors including myco‐bacteraemia (23), anergy to purified protein derivative (PPD) (42) and signs of impaired immune response (Kaposi sarcoma, oral candidiasis) were found to be associated with an increased mortality risk in a few studies.  Malnutrition (9,43,44) and related factors such as low BMI (23,42), low weight (18,45) and weight loss (46) were shown to increase the risk of death in several studies. Other concomitant illnesses such as malignancy (6,11,14) renal disease (43,44) and liver disease (44,46) were less consistently reported as being associated with an increased mortality risk. A further reported co‐morbid condition that increased the risk of mortality was silicosis (7), respiratory failure (43), chronic lung disease (14) and congestive heart failure (14). Several studies did not evaluate specific co‐morbid conditions, but rather reported them as a whole such as coexisting pathology (25), other diseases (15,19) and co‐morbidity index (24). Among behavioral factors, alcoholism (6,10‐12,19) was consistently reported as a significant risk factor while drug abuse (6) and injection drug use (47) were infrequently reported as being associated with an increased mortality risk.  114  Table 4.3: Socio‐demographic risk factors for dying in patients with Tuberculosis (TB)  Name of factors  Increasing age/older age  Decreasing age /young age Male gender  Study ID/Reference Unadjusted/Adjusted Estimate  Only unadjusted estimate  Adjusted Estimate  Barker et al.2002, Borgdorff et al.1998, Bustamante‐ Montes et al.2000, Connolly et al.1998, Cullinan et al.1991, Garin et al.1997, Greenway et al.2002, Hansel et al.2004, Kangombe et al.2000 , Kartaloglu et al.2003, Kolappan et al.2006, Lawn et al.1999, Lee et al. 2005,Mathew et al.2006, Rabaud et al.1997, Rao et al.1998, Sterling et al.2006, Zachariah et al.2002, Vree et at.2007 (2), Glynn et al.1998, Harries et al. 1998, Harries et al.1999, Helbling et al.2002, Kangombe et al.2004, Pablos‐Mendez et al.1996, Quy et al.2006, Van den Broek et al.1998, Walpola et al.2003, Khan et al.2006 Kassim et al.1995  Cullinan et al.1991, Kartaloglu et al.2003, Lawn et al.1999, Lee et al. 2005, Mathew et al.2006, Vree et at.2007 (2), Helbling et al.2002, Kassim et al.1995, Walpola et al.2003  Barker et al.2002, Borgdorff et al.1998, Bustamante‐Montes et al.2000, Connolly et al.1998, Garin et al.1997, Greenway et al.2002, Hansel et al.2004, Kangombe et al.2000, Kolappan et al.2006, Rabaud et al.1997, Rao et al.1998, Sterling et al.2006, Zachariah et al.2002, Glynn et al.1998, Harries et al. 1998, Kangombe et al.2004, Pablos‐Mendez et al.1996, Quy et al.2006, Van den Broek et al.1998, Khan et al.2006  Indigenous Population  Borgdorff et al.1998, Quy et al.2006 Sacks et al. 1998, Abos‐ Hernandez et al.2002, Dewan et al.2004, Rabaud et al. 1997, Sterling et al. 2006 Dewan et al.2004, Khan et al. 200617 Xie et al. 1992  White/Caucasian  Bakshi et al. 1998  Non‐white (Asian)  Xie et al. 1992  Female gender Unemployment  Homelessness  17  Kassim et al.1995  Sacks et al.1998  Borgdorff et al. 1998, Cr35, Quy et al. 2006 Abos‐Hernandez et al.2002 Dewan et al.2004, Rabaud et al. 1997, Sterling et al. 2006 Dewan et al.2004, Khan et al. 2006 Xie et al. 1992  Bakshi et al. 1998 Xie et al. 1992  ‐Reported P value was 0.07  115  Name of factors  Study ID/Reference Unadjusted/Adjusted Estimate  Only unadjusted estimate  Adjusted Estimate  Local –born (French‐origin/US‐ born/Canadian‐born, etc.)  Rabaud et al. 1997, Helbling et al.2002  Helbling et al.2002  Rabaud et al. 1997  Illegal immigrant  Borgdorff et al. 1998  Borgdorff et al. 1998  Schooling (<8 years, no schooling) Hospital diagnosis  Aerts et al.2004, Quy et al. 2006 Aerts et al.2004  Aerts et al.2004, Quy et al. 2006 Aerts et al.2004  Distance from hospital Living in rural setting  Barker et al.2002 Lawn et al. 1999  Lawn et al. 1999  District hospitals  Harries et al. 2001  Harries et al. 2001  Barker et al.2002  Table 4.5 Summarizes risk factors related to treatment and the particular health care system that was in place. Mortality risk was elevated among patients whose treatment was delayed (13,48,49) or where there was a delay in diagnosis (13). History of none (34) or incomplete treatment (12,28,34) and treatment default (12,17,46) or failed treatment (32) have been documented as being associated with increased mortality in several studies. In one study the use of DOT, was found to decrease the risk of dying (30), while non‐compliance increased the mortality risk in another study (15). Patients who were being treated for their TB (10,17,42,45) and patients with a prior history of TB (19,41) were more likely to die. Less frequently other significant risk factors reported were consultation with a traditional healer (48), lack of physician experience with TB (32) diagnosis by internal medicine specialists/surgeon/other specialists (6) and delayed ICU admission or admission to a non‐medical ward (13).  116  Table 4.4: Co‐morbidities associated with mortality in patients with Tuberculosis (TB)  Name of factors  HIV/AIDS  Malignancy  Malnutrition  Prognostic Nutritional Index Weight loss/Low weight  BMI (Low) Mid Upper Arm Circumference (<200 mm) Liver cirrhosis /liver failure Renal failure/diseases Respiratory failure  Study ID/Reference Unadjusted/Adjusted Estimate  Only unadjusted estimate  Adjusted Estimate  Aerts et al.2004, Borgdorff et al. 1998, Churchyard et al.2000, Connolly et al. 1998, Elliott et al. 1995, Garcia‐Garcia et al.2002,Gustafson et al.2007, Kangombe et al.2000, Lawn et al. 1999, Malkin et al. 1997, Murray et al. 1999, Norrgren H et al. 2001 Nunn et al. 1992, Rabaud et al. 1997, Richter et al. 1995, Soeters et al. 2005, Sterling et al. 2006, Zachariah et al.2002, Ackah et al. 1995 Banergee et al. 1997, Fischl et al. 1992, Glynn et al. 1998, Hargreaves et al. 2001, Harries et al. 1998, Kelly et al. 1998, Pablos‐Mendez et al. 1996, Perriens et al. 1991, Quy et al. 2006, Van den Broek et al. 1998 Borgdorff et al. 1998, Dewan et al. 2004, Sterling et al. 2006 Rao et al. 1998, Zachariah et al.2002, Walpola et al.2003  Lawn et al. 1999, Malkin et al. 1997, Murray et al. 1999, Richter et al. 1995, Soeters et al. 2005, Ackah et al. 1995, Banergee et al. 1997, Fischl et al. 1992, Perriens et al. 1991  Aerts et al.2004, Borgdorff et al. 1998, Churchyard et al.2000, Connolly et al. 1998, Elliott et al. 1995, Garcia‐Garcia et al.2002,Gustafson et al.2007, Kangombe et al.2000, Norrgren 1995 et al. 2001, Nunn et al. 1992, Rabaud et al. 1997, Sterling et al. 2006, Zachariah et al.2002, Abos‐ Hernandez et al.2002, Hargreaves et al. 2001, Harries et al. 1998, Kelly et al. 1999, Pablos‐ Mendez et al. 1996, Quy et al. 2006, Van den Broek et al. 1998  Kartaloglu et al. 2003 Garcia‐Garcia et al. 2002, Sacks et al. 1998, Santha et al. 2000 Gustafson et al. 2007, Hargreaves et al. 2001 Gustafson et al. 2007  Kartaloglu et al. 2003  Garcia‐Garcia et al.2002, Walpola et al.2003 Rao et al.1998, Walpola et al.2003 Rao et al. 1998  Borgdorff et al. 1998, Dewan et al. 2004, Sterling et al. 2006 Rao et al. 1998, Zachariah et al. 2002, Walpola et al. 2003 Garcia‐Garcia et al. 2002, Sacks et al. 1998, Santha et al. 2000 Gustafson et al. 2007 Gustafson et al. 2007 Garcia‐Garcia et al.2002, Walpola et al.2003 Rao et al.1998, Walpola et al.2003 Rao et al. 1998  117  Name of factors  Study ID/Reference Unadjusted/Adjusted Estimate  Chronic lung disease Congestive heart failure DCI Co‐morbidity Index Other diseases /risk factors  Only unadjusted estimate  Adjusted Estimate  Co‐existing pathology  Dewan et al.2004 Dewan et al.2004 Hansel et al.2004 Bustamante‐Montes et al.2000, Zafran et al. 1994 Abos‐Hernandez et al.2002  Immune‐ suppression  Rao et al. 1998  Dewan et al.2004 Dewan et al.2004 Hansel et al.2004 Bustamante‐Montes et al.2000 Abos‐Hernandez et al.2002 Rao et al. 1998  Signs of impaired immune defense (Kaposi’s Sarcoma, oral candidiasis, leukoplakia) Mycobacteraemia  Gustafson et al.2007, Hargreaves et al. 2001  Gustafson et al.2007, Hargreaves et al. 2001  Hargreaves et al. 2001  Hargreaves et al. 2001  Anergy to tuberculin  Gustafson et al.2007  Gustafson et al.2007  Silicosis  Churchyard et al.2000  Churchyard et al.2000  Drug addiction  Borgdorff et al. 1998  Borgdorff et al. 1998  Alcoholism  Injection drug abuse  Borgdorff et al. 1998, Mathew et al. 2006, Sterling et al. 2006, Zafran et al. 1994 Kourbatova et al. 2006 (1)  Smokers and alcoholics  Kolappan et al. 2006  Zafran et al. 1994  Mathew et al. 2006, Zafran et al. 1994  Borgdorff et al. 1998, Sterling et al. 2006  Kourbatova et al. 2006 (1) Kolappan et al. 2006  118  Table 4.5: Health‐care delivery and treatment‐ related risk factors of mortality  Name of factors  Directly Observed Therapy (DOT)18 Non‐compliance  Study ID/Reference Unadjusted/Adjusted Estimate  Only unadjusted estimate  Adjusted Estimate  Jasmer et al. 200419, Khan et al. 2006 Bustamante‐Montes et al. 2000  Jasmer et al. 2004  Khan et al. 2006 Bustamante‐ Montes et al. 2000 Glynn et al. 1998  BCG status (negative)  Glynn et al. 1998  Co‐trimoaxazole prophylaxis20  Grimwade et al. 2005  ICU admission/Admitted in non‐ medical ward Hospital with low TB admission rates  Greenway et al. 2002  Admission to emergency department (vs. routine hospital admission) Diagnosis by a specialist  Hansel et al. 2004  Hansel et al. 2004  Borgdorff et al. 1998  Physician experience with TB Delay in starting treatment  Khan et al. 200621 Greenway et al. 2002, 80, 81  Initially missed diagnosis  Greenway et al. 2002  Incomplete Rx  Garin et al. 1997, Kolappan et al. 2006, Pablos‐Mendez et al. 199622  No treatment (vs. Appropriate Rx)  Pablos‐Mendez et al. 1996  Rx defaulted  Garcia‐Garcia et al.200223, Kolappan et al. 2006, Lawn et al. 1999  Borgdorff et al. 1998 Khan et al. 2006 Greenway et al. 2002 Greenway et al. 2002 Kolappan et al. 2006, Pablos‐ Mendez et al. 1996 Pablos‐Mendez et al. 1996 Garcia‐Garcia et al. 2002, Kolappan et al. 2006  Grimwade et al. 2005 Greenway et al. 2002 Greenway et al. 2002  Greenway et al. 2002  80, 81  Garin et al. 1997  Lawn et al. 1999  18  ‐DOT was associated with increased survival ‐The effect of DOT was evaluated in patients who died due to TB and who died due to other causes 20 ‐Associated with increased survival 21 ‐Associated with decreased mortality 22 ‐This study evaluated the effect of incomplete treatment among post‐treatment survivors 23 ‐Associated with increased survival in patients who died due to TB 19  119  Name of factors  Study ID/Reference Unadjusted/Adjusted Estimate  Only unadjusted estimate  Rx failed  Kolappan et al. 2006  Consultation with traditional healer  Barker et al. 2006  Barker et al. 2006  Previous TB/Retreatment  Lawn et al. 1999, Mathew et al. 2006, Zafran et al. 1994  <3 effective anti‐TB drug  Gustafson et al.2007, Lawn et al. 1999, Mathew et al. 2006, Santha et al.2000, Xie et al. 1992, Zafran et al. 1994 Fischl et al. 1992  Length of hospital stay  Kartaloglu et al. 2003  Kartaloglu et al. 2003  Refused health service attention  Bustamante‐Montes et al. 2000  Desire to receive health service24 (↓ Mortality)  Bustamante‐Montes et al. 2000  24  Adjusted Estimate Kolappan et al. 2006 Gustafson et al.2007, Santha et al.2000, Xie et al. 1992  Fischl et al. 1992  Bustamante‐ Montes et al. 2000 Bustamante‐ Montes et al. 2000  ‐Associated with decreased mortality  120  Table 4.6: Risk factors associated with diagnosis and the type of disease  Name of Factors  Study ID/Reference Unadjusted/Adjusted Estimate  Only unadjusted estimate  Meningeal TB  Kourbatova et al. 2006 (2), Xie et al. 1992 Kourbatova et al. 2006 (2)  Kourbatova et al. 2006 (2), Xie et al. 1992 Kourbatova et al. 2006 (2)  Advanced PTB  Xie et al. 1992  Far advanced PTB on CXR  Kartaloglu et al. 2003  Kartaloglu et al. 2003  Pulmonary TB (PTB)  Rabaud et al. 1997, Bakshi et al. 1998 Borgdorff et al. 199825, Kangombe et al. 2000, Harries et al. 1998, Harries et al. 2001, Kangombe et al. 2004, Walpola et al. 2003  Bakshi et al. 1998  Miliary TB  Extra‐pulmonary TB (EPTB)  Xie et al. 1992  Harries et al. 2001  Both Pulmonary and Extra‐ pulmonary  Aerts et al.2004, Borgdorff et al. 1998  Smear +ve PTB  Xie et al. 1992  Smear ‐ve TB  Connolly et al. 1998, Gustafson et al. 2007, Kangombe et al. 2000, Lawn et al. 1999, Harries et al. 1998, Harries et al. 2001, Harries et al. 1999, Kangombe et al. 2004  Lawn et al. 1999, Harries et al. 2001, Harries et al. 1999  Extensive disease/infiltrate on CXR  Cullinan et al. 199126, Lawn et al. 1999, Sacks et al. 1998, Zafran et al. 1994  Bi‐lateral disease on CXR  Dewan et al.2004, Kourbatova et al. 2006 (1) Abos‐Hernandez et al. 2002  Cullinan et al. 1991, Lawn et al. 1999, Sacks et al. 1998, Zafran et al. 1994 Kourbatova et al. 2006 (1)  Number of lobes involved on CXR Cavitary lesions on CXR  25 26  Kourbatova et al. 2006 (1)  Adjusted Estimate  Rabaud et al. 1997 Borgdorff et al. 1998, Kangombe et al. 2000, Harries et al. 1998, Kangombe et al. 2004, Walpola et al. 2003 Aerts et al.2004, Borgdorff et al. 1998 Xie et al. 1992 Connolly et al. 1998, Gustafson et al. 2007, Kangombe et al. 2000, Harries et al. 1998, Kangombe et al. 2004  Dewan et al. 2004 Abos‐Hernandez et al. 2002  Kourbatova et al. 2006 (1)  ‐EPTB was associated with decreased mortality ‐Severity measured by number of zones (2‐3 zones)  121  Name of Factors  Study ID/Reference Unadjusted/Adjusted Estimate  Only unadjusted estimate  Adjusted Estimate  Multidrug resistant (MDR‐ TB)  Garcia‐Garcia et al. 200227, Mathew et al. 2006, Fischl et al. 1992, Pablos‐Mendez et al. 1996, Quy et al. 2006  Mathew et al. 2006, Fischl et al. 1992  Any resistance  Matos et al. 2007  Poly‐resistance  Quy et al. 2006  Garcia‐Garcia et al.2002, Pablos‐ Mendez et al. 1996, Quy et al. 2006 Matos et al. 2007 Quy et al. 2006  Absence of cough  Greenway et al. 2002  Prolonged symptoms (prior to initial diagnosis) Duration of symptoms before treatment started ( > 1 month)  Lawn et al. 1999  Lawn et al. 1999  Kourbatova et al. 2006 (1), Hargreaves et al. 2001  Kourbatova et al. 2006 (1)  Greenway et al. 2002  Hargreaves et al. 2001  Studies that reported risk factors related to type of TB, drug resistance and diagnosis are listed in Table 4.6. However, the association between type of TB and risk of death (pulmonary vs. extra‐pulmonary) was not clear‐cut. Extra‐pulmonary TB (22,44,50‐52) was shown to be a risk factor in a number of studies, while other studies reported it as not being significant (6,37,40). In addition, severe forms of EPTB such as miliary (41,53) and meningeal TB (53) were found to be associated with an increased mortality risk in several studies. Severity based on chest radiograph findings, (14,16‐18,25,41,47,54) was measured and termed differently in reported studies including extensive disease (54), bi‐lateral disease (14,47), greater severity (17), extensive infiltrates (18), far advanced PTB (16), extent of lung tissue involved (25) and cavitary lesions (47) but were associated with a significantly increased risk of death. In terms of bacteriological status, smear‐negative TB was consistently reported as a significant risk factor (8,17,22,42,50‐52,55) while one study demonstrated an increased risk of dying among smear 27  ‐MDR was also associated with TB‐caused mortality  122  positive TB patients (41). With regard to drug resistance, the elevated risk of dying among patients with MDR‐TB (10,20,34,35,46) was documented in a number of studies.  4.4 DISCUSSION This review demonstrates that mortality in TB patients occurs in a significant minority of patients (median death: overall‐ 12.8% and during treatment‐ 10.3%). A wide variation in the proportion of patients dying from TB is observed (ranging from 2.2% to 65.2% across studies). Several factors such as the quality of the health care delivery system, availability of resources, background incidence and prevalence of HIV might account for these differences. When we categorized mortality according to background burden of disease, mortality in high‐burden settings was substantially higher than that in low‐burden settings. The observed differences between low and high‐burden setting are possibly related to several important factors. Countries with a high‐burden of TB are generally resource–poor, which affects operational activities such as prompt diagnosis, institution of adequate chemotherapy and subsequent follow‐up. Therefore, the quality of health care service is frequently compromised in this setting, leading to unfavorable treatment outcomes including a relatively higher mortality.  This review has shown a substantially higher mortality among persons, infected with HIV which has previously been reported. There was also a wide variation of death rates across studies and by setting, which reflects the differences in background prevalence of HIV and availability of resources. As expected, mortality among HIV infected persons living in high‐burden countries was higher than subjects from low‐prevalence countries. Most of studies in high‐prevalence settings included in this review were conducted in sub‐Saharan Africa (such as 11 study in Malawi and 7 study in South Africa), which continue to experience an explosive HIV epidemic. The strikingly higher mortality in HIV‐TB co‐infected patients could be another notable explanation for overall higher mortality in high–burden settings. For instances, the highest mortality was reported in a study (29) from Malawi which included a population with a high prevalence of HIV (77%). In another study from Malawi (22) with the 2nd highest proportion of subjects co‐infected with HIV (48.7% at 2 years) reported a HIV prevalence of 68.7% for. The 123  Sudanese study which had the lowest overall mortality (2.2%), had a 4% prevalence of co‐ infection (TB and HIV).  Mortality among countries with an intermediate burden of TB had a slightly lower rate than that of low‐burden countries. The small number of studies in this setting might account for such a difference.  Included studies had different follow‐up periods which made comparisons  between studies difficult. Mortality during treatment, whenever available was calculated. However, the rates were similar according to burden of disease, reflecting the overall mortality trend. Death during hospitalization among hospitalized TB patients was relatively low (7.8%). Shorter follow up and a low number of studies may be confounding in this situation. Besides this, improved care and management of TB patients in the acute care setting will likely will result in a better survival for those patients.  The potential mechanism of death could not be explicitly stated due to the limited number of studies that investigated the causal relationship between death and TB. Based on 13 studies that reported death due to TB (underlying/primary cause) and total TB cases, the global observed TB‐caused death was quite low compared to the overall CDR. This demonstrated that not all patients were dying from TB. In high‐burden countries, most of studies did not provide any information on specific cause of death. Therefore, the reporting frequency of TB‐related death was much higher in low‐burden countries (6 out of 16 in low burden and 4 out of 34 in high‐burden countries). Despite a low level of representation from the high‐burden setting, the proportion was relatively lower in low‐incidence countries than that of high‐incidence countries (2.8%).  However, TB‐related death was relatively lower (1.2%) in intermediate‐ burden  countries, which was highly influenced by a Turkish study (0.6%) with low mortality. However, this Turkish study only reported in‐hospital mortality from PTB (16).  This review has shown that only one third of deaths (34.7%) among TB patients were primarily due to TB. A wide discrepancy in the proportion of TB related deaths (which ranged from 1.4% to 75.3%) was found across studies, reflecting the differences in methodology (such as 124  definition of TB‐caused death) and study population. The higher proportion of TB‐related deaths in high burden countries was supportive of this observation. These findings have a potential implication for TB‐control programs. Evaluating the performance of TB control program based on all‐cause mortality might be misleading. Interventions based on such findings will have limited applicability in preventing deaths related to other causes. An integrated approach, emphasizing other co‐morbidities that primarily or secondarily contribute to mortality would be valuable in order to prevent death from TB.  This review identified several important socio‐demographic risk factors that are independently associated with an increased risk of death from TB. Among them, older age was consistently identified as an important risk factor for mortality. However, the relationship between gender and mortality was not clear‐cut. A few studies reported increased risk among males, while others found the reverse, where as other studies did not find a significant association between gender and mortality.  TB was more frequent among foreign‐born persons especially in countries with a high level of immigration. However, the mortality risk was not proportional in this population group. Moreover, local‐born TB patients (such as the USA‐born and France‐born) experienced a higher risk of dying. Healthy life style, decreased prevalence of HIV and other co‐morbidities among the foreign‐born could be considered in support of this finding. Mortality was higher among socially disadvantaged TB patients (such as patients who were homeless, unemployed or had a low standard of education). Homelessness and unemployment are indicators of low socio‐ economic status, which have the potential to increase the risk of dying due to its relationship with reduced access to health care. Some of these social determinants are potentially modifiable. Interventions (such as free access to health care for socially marginalized patients, availability of shelters/affordable housing for homelessness patients and employment training program targeting these high risk populations should be considered.  125  Another major observation from this review was a strong relationship between HIV and mortality among TB patients. Besides age, HIV sero‐positivity was consistently found to be associated with an increased mortality risk. It should be noted that not all studies that investigated the impact of HIV on mortality found a significant association between them (10,13,18,24,43,47,53,56). However, the global mortality among HIV co‐infected patients was much higher, approximately five‐fold higher that non‐HIV infected individuals. This clearly demonstrates strong evidence in support of the independent association between HIV infection and mortality in TB patients. The increased risk associated with HIV‐TB co‐infection was consistently found in low‐burden, intermediate‐burden and high‐burden countries. Due to presence of strong interaction between HIV and TB, effective interventions for the prevention of HIV as well as prevention of TB among people living with HIV are urgently required.  The relative contribution of TB itself and HIV to mortality among dually infected patients is not clearly understood. The specific–cause of death (related to TB) was available in only a few studies. The previously cited clinical trial (11) reported only one death due to TB and but 11 due to HIV/AIDS among 71 observed deaths. Deaths among HIV infected TB patients tend to increase over time, and is likely related to the underlying degree of immune‐suppression. This clinical trial reported an increased mortality among HIV infected individuals at the end of the follow‐up period (11). This study (11) reported no significant difference in mortality between HIV positive and negative TB patients in the first month of treatment, but mortality was significantly higher among HIV infected persons during the subsequent 2‐6 month treatment period. The strong association between being HIV positive and mortality has important implications on TB control programs as it clearly demonstrates a highly vulnerable population and offers an important opportunity for collaborative activities between TB and HIV control programs targeting dually infected persons. Aggressive monitoring of patients, long‐term‐follow up and introduction of anti‐retroviral therapy, especially HARRT have a significant potential to improve the outcomes among HIV/TB co‐infected persons (57,58). Therefore, establishment of a longitudinal surveillance system to identify patients with HIV at an earlier stage and subsequent introduction of HARRT would be valuable in these circumstances. 126  This review has shown that several concomitant illnesses (in addition to HIV) likely independently increase the risk of death among TB patients. Since all TB patients do not die from TB, this is suggestive that there are co‐morbidities responsible for deaths. Although several co‐morbid conditions such as malignancy, malnutrition and renal diseases are well‐ known risk factors for the development of active TB, their impact on mortality has been less clear cut. However, all these factors were found to be associated with an increased mortality in several studies reviewed in our SR. These findings are not surprising, since these conditions have the potential to impair cellular immunity in TB patients, which not surprisingly might increase the risk of dying. The superimposition of TB on chronic illness could be another potential explanation for the increased risk of dying in these patients. Some of these co‐ morbidities such as malnutrition, anemia and respiratory failure are potentially reversible. Specific interventions (such as nutritional supplement for patients with malnutrition or weight loss and comprehensive treatment program for patients with malignancy) targeting these high‐ risk populations might be beneficial in order to improve the survival of TB patients.  Several behavioral factors such as alcoholism and drug abuse were significantly associated with the increased mortality observed in this review. All these factors are indicative of a high risk life style and were associated with increased morbidity and mortality (59‐61). In addition, use of alcohol and other illicit drugs during treatment likely may attribute to reduced efficacy of drugs due to poor adherence. Injection drug use is well‐recognized and a fundamental cause of several blood‐borne diseases including HIV. However, these behavioral factors are subjects to modifications. Interventions targeting those high risk populations are urgently required. Establishment of a systematic approach in order to identify TB patients with alcoholism and drug abuse, followed by the initiation of appropriate drug or alcoholism treatment programs should be prioritized in the care plan of TB patients.  Severity of disease was found to be an important risk factor for mortality in this review. In the case of PTB, severity of disease was quantified and termed differently (such as far advanced, extensive disease, bilateral disease and cavitation etc) in different studies. All these descriptors 127  are associated with extensive pulmonary TB and tend to increase the likelihood of an unfavorable outcome. Paradoxically mortality was higher among patients with smear negative TB. However, this might be confounded by HIV infection, as an increased prevalence of HIV was observed among this group of patients (55). In addition, EPTB including miliary and meningeal TB (both represent life threatening forms of TB) was found to increase the mortality risk among TB patients. Due to atypical presentation, detection of smear negative TB and EPTB specially disseminated and meningeal TB becomes quite challenging, which further increases  the  likelihood of missed or delayed diagnosis, eventually placing the TB patients at an increased risk of dying without treatment. A high index of suspicion and if appropriate, initiation of therapy before the confirmed diagnosis is critical in order to prevent morbidity and mortality from these devastating forms of TB. An awareness regarding atypical presentations and the increased mortality related to EPTB (especially miliary and CNS/meningeal TB) in the community as well as in health care workers is required.  Several treatment‐related factors were found to be associated with an increased risk of death. Incomplete treatment and no treatment at all were also found to increase the risk of death among TB patients. Mortality was higher among patients whose diagnosis was missed initially and whose treatment was delayed. In low‐burden settings, physicians tend to become less aware of the possibility of TB as a diagnosis due to lack to TB cases, especially among patients with atypical presentations, which increases the risk of a missed diagnosis. This observation was seen in two studies. A Dutch study demonstrated an independent association between diagnosis by type of physicians and mortality. Another study in Canada reported better survival outcomes associated with physician experience in managing TB. These factors are potentially reversible with improved education of physicians.  This review has several limitations. Our search strategy was limited to English literature. Therefore it is possible we missed some important non‐English studies were not included in this review. Only very few studies reported mortality in terms of person years. Due to the unavailability of mortality in person years, we could not make an overall estimate of mortality 128  in a standardized time unit despite differential follow up times used by several studies. Wide variation was observed across studies in terms of study design, population and setting, which preclude us to yield an overall robust quantitative assessment. Autopsy is considered as a gold standard for specific cause of death. Definition of TB‐related deaths varied widely across studies due to the limited number of autopsy‐proven cases in the majority of studies. Cautions should be considered in interpreting the findings resulting from TB‐caused deaths.  4.5 CONCLUSIONS In conclusion, mortality among TB patients varies significantly but in some settings is high. Mortality was twice as high in high‐burden countries compared to subjects in a low‐burden setting. The higher mortality among HIV positive TB patients compared to HIV non‐infected patients was consistently observed in many studies. Besides HIV, older age was widely documented to be a significant risk factor. Other important risk factors were unemployment, homelessness, co‐morbid conditions (such as malignancy, renal failure and malnutrition), EPTB, smear negative TB and. MDR TB. These findings have important implications for TB control programs. Several risk factors are potentially modifiable. Interventions targeting high risk populations should be considered.  129  4.6 REFERENCES (1) Maher D, Raviglione M. Global epidemiology of tuberculosis. Clin.Chest Med. 2005 v; Jun;26(2):167‐182. (2) World Health Organization. Global Tuberculosis Control 2009: Epidemiology Strategy Financing. 2009; Available at: http://www.who.int/tb/publications/global_report/2009/pdf/full_report.pdf. Accessed April/02, 2010. (3) World Health Organization. Tuberculosis Facts 2009 Update. 2009; Available at: http://www.who.int/tb/publications/2009/factsheet_tb_2009update_dec09.pdf. Accessed April/02, 2010. (4) World Health Organization. TB Data: Table A‐1 Estimated epidemiological burden of TB, all forms, 1990–2008. 2009; Available at: http://www.who.int/tb/publications/global_report/2009/update/a‐1_full.pdf. Accessed June/12, 2010. (5) Barker RD, Nthangeni ME, Millard FJ, Barker RD, Nthangeni ME, Millard FJC. Is the distance a patient lives from hospital a risk factor for death from tuberculosis in rural South Africa? International Journal of Tuberculosis & Lung Disease 2002 Feb;6(2):98‐103. (6) Borgdorff MW, Veen J, Kalisvaart NA, Nagelkerke N. Mortality among tuberculosis patients in the netherlands in the period 1993‐1995. European Respiratory Journal 1998 Apr;11(4):816‐ 820. (7) Churchyard GJ, Kleinschmidt I, Corbett EL, Murray J, Smit J, De Cock KM. Factors associated with an increased case‐fatality rate in HIV‐infected and non‐infected South African gold miners with pulmonary tuberculosis. International Journal of Tuberculosis and Lung Disease 2000;4(8):705‐712. 130  (8) Connolly C, Davies GR, Wilkinson D. Impact of the human immunodeficiency virus epidemic on mortality among adults with tuberculosis in rural South Africa, 1991‐1995. International Journal of Tuberculosis and Lung Disease 1998;2(11):919‐925. (9) Zachariah R, Spielmann MP, Harries AD, Salaniponi FML. Moderate to severe malnutrition in patients with tuberculosis is a risk factor associated with early death. Trans.R.Soc.Trop.Med.Hyg. 2002;96(3):291‐294. (10) Mathew TA, Ovsyanikova TN, Shin SS, Gelmanova I, Balbuena DA, Atwood S, et al. Causes of death during tuberculosis treatment in Tomsk Oblast, Russia. International Journal of Tuberculosis and Lung Disease 2006 Aug;10(8):857‐863. (11) Sterling TR, Zhao Z, Khan A, Chaisson RE, Schluger N, Mangura B, et al. Mortality in a large tuberculosis treatment trial: Modifiable and non‐modifiable risk factors. International Journal of Tuberculosis and Lung Disease 2006 May;10(5):542‐549. (12) Kolappan C, Subramani R, Karunakaran K, Narayanan PR. Mortality of tuberculosis patients in Chennai, India. Bull.World Health Organ. 2006;84(7):555‐560. (13) Greenaway C, Menzies D, Fanning A, Grewal R, Yuan L, FitzGerald JM. Delay in diagnosis among hospitalized patients with active tuberculosis ‐ Predictors and outcomes. American Journal of Respiratory and Critical Care Medicine 2002 01 Apr;165(7):927‐933. (14) Dewan PK, Arguin PM, Kiryanova H, Kondroshova NV, Khorosheva TM, Laserson K, et al. Risk factors for death during tuberculosis treatment in Orel, Russia. International Journal of Tuberculosis and Lung Disease 2004 May;8(5):598‐602. (15) Bustamante‐Montes LP, Escobar‐Mesa A, Borja‐Aburto VH, Gomez‐Munoz A, Becerra‐ Posada F, Bustamante‐Montes LP, et al. Predictors of death from pulmonary tuberculosis: the case of Veracruz, Mexico. International Journal of Tuberculosis & Lung Disease 2000 Mar;4(3):208‐15.  131  (16) Kartaloglu Z, Ilvan A, Kilic E, Okutan O, Cerrahoglu K, Ciftci F, et al. Deaths in patients with pulmonary tuberculosis: an analysis of a chest diseases hospital in Istanbul, Turkey. Medical Principles & Practice 2003 Jan‐Mar;12(1):30‐3. (17) Lawn SD, Acheampong JW, Lawn SD, Acheampong JW. Pulmonary tuberculosis in adults: factors associated with mortality at a Ghanaian teaching hospital. West African Journal of Medicine 1999 Oct‐Dec;18(4):270‐4. (18) Sacks LV, Pendle S, Sacks LV, Pendle S. Factors related to in‐hospital deaths in patients with tuberculosis. Archives of Internal Medicine 1998 Sep 28;158(17):1916‐22. (19) Zafran N, Heldal E, Pavlovic S, Vuckovic D, Boe J, Zafran N, et al. Why do our patients die of active tuberculosis in the era of effective therapy? Tubercle & Lung Disease 1994 Oct;75(5):329‐ 33. (20) Fischl MA, Daikos GL, Uttamchandani RB, Poblete RB, Moreno JN, Reyes RR, et al. Clinical presentation and outcome of patients with HIV infection and tuberculosis caused by multiple‐ drug‐resistant bacilli. Ann.Intern.Med. 1992 08/01;117(3):184‐190. (21) El‐Sony AI, Khamis AH, Enarson DA, Baraka O, Mustafa SA, Bjune G. Treatment results of DOTS in 1797 Sudanese tuberculosis patients with or without HIV co‐infection. International Journal of Tuberculosis & Lung Disease 2002 Dec;6(12):1058‐1066. (22) Kang'ombe CT, Harries AD, Ito K, Clark T, Nyirenda TE, Aldis W, et al. Long‐term outcome in patients registered with tuberculosis in Zomba, Malawi: mortality at 7 years according to initial HIV status and type of TB. International Journal of Tuberculosis & Lung Disease 2004 Jul;8(7):829‐836. (23) Hargreaves NJ, Kadzakumanja O, Whitty CJ, Salaniponi FM, Harries AD, Squire SB. 'Smear‐ negative' pulmonary tuberculosis in a DOTS programme: poor outcomes in an area of high HIV seroprevalence. International Journal of Tuberculosis & Lung Disease 2001 Sep;5(9):847‐854.  132  (24) Hansel NN, Merriman B, Haponik EF, Diette GB, Hansel NN, Merriman B, et al. Hospitalizations for tuberculosis in the United States in 2000: predictors of in‐hospital mortality. Chest 2004 Oct;126(4):1079‐86. (25) Abos‐Hernandez R, Olle‐Goig JE. Patients hospitalised in Bolivia with pulmonary tuberculosis: risk factors for dying. International Journal of Tuberculosis & Lung Disease 2002 Jun;6(6):470‐474. (26) L'Ecuyer PB, Woeltje KF, Seiler SM, Fraser VJ. Management and outcome of tuberculosis in two St Louis hospitals, 1988 to 1994. Infection Control & Hospital Epidemiology 1998 Nov;19(11):836‐841. (27) Elliott AM, Halwiindi B, Hayes RJ, Luo N, Mwinga AG, Tembo G, et al. The impact of human immunodeficiency virus on mortality of patients treated for tuberculosis in a cohort study in Zambia. Transactions of the Royal Society of Tropical Medicine & Hygiene 1995 Jan‐ Feb;89(1):78‐82. (28) Garin B, Glaziou P, Kassa‐Kelembho E, Yassibanda S, Mbelesso P, Morvan J. High mortality rates among patients with tuberculosis in Bangui, Central African Republic. Lancet 1997 01 Nov;350(9087):1298. (29) Kelly PM, Cumming RG, Kaldor JM. HIV and tuberculosis in rural sub‐Saharan Africa: a cohort study with two year follow‐up. Transactions of the Royal Society of Tropical Medicine & Hygiene 1999 May‐Jun;93(3):287‐293. (30) Jasmer RM, Seaman CB, Gonzalez LC, Kawamura LM, Osmond DH, Daley CL. Tuberculosis treatment outcomes: Directly observed therapy compared with self‐administered therapy. American Journal of Respiratory and Critical Care Medicine 2004 01 Sep;170(5):561‐566. (31) Ackah AN, Coulibaly D, Digbeu H, Diallo K, Vetter KM, Coulibaly IM, et al. Response to treatment, mortality, and CD4 lymphocyte counts in HIV‐infected persons with tuberculosis in Abidjan, Cote d'Ivoire. Lancet 1995 Mar 11;345(8950):607‐610. 133  (32) Khan K, Campbell A, Wallington T, Gardam M. The impact of physician training and experience on the survival of patients with active tuberculosis. CMAJ Canadian Medical Association Journal 2006 September 26;175(7):749‐753. (33) Murray J, Sonnenberg P, Shearer SC, Godfrey‐Faussett P. Human immunodeficiency virus and the outcome of treatment for new and recurrent pulmonary tuberculosis in African patients. American Journal of Respiratory and Critical Care Medicine 1999;159(3):733‐740. (34) Pablos‐Mendez A, Sterling TR, Frieden TR. The relationship between delayed or incomplete treatment and all‐cause mortality in patients with tuberculosis. JAMA 1996 Oct 16;276(15):1223‐1228. (35) Quy HT, Cobelens FG, Lan NT, Buu TN, Lambregts CS, Borgdorff MW. Treatment outcomes by drug resistance and HIV status among tuberculosis patients in Ho Chi Minh City, Vietnam. International Journal of Tuberculosis & Lung Disease 2006 Jan;10(1):45‐51. (36) Vree M, Huong NT, Duong BD, Sy DN, Van LN, Hung NV, et al. Survival and relapse rate of tuberculosis patients who successfully completed treatment in Vietnam. Int.J.Tuberc.Lung Dis. 2007 Apr;11(4):392‐397. (37) Rabaud C, Engozogho M, Dailloux M, Hoen B, May T, Canton P, et al. Tuberculosis in Lorraine, France: study of prognostic factors. International Journal of Tuberculosis & Lung Disease 1997 Jun;1(3):246‐9. (38) Aerts D, Jobim R, Aerts D, Jobim R. The epidemiological profile of tuberculosis in southern Brazil in times of AIDS. International Journal of Tuberculosis & Lung Disease 2004 Jun;8(6):785‐ 91. (39) Helbling P, Medinger C, Altpeter E, Raeber PA, Beeli D, Zellweger JP. Outcome of treatment of pulmonary tuberculosis in Switzerland in 1996. Swiss Medical Weekly 2002 Sep 7;132(35‐ 36):517‐522.  134  (40) Bakhshi SS, Hawker J, Ali S. Tuberculosis mortality in notified cases from 1989‐1995 in Birmingham. Public Health 1998 May;112(3):165‐168. (41) Xie HJ, Enarson DA, Chao CW, Allen EA, Grzybowski S, Xie HJ, et al. Deaths in tuberculosis patients in British Columbia, 1980‐1984. Tubercle & Lung Disease 1992 Apr;73(2):77‐82. (42) Gustafson P, Gomes VF, Vieira CS, Samb B, Naucler A, Aaby P, et al. Clinical predictors for death in HIV‐positive and HIV‐negative tuberculosis patients in Guinea‐Bissau. Infection 2007 Apr;35(2):69‐80. (43) Rao VK, Iademarco EP, Fraser VJ, Kollef MH, Rao VK, Iademarco EP, et al. The impact of comorbidity on mortality following in‐hospital diagnosis of tuberculosis. Chest 1998 Nov;114(5):1244‐52. (44) Walpola HC, Siskind V, Patel AM, Konstantinos A, Derhy P. Tuberculosis‐related deaths in Queensland, Australia, 1989‐1998: characteristics and risk factors. International Journal of Tuberculosis & Lung Disease 2003 Aug;7(8):742‐750. (45) Santha T, Garg R, Frieden TR, Chandrasekaran V, Subramani R, Gopi PG, et al. Risk factors associated with default, failure and death among tuberculosis patients treated in a DOTS programme in Tiruvallur District, South India, 2000. International Journal of Tuberculosis and Lung Disease 2002 Sep;6(9):780‐788. (46) Garcia‐Garcia Mde L, Ponce‐De‐Leon A, Garcia‐Sancho MC, Ferreyra‐Reyes L, Palacios‐ Martinez M, Fuentes J, et al. Tuberculosis‐related deaths within a well‐functioning DOTS control program. Emerging Infectious Diseases 2002 Nov;8(11):1327‐33. (47) Kourbatova EV, Borodulin BE, Borodulina EA, del Rio C, Blumberg HM, K. LM,Jr, et al. Risk factors for mortality among adult patients with newly diagnosed tuberculosis in Samara, Russia. International Journal of Tuberculosis & Lung Disease 2006 Nov;10(11):1224‐30.  135  (48) Barker RD, Millard FJ, Malatsi J, Mkoana L, Ngoatwana T, Agarawal S, et al. Traditional healers, treatment delay, performance status and death from TB in rural South Africa. International Journal of Tuberculosis & Lung Disease 2006 Jun;10(6):670‐675. (49) Alpert PL, Munsiff SS, Gourevitch MN, Greenberg B, Klein RS. A prospective study of tuberculosis and human immunodeficiency virus infection: clinical manifestations and factors associated with survival. Clinical Infectious Diseases 1997 Apr;24(4):661‐668. (50) Kang'ombe C, Harries AD, Banda H, Nyangulu DS, Whitty CJ, Salaniponi FM, et al. High mortality rates in tuberculosis patients in Zomba Hospital, Malawi, during 32 months of follow‐ up. Transactions of the Royal Society of Tropical Medicine & Hygiene 2000 May‐Jun;94(3):305‐ 9. (51) Harries AD, Nyangulu DS, Kang'ombe C, Ndalama D, Glynn JR, Banda H, et al. Treatment outcome of an unselected cohort of tuberculosis patients in relation to human immunodeficiency virus serostatus in Zomba Hospital, Malawi. Transactions of the Royal Society of Tropical Medicine & Hygiene 1998 May‐Jun;92(3):343‐347. (52) Harries AD, Hargreaves NJ, Gausi F, Kwanjana JH, Salaniponi FM. High early death rate in tuberculosis patients in Malawi. International Journal of Tuberculosis & Lung Disease 2001 Nov;5(11):1000‐1005. (53) Kourbatova EV, K. LM,Jr, Romero J, Kraft C, del Rio C, Blumberg HM, et al. Risk factors for mortality among patients with extrapulmonary tuberculosis at an academic inner‐city hospital in the US. European Journal of Epidemiology 2006;21(9):715‐21. (54) Cullinan P, Meredith SK, Cullinan P, Meredith SK. Deaths in adults with notified pulmonary tuberculosis 1983‐5. Thorax 1991 May;46(5):347‐50. (55) Harries AD, Nyirenda TE, Banerjee A, Boeree MJ, Salaniponi FM. Treatment outcome of patients with smear‐negative and smear‐positive pulmonary tuberculosis in the National  136  Tuberculosis Control Programme, Malawi. Transactions of the Royal Society of Tropical Medicine & Hygiene 1999 Jul‐Aug;93(4):443‐446. (56) Nunn P, Brindle R, Carpenter L, Odhiambo J, Wasunna K, Newnham R, et al. Cohort study of human immunodeficiency virus infection in patients with tuberculosis in Nairobi, Kenya: Analysis of early (6‐month) mortality. Am.Rev.Respir.Dis. 1992;146(4):849‐854. (57) Burman W, Benator D, Vernon A, Khan A, Jones B, Silva C, et al. Acquired rifamycin resistance with twice‐weekly treatment of HIV‐related tuberculosis. American Journal of Respiratory & Critical Care Medicine 2006 Feb 1;173(3):350‐356. (58) Dheda K, Lampe FC, Johnson MA, Lipman MC. Outcome of HIV‐associated tuberculosis in the era of highly active antiretroviral therapy. J.Infect.Dis. 2004 Nov 1;190(9):1670‐1676. (59) Oyefeso A, Ghodse H, Clancy C, Corkery J, Goldfinch R. Drug abuse‐related mortality: a study of teenage addicts over a 20‐year period. Social Psychiatry & Psychiatric Epidemiology 1999 Aug;34(8):437‐441. (60) Patra J, Taylor B, Rehm JT, Baliunas D, Popova S. Substance‐attributable morbidity and mortality changes to Canada's epidemiological profile: measurable differences over a ten‐year period. Canadian Journal of Public Health.Revue Canadienne de Sante Publique 2007;98(3):228‐ 234. (61) Paulozzi LJ, Annest JL. US data show sharply rising drug‐induced death rates. Injury Prevention 2007 Apr;13(2):130‐132.  137  CHAPTER 5: AN ALL CAUSE MORTALITY STUDY AMONG PATIENTS WITH TUBERCULOSIS IN BRITISH COLUMBIA28  28  ‐“A version of this chapter will be submitted for publication. Moniruzzaman, A. Kazanjian, A. Wong, H. Elwood, R.K. and FitzGerald, J. M. (2010) Mortality (all‐cause) in TB patients in British Columbia ‐ a population‐based study.”  138  5.1 INTRODUCTION Mortality in TB patients continues to pose a significant public health challenge throughout the word. Globally, TB accounts for 6% of all deaths and more than one fourth of all preventable adult deaths (1‐3). An estimated two million people die of TB each year, (1.8 million deaths in 2008) with a case fatality rate (CFR) of up to 53% (overall global CFR is 23%) (4‐6). The number of deaths caused by M. tuberculosis was higher than that of any other infectious agent in the pre‐HIV era (7,8). Currently, TB remains the second commonest cause of death due to a single infectious agent (HIV ranks first). In the modern era, all TB‐related deaths are potentially preventable. This raises a simple question, why are we experiencing so many deaths related to TB despite the presence of effective chemotherapy? If appropriate interventions are not effectively implemented, many patients with this preventable and treatable disease will continue to have unnecessary morbidity and mortality. It is widely acknowledged that more epidemiological research is required on the etiology and treatment of TB including identifying the determinants of mortality among TB patients.  Although studies conducted in this area have identified several factors (e.g., older age, HIV infection, multi‐drug resistance, co‐morbid conditions, and immune‐suppression status) as being associated with TB related mortality (2,9‐15), the epidemiological risk factors for mortality requires further exploration. In addition, there is a scarcity of longitudinal, population‐based studies. Most studies have been conducted on hospitalized patients or in selected population (2,10,11,16‐23).  Studies based on narrowly defined populations are prone to have limited external validity, which often affects their application to the general population. Research focused on examining the impact of preventable risk factors on mortality among TB patients in a general population setting is clearly required to address these information gaps. There is a scarcity of epidemiological and clinical research in Canada on TB‐related mortality. Recently, TB mortality has caused a growing concern in many industrialized countries of world due to an increase in the number of TB related deaths (16,24,25). This situation reinforces the need for an 139  investigation of the risk factors for TB related mortality in a Canadian setting. The purpose of this study was to describe the characteristics of TB patients who died from any cause, to investigate epidemiological trends for deaths compared to the general population in British Columbia and identify the risk factors associated with mortality among TB patients.  5.2 METHODS The provincial TB service (TB Control Division) at the British Columbia Centre for Disease Control (BCCDC) acts as a referral centre for the prevention, control and treatment of TB in British Columbia (BC). TB control activities in BC are centralized; this center maintains a central registry that includes information on all TB patients diagnosed in the province. In addition, all mycobacteriology tests are coordinated by this center. The BCCDC TB laboratory, which performs all culture and drug susceptibility tests, also severs as the reference laboratory for this province. This study used a retrospective cohort study design and included all TB patients that were diagnosed in BC from 1990 to 2006. For descriptive statistics such as ASMR, we used all TB patients. However, we used a restricted sample to indentify risk factors associated with mortality among TB patients. Patients who had a history of TB diagnosis before 1990 were not included in the Cox regression analysis.  5.2.1 Variables Description Age was estimated at the time of diagnosis. Age was also analyzed as a continuous variable as well as a categorical variable. Four age groups were constructed and 0‐40 years was used as the reference group. Based on year of birth, several birth‐cohorts were constructed for each decade (for examples, 1901‐1910, 1911‐1920, 1921‐1930 etc.). Death was defined as a patient who had an active TB diagnosis and died from any cause during the study period (1990 to 2006). A detailed description of several important variables was provided in Chapter 2.  140  5.2.2 Statistical Analysis An initial descriptive analysis for the study population was performed. Comparisons of continuous variables were done using student t test and Wilcoxon rank sum test whenever appropriate. Categorical variables were compared using the chi‐square test and Fisher’s exact test (if one of the expected cell values were less than 5).  For each calendar year, a crude death rate per 1000 from 1990 to 2006 was calculated for the TB cohort, and was compared to the crude death rate in the general population of BC. The crude mortality rate ratio was also calculated for this purpose. Since the TB cohort consists of a selective group from the general population, the age standardized mortality ratio (ASMR) was also estimated for each year. The estimated population of BC for each year was considered as the general population. The age specific death rate of the general population of BC was used to estimate ASMR for the TB cohort. The age groups considered to estimate ASMR were as follows: 0‐24 years, 25‐44 years, 45‐64, years, 65‐84 years, and 85 plus years. To calculate age specific death rate of the TB cohort, total active TB cases of each calendar year were considered as denominator and total deaths among TB patients in the same calendar year were used as the numerator. Age for active TB cases (denominator) that were diagnosed in the each calendar year was recalculated at midpoint of the same calendar year. Age of death was used for the active TB cases that died. All general population data were downloaded from the BC Vital Statistics website for the years of 1990‐2006.  To identify risk factors associated with mortality among TB patients, single variable and multi‐ variable Cox regression was chosen as the primary analytic method. Patients with a history of diagnosis before 1990 were not eligible for follow up and excluded from the Cox regression analysis. Hazard ratio (HR) with the corresponding 95% Confidence Interval was reported as a measure of association. All reported p values were two sided. For this time to event analysis, deaths among TB patients during follow the up period was the primary outcome. Follow up time was estimated as the difference between time 1 and time 0. Date of diagnosis of the  141  primary episode of TB patients was the time 0. Patients were censored if they died (event of interest), or were lost to follow up, or left the Province, or reached study end point (December 31st, 2006). These times were considered as time 1.  Due to missing observations that was associated with co‐morbid conditions (such as HIV/AIDS, diabetes mellitus, malignancy, malnutrition, alcoholism, drug abuse, and chronic renal failure), the effects of these co‐morbidities was also assessed with and without missing observations in univariate and multivariable setting. While evaluating the effect of a co‐morbid condition in presence of missing observations, patients without the co‐morbidity and with unknown status were used as the reference group.  142  Figure 5.1: Study population for all‐cause mortality among TB patients (n=5408)  5433 TB Cases Registered with TB Division between 1990 and 2006  Excluded (n=25)  TB patients who were originally diagnosed in 1989  Study Cohort (n= 5408)  Excluded: TB patients with primary episode before 1990 or unknown status  Eligible for follow up (n=5021) January 1st, 1990: Follow up starts  Outcome of interest: Death (953 events between 1990 and 2006)  December 31st, 2006: Follow up ends  1) Primary Analysis: Cox regression among patients with primary episode 1990 or after (deaths vs. survivors, n=5021) 2) Additional analysis: patients who died during treatment (n=309) vs. who did not die during treatment (4904)  143  Several criteria were followed to select variables for multi‐variable Cox regression analysis. In addition to statistical significance in bi‐variate analysis, evidence from the literature and items of clinical relevance were also considered to select variables for multi‐variable models. Proportionality assumptions for several important variables (for instances, birth place and ethnicity) were also assessed. Two multivariable models (Model 1 and Model 2) were constructed. Variables that were considered in the first multivariable model (Model 1) were as follows: age, gender, birthplace, TB recurrence, culture confirmed TB, disseminated TB, CNS/meningeal TB, far advanced TB, moderately advanced TB, HIV/AIDS status, malignancy, renal failure, diabetes mellitus, immunosuppressive medication, malnutrition, alcoholism and substance abuse. We constructed a separate multivariable model (Model 2) which included ethnicity and all the above listed variables except birthplace. Both multivariable models evaluated the effects of co‐morbid conditions in presence of missing observations (no/unknown status was used as the reference group).  Additional Univariate and Multivariable Cox regression models were constructed for each co‐ morbid and behavioral condition. This analysis was restricted to patients who provided valid responses (Yes or No) to co‐morbidity status. A more detailed description of multivariable models for co‐morbidities without missing observations was presented in the Appendix A (Table A.7). All these multivariable models (Model 5 to Model 11) were controlled for age, gender, ethnicity, recurrence, culture status, diagnosis‐related factors, and HIV/AIDS status. In all these multivariable models (Model 5 to Model 11), only the adjusted hazard ratio (AHR) of the co‐ morbid condition was reported. For example, the variables that were used for Model 5 were: any malignancy, age, gender, birthplace, TB recurrence, culture confirmed TB, disseminated TB, CNS TB, far advanced TB, moderately advanced TB and HIV/AIDS status. The corresponding AHR of any malignancy was only reported.  Most of the published literature reported mortality among TB patients who died during treatment. Therefore, we conducted additional analysis to identify risk factors for mortality during treatment in this TB cohort. Death during TB treatment was the primary outcome and 144  patients who did not die during treatment were the comparison group for this additional analysis. Univariate and multivariable logistic regression was conducted in this setting (since patients were observed for a fixed length of time). To select variables for multivariable models, a similar strategy was followed as used for Cox regression method.  5.3 RESULTS 5.3.1 Description of Study Cohort and Mortality The study population included a total of 5408 TB patients, who had at least one active episode of TB during the study period (1990 to 2006). The characteristics of the study population are presented in Table 5.1. The mean age of the study cohort was 49 years while the median was 47 (IQR: 31‐67) years. There were more males (55%) than females. Foreign‐born subjects accounted for two‐third (67%) of the total cases. In terms of ethnicity, Chinese subjects represented the highest percentage (24%) of all TB patients followed by Caucasian (17%). Aboriginal people represented 13% of total TB cases. This TB cohort was over‐represented by people who were born in western‐Pacific regions (40%) while people born in the WHO Pan‐ American Health Regions represented 31% of subjects. The socio‐demographic characteristics of patients who died are also presented in Table 5.1.  145  Table 5.1: Socio‐demographic characteristics of study cohort (n=5408)  Variable  Age at diagnosis Mean (SD) Median (IQR) Age at death Mean (SD) Median (IQR, Min‐ Max) Age groups 0‐40 years 41‐60 years 61‐80 years 80 plus years Biological gender Male Female Birth place Canadian‐born non‐ Aboriginal Canadian‐born Aboriginal Foreign‐born Unknown Birth Country regions PAHO African Region Eastern Mediterranean Europe Southeast Asia Region Western Pacific Region Unknown /missing Ethnicity Caucasian Aboriginals Chinese South‐East Asian Filipino Vietnamese Others Unknown /Missing Marital status Single Not single Other /Unknown/missing  All cases (n=5408) N (%)  Patients who died (n =1069) N (%)  Years 48.7 (21.8) 46.7 (31‐67)  Years 65.2 (17.6) 68.6 (51‐79) Years 67.7 (17.8) 71.6 (53‐82, 15‐107))  _  2213 (41) 1421 (26) 1366 (25) 408 (8)  109 (10) 235 (22) 428 (40) 297 (28)  2951 (55) 2457 (45)  713 (67) 356 (33)  876 (16) 709 (13) 3606 (67) 217 (4)  227 (21) 203 (19) 476 (45) 163 (15)  1686 (31) 97 (2) 138 (3) 270 (5) 762 (14) 2181 (40) 274 (5)  446 (42) 7 (1) 9 (1) 66 (6) 99 (9) 242 (22) 200 (19)  938 (17) 709 (13) 1293 (24) 830 (15) 428 (8) 306 (6) 558 (10) 346 (6)  304 (28) 203 (19) 213 (20) 108 (10) 25 (2) 14 (1) 44 (4) 158 (15)  1111 (20) 1564 (29) 2733 (51)  194 (18) 336 (31) 539 (51)  146  Table 5.2: Comparisons of crude death rates (CDR) and age standardized mortality ratios (ASMR) in British Columbia from 1990 to 2006 (TB cohort vs. general population)  TB Cohort  1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Total  Total TB cases  Total deaths among TB patients  Crude Death Rate/1000  General Population (BC) Crude Death Rate/1000  275 284 336 340 332 324 326 418 333 327 314 368 300 325 310 278 335 5525  27 31 40 48 53 45 66 68 70 89 69 92 87 89 99 46 50 1069  98.2 109.2 119.0 141.2 159.6 138.9 202.5 162.7 210.2 272.2 219.7 250.0 290.0 273.8 319.4 165.5 149.3 Median: 166  7.1 7.1 7.1 7.2 7.0 6.9 7.1 6.9 7.0 7.0 6.8 6.9 7.0 7.0 7.1 7.1 7.1 Median: 7  Rate Ratio (TB cohort/General Population) Crude Death Rate Ratio (Column 4/ Column 5) 14 16 17 20 23 20 29 24 30 39 33 36 41 39 45 23 21 Median: 24  ASMR  9 9 12 13 15 14 20 15 19 24 22 22 29 24 27 16 15 Median: 16  During the entire study period, 1069 (19.8%, 95% CI: 18.7%, 20.9%) deaths were observed among all notified TB cases. Out of the 1069 deaths, 178 patients (17%) died before treatment could be commenced, 309 patients (29%) died during treatment and 565 patients (53%) died after treatment. A small minority 17 (1%) patients could not be categorized into treatment categories due to missing information. Among patients who did not receive any treatment, 173 cases were diagnosed at post mortem or with culture results that became available after the subject died. Although TB was discovered before death in the remaining five cases, treatment could be initiated in those cases. Therefore, deaths in the treatment groups were as follows: no 147  treatment category‐ 178/5408=3.2%; during treatment category‐ 309/5408=5.7%; and post‐ treatment category‐ 565/5408=10.5%.  Figure 5.2: Trend of crude death rates for TB cohort in British Columbia from 1990 to 2006  148  Figure 5.3: Trend of age standardized mortality ratios (ASMR) for TB cohort in British Columbia from 1990 to 2006  149  Table 5.3: Number of deaths among TB patients in BC over time (1990‐2006) by birth cohort Year of Birth Year of Death 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Total deaths Total cases % of death  1910 or before 9 11 6 8 9 7 7 4 4 10 3 7 6 5 5 2 1  1911‐ 20  1921‐ 30  1931‐ 40  1941‐ 50  1951‐ 60  1961‐ 70  1971‐ 80  1981‐ 90  1991‐ 06  Total deaths  2 5 13 13 13 8 9 23 20 19 19 24 14 25 24 8 11  8 9 9 9 10 12 22 11 20 24 14 24 19 17 31 10 7  3 3 2 2 3 2 8 12 7 12 15 11 10 16 15 8 5  3 2 7 7 9 5 6 6 5 5 7 11 11 9 9 7 7  2 2 2 6 5 9 10 7 9 11 5 9 16 8 9 8 7  0 0 1 3 4 2 3 5 5 6 2 6 8 7 4 3 9  0 0 0 0 0 0 1 0 0 1 4 0 2 1 2 0 3  0 0 0 0 0 0 0 0 0 0 0 0 1 1 0 0 0  0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0  27 32 40 48 53 45 66 68 70 88 69 92 87 89 99 46 50  104  250  256  134  116  125  68  14  2  0  1069  154  474  739  631  648  828  941  621  274  98  5408  0.68  0.53  0.35  0.21  0.18  0.15  0.07  0.02  0.01  0.00  0.20  150  Figure 5.4: Trend of deaths among TB patients in British Columbia from 1990 to 2006 by birth cohort  151  5.3.2 Comparisons of Mortality between TB Cohort and General Population Table 5.2 shows the crude death rate (per 1000) for the TB cohort for each calendar year (Range: 98‐319, Median: 166). The CDR was lowest in 1990 (98) and was highest in 2004 (319). The trend of CDR showed an increasing trend from 1990 to 2004 (Figure 5.2). There was a substantial drop in CDR in 2005 (166) and in 2006 (149) compared to 2004 (314), but these rates were still much higher than the rates in the initial period (1990 to 1994). On the other‐ hand, the CDR for the general population was stable during the entire study period and was approximately 7 per 1000 (Table 5.2 and Figure 5.2). The ratio of these crude death rates (TB cohort vs. the general population) is also presented in Table 5.2. The CDR for the TB cohort was much higher for each calendar year. In 1990, the CDR was 14 times higher and in 2004, it was 45 times higher compared to general population. In 2005 and 2006, the ratios were 23 and 21 respectively.  The age standardized mortality ratio (ASMR) was estimated for each year, which is shown in Table 5.2 and Figure 5.3. These ratios were lower compared to the Crude Death Rate ratio, but much higher than one. Overall, ASMR showed an increasing trend from 1990 to 2006 (Range: 9‐ 29, Median: 16). The peak ASMR was observed during 2002 to 2004. The ASMR was 9 in 1990 and 15 in 2006. The ASMR was the highest in 2002 (29) and lowest in 1990 and 1991 (9). In 2004, the ASMR was 27 while CDR ratio was 45 in the same calendar year. Table 5.3 shows the number of deaths according 10‐years birth‐cohort. As expected, TB patients who were born in 1910 or before had the highest death rate (68%). Overall, the cumulative proportion of patients who died decreased among younger patients in the cohort. No deaths were observed among TB patients who were born after 1990. Patients born between 1961 and 1970 accounted for the highest percentage (17%) of all TB cases and had a mortality rate of 7%. The preceding birth‐cohort (1951 to 1960) had a two‐fold higher mortality rate (15%) compared to the birth‐cohort of 1961‐70. The death patterns among TB patients according to birth‐cohort were graphically presented in Figure 5.4.  152  Figure 5.5: Histogram of age at death for TB patients who died between 1990 and 2006 (n=1069)  5.3.3 Follow­up and Risk Factors for All­cause Mortality among TB Patients Out of 5408 TB patients, 5021 (93%) TB patients were eligible for follow up, which contributed 953 events and in‐total 33,035 person‐years (PYs) during the entire observation period. The overall occurrence of death was 29 per 1000 PYs. The median follow up was 5.9 years (Range:  153  0‐17 years, Inter‐Quartile Range: 1.6‐10.9 years). The median time of death (since the initial diagnosis) was 1.1 years (Range: 0‐15.6 years, IQR: 0.1‐3.8 years). During the initial period, the cumulative mortality at the first 6 (393 deaths) and 12 months (466 deaths) was 8% and 10% respectively. Mortality at 5 years (779 deaths), 10 years (903 deaths), and 15 years (952 deaths) were 17%, 22% and 26% respectively.  Table 5.4 shows the incidence of mortality in terms of person years and the results from univariate Cox regression (unadjusted HR estimates) for socio‐demographic characteristics. The risk of dying increased significantly with age (UHR: 1.05 per year, 95% CI: 1.04, 1.05) and was the highest among the older age groups. Males had an increased likelihood (UHR=1.79, 95% CI: 1.57, 2.05) to die compared to females. The risk of dying was twice as high among Canadian‐ born non‐aboriginal subjects (Unadjusted HR=1.93, 95% CI: 1.64, 2.28) and Aboriginal people (UHR: 2.08, 95% CI: 1.74, 2.49) compared to the foreign‐born. In terms of ethnicity, Caucasian (Unadjusted HR=2.78) and Aboriginals (Unadjusted HR=2.35) had a significantly increased risk of dying compared to the reference group (Chinese, Filipino and Vietnamese).  Table 5.5 presents unadjusted HR estimates for the diagnosis related factors. TB patients who died were less likely to have presented with recurrent disease (Unadjusted HR=0.59) and more likely to have culture confirmed TB (Unadjusted HR=2.50). Overall, patients with extra‐ pulmonary TB had a decreased likelihood (Unadjusted HR=0.68) of dying compared patients with pulmonary TB (PTB). However, a number of specific TB diagnoses were significantly associated with a greater risk of mortality. For examples, patients who had disseminated TB (Unadjusted HR=4.40), far advanced PTB (Unadjusted HR=2.16), CNS/meningeal TB (Unadjusted HR: 1.69) or moderately advanced PTB (Unadjusted HR=1.33) were more likely to die. Patients with MDR TB did not have a significant impact on mortality (Unadjusted HR=0.85, 95% CI: 0.35, 2.04) compared to patients with drug sensitive TB.  154  Figure 5.6: Survival curve among aboriginal (bottom line), canadian‐born non‐aboriginal and foreign‐born (top line) TB patients29  29  ‐p value from Log Rank test was <0.001.  155  Table 5.4: Comparisons of socio‐demographic characteristics between TB patients who died during follow up period (n=953) and who survived (n=4068)  Variable  Number of death  Total Person Years (PYs)  Death/1000 PYs  Unadjusted Hazard Ratio (95% CI)  Overall  953  33035  29  ‐  Age at initial diagnosis Mean (SD) Age groups (at first diagnosis) 0‐40 years 41‐60 years 61‐80 years 80 plus years Biological gender Male Female Birth place CB non‐aboriginal CB Aboriginal Foreign‐born Birth Country regions PAHO African Region Eastern Mediterranean European Region Southeast Asia Region Western Pacific Region Ethnicity Caucasian Aboriginals Southeast Asian Other Ethnicity Chinese/Filipino/Vietnamese Marital status Not single Single  1.05 (1.04, 1.05) 127 222 413 191  16316 8583 6775 1361  8 26 61 140  Reference 3.11 (2.50, 3.87) 7.01 (5.74, 8.55) 13.59 (10.85, 17.03)  627 326  16799 16236  37 20  1.79 (1.57, 2.05) Reference  207 170 419  5821 4600 22100  36 37 19  1.93 (1.64, 2.28) 2.08 (1.74, 2.49) Reference  391 7 9 59 96 200  10999 604 837 1709 4632 13619  36 12 11 35 21 15  2.49 (2.10, 2.96) 0.77 (0.36, 1.64) 0.72 (0.37, 1.39) 2.39 (1.79, 3.19) 1.35 (1.06, 1.72) Reference  273 170 104 40 213  6112 4600 5283 3072 12933  45 37 20 13 16  2.78 (2.33, 3.33) 2.35 (1.92, 2.88) 1.16 (0.92, 1.46) 0.74 (0.53, 1.04) Reference  271 169  12043 9259  23 18  1.21 (1.00, 1.47) Reference  156  Table 5.5: Comparisons of diagnosis type and related factors between TB patients who died during follow up period (n=953) and who survived (n=4068)  Variable  Number of death  Total Person Years (PYs)  Death/1000 PYs  Unadjusted Hazard Ratio (95% CI)  Overall  953  33035  29  ‐  938 15  32071 963  29 16  Reference 0.59 (0.36, 0.99)  119 834  9385 23650  13 35  Reference 2.50 (2.06, 3.02)  779 50 5  21594 2092 143  36 24 35  Reference 0.68 (0.51, 0.91) 0.85 (0.35, 2.04)  703 185 65  21604 9905 1526  33 19 43  Reference 0.59 (0.50, 0.70) 1.23 (0.95, 1.58)  840 113  32254 781  26 145  Reference 4.40 (3.62, 5.36)  933 20  32656 378  29 53  Reference 1.69 (1.09, 2.64)  909 44  32188 847  28 52  Reference 2.16 (1.59, 2.92)  848 305  29645 3390  28 90  Reference 1.33 (1.08, 1.63)  904 49  31945 1090  29 45  Reference 1.09 (0.82, 1.45)  Case type New Recurrent Culture results Negative/no culture Positive Drug resistance status Sensitive Mono/poly resistance MDR Disease type of current episode Pulmonary Extra‐pulmonary Both Miliary/disseminated TB No Yes CNS/Meningeal TB No Yes Far advanced pulmonary TB No Yes Moderately advanced pulmonary TB No Yes Cavitary TB No Yes  157  Table 5.6: Comparisons of co‐morbidities between TB patients who died during follow up period (n=953) and those who survived (n=4068)  Variable  Number of death  Total Person Years (PYs)  Death/1000 PYs  Unadjusted Hazard Ratio (95% CI)  Overall  953  33035  29  ‐  861 92  31079 1956  28 47  Reference 1.61 (1.30, 1.99)  920 33  32753 282  28 117  Reference 3.49 (2.46, 4.94)  848 105  31414 1621  27 65  Reference 2.36 (1.93, 2.89)  867 86  32119 916  27 94  Reference 2.87 (2.30, 3.58)  824 129  31931 1103  26 117  Reference 3.42 (2.83, 4.12)  873 80  32574 460  27 174  Reference 4.96 (3.94, 6.24)  916 37  32446 588  28 63  Reference 2.02 (1.45, 2.80)  914 39  32732 303  28 129  Reference 3.69 (2.68, 5.08)  Diabetes Mellitus No/unknown Yes Malnutrition No/unknown Yes Alcoholism No/unknown Yes Substance abuse No/unknown Yes Either HIV or AIDS No/Unknown Yes Any Malignancy No/unknown Yes Immunosuppressive medication No/unknown Yes Chronic Renal failure No/unknown Yes  Table 5.6 presents unadjusted hazard ratio estimates for behavioral factors and co‐morbid diseases. TB Patients who had a history of alcoholism (Unadjusted HR=2.36) and who used drugs (Unadjusted HR=2.87) were significantly more likely to die. In addition, presence of diabetes mellitus (Unadjusted HR=1.61), malnutrition (Unadjusted HR=3.49), HIV/AIDS (Unadjusted HR=3.42), malignancy (Unadjusted HR=4.96), immunosuppressive medication (Unadjusted HR=2.02) and renal failure (Unadjusted HR=3.69) were also significantly associated with mortality among TB patients. 158  Table 5.7: Multivariable cox regression analysis for risk factors of mortality among TB patients Variable  Model 1 (n=4811) Adjusted HR (95% CI)  Model 2 (n=4684) Adjusted HR (95% CI)  Age at diagnosis (per decade)  1.06 (1.05, 1.06)  1.06 (1.05, 1.06)  Male gender Birth place Foreign‐born Canadian‐born non‐Aboriginal Canadian‐born Aboriginal Ethnicity Caucasian Aboriginals Southeast Asian Others Chinese/Filipino/Vietnamese Recurrent TB Culture confirmed TB Miliary /disseminated TB CNS TB Far advanced pulmonary TB Moderately advanced pulmonary TB Either HIV or AIDS30 Any malignancy31 Chronic renal failure32 Diabetes mellitus33 Immunosuppressive medication34 Malnutrition35 Alcoholism36 Drug abuse37  1.42 (1.22, 1.65)  1.45 (1.25, 1.69)  Reference 1.70 (1.42, 2.04) 2.65 (2.16, 3.26)  0.73 (0.43, 1.22) 1.42 (1.14, 1.77) 3.18 (2.48, 4.06) 1.43 (0.80, 2.56) 1.81 (1.29, 2.55) 1.06 (0.85, 1.33) 3.53 (2.70, 4.62) 2.58 (1.99, 3.35) 2.15 (1.40, 3.31) 1.14 (0.91, 1.44) 1.45 (1.00, 2.10) 1.07 (0.71, 1.63) 1.49 (1.16, 1.91) 1.95 (1.43, 2.65)  1.72 (1.42, 2.08) 2.60 (2.07, 3.27) 0.85 (0.67, 1.08) 1.01 (0.72, 1.43) Reference 0.65 (0.36, 1.15) 1.30 (1.04, 1.61) 3.40 (2.67, 4.34) 1.63 (0.95, 2.78) 1.99 (1.44, 2.77) 1.09 (0.87, 1.36) 3.55 (2.72, 4.64) 2.36 (1.81, 3.07) 2.14 (1.39, 3.29) 1.22 (0.97, 1.55) 1.35 (0.93, 1.95) 1.29 (0.87, 1.93) 1.50 (1.17, 1.93) 1.89 (1.39, 2.58)  30  ‐ Patients with negative HIV or unknown status was used as reference group ‐Patients without disease condition or unknown status was used as the reference group 32 ‐ Patients without disease condition or unknown status was used as the reference group 33 ‐ Patients without disease condition or unknown status was used as the reference group 34 ‐ Patients without disease condition or unknown status was used as the reference group 35 ‐ Patients without disease condition or unknown status was used as the reference group 36 ‐ Patients without disease condition or unknown status was used as the reference group 37 ‐ Patients without disease condition or unknown status was used as the reference group 31  159  The adjusted HR estimates for mortality among TB patients are presented in Table 5.7. Patterns of significant risk factors were similar in both multivariable Cox models (Model 1 and Model 2). In multivariable Cox regression models (Model 1), significant risk factors for mortality included: increasing age (Adjusted HR=1.06 per year), male gender (Adjusted HR=1.42), Canadian‐born non‐aboriginal (Adjusted HR=1.70), Aboriginal people (Adjusted HR=2.65), culture confirmed TB (Adjusted HR=1.42), Miliary TB (Adjusted HR=3.18), far advanced PTB (Adjusted HR=1.81), HIV/AIDS (Adjusted HR=3.53), malignancy (Adjusted HR=2.58), renal failure (Adjusted HR=2.15), history of alcoholism (Adjusted HR=1.49), and substance abuse (Adjusted HR=1.95). Caucasian (Adjusted HR=1.72) also had an increased risk of mortality (Model 2). The effects of recurrence, moderately advanced PTB, CSN/meningeal TB, diabetes mellitus, immunosuppressive medication and malnutrition were not significant in multivariable models.  The frequency of missing observations was relatively higher for several co‐morbidities. The mortality risk associated with each co‐morbid condition was evaluated separately with univariate and multivariate setting. For this analysis, only TB patients with valid responses for the corresponding comorbidity status were considered. The results of this sensitivity analysis is presented in Appendix A (Table A.9 and Table A.10).  The risk factors for mortality were further investigated among TB patients who died during treatment (comparison group: TB patients who did not die during treatment). The results of this additional analysis are presented in Table 5.8. Overall, the variables which were significant in the multivariable setting were similar to the primary analysis (Table 5.6). Increasing age (AOR: 1.06 per year), male gender (AOR: 1.64), Aboriginal people (AOR: 1.96), culture confirmed TB (AOR: 2.13), disseminated TB (AOR: 4.64), HIV/AIDS (AOR: 5.62), malignancy (AOR: 4.01), malnutrition (AOR: 2.34), renal failure (AOR: 4.18), alcoholism (AOR: 1.83) and drug abuse (AOR: 2.24) were significant risk factors in multivariable models.  160  Table 5.8: Univariate and multivariable logistic regression analysis for risk factors of mortality for patients who died during treatment (n=309) and who did not (n=4904)  Variable  Age at diagnosis Male gender Birth place Foreign‐born Canadian‐born non‐Aboriginal Canadian‐born Aboriginal Birth Country regions PAHO African Region Eastern Mediterranean European Region Southeast Asia Region Western Pacific Region Ethnicity Caucasian Aboriginals Southeast Asian Others Chinese/Filipino/Vietnamese Recurrent TB Culture confirmed TB Drug resistance status Sensitive Mono/poly resistance MDR Miliary/disseminated TB CNS/Meningeal TB Far advanced PTB Moderately advanced PTB Either HIV or AIDS38 Any Malignancy39 Chronic renal failure40 Immunosuppressive medication41  Patients died during Treatment (n=5213)  Patients died during Treatment (n=5073)  Patients died during Treatment (n=4934)  Unadjusted Odds Ratio (95% CI)  Adjusted Odds Ratio (95% CI)  Adjusted Odds Ratio (95% CI)  1.05 (1.04, 1.05) 2.03 (1.58, 2.61)  1.06 (1.05, 1.07) 1.59 (1.18, 2.14)  1.06 (1.05, 1.07) 1.64 (1.20, 2.23)  Reference 1.40 (1.00, 1.95) 2.14 (1.56, 2.94)  Reference 1.08 (0.74, 1.58) 2.12 (1.40, 3.20)  ‐  ‐  ‐  ‐  1.13 (0.77, 1.66) 2.73 (1.56, 4.78)  1.12 (0.77, 1.64) 1.96 (1.24, 3.09) 0.74 (0.46, 1.18) 0.67 (0.35, 1.30) Reference 1.04 (0.70, 1.56) 2.13 (1.25, 3.63)  3.90 (2.45, 6.23)  4.64 (2.91, 7.40)  5.26 (3.14, 8.81) 4.26 (2.70, 6.73) 4.72 (2.46, 9.06) 1.12 (0.55, 2.30)  5.62 (3.31, 9.56) 4.01 (2.52, 6.38) 4.18 (2.12, 8.24) 1.38 (0.69, 2.73)  2.03 (1.51, 2.73) 0.86 (0.27, 2.78) 0.59 (0.19, 1.90) 1.72 (0.99, 2.99) 0.84 (0.52, 1.35) Reference 1.95 (1.39, 2.73) 2.26 (1.59, 3.21) 0.86 (0.55, 1.33) 0.55 (0.29, 1.00) Reference 1.54 (1.09, 2.18) 5.63 (3.44, 9.22) Reference 0.59 (0.35, 1.00) 1.58 (0.55, 4.50) 5.78 (4.06, 8.24) 1.26 (0.51, 3.17) 1.79 (0.92, 3.47) 1.29 (0.87, 1.92) 4.43 (3.20, 6.13) 6.03 (4.05, 8.97) 7.29 (4.45, 11.93) 2.40 (1.38, 4.18)  38  ‐ Patients with negative or unknown HIV status was used as reference group ‐ Patients without malignancy or unknown status was used as reference group 40 ‐ Patients without renal failure or unknown status was used as reference group 41 ‐ Patients without immune‐suppressive medication or unknown status was used as reference group 39  161  Variable Diabetes mellitus42 Malnutrition43 Alcoholism44 Drug abuse45  Patients died during Treatment (n=5213)  Patients died during Treatment (n=5073)  Patients died during Treatment (n=4934)  Unadjusted Odds Ratio (95% CI)  Adjusted Odds Ratio (95% CI)  Adjusted Odds Ratio (95% CI)  1.45 (0.97, 2.18) 5.41 (3.04, 9.63) 2.76 (1.93, 3.93) 3.57 (2.42, 5.27)  1.46 (0.67, 3.16) 1.66 (1.02, 2.69) 2.27 (1.28, 4.03)  2.34 (1.16, 4.70) 1.83 (1.13, 2.97) 2.24 (1.25, 4.02)  5.4 DISCUSSION This study demonstrates that the overall crude death rate was 20% in a population‐based TB cohort over a period of 17 years. When stratified by treatment, the death before treatment category was 2.3% and during treatment category was 5.7%. The combined death rate (before or during treatment) was 9.0%. The Kaplan Meier survival analysis demonstrated that the cumulative mortality at 6 and 12 months were 8% and 10% respectively (overall mortality was 29 per 1000 PYs). In a previous study (26) that was conducted in BC for the period between 1980‐1984 a total of 201 deaths were identified among 1884 TB patients (CDR= 201/1881=10.6%). The current mortality rate is slightly lower compared to this previous BC study. The CDR varies widely across studies, mostly due to differences in background incidence, setting, methodology and observation period. However, the morality rate in the current study was consistent with studies conducted in similar settings. For instance, a population‐based study in Netherlands (12) in the period 1993‐1995 reported a mortality of 7% (258/4340) during one year of follow‐up period while on treatment and 2% (105/4981) mortality at the time of diagnosis. A recent study in Queensland, Australia, which has similar TB epidemiological feature to Canada have shown that 12.7% of (127/1003=12.7%) of notified cases die before completing treatment (27). The overall mortality in Australian study was slightly higher compared to our 42  ‐ Patients without diabetes mellitus or unknown status was used as reference group ‐ Patients without malnutrition or unknown status was used as reference group 44 ‐ Patients without alcoholism or unknown status was used as reference group 45 ‐ Patients without drug abuse or unknown status was used as reference group 43  162  study (9%), but comparable to the prior BC study (10.6 %.). Another population‐based study in UK (28) reported a CFR (all‐cause) of 12.9% (158/1222) during 1980s among patients with pulmonary TB while on treatment during a period of two years. This study also observed 50 deaths (50/1693=3%) who were diagnosed with TB on post‐mortem examination, slightly higher compared to the current study. A Canadian study reported an all‐cause mortality rate of 12% among hospitalized TB patients (23). A USA study has shown an all‐cause mortality rate of 15% (31/212) among patients with extra‐pulmonary TB at one year (24).  This study demonstrated an excess mortality among TB patients compared with the general population. In terms of crude death rate, TB patients had 26 (median) times higher mortality compared to the general population of BC. Even controlling for age, the standardized mortality ratio was much higher compared to the general population of BC. In the current study, the age standardized mortality ratio (ASMR) which was estimated for each calendar year (1990 to 2006) ranged from 9 to 29‐ lowest in 1990 and 1991 and highest in 2002. The median ASMR was 16. Overall, the ASMR showed an increasing trend from 1990 to 2004 and had a peak in the period 1999‐2004. During 1999‐2004, each calendar year had an ASMR over 20. This excess mortality during 1999‐2004 might be related to enhanced notifications of deaths among TB patients and to HIV epidemic which was observed in this province during late 1990s. During 1990s, the incidence of HIV in BC was very high which started to decrease in the following decade (29). For example, the HIV incidence in BC was 22.7 in 1992, which fell to 9.4 in 2005 (29). In addition, over‐representation by elderly and Aboriginal people, presence of concurrent disease and increased frequency of drug abuse and alcoholism in certain ethnic groups (Aboriginal people and Caucasians) could be other potential explanations for this excess mortality among TB patients in this province.  A small number of studies have also compared mortality among TB patients with the general population and reported SMR as an indicator of excess mortality. However, our SMR was higher compared to the Netherland study although CDR was similar for both studies. The Dutch study showed that TB patients had eight times higher mortality (SMR=8.3) at one year while on 163  treatment compared to the general population (12). This Dutch study excluded patients who died at the time of diagnosis. Another study conducted in England and Wales also reported ten‐ time excess mortality (all‐cause) among patients with pulmonary TB compared to the general population after controlling for the effect of age and gender before the completion of chemotherapy (28). The UK study (28) also excluded TB patients that were diagnosed post‐ mortem. Since the current study included mortality during the post‐treatment period that contributed 50% of the total deaths, the excess mortality in the current study compared to Dutch and UK study (12,28) might be due to inclusion of patients with post‐ treatment mortality. A recent study which was conducted in Chennai, India found six‐times (SMR: 6.1) excess mortality among TB patients compared to the general population at 20 months of follow up period (30). Despite the Indian study having a higher CDR (9.4% at 20 months, our study: 5.7% during treatment), their standardized mortality ratio was much lower compared to this study. The excess mortality among TB patients in this province compared to TB patients in India might be due to differences in observation period and TB epidemiology in two settings. TB in the industrialized nations mostly occurs among vulnerable individuals such as elderly people, patients with co‐morbidities and marginalized populations (31).  This study identified a number of important risk factors of mortality among TB patients in British Columbia, Canada. Increasing age, male gender, Canadian‐born non‐aboriginal status, Aboriginal ethnicity, Caucasian ethnicity, culture confirmed TB, miliary TB, far advanced PTB, HIV/AIDS, malignancy, renal failure, alcoholism and substance abuse were significantly associated with mortality in multivariate analyses.  The increased likelihood of death among elderly people has been shown in previous studies (9,12,28,31‐33). Elderly people are more likely to die from their diseases due to a decreased immune response. In this TB cohort, one third of TB cases were diagnosed when they were more than 60 years or older. Survivors were twenty years younger (at the time of last episode) compared to patients who died. In addition, detection and management of TB among elderly  164  patient is more challenging. A Canadian study conducted among hospitalized patients reported a significant association between initially missed diagnosis and older age (23).  Our study demonstrated a predominance of death among males. This finding is consistent with previous studies. Other studies have reported an increased risk of dying among males (12,28)(30,34). Increased risk of death among males might be attributable to a higher prevalence of concurrent diseases among men and/or due to predominance of risky behavior such as alcoholism and drug abuse among males. Another potential reason might be an increased frequency of more severe forms of TB among males. This was evident in the current study. Both the frequency of drug abuse (10% vs. 6%, p=0.003) and alcoholism (16% vs. 5%, p<0.001) was significantly higher among males compared to females. Several prognostic factors /co‐morbidities such as HIV/AIDS (13% vs. 7%, p<0.001), malignancy (7% vs. 4%, p=0.002) and malnutrition (4% vs. 2%, p=0.039) were significantly more frequent among males. More severe forms of TB that were significantly associated with greater mortality were more likely to occur in males (miliary TB: 5% vs. 4%, p=0.068; meningeal TB: 2% vs. 1%, p=0.251; far advanced PTB: 3% vs. 1%; p<0.001 and moderately advanced PTB: 9% vs. 6%, p=0.001).  Our study demonstrated an increased likelihood of dying among non‐aboriginal Canadian‐born TB patients compared to foreign‐born subjects even after adjusting for the effects of age, gender and other factors. Despite an over‐representation of foreign‐born people in the TB cohort, this finding was not unexpected. Similar data has also been shown in many studies conducted in industrialized countries with a high level of immigration. In the Dutch study, the risk of dying was significantly lower among the non‐Dutch population compared with the Dutch (12). The UK study (28) showed a very low mortality rate among TB patients who had immigrated from the Indian sub‐continental (3.4%, overall: 12.9%). Another study in Lorraine, France has shown that French origin (compared to foreign‐born) as a risk factor of mortality during treatment (32).  165  Several reasons such as life style and absence of concurrent diseases might be associated with the decreased risk of mortality among foreign‐born population. The frequency of drug abuse (10% vs. <1%) and alcoholism (10% vs. 1%) was much higher among Canadian‐born non‐ aboriginal subjects compared to the foreign‐born (Table A.11). An increased prevalence of HIV/AIDS was also evident in this study (13% vs. 2%). In addition, the prevalence of other concurrent diseases such as malignancy (5% vs. 2%) and malnutrition (2% vs. <1%) was more frequent among Canadian‐born non‐aboriginal people compared to foreign‐born people. The healthy migration effect might be another potential explanation for the reduced risk of mortality among the foreign‐born. Despite immigrants coming from high prevalence countries, immigrants are not true representatives of their birth country. Immigrants have a number of positive factors that distinguishes then from non immigrants and makes them less likely to die. This was evident from the current data as the average age of death for the Canadian‐born non‐ aboriginal people was twelve years lower (62.8 vs. 74.5 years) compared to foreign‐born people. Even the average age of death for the general population of BC (range: 70.2 ‐73.9 years during 1990‐2006) was lower than the average age of death of foreign‐born TB patients. The immigration surveillance for newly arrived immigrants might lead to earlier detection of disease, thus eventually leading to better prognosis.  This study demonstrated a disproportionate increased rate of mortality among TB patients in several ethnic subgroups. Aboriginals had the highest level of mortality among all the ethnic groups. When compared to Chinese/Vietnamese/Filipino, Aboriginals had three times more likelihood to die. Not only that, the mean age of death for aboriginal people was 55 years, 8 years earlier compared to Canadian‐born non‐aboriginal and 20 years earlier compared to foreign‐born subjects with TB. Aboriginal people were over‐presented in younger age (at death) group – 20% of Aboriginal people were dying before the age of 41 compared to 14% among Canadian‐born non‐aboriginal and 5% foreign‐born people. In addition, TB was diagnosed at much earlier age among Aboriginal communities (average age of diagnosis was 41 years for Aboriginals, 46 years for CB non‐aboriginals and 50 years for foreign‐born). The prior BC study also reported a significantly elevated risk of mortality among indigenous TB patients (26). On 166  the other hand, the Australian study reported that CFR did not differ significantly among ethnic groups and no increased risk of TB‐related mortality among Indigenous Australians (27).  The excess mortality among Aboriginal people might be attributable to life style that increases their vulnerability to death. In the current TB cohort, prevalence of drug abuse and alcoholism among aboriginal people was 15% and 23% respectively, which were higher compared to Canadian‐born non‐aboriginal (10% and 10%) and much higher compared to the foreign‐born (1% and <1%). Overall, Canadian aboriginal people were over‐represented in HIV epidemic (35,36), which could be related to increased frequency of substance abuse, eventually leading to a worse prognosis. Despite universal access to health care, Aboriginal people were also over‐ represented in marginalized populations who were also hard to reach. As result, the early detection of diseases becomes more challenging in Aboriginal population (might be related to failure in seeking care). Strengthening of TB control initiatives targeting Aboriginal Canadians should be introduced immediately in order to reduce the risk of premature death. Culturally more appropriate interventions in addition to the current TB care might be valuable in this setting.  This study identified an increased likelihood of death among Caucasians. This association is not unexpected. Studies conducted in similar settings have shown an increased risk among Caucasians. The UK study (28) showed a higher mortality among Caucasians compared to patients who had immigrated from the Indian sub‐continental (16% vs. 3.4%, p=0.09). However, the results of the current study contrasts with previous mortality study from BC, which showed an increased risk of dying among non‐white (mostly Asians) compared to Caucasians. Black and Hispanic had the highest mortality rate in a USA study (37). Mortality among Caucasians was 45 per 1000 PYs, even higher than that Aboriginal people in the current study. The increased risk of death among Caucasians might be related to the predominance of males and elderly population as well as increased presence of co‐morbid conditions in the Caucasian population. In this study, among Caucasians, 68% were male and 41% was more than 60 years old at the time diagnosis. Prevalence of drug abuse (18% vs. <1%), alcoholism (20% vs. 2%), HIV/AIDS (21% vs. 167  2%) and malignancy (13% vs. 5%) was significantly higher among Caucasians compared to Asian. Miliary TB (5% vs. 3%), far advanced TB (4% vs. 2%) and moderately advanced TB (13% vs. 5%) were significantly more frequent among Caucasians compared to Asian subjects.  Severe forms of TB, disseminated/miliary TB and far advanced TB were significant risk factors for mortality among TB patients in the current study. Other studies conducted in similar setting confirmed this association (24,38,39). For example, disseminated TB was associated with a poor prognosis among patients with extra‐pulmonary TB in a study conducted at an inner‐city USA hospital (24). Previous studies reported a very high short‐term mortality among patients with disseminated (20‐86%) (24,38‐41) and meningeal (22‐70%) TB (24,42‐45). In the current study, the death rate was very high among patients with miliary TB (145 deaths per 1000 PYs and CFR was 50%). Due to atypical presentation, detection of disseminated TB is often difficult, which increases the likelihood of a missed or delayed diagnosis, eventually placing the TB patients at an increased risk of dying without treatment. Disease severity for pulmonary TB, which was measured by radiological extent of disease, (2,3,28) has been found to be significant in other studies (28). Advanced TB (bilateral lung involvement and cavitary lesions) was a significant predictor of mortality in multivariable model in the study conducted in Samara, Russia (46). A high index of suspicion, for earlier diagnosis and subsequent initiation of chemotherapy before the diagnosis is confirmed would be valuable under these circumstances. Awareness including increased education regarding the atypical presentation and the increased mortality related to miliary and other severe forms of TB should be considered in order to prevent morbidity and mortality from these devastating types of TB.  Our study demonstrated that patients with culture‐confirmed TB had an increased risk of dying. This finding was consistently observed in multivariate model as well as in univariate model conducted for each treatment group. Although few studies investigated this type of association, the relationship between culture‐confirmed TB and mortality was not clear‐cut. The bacteriological confirmation was not significant in multivariable setting in the Dutch study. In a study from Guinea‐Bissau, a positive smear result was negatively associated with mortality (9). 168  The relationship between positive smear result and mortality was insignificant in South African studies using a multivariate analysis. (47,48). However, the study conducted in India (30) identified smear positive disease as an independent risk factor for TB mortality (all cause, on treatment). A study in Tomsk Oblast has shown that culture positivity was associated with mortality (31). The presence of AFB in sputum was positively associated with all‐cause mortality in the prior BC study (26).  Furthermore, this association might be due to the increased frequency of severe forms of TB among culture positive patients and an increased presence of co‐morbidity. This was evident in the current data. The frequency of miliary TB (5% vs. 2%, p=<0.001), meningeal TB (2% vs. 1%, p=0.006), far advanced PTB (2% vs. <1%; p<0.001) and moderately advanced PTB (9 % vs. 3%, p<0.001) was significantly higher among culture positive patients compared to TB patients with no culture or negative culture results. The prevalence of drug abuse (10% vs. 2%, p<0.001) and alcoholism (13% vs. 5%, p=<0.001) as well as other prognostic factors such as chronic renal failure (4% vs. 1%, p=0.002), malnutrition (3% vs. 1%, p=0.001) and HIV/AIDS (12% vs. 2%, p<0.001) was significantly higher among culture‐confirmed TB cases in comparison to patients with negative cultures.  In the current study, multi‐drug resistance (MDR) was not associated with mortality in univariate Cox regression analysis. The small number of MDR cases in this cohort as well as a low rate of resistance among Aboriginals or Canadian‐born non‐aboriginal likely contributes to this result. Although several studies have shown MDR as an important prognostic factor for death among TB patients (16,31,49), other studies did not find a significant association between drug resistance and mortality. For examples, the Australian study also reported no increased risk of mortality among patients with MDR (27). MDR‐TB did not have a significant impact on mortality in a US study (24). Another study in high burden setting like South Africa demonstrated no increased risk associated with drug resistance, especially with MDR (47,48).  169  The current study demonstrated a significantly increased likelihood of death among HIV positive patients (117 deaths per 1000 PYs). The increased risk of death associated with HIV has been consistently observed in previous studies (9,10,12,32,47,50,51). A Dutch study observed a tenfold increased risk of dying among HIV/TB co‐infected patients. The effect of HIV was more pronounced in most African studies, where the prevalence of HIV/TB co‐infection was considerably higher (9,47,48,52‐55). For examples, a study in South Africa (47) reported that HIV positive patients were fifteen times more likely to die from TB compared to HIV negative TB patients. In another prospective cohort study in South Africa, mortality at 6 months (all‐cause) was 13.7% among HIV positive patients and was 0.5% among HIV‐negative patients (48).  In this study, 45% of all HIV co‐infected subjects died and the mean age of these patients at death was 42 years. The high mortality among HIV/TB co‐infected patients is related to the immune ‐suppression caused by HIV itself. As a result, HIV/AIDS is considered as an important prognostic factor for TB patients. Due to its association with immune‐suppression, the HIV epidemic has significantly changed TB epidemiology globally. In North America, morbidity and mortality related to TB has increased substantially since 1990, mostly related to increased prevalence of HIV/TB co‐infection during the post‐HIV era (56). In this study, 5.5% of all TB cases were HIV positive. During 1990s, the incidence of HIV in this province was relatively high (29). In 1992, the HIV incidence in BC was 22.7, which dropped to 9.4 in 2005 (29).  This study demonstrated an increased risk among patients with malignancy and chronic renal failure. As a marker of immune‐suppression, malignancy increases the likelihood of death in TB patient. The increased risk associated with this co‐morbidity was consistently found in other studies. Malignancy was a strong predictor (AHR: 3.1) of mortality in the Dutch study (12). 61% (n=97) of TB patients who had malignancy died during the study period and the mean age of death of these patients was 71 years (IQR: 66‐81 years). Among patients with renal failure, CFR was 47% and mean age of death was 67 years (IQR: 59‐78 years). Deaths among older subjects were largely attributable to co‐morbid conditions such as malignancy and renal failure, rather  170  than drug abuse, alcoholism and HIV/AIDS. However, coexisting chronic diseases were not significant in the USA study (24) possibly due to the small number of cases.  This study demonstrates an increased likelihood of death among TB patients with alcoholism and patients who had history of substance abuse. Similar findings were observed in several studies (46). Alcoholism (AHR=2.4) and substance abuse (AHR=3.5) was associated with increased risk of mortality in the Dutch study (12). The combination of smoking and alcoholism was identified as a risk factor of mortality in Chennai, India study (30). The study in Samara, Russia has shown both substance abuse and alcoholism as significant predictors of mortality (46). Another Russian study in Tomsk Oblast has shown a strong relationship (AHR: 2.7) between alcoholism and mortality (31). The study of White and others in USA has shown that substance abuse including alcoholism could contribute to additional deaths even among TB patients without AIDS (37).  In the current TB cohort, the prevalence of substance abuse and alcoholism was 4% and 8% respectively (including patients with missing information on alcoholism and drug abuse). The case fatality rate was 40% among patients with alcoholism (65 deaths per 1000 PYs) and the mean age of death of these patients was 55 years. The CFR was 41% among TB patients with substance abuse (94 deaths 1000 PYs) and the mean age at death of these patients was 43 years. On the other hand, the mean age of death for the entire TB cohort was 68 years and for the general population during same the time period was about 72 years. Thus TB patients with substance abuse were dying at a much younger age. Although not clearly distinct, a bimodal pattern of TB related mortality in respect to age was evident in the current study‐ one peak at a younger age (40‐50 years) and another expected peak among older subjects (75‐90 years). TB patients dying at a younger age were largely attributable to drug abuse and related vulnerabilities including AIDS. Similar bimodal patterns of age were observed in a USA study (56). So both substance abuse and alcoholism not only increased the risk of mortality, but also increased the likelihood of death at much younger age.  171  According to the national addiction survey in 2004, the prevalence of lifetime and past year illicit drug use (other than cannabis) in BC was 23% and 4% respectively (57). BC has the highest prevalence of illicit drug use compared to other provinces in Canada. Life term harm associated with drug abuse was also higher among British Columbians (57). The frequency of drug use was relatively low in the TB cohort due to over‐representation of foreign‐born people who use illicit drugs very infrequently. There was evidence suggesting an increased morbidity (harm) and mortality among cases that used illicit drug and alcohol (58‐60). Both alcoholism and drug use are subject to modifications. Interventions targeting these high risk populations are urgently required. Establishment of a systematic approach in order to identify TB patients with alcoholism and drug abuse, enrolling them in appropriate drug or alcoholism treatment programs should be prioritized as part of current TB care.  This study is subject to several limitations including incomplete reporting of co‐morbidities (which range from 40% to 52%) and passive nature of follow up (Please see the Table A.1 in the Appendix for expected frequencies of missing observations). However, Cox regression analysis was conducted in two settings for co‐morbidities: with and without missing observations. Patterns of significant variables were similar in both scenarios (please see Appendix A: Table A.8 and Table A.9). Previous studies conducted among TB contact in this province demonstrated a low level of migration among TB patients (61). We therefore do not consider that missing observations and migration effects would bias the HR estimates substantially.  5.5 CONCLUSIONS In conclusion, this study has demonstrated that a relatively higher excess mortality exists in the TB cohort compared to the general population of BC. TB patients die at a younger age compared to the general population. Aboriginal people had not only one of the highest rates of mortality but also died at a much younger age. Deaths at a young age in TB patients were largely attributable to behavioral factors such as by drug abuse and alcoholism and HIV/AIDS, whereas deaths at an older age were associated with other co‐morbid‐conditions including 172  malignancy and renal failure. In addition, severity of disease, Caucasian ethnicity, older age and male gender were significantly associated with greater mortality. These findings suggest that TB patients in this province require more rigorous interventions targeting high risk populations than are not routinely provided through conventional TB programs.  173  5.5 REFERENCES (1) Raviglione MC, Snider DE,Jr, Kochi A. Global epidemiology of tuberculosis. Morbidity and mortality of a worldwide epidemic. JAMA 1995 Jan 18;273(3):220‐226. (2) Abos‐Hernandez R, Olle‐Goig JE. Patients hospitalised in Bolivia with pulmonary tuberculosis: risk factors for dying. International Journal of Tuberculosis & Lung Disease 2002 Jun;6(6):470‐474. (3) Olle‐Goig JE. Patients with tuberculosis in Bolivia: why do they die?. Pan American Journal of Public Health 2000 Sep;8(3):151‐155. (4) Dye C, Scheele S, Dolin P, Pathania V, Raviglione MC. Consensus statement. Global burden of tuberculosis: estimated incidence, prevalence, and mortality by country. WHO Global Surveillance and Monitoring Project. JAMA 1999 Aug 18;282(7):677‐686. (5) Dolin PJ, Raviglione MC, Kochi A. Global tuberculosis incidence and mortality during 1990‐ 2000. Bull.World Health Organ. 1994;72(2):213‐220. (6) World Health Organization. Tuberculosis Facts 2009 Update. 2009; Available at: http://www.who.int/tb/publications/2009/factsheet_tb_2009update_dec09.pdf. Accessed April/02, 2010. (7) Kochi A. The global tuberculosis situation and the new control strategy of the World Health Organization. 1991. Bull.World Health Organ. 2001;79(1):71‐75. (8) Kochi A. The global tuberculosis situation and the new control strategy of the World Health Organization. Tubercle 1991 Mar;72(1):1‐6. (9) Gustafson P, Gomes VF, Vieira CS, Samb B, Naucler A, Aaby P, et al. Clinical predictors for death in HIV‐positive and HIV‐negative tuberculosis patients in Guinea‐Bissau. Infection 2007 Apr;35(2):69‐80. 174  (10) Connolly C, Davies GR, Wilkinson D. Impact of the human immunodeficiency virus epidemic on mortality among adults with tuberculosis in rural South Africa, 1991‐1995. International Journal of Tuberculosis and Lung Disease 1998;2(11):919‐925. (11) Rao VK, Iademarco EP, Fraser VJ, Kollef MH, Rao VK, Iademarco EP, et al. The impact of comorbidity on mortality following in‐hospital diagnosis of tuberculosis. Chest 1998 Nov;114(5):1244‐52. (12) Borgdorff MW, Veen J, Kalisvaart NA, Nagelkerke N. Mortality among tuberculosis patients in the netherlands in the period 1993‐1995. European Respiratory Journal 1998 Apr;11(4):816‐ 820. (13) Harries AD, Nyangulu DS, Kang'ombe C, Ndalama D, Glynn JR, Banda H, et al. Treatment outcome of an unselected cohort of tuberculosis patients in relation to human immunodeficiency virus serostatus in Zomba Hospital, Malawi. Transactions of the Royal Society of Tropical Medicine & Hygiene 1998 May‐Jun;92(3):343‐347. (14) Glynn JR, Warndorff DK, Fine PE, Munthali MM, Sichone W, Ponnighaus JM. Measurement and determinants of tuberculosis outcome in Karonga District, Malawi. Bull.World Health Organ. 1998;76(3):295‐305. (15) van den Broek J, Mfinanga S, Moshiro C, O'Brien R, Mugomela A, Lefi M. Impact of human immunodeficiency virus infection on the outcome of treatment and survival of tuberculosis patients in Mwanza, Tanzania. International Journal of Tuberculosis & Lung Disease 1998 Jul;2(7):547‐552. (16) Pablos‐Mendez A, Sterling TR, Frieden TR. The relationship between delayed or incomplete treatment and all‐cause mortality in patients with tuberculosis. JAMA 1996 Oct 16;276(15):1223‐1228.  175  (17) Zahar JR, Azoulay E, Klement E, De Lassence A, Lucet JC, Regnier B, et al. Delayed treatment contributes to mortality in ICU patients with severe active pulmonary tuberculosis and acute respiratory failure. Intensive Care Med. 2001 Mar;27(3):513‐520. (18) Mathur P, Sacks L, Auten G, Sall R, Levy C, Gordin F. Delayed diagnosis of pulmonary tuberculosis in city hospitals. Arch.Intern.Med. 1994 Feb 14;154(3):306‐310. (19) Katz I, Rosenthal T, Michaeli D. Undiagnosed tuberculosis in hospitalized patients. Chest 1985 Jun;87(6):770‐774. (20) Sacks LV, Pendle S, Sacks LV, Pendle S. Factors related to in‐hospital deaths in patients with tuberculosis. Archives of Internal Medicine 1998 Sep 28;158(17):1916‐22. (21) Penner C, Roberts D, Kunimoto D, Manfreda J, Long R. Tuberculosis as a primary cause of respiratory failure requiring mechanical ventilation. American Journal of Respiratory & Critical Care Medicine 1995 Mar;151(3 Pt 1):867‐872. (22) Zafran N, Heldal E, Pavlovic S, Vuckovic D, Boe J, Zafran N, et al. Why do our patients die of active tuberculosis in the era of effective therapy? Tubercle & Lung Disease 1994 Oct;75(5):329‐ 33. (23) Greenaway C, Menzies D, Fanning A, Grewal R, Yuan L, FitzGerald JM. Delay in diagnosis among hospitalized patients with active tuberculosis ‐ Predictors and outcomes. American Journal of Respiratory and Critical Care Medicine 2002 01 Apr;165(7):927‐933. (24) Kourbatova EV, K. LM,Jr, Romero J, Kraft C, del Rio C, Blumberg HM, et al. Risk factors for mortality among patients with extrapulmonary tuberculosis at an academic inner‐city hospital in the US. European Journal of Epidemiology 2006;21(9):715‐21. (25) de Meer G, van Geuns HA. Rising case fatality of bacteriologically proven pulmonary tuberculosis in The Netherlands. Tubercle & Lung Disease 1992 Apr;73(2):83‐86.  176  (26) Xie HJ, Enarson DA, Chao CW, Allen EA, Grzybowski S, Xie HJ, et al. Deaths in tuberculosis patients in British Columbia, 1980‐1984. Tubercle & Lung Disease 1992 Apr;73(2):77‐82. (27) Walpola HC, Siskind V, Patel AM, Konstantinos A, Derhy P. Tuberculosis‐related deaths in Queensland, Australia, 1989‐1998: characteristics and risk factors. International Journal of Tuberculosis & Lung Disease 2003 Aug;7(8):742‐750. (28) Cullinan P, Meredith SK, Cullinan P, Meredith SK. Deaths in adults with notified pulmonary tuberculosis 1983‐5. Thorax 1991 May;46(5):347‐50. (29) BC Center for Disease Control. Annual Update Report, HIV and AIDS, 2008. 2008; Available at: http://www.bccdc.ca/NR/rdonlyres/8D247951‐2207‐46AB‐AFF5‐ 9FBC7467DC9F/0/STI_HIVReport_HIVAIDSUpdate2008_20090909.pdf. Accessed March/31, 2010. (30) Kolappan C, Subramani R, Karunakaran K, Narayanan PR. Mortality of tuberculosis patients in Chennai, India. Bull.World Health Organ. 2006;84(7):555‐560. (31) Mathew TA, Ovsyanikova TN, Shin SS, Gelmanova I, Balbuena DA, Atwood S, et al. Causes of death during tuberculosis treatment in Tomsk Oblast, Russia. International Journal of Tuberculosis and Lung Disease 2006 Aug;10(8):857‐863. (32) Rabaud C, Engozogho M, Dailloux M, Hoen B, May T, Canton P, et al. Tuberculosis in Lorraine, France: study of prognostic factors. International Journal of Tuberculosis & Lung Disease 1997 Jun;1(3):246‐9. (33) Barker RD, Nthangeni ME, Millard FJ, Barker RD, Nthangeni ME, Millard FJC. Is the distance a patient lives from hospital a risk factor for death from tuberculosis in rural South Africa? International Journal of Tuberculosis & Lung Disease 2002 Feb;6(2):98‐103.  177  (34) Akksilp S, Karnkawinpong O, Wattanaamornkiat W, Viriyakitja D, Monkongdee P, Sitti W, et al. Antiretroviral therapy during tuberculosis treatment and marked reduction in death rate of HIV‐infected patients, Thailand. Emerging Infectious Diseases 2007 Jul;13(7):1001‐1007. (35) Public Health Agency of Canada. HIV/AIDS in Canada‐ Surveillance Report to June 30, 2006. 2006; Available at: http://www.phac‐aspc.gc.ca/publicat/aids‐sida/haic‐vsac0606/pdf/haic‐ vsac0606.pdf. Accessed March/31, 2010. (36) Public Health Agency of Canada. HIV/AIDS Epi Updates May 2005. 2005; Available at: http://www.phac‐aspc.gc.ca/publicat/epiu‐aepi/epi‐05/pdf/epi_05_e.pdf. Accessed March/31, 2010. (37) White MC, Portillo CJ, White MC, Portillo CJ. Tuberculosis mortality associated with AIDS and drug or alcohol abuse: analysis of multiple cause‐of‐death data. Public Health 1996 May;110(3):185‐9. (38) Martineau AR, Lowey H, Tocque K, Davies PD. Decreasing tuberculosis case fatality in England and Wales, 1988‐2001. International Journal of Tuberculosis & Lung Disease 2004 Jun;8(6):737‐742. (39) Long R, O'Connor R, Palayew M, Hershfield E, Manfreda J. Disseminated tuberculosis with and without a miliary pattern on chest radiograph: a clinical‐pathologic‐radiologic correlation. International Journal of Tuberculosis & Lung Disease 1997 Feb;1(1):52‐58. (40) Maartens G, Willcox PA, Benatar SR. Miliary tuberculosis: rapid diagnosis, hematologic abnormalities, and outcome in 109 treated adults. Am.J.Med. 1990 Sep;89(3):291‐296. (41) Vijayan VK. Disseminated tuberculosis. J.Indian Med.Assoc. 2000 Mar;98(3):107‐109. (42) Porkert MT, Sotir M, Parrott‐Moore P, Blumberg HM. Tuberculous meningitis at a large inner‐city medical center. Am.J.Med.Sci. 1997 Jun;313(6):325‐331.  178  (43) Verdon R, Chevret S, Laissy JP, Wolff M. Tuberculous meningitis in adults: review of 48 cases. Clinical Infectious Diseases 1996 Jun;22(6):982‐988. (44) Yechoor VK, Shandera WX, Rodriguez P, Cate TR. Tuberculous meningitis among adults with and without HIV infection. Experience in an urban public hospital. Arch.Intern.Med. 1996 Aug 12‐26;156(15):1710‐1716. (45) Ogawa SK, Smith MA, Brennessel DJ, Lowy FD. Tuberculous meningitis in an urban medical center. Medicine 1987 Jul;66(4):317‐326. (46) Kourbatova EV, Borodulin BE, Borodulina EA, del Rio C, Blumberg HM, K. LM,Jr, et al. Risk factors for mortality among adult patients with newly diagnosed tuberculosis in Samara, Russia. International Journal of Tuberculosis & Lung Disease 2006 Nov;10(11):1224‐30. (47) Churchyard GJ, Kleinschmidt I, Corbett EL, Murray J, Smit J, De Cock KM. Factors associated with an increased case‐fatality rate in HIV‐infected and non‐infected South African gold miners with pulmonary tuberculosis. International Journal of Tuberculosis and Lung Disease 2000;4(8):705‐712. (48) Murray J, Sonnenberg P, Shearer SC, Godfrey‐Faussett P. Human immunodeficiency virus and the outcome of treatment for new and recurrent pulmonary tuberculosis in African patients. American Journal of Respiratory and Critical Care Medicine 1999;159(3):733‐740. (49) Garcia‐Garcia Mde L, Ponce‐De‐Leon A, Garcia‐Sancho MC, Ferreyra‐Reyes L, Palacios‐ Martinez M, Fuentes J, et al. Tuberculosis‐related deaths within a well‐functioning DOTS control program. Emerging Infectious Diseases 2002 Nov;8(11):1327‐33. (50) Sterling TR, Zhao Z, Khan A, Chaisson RE, Schluger N, Mangura B, et al. Mortality in a large tuberculosis treatment trial: Modifiable and non‐modifiable risk factors. International Journal of Tuberculosis and Lung Disease 2006 May;10(5):542‐549.  179  (51) Zachariah R, Spielmann MP, Harries AD, Salaniponi FML. Moderate to severe malnutrition in patients with tuberculosis is a risk factor associated with early death. Trans.R.Soc.Trop.Med.Hyg. 2002;96(3):291‐294. (52) De Cock KM, Gnaore E, Adjorlolo G, Braun MM, Lafontaine MF, Yesso G, et al. Risk of tuberculosis in patients with HIV‐I and HIV‐II infections in Abidjan, Ivory Coast. BMJ 1991 Mar 2;302(6775):496‐499. (53) Elliott AM, Halwiindi B, Hayes RJ, Luo N, Mwinga AG, Tembo G, et al. The impact of human immunodeficiency virus on mortality of patients treated for tuberculosis in a cohort study in Zambia.[see comment]. Transactions of the Royal Society of Tropical Medicine & Hygiene 1995 Jan‐Feb;89(1):78‐82. (54) Ackah AN, Coulibaly D, Digbeu H, Diallo K, Vetter KM, Coulibaly IM, et al. Response to treatment, mortality, and CD4 lymphocyte counts in HIV‐infected persons with tuberculosis in Abidjan, Cote d'Ivoire. Lancet 1995 Mar 11;345(8950):607‐610. (55) Garin B, Glaziou P, Kassa‐Kelembho E, Yassibanda S, Mbelesso P, Morvan J. High mortality rates among patients with tuberculosis in Bangui, Central African Republic. Lancet 1997 01 Nov;350(9087):1298. (56) Braun MM, Cote TR, Rabkin CS. Trends in death with tuberculosis during the AIDS era. JAMA 1993 Jun 9;269(22):2865‐2868. (57) Health Canada. Canadian Addiction Survey (CAS), Detailed Report, March 2005. 2005; Available at: http://www.ccsa.ca/2005%20CCSA%20Documents/ccsa‐004028‐2005.pdf. Accessed March/31, 2010. (58) Oyefeso A, Ghodse H, Clancy C, Corkery J, Goldfinch R. Drug abuse‐related mortality: a study of teenage addicts over a 20‐year period. Social Psychiatry & Psychiatric Epidemiology 1999 Aug;34(8):437‐441.  180  (59) Patra J, Taylor B, Rehm JT, Baliunas D, Popova S. Substance‐attributable morbidity and mortality changes to Canada's epidemiological profile: measurable differences over a ten‐year period. Canadian Journal of Public Health.Revue Canadienne de Sante Publique 2007;98(3):228‐ 234. (60) Paulozzi LJ, Annest JL. US data show sharply rising drug‐induced death rates. Injury Prevention 2007 Apr;13(2):130‐132. (61) A. Morán Mendoza. The value of the tuberculin skin test size in predicting the development of tuberculosis in contacts of active cases. University of British Columbia: University of British Columbia; 2004.  181  CHAPTER 6: A TB­RELATED MORTALITY STUDY IN BRITISH COLUMBIA, CANADA FROM 1990­200646  46  ‐ “A version of this chapter will be submitted to a journal for publication. Moniruzzaman, A. Kazanjian, A. Wong, H. Elwood, R.K. and FitzGerald, J. M. (2010) Patients dying from Tuberculosis ‐ a population‐based study.”  182  6.1 INTRODUCTION TB‐related mortality remains a significant public health challenge locally and internationally. Globally, TB accounts for 6% of all deaths and more than one fourth of all preventable adult deaths (1‐3). In 2008, an estimated 1.8 million people died of TB including 0.5 million deaths among HIV/TB co‐infected patients (4). In pre‐HIV era, the number of deaths caused by Mycobacterium tuberculosis was higher than that of any other infectious agent (5). Since then, TB epidemiology has changed significantly especially in relationship to the HIV epidemic, change in TB‐management practice and emergence of drug resistant diseases (6). However, TB remains the second commonest cause of death due to a single infectious agent (HIV ranks first). In the modern era, all TB related deaths are potentially preventable. This raises an important question, why are there so many deaths related to TB despite the availability of effective chemotherapy? If appropriate interventions are not effectively implemented, many patients with this treatable and preventable disease will continue to have unnecessary morbidity and mortality. It is widely acknowledged that more epidemiological research is required on the etiology and treatment of TB including specific investigations of the determinants of mortality among TB patients.  According to World Health Organization, TB is considered as the cause of death if patients died before completion of chemotherapy (7). This definition which includes all causes of death regardless of the primary cause of death provides an overestimation of TB‐related death. Based on this definition patients who die during treatment due to other causes (such as a motor vehicle accident or co‐morbidity) would be considered to be a TB‐related death. Despite a substantial literature on TB mortality, most of the studies have reported deaths that occurred during the period of treatment (8,9,9‐12), and only a limited number of studies (13‐15) have investigated the specific causes of death among TB patients, as well as identifying risk factors among TB patients whose deaths were attributable to TB. These studies demonstrated that a substantial proportion of deaths while on treatment were not related to TB, rather that the deaths were attributed to other illnesses or causes. Evaluating the performance of TB control program based upon deaths during treatment might therefore be misleading. Interventions 183  based on such finding would have limited applicability in preventing deaths in those who died from other causes.  Despite these issues related to the definition of TB‐related mortality, it remains a cause of growing concern in many industrialized countries that there is an overall increase in the number of TB related deaths (16‐19). Because of uncertainties with regard to definitions and an apparent increase in mortality there is a need to reassess the risk factors for TB related mortality in a Canadian setting. Although this question was previously assessed in BC in the 1980’s (14,20), both these studies were limited in scope, especially in terms of the number of cases, analytical techniques, and study duration. One study was case‐series in nature (20). Moreover, both studies were conducted in the pre‐HIV era. Since then, epidemiology of TB has changed greatly, particularly due to the HIV epidemic and an increased prevalence of TB among foreign‐born individuals.  In addition, there is a significant absence of large, longitudinal, population‐based studies. Most all‐cause mortality studies have been conducted on hospitalized patients or in selected populations (3,10,19,21‐27). Studies based on narrowly defined populations are prone to have limited external validity, which affects their applicability to the general population. Research focused on examining the impact of preventable risk factors on mortality among TB patients in a general population setting is clearly required to address these gaps. The objectives of this study were to characterize mortality among TB patients, to identify potentially modifiable or preventable risk factors for patients whose deaths were attributable to TB and to determine if there are differences in risk factor characteristics among two types of TB‐related deaths (those where TB was the underlying cause of death or only a contributing factor).  6.2 METHODS The provincial TB service (TB Control Division) at the British Columbia Centre for Disease Control (BCCDC) acts as a referral centre for the prevention, control and treatment of TB infection occurring in British Columbia (BC). TB control activities in BC are centralized; 184  therefore, this center maintains a central registry that includes information on all TB patients diagnosed in this province. In addition, all myco‐bacteriology tests are coordinated by this center. The BCCDC TB laboratory, which performs all culture and drug susceptibility tests, also serves as the reference laboratory for the province. This study used a retrospective cohort study design and included all TB patients that were diagnosed in British Columbia from 1990 to 2006. General population data for the Province of BC were extracted from the Vital Statistics Canada website.  6.2.1 Cause of Death The cause of death was defined as a patient who had a diagnosis of active TB and died from any cause during the study period (1990 to 2006). The cause of death in relationship to TB has been categorized in a systematic manner for the study period using the following categories.   TB was the underlying factor    TB was a contributing factor    TB was unrelated to death    Unknown  185  Figure 6.1: Study population for TB‐related mortality in British Columbia (n=5408)  5433 TB Cases Registered with TB Division between 1990 and 2006  Excluded (n=25) TB patients who were originally diagnosed in 1989  Study Cohort (n= 5408)  TB as the underlying cause (n=109): Outcome of interest  TB as a contributing factor (n=177): Outcome of interest  Cause of death unknown (n=551) or TB unrelated to death (n=232) or did not die (n=4339): Comparison group  Outcome: TB‐related death (TB was the underlying cause or a contributing factor); Comparison group: Survivors and patients whose death was unrelated to TB or unknown Primary Analysis: Multinomial Logistic Regression  186  The TB Control Division is informed of death of TB cases from several sources. If the death occurred in hospital and TB was listed as a cause of death or TB was discovered at or after death either by autopsy or ante‐mortem or post‐mortem specimen, the TB control program is notified about this patient by the concerned agency or physicians. In TB patients who died when they were on active follow up (during treatment) or died after treatment, the TB control program is usually notified about these deaths by their care‐providers or by the concerned authority. In addition, the electronic TB registry is linked to Vital Statistics Agency (VSA) and this registry does a periodic update from Vital Statistics Agency to obtain date of death. This additional safeguard ensures that the provincial TB control program is informed of each death occurring among TB patients. Subsequently, medical records including death certificates and autopsy reports related to each death are requested. Each death is reviewed by a TB physician and a decision is made based on the categories listed above.  In the current study, if TB was the main cause of death, this death was termed as a ‘TB‐caused’ death and if TB was one of the causes of death, then it was termed as ‘TB‐contributed’ death. When both types of deaths combined, it was termed as TB‐related death (TB was the underlying factor or a contributing factor). Although, in this study, we reported all deaths in patients with TB, the primary outcome of interest was TB‐related deaths.  6.2.2 Variables Description Variables that were used in this study are presented in three broad categories: socio‐ demographic characteristics; diagnosis‐related factors and co‐morbidity. Under the socio‐ demographic heading, variables such as: age, gender, birth‐country, marital status, and ethnicity are listed. Age was estimated at the date of diagnosis. In patients who had multiple recurrence of TB, the diagnosis date during last episode was used. Age was also analyzed as a continuous variable and the effect of age was reported per decade. Diagnosis type variables included case type (new or recurrent), culture‐confirmed TB, resistance to anti‐TB chemotherapy, miliary/disseminated TB, meningeal TB, far advanced pulmonary TB etc. Criteria 187  set by Health Canada were followed to define the types of TB diagnosed. In patients wh had multiple episodes of TB, we used information from the last episode.  Co‐morbidities (such as diabetes mellitus, HIV/AIDS and malignancy etc) and behavioral factors (such as substance abuse and alcoholism) are also presented in this category. Detailed descriptions of these variables have been provided in the Chapter 2. TB which was diagnosed post‐mortem was defined if TB was notified on or after death. These post‐mortem deaths also included patients whose TB diagnosis was suspected during life, but the diagnosis was only confirmed after death. TB diagnosed post mortem was used a proxy indicator of failure of diagnosis in this study.  6.2.3 Statistical Analysis An initial descriptive analysis for the study population was performed. Comparisons of continuous variables between groups were done using student t test and Wilcoxon rank sum test whenever appropriate. Categorical variables were compared using the Chi‐square test. TB‐ related death rate per 100,000 for each calendar year from 1990 to 2006 was calculated and the BC general population of the corresponding calendar year was used as denominators for this purpose.  Both TB ‐caused and TB‐contributed deaths were the primary interest of outcome. In addition to TB‐related deaths, there were two more types of deaths‐ deaths unrelated to TB and unknown. These two types of death and patients who did not die during the study period were included in the comparison group. Since there were three nominal groups (TB‐caused deaths, TB‐contributed deaths and comparison group), Multi‐nominal logistic regression has been chosen as the primary analytical method. Both single variable and multi‐variable multinomial logistic regression was used to identify risk factors associated with TB‐related mortality. The strength of the association between risk factors and death are reported using odds ratios (OR) with 95% confidence intervals (CI). All reported p values were two sided. 188  Several criteria were followed to select variables for multi‐variable multinomial logistic regression analysis. In addition to statistical significance in bi‐variate analysis, evidence from literature and clinical relevance was also considered to select variable for multi‐variable model. Due to universal confounding effect, both age and gender were included in multi‐variable model regardless of its statistical significance in bi‐variate setting. To evaluate the effect of co‐ morbid conditions (such as HIV/AIDS, malignancy and diabetes mellitus) on TB‐related mortality, patients without co‐morbidities and patients with unknown co‐morbid status were combined and used as the comparison group.  6.3 RESULTS During the entire study period, 5408 TB patients were registered with the TB Control Division. The mean age of the cohort at the time of diagnosis was 48.7 years. One third of patients were 60 years or older when they were diagnosed with active TB. There was a predominance of males (55%). Foreign‐born individuals (67%) accounted two‐third of all patients and Canadian‐ born Aboriginals represented another 13% of cases. In terms of ethnicity, the major contributors to this TB cohort were Chinese (24%) followed by Caucasian (17%) and Southeast Asian (15%). Majority of TB patients were born in the Western Pacific (40%) and Pan‐American Health Regions (31%). The socio‐demographic characteristics of the entire TB cohort are presented in Table 6.1.  189  Table 6.1: Socio‐demographic characteristics of study cohort (n=5408) Variable Age at diagnosis Mean (SD) Median (Range) Age groups 0‐40 years 41‐60 years 61‐80 years 80 plus years Biological gender Male Female Birth place Canadian‐born non‐Aboriginal Canadian‐born Aboriginal Foreign‐born Unknown Birth Country regions PAHO African Region Eastern Mediterranean Europe Southeast Asia Region Western Pacific Region Unknown /missing Ethnicity Caucasian Aboriginals Chinese South‐East Asian Filipino Vietnamese Others Unknown /Missing Marital status Single Not single Other /Unknown/missing  n (%) Years 48.7 (21.8) 46.7 (31‐67) 2213 (41) 1421 (26) 1366 (25) 408 (8) 2951 (55) 2457 (45) 876 (16) 709 (13) 3606 (67) 217 (4) 1686 (31) 97 (2) 138 (3) 270 (5) 762 (14) 2181 (40) 274 (5) 938 (17) 709 (13) 1293 (24) 830 (15) 428 (8) 306 (6) 558 (10) 346 (6) 1111 (20) 1564 (29) 2733 (51)  190  Table 6.2: Characteristics of TB patients who died during 1990 to 2006 in British Columbia (N=1069) Variable Age at death Mean (SD) Median (IQR) Age groups (at death) 0‐40 years 41‐60 years 61‐80 years 80 plus years Age groups (at diagnosis) 0‐40 years 41‐60 years 61‐80 years 80 plus years Biological gender Male Female Birth place Canadian‐ born non‐Aboriginal Canadian‐ born Aboriginal Foreign‐born Unknown Birth Country regions PAHO African Region Eastern Mediterranean & Europe Southeast Asia Region Western Pacific Region Unknown /missing Ethnicity Caucasian Aboriginals Chinese South‐East Asian Filipino/Vietnamese Others Unknown/missing Marital status Single Not single Other/Unknown/missing Cause of death TB as a principal cause TB as a contributing factor TB was unrelated to death Unknown  (n =1069) Years 67.7 (17.8) 71.5 (53‐82) 109 (10) 235 (22) 428 (40) 297 (28) 134 (12) 243 (23) 472 (44) 220 (21) 713 (67) 356 (33) 227 (21) 203 (19) 476 (45) 163 (15) 446 (42) 7 (1) 75 (7) 99 (9) 242 (22) 200 (19) 304 (28) 203 (19) 213 (20) 108 (10) 39 (3) 44 (4) 158 (15) 194 (18) 336 (31) 539 (29) 109 (10) 177 (16) 232 (22) 551 (52)  191  Among these TB cases, 1069 patients died during the study period (1990 to 2006). Out of all these deaths, 109 deaths were attributable to TB (main cause of death) and another 177 deaths were contributed by TB (one of the causes of death). TB was not related to death for another 232 cases. For approximately one‐half of the deaths, the relationship to TB could not be determined (termed as unknown). Overall, TB‐caused death was 2% (109/5408) and TB‐ contributed death was 3.3% (177/5408). The combined TB‐related death (whether TB was directly or indirectly related to TB) was 5.3% (286/5408) and accounted for 26% of all deaths. The socio‐demographic characteristics of TB patients who died during 1990 to 2006 are presented in Table 6.2.  The numbers of cause‐specific deaths during the entire study period are presented in Table 6.3. Figure 6.2 and Figure 6.3 show the trends in death overtime according to causes of deaths. Overall, there was an increasing trend of proportions of all‐cause deaths over time. However, the proportions of TB‐caused death (range: 1% to 4%; highest in 1991 and 1992 and lowest in 2002 to 2006) and TB‐contributed death (range: 3% to 8%; lowest in 1993 and 1997 and highest in 1991 and 2001) of all TB patients did not increase proportionately. Moreover, the proportions of TB‐caused death showed a slightly decreasing trend and the TB‐contributed death proportions remained stable over the study period. The increasing overall death trends were mainly due to increasing number of deaths due to unknown cause. TB‐caused death rate, which ranged from 0.05 to 0.40 per 100, 000 (BC population) during 1990 to 2006, showed a slightly decreased trend (Figure 6.3). TB‐contributed death rate (range: 0.1 to 0.5 per 100, 000) did not change significantly over time.  192  Table 6.3: Number of cause‐specific deaths by calendar year (1990 to 2006)  Year  Total Deaths  TB‐ caused47  TB‐ contributed48  TB‐caused or contributed  1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Total  27 32 40 48 53 45 66 68 70 88 69 92 87 89 99 46 50 1069  9 10 14 6 5 9 10 5 5 4 5 11 4 2 4 3 3 109  11 12 6 5 12 5 4 8 8 11 10 18 15 10 14 7 21 177  20 22 20 11 17 14 14 13 13 15 15 29 19 12 18 10 24 286  Unrelated Unknown49 to TB 7 5 9 19 18 10 10 22 15 25 12 10 21 16 11 13 9 232  0 5 11 18 18 21 42 33 42 48 42 53 47 61 70 23 17 551  Total TB cases50 275 284 336 340 332 324 326 418 333 327 314 368 300 325 310 278 335 5525  Table 6.4 shows occurrence of TB‐related deaths during the treatment period. About 56% of all TB‐caused deaths and 34% of TB‐contributed deaths were diagnosed post‐mortem. Post‐ mortem TB cases mostly did not receive any treatment. Among patients who died during treatment, the majority of patients died within one month of treatment initiation (68% for TB‐ caused deaths and 40% for TB‐contributed deaths). Out of all TB‐caused deaths, only one patient (1%) died after 6 months of treatment. Among patients with TB‐contributed deaths, 11 patients (11/177=17%) died 6 months after initiation of anti‐TB chemotherapy.  47  TB was the principal cause of death TB was not the primary cause of death, but was a contributing factor to death 49 The cause of death in relationship to TB was unknown 50 Total number of cases includes new and recurrent cases for each year. TB patients with multiple active episodes between 1990 and 2006 were accounted more than once. 48  193  Figure 6.2: Trend of cause‐specific death percentage for TB cohort in British Columbia from 1990 to 2006  194  Figure 6.3: TB‐related death rate (per 100, 000 BC population) in BC from 1990‐2006  Table 6.5 presents the socio‐demographic characteristics, diagnosis‐related factors and co‐ morbid conditions according to types of death (such as TB‐caused death, TB‐contributed death, unrelated, and unknown) and survivors. Aboriginal People were over‐represented among patients whose deaths were directly caused by TB. There was a disproportionate increase of milliary TB, meningeal TB and far advanced TB among patients who died from TB (both TB‐ caused and TB‐contributed deaths). Several behavioral factors (drug abuse and alcoholism) and co‐morbidities such as malnutrition, HIV/AIDS and renal failure were also more frequent among patients whose deaths were directly or indirectly attributed by TB.  195  Table 6.4: Occurrence of TB‐related death during the treatment period  Number of cases (TB was the principal cause)51 Percentage of total TB‐caused deaths on treatment Percentage of total TB‐caused deaths Cumulative Percentage Number of cases (TB was the contributing cause)52 Percentage of total TB‐ contributed deaths on treatment Percentage of total TB‐ contributed deaths Cumulative Percentage  51 52  Post‐ mortem/ before treatment 59  First month  2nd month  3rd month  4th month  5th month  6th month  After 6 months  32  6  3  1  2  2  1  68%  13%  7%  2%  4%  4%  2%  56%  30%  5%  3%  1%  2%  2%  1%  56%  86%  91%  94%  95%  97%  99%  100%  58  45  23  12  6  5  2  19  40%  20%  11%  5%  5%  2%  17%  34%  27%  13%  7%  4%  3%  34%  61%  74%  81%  85%  88%  1%  11%  89%  100%  ‐Out of 109 TB‐caused deaths, treatment‐related info was unavailable for 3 cases. ‐ Out of 177 TB‐contributed deaths, treatment‐related info was unavailable for 7 cases.  196  Table 6.5: Socio‐demographic and diagnosis‐related characteristics and prevalence of associated illness across several types of death (n=1069) and survivors (n=4339) Variable  Age at diagnosis Mean (SD) Age groups at diagnosis 0‐40 years 41‐60 years 61‐80 years 80 plus years Age at death Mean (SD) Age groups at death 0‐40 years 41‐60 years 61‐80 years 80 plus years Biological gender Male Female Ethnicity Caucasian Aboriginals Southeast Asian Vietnamese/Filipino/Chinese Others Unknown Birth place CB non‐Aboriginal CB Aboriginal Foreign born Unknown Drug resistance Sensitive Mono/poly‐resistant MDR No culture Recurrence Clinical TB Milary TB Meningeal TB Far advanced PTB Moderately advanced TB Cavitary TB Diabetes mellitus  Survivors (n=4339)  TB was the principal cause (n=109)  TB was a contributin g factor (n=177)  TB was unrelated to death (n=232)  Unknow n (n=551)  44.7 (20.8)  60.1 (18.1)  64.9 (19.4)  67.2 (17.7)  65.4 (16.7)  2079 (48) 1178 (27) 894 (21) 188 (4)  21 (19) 31 (28) 42 (39) 15 (14)  28 (16) 39 (22) 62 (35) 48 (27)  26 (11) 48 (21) 97 (42) 61 (26)  59 (11) 125 (23) 271 (49) 96 (17)  ‐  60.2 (18.2)  65.1 (19.4)  67.6 (17.6)  70.1 (16.9)  21 (19) 31 (28) 41 (38) 16 (15)  28 (16) 39 (22) 62 (35) 48 (27)  25 (11) 49 (21) 96 (41) 62 (27)  35 (6) 116 (21) 229 (42) 171 (31)  2238 (52) 2101 (48)  63 (58) 46 (42)  113 (64) 64 (36)  166 (72) 66 (28)  371 (67) 180 (33)  634 (15) 506 (12) 722 (17) 1775 (41) 514 (12) 188 (4)  24 (22) 28 (26) 15 (14) 13 (12) 5 (5) 24 (22)  32 (18) 36 (20) 7 (4) 35 (20) 9 (5) 58 (33)  58 (25) 35 (25) 17 (7) 62 (27) 8 (3) 52 (22)  190 (34) 104 (19) 69 (13) 142 (26) 22(4) 24 (4)  649 (15) 506 (12) 3130 (72) 54 (1)  10 (9) 28 (26) 37 (34) 34 (31)  24 (14) 36 (20) 64 (36) 53 (30)  47 (20) 35 (15) 104 (45) 46 (20)  146 (26) 104 (19) 271 (49) 30 (5)  2909 (67) 297 (7) 29 (1) 1104 (25) 361 (8) 1130 (26) 118 (3) 56 (1) 60 (1) 272 (6) 267 (6) 253 (6)  84 (77) 3 (3) 3 (3) 19 (17) 11 (10) 17 (16) 48 (44) 8 (7) 9 (8) 9 (8) 8 (7) 4 (4)  158 (89) 11 (6) 0 (0) 8 (5) 16 (9) 10 (6) 36 (20) 5 (3) 3 (2) 10 (6) 11 (6) 15 (8)  191 (82) 15 (6) 1 (<1) 25 (11) 31 (13) 24 (10) 13 (6) 2 (1) 4 (2) 17 (7) 9 (4) 17 (7)  438 (80) 29 (5) 2 (<1) 82 (15) 71 (13) 81 (15) 22 (4) 6 (1) 30 (5) 92 (17) 23 (4) 64 (12)  ‐  197  Variable  Malnutrition Either HIV or AIDS Alcoholism Drug Abuse Malignancy Immunosuppressive medication Renal failure  Survivors (n=4339)  TB was the principal cause (n=109)  TB was a contributin g factor (n=177)  TB was unrelated to death (n=232)  Unknow n (n=551)  39 (1) 164 (4) 188 (4) 130 (3) 62 (1) 81 (2) 49 (1)  8 (7) 17 (16) 21 (19) 12 (11) 1 (1) 3 (3) 7 (6)  12 (7) 40 (23) 28 (16) 30 (17) 30 (17) 13 (7) 24 (14)  6 (3) 28 (12) 15 (6) 16 (7) 26 (11) 7 (3) 6 (3)  9 (2) 47 (8) 62 (11) 31 (6) 40 (7) 19 (3) 6 (1)  The results of univariate Multinomial Logistic Regression are presented in Table‐6.6, Table‐6.7 and Table‐6.8. The risk of death significantly increased with age (31% per decade for TB‐caused death and 47% per decade for TB‐contributed deaths). Males had a significantly increased risk of death (UOR: 1.49, 95% CI: 1.09, 2.04) among patients whose deaths were indirectly contributed by TB. Aboriginal people were three to four‐times (TB‐caused deaths: UOR: 4.11, 95% CI: 2.50, 6.77; TB‐contributed deaths: UOR: 3.06, 95% CI: 2.02, 4.64) more likely to die from TB compared to the foreign‐born. There was an increased likelihood of deaths among Caucasians in patients with TB‐caused and TB‐contributed deaths (TB‐caused deaths: UOR: 4.14, 95% CI: 2.10, 8.17; TB‐contributed deaths: UOR: 2.05, 95% CI: 1.26, 3.34) compared to the reference group (mostly Chinese). In patients with TB‐caused deaths and patients with TB‐ contributed deaths, there was no significant difference in risk between foreign‐born and Canadian–born (CB) non‐aboriginal people.  Miliary TB (Table 6.7) was one of the strongest risk factor for both TB‐caused (UOR: 25.56, 95% CI: 16.94, 38.55) and TB‐contributed mortality (UOR: 8.29, 95% CI: 5.56, 12.37). In addition, other severe forms of tuberculosis such as MDR (UOR: 3.95, 95% CI: 1.19, 13.15), culture confirmed TB (UOR: 1.72, 95% CI: 1.02, 2.90), meningeal TB (UOR: 6.26, 95% CI: 2.93, 13.40) and far advanced PTB (4.81, 95% CI: 2.36, 9.81) were significantly associated with TB‐caused deaths. The effect of miliary TB (UOR: 8.29, 95% CI: 5.56, 12.37) and culture‐confirmed TB (UOR: 5.31, 95% CI: 2.79, 10.08) on TB‐contributed mortality was also significant in the univariate model (Table 6.7). 198  Table 6.8 presents the risk associated with TB‐related mortality from accompanied illness/co‐ morbid conditions. Malnutrition (UOR: 7.43, 95% CI: 3.45, 16.03), HIV/AIDS (UOR: 3.78, 95% CI: 2.21, 6.44) and renal failure (UOR: 5.69, 95% CI: 2.54, 12.75) were the significant risk factors of mortality in patients who died from TB. Behavioral factors such as alcoholism (UOR: 4.37, 95% CI: 2.68, 7.15), and drug abuse (UOR: 3.46, 95% CI: 1.86, 6.4) were also significant risk factor of TB‐caused mortality. All these factors including HIV/AIDS (UOR: 5.97, 95% CI: 4.10, 8.68) and malignancy (UOR: 7.96, 95% CI: 5.18, 12.24) were significantly associated with TB‐contributed deaths. However, the effects of co‐morbid conditions were much higher for TB‐contributed deaths than that of TB‐caused deaths (for examples, renal failure, UOR: 13.01, 95% CI: 7.90, 21.43; drug abuse, UOR: 5.70, 95% CI: 3.75, 8.68). The strongest risk factors for TB‐contributed deaths were presence of renal failure and also malignancy, which were not risk factors deaths caused by TB‐.  The results of multivariable multi‐nominal logistic regression are presented in Table 6.9. The multivariable multinomial logistic regression demonstrated that miliary TB (AOR: 32.37, 95% CI: 18.66, 56.13), far advanced PTB (AOR: 10.73, 95% CI: 4.72, 24.41), CNS/meningeal TB (AOR: 4.75, 95% CI: 1.25, 18.08) and Aboriginal ethnicity (AOR: 4.38, 95% CI: 2.00, 9.60) were the strongest predictors of TB‐caused mortality. In addition, increasing age (per decade, AOR: 1.72, 95% CI: 1.48, 2.02), renal failure (AOR: 3.46, 95% CI: 1.10, 10.88), malnutrition (AOR: 3.28, 95% CI: 1.25, 8.62), alcoholism (AOR: 2.70, 95% CI: 1.37, 5.33) and Caucasian ethnicity (AOR: 2.49, 95% CI: 1.18, 5.27) were other significant risk factors of TB‐caused mortality.  199  Table 6.6: Univariate multinomial logistic regression analysis for socio‐demographic risk factors of TB‐related mortality  Variable Age at diagnosis (per decade) Age groups (at diagnosis) 0‐40 years 41‐60 years 61‐80 years 80 plus years Biological gender Male Female Birth place Canadian‐born non‐Aboriginal Canadian‐born Aboriginal Foreign‐born Birth Country regions PAHO African Region Eastern Mediterranean and Europe Southeast Asia Region Western Pacific Region Ethnicity Caucasian Aboriginals Southeast Asian Other Ethnicity Chinese/Filipino/Vietnamese Marital status Single Not single  TB was the underlying cause Unadjusted OR (95% CI)  TB was a contributing factor Unadjusted OR (95% CI)  1.31 (1.20, 1.43)  1.47 (1.36, 1.60)  Reference 2.37 (1.35, 4.13) 3.43 (2.02, 5.82) 4.48 (2.29, 8.78)  Reference 2.23 (1.37, 3.64) 3.79 (2.42, 5.96) 10.75 (6.66, 17.37)  1.16 (0.79, 1.70) Reference  1.49 (1.09, 2.04) Reference  1.13 (0.56, 2.27) 4.11 (2.50, 6.77) Reference  1.56 (0.97, 2.51) 3.06 (2.02, 4.64) Reference  5.26 (2.62, 10.57) 2.27 (0.29, 17.94) 3.25 (1.18, 9.01)  2.50 (1.64, 3.81) 1.34 (0.32, 5.65) 1.28 (0.59, 2.78)  2.59 (1.05, 6.39) Reference  0.76 (0.36, 1.59) Reference  4.14 (2.10, 8.17) 6.61 (3.40, 12.83) 2.83 (1.34, 5.97) 1.40 (0.50, 3.94) Reference  2.05 (1.26, 3.34) 3.16 (1.97, 5.07) 0.49 (0.22, 1.11) 0.94 (0.45, 1.96) Reference  Reference 1.09 (0.56, 2.10)  Reference 0.85 (0.50, 1.43)  200  Table 6.7: Univariate multinomial logistic regression analysis for diagnosis type and related risk factors of TB‐related mortality  Variable Case type New Recurrent Culture results Negative/no culture Positive Drug resistance status Sensitive Mono/poly resistance MDR Disease type of current episode Pulmonary Extra‐pulmonary Both Millary/disseminated TB No Yes CNS/Meningeal TB No Yes Far advanced PTB No Yes Moderately advanced PTB No Yes Cavitary TB No Yes  TB was the underlying cause Unadjusted OR (95% CI)  TB was a contributing factor Unadjusted OR (95% CI)  Reference 1.13 (0.60, 2.12)  Reference 1.01 (0.60, 1.70)  Reference 1.72 (1.02, 2.90)  Reference 5.31 (2.79, 10.08)  Reference 0.37 (0.12, 1.18) 3.95 (1.19, 13.15)  Reference 0.72 (0.39, 1.34) ‐  Reference 1.36 (0.89, 2.06) 2.60 (1.38, 4.88)  Reference 0.87 (0.60, 1.23) 2.45 (1.51, 3.97)  Reference 25.56 (16.94, 38.55)  Reference 8.29 (5.56, 12.37)  Reference 6.26 (2.93, 13.40)  Reference 2.30 (0.91, 5.78)  Reference 4.81 (2.36, 9.81)  Reference 0.92 (0.30, 2.94)  Reference 1.12 (0.56, 2.23)  Reference 0.75 (0.39, 1.42)  Reference 1.28 (0.62,2.65)  Reference 1.07 (0.57, 1.99)  201  Table 6.8: Univariate multinomial logistic regression analysis for co‐morbid risk factors of TB‐ related mortality  Variable Diabetes Mellitus No/unknown Yes Malnutrition No/unknown Yes Alcoholism No/unknown Yes Substance abuse No/unknown Yes Any Malignancy No/unknown Yes Immunosuppressive medication No/unknown Yes Chronic Renal failure No/unknown Yes Either HIV or AIDS No/unknown Yes  TB was the underlying cause Unadjusted OR (95% CI)  TB was a contributing factor Unadjusted OR (95% CI)  Reference 0.55 (0.20, 1.49)  Reference 1.33 (0.78, 2.28)  Reference 7.43 (3.45, 16.03)  Reference 6.83 (3.58, 13.00)  Reference 4.37 (2.68, 7.15)  Reference 3.44 (2.26, 5.25)  Reference 3.46 (1.86, 6.41)  Reference 5.70 (3.75, 8.68)  Reference 0.36 (0.05, 2.61)  Reference 7.96 (5.18, 12.24)  Reference 1.33 (0.41, 4.25)  Reference 3.72 (2.05, 6.75)  Reference 5.69 (2.54, 12.75)  Reference 13.01 (7.90, 21.43)  Reference 3.78 (2.21, 6.44)  Reference 5.97 (4.10, 8.68)  The greatest risk factors for TB‐contributed mortality were renal failure (AOR: 6.77, 95% CI: 3.13, 14.61), HIV/AIDS (AOR: 6.73, 95% CI: 3.44, 13.15), miliary TB (AOR: 6.09, 95% CI: 3.44, 13.15) and malignancy (AOR: 5.31, 95% CI: 2.99, 9.43). Increasing age (AOR: 1.84, 95% CI: 1.60, 2.14), malnutrition (AOR: 2.96, 95% CI: 1.34, 6.53), alcoholism (AOR: 2.82, 95% CI: 1.55, 5.13) and drug abuse (AOR: 2.75, 95% CI: 1.36, 5.56) and Aboriginal ethnicity (AOR: 2.13, 95% CI: 1.14, 3.95) were also significantly associated with TB‐contributed mortality. Gender, immunosuppressive medication and culture confirmed TB did not reach statistical significance (95% CI of the corresponding odds ratio included 1) in the multivariate model (Table 6.9).  202  Table 6.9: Multivariable multinomial logistic regression analysis for risk factors of TB‐related mortality among TB patients  Variable  TB was the underlying  TB was a contributing  cause  factor  AOR (95% CI)  AOR (95% CI)  Age at diagnosis (per decade)  1.72 (1.48, 2.02)  1.84 (1.60, 2.14)  Male gender  0.67 (0.41, 1.10)  1.05 (0.70, 1.59)  Caucasian  2.49 (1.18, 5.27)  1.10 (0.64, 1.89)  Aboriginals  4.38 (2.00, 9.60)  2.13 (1.14, 3.95)  Southeast Asian  2.19 (0.99, 4.80)  0.45 (0.20, 1.06)  Others  1.63 (0.54, 4.92)  1.17 (0.53, 2.59)  Reference  Reference  1.17 (0.56, 2.46)  1.75 (0.83, 3.72)  32.37 (18.66, 56.13)  6.09 (3.44, 10.77)  CNS/ meningeal TB  4.75 (1.25, 18.08)  2.33 (0.63, 8.66)  Far advanced pulmonary TB  10.73 (4.72, 24.41)  1.65 (0.48, 5.64)  Either HIV or AIDS  1.97 (0.85, 4.60)  6.73 (3.44, 13.15)  Malnutrition  3.28 (1.25, 8.62)  2.96 (1.34, 6.53)  Alcoholism  2.70 (1.37, 5.33)  2.82 (1.55, 5.13)  Drug abuse  1.56 (0.64, 3.82)  2.75 (1.36, 5.56)  Any Malignancy  0.39 (0.05, 3.17)  5.31 (2.99, 9.43)  Chronic Renal Failure  3.46 (1.10, 10.88)  6.77 (3.13, 14.61)  Immunosuppressive Medication  0.49 (0.13, 1.88)  1.56 (0.68, 3.60)  Ethnicity  Filipino/Vietnamese/Chinese Culture confirmed TB Millary/disseminated TB  203  Table 6.10: Comparisons of socio‐demographic and clinical characteristics between TB patients who were diagnosed postmortem (137) and who were diagnosed alive (n=5235) Variable  Age at diagnosis Mean (SD) Age groups (at diagnosis, last episode) 0‐40 years 41‐60 years 61‐80 years 80 plus years Biological gender Male Female Ethnicity Caucasian Aboriginals South‐East Asian Vietnamese/Filipino/Chinese Others Birth place Canadian born non‐Aboriginal Canadian born Aboriginal Foreign born Cause of Death TB was the underlying factor TB was a contributing factor TB was unrelated to death/unknown /did not die Drug resistance Sensitive Mono/poly‐resistant MDR Recurrence Absence of positive culture Millary TB Meningeal TB Cavitary TB Far advanced TB Moderately advanced TB Either HIV or AIDS Drug abuse Alcoholism Diabetes mellitus  Diagnosed alive (5235) N (%)  Post‐mortem TB (173) N (%)  P value  Years 48.2 (21.7)  Years 64.9 (17.9)  2193 (42) 1374 (26) 1294 (25) 374 (7)  20 (12) 47 (27) 72 (42) 34 (20)  <0.001  2847 (54) 2388 (46)  104 (60) 69 (40)  0.136  905 (18) 676 (14) 821 (17) 2002 (40) 548 (11)  33 (30) 33 (30) 9 (8) 25 (23) 10 (9)  855 (17) 676 (13) 3560 (53)  21 (21) 33 (33) 46 (46)  <0.001  50 (1) 123 (2) 5062 (97)  59 (34) 54 (31) 60 (35)  <0.001  3654 (91) 346 (9) 35 (1) 478 (9) 1228 (24) 194 (4) 68 (1) 315 (6) 100 (2) 395 (7) 275 (5) 203 (8) 294 (11) 346 (13)  126 (93) 9 (7) 0 (0) 12 (7) 34 (20) 43 (25) 9 (5) 3 (2) 6 (3) 5 (3) 21 (12) 16 (14) 20 (17) 7 (6)  <0.001  <0.001  0.398 0.342 0.244 <0.001 <0.001 0.018 0.146 0.021 <0.001 0.027 0.050 0.023  204  Table 6.10 presents bi‐variate comparisons between patients who had TB diagnosed post‐ mortem and patients whose TB was diagnosed while they were still alive. TB patients whose disease was diagnosed at autopsy were significantly more likely to be older, Aboriginals, Caucasians, and to die from TB (TB‐caused death and TB‐contributed death). The diagnosis of miliary TB and meningeal TB were also significantly associated with TB diagnosed at autopsy. Several co‐morbidities including diabetes mellitus, HIV/AIDS, malnutrition and malignancy as well as behavioral factors (such as alcoholism and substance abuse) were significantly more frequent among patients diagnosed at autopsy (Table 6.10).  Table 6.11: Relationship between HIV and cause of death Cause of death  HIV‐positive N (%)  HIV‐negative N (%)  HIV‐Unknown N (%)  132 (100)  448 (100)  489 (100)  TB as the underlying cause  17 (13)  92 (21)  ‐  TB as a contributing factor  40 (30)  137 (31)  ‐  TB was unrelated to death  28 (21)  50 (11)  154 (31)  Unknown  47 (36)  169 (37)  335 (69)  All deaths  6.4 DISCUSSION This study demonstrated that overall 1069 patients died (1069/5408=20%) during the entire study period (1990‐2006). However, the TB‐caused death and TB‐contributed death was 2.0% (109/5408) and 3.3% (177/5408) respectively. In total, 286 deaths were attributable to TB directly or indirectly, which provided an overall crude death of 5.3% (286/5408). Wide discrepancies in TB‐related death proportions exist across studies, largely due to differences in methodology, observation period, geographical locations and background prevalence. In addition, the majority of TB‐mortality studies only reported deaths among TB patients with  205  which occurred during treatment regardless of the specific cause of death, which make the comparisons to others studies somewhat difficult.  However, our TB‐related death was relatively lower than studies conducted in similar setting. The previous BC studies (14,20) reported a CFR of over 3% (99/2997=3.3% during 1970‐74; 58/1884=3.1% during 1980‐84) which is higher than in the current study. A national study in Netherlands (11) demonstrated that 35 deaths (TB‐caused death: 35/4340=0.8%) were caused by TB when first year of treatment had been completed. This Dutch study did not include patients who died at the time of diagnosis. In the current study, 50 of out 109 TB‐caused occurred during treatment, which would provide a TB‐caused death of 0.9% on treatment. In prior BC studies, CFR during treatment was 1.7% (49/2997) and 1.6% (30/1884) respectively (14,20). A study in Queensland, Australia reported deaths among TB patients similar to ours (15). They reported overall 127 deaths (127/1003= 12.7%) before completion of chemotherapy, of which 53 (CFR=53/1003=5.3%) were primarily due to TB and 34 (CFR=34/1003=3.4%) were partially attributable to TB. In our study, 487 patients (487/5408=9%) died before the completion of chemotherapy.  A national study in England and Wales (28) reported CFR from a period of 21 years, which ranged from 11.2% (in 1972) to 7.21% (in 1992). Another study in Liverpool, UK observed a death rate of 23.7% (all cause, 104/439) over an 8‐year period (29). However, TB was listed as a cause of death for 34 cases that would provide a TB‐contributed death rate of 7.7% (29). A relatively lower TB‐caused death was observed in clinical trials. The tuberculosis trials consortium (TBTC) study 22 reported a CFR (all‐cause) of 6.6% (71/1075) among patients with PTB thorough a two‐year follow up phase. However, in this clinical trial, only one death (1/1075 =0.09 %) was attributed to TB (9).  Another study in South African gold miners found an overall CFR rate of 3.6% (81/2236) among patients with culture‐confirmed PTB during a 6 month of treatment period (30). However, their TB‐caused CFR was 25/2236=1.1% on treatment (30). However, the proportion of deaths 206  attributed to TB was much higher in several Russian studies. A recent study in Tomsk Oblast found relatively higher TB‐related deaths (138/1916=7.2%) among patients while on treatment and reported that about three quarters of all deaths (138/183=75%) was directly attributable to TB (8). Another study in Samara, Russia (31) reported a high proportion of deaths attributed to TB (89%) despite an overall low case fatality rate (all cause: 5.9%; on treatment: 3.6%). The Russian study in Orel (32) also reported that TB was a probable cause of death for 73% (46/55) of TB patients who died after treatment initiation during an 8‐month follow up period.  The TB‐caused death rate ranged from 0.05 to 0.40 per 100, 000 (general population) in our study during the entire observation period. A study in Veracruz, Mexico reported a death rate of 0.9 per 100, 000 for the year of 1993 (13). Globally, the TB mortality rate was 21 (per 100, 000) in 2008 and 30 in 1990 (33), which was much higher than that of our study. Although the global TB mortality is falling the rate of decline is very low. Low CFR in the current study might be related to quality of care provided by TB control program, centralized TB control activities, universal access to health care, healthy migration effect, low level of recurrence and resistance.  This study also identified several important risk factors for TB‐caused morality and TB‐ contributed mortality. The significant risk factors for TB‐caused mortality in multivariable analysis were older age, Aboriginal ethnicity, Caucasian ethnicity, miliary/meningeal TB, far advanced PTB, malnutrition, renal failure and alcoholism. On other hand, older age, Aboriginal ethnicity, miliary TB, HIV/AIDS, malnutrition, alcoholism, substance abuse, malignancy and renal failure were significantly associated with TB‐contributed mortality.  6.4.1 Older Age This study demonstrated a very strong age effect on TB‐related mortality. Elderly people were significantly more likely to have TB‐caused and TB‐contributed mortality. This finding was anticipated and consistent with the published literature. The Mexican study observed an increased effect of age (AOR: 1.02 per year) among patients dying from TB (13). A Dutch study 207  reported that the mortality rate due to PTB was significantly higher among patients older than 75 years (17). The Australian study also observed increased occurrence of TB‐related deaths in elderly TB patients (15). An increased mortality among elderly TB patients was reported in many all‐cause mortality studies (9‐11).  A national (all‐cause mortality) study in Netherlands reported a high relative risk of death (AHR=45) among patients 65 years and older compared to patients younger than 25 years (11). The North American clinical trial (all cause mortality) found an increased effect of age (UOR: 1.04) in univariate analysis (9). The current TB cohort is over‐represented by elderly people. About one‐third of TB patients were more than 60 years older at the time of diagnosis (current episode). Elderly people are more vulnerable to die from the disease due to a decreased immune response, increased presence of co‐morbid diseases and more advanced TB. In addition, the diagnosis of TB is often difficult among elderly people due to atypical presentation of TB and superimposition of other chronic illness, increasing the probability of delayed diagnosis and initiation of appropriate treatment. Therefore, a high index of suspicion is required in order to make a prompt diagnosis and commence treatment of elderly TB patients.  6.4.2 Failure of Diagnosis/Missed Diagnosis The key finding in the current study was the post‐mortem TB (missed diagnosis/failure of diagnosis) as an increasingly important contributor to TB‐caused death. In the current study, 56% of TB‐caused deaths and 34% of TB‐contributed deaths were diagnosed post mortem. The significant contribution of failure of diagnosis to TB‐related death has been reported consistently in previous studies (14,20,34). The relative contribution of missed diagnosis to TB‐ related deaths in the Australian study (15) was 28.7% (25/87), which was relatively lower that in our study (113/286=39.5%). Compared to prior BC studies, the current study demonstrated a decline in CFR (dying from TB), but the relative contribution of post‐mortem TB did not decrease proportionately. Even this study showed a relatively higher proportion of TB‐caused  208  deaths due to missed diagnosis than that of earlier BC studies (during 1970‐74: 51%, during 1980‐84: 48%).  Failure to diagnose TB might be related to increased prevalence of TB in certain demographic groups (such as elderly persons and aboriginal people) as well as higher proportion of TB with atypical presentations (such as miliary TB and meningeal TB) among post‐mortem cases. In the current study, in addition to TB‐caused death, several factors such as older age, aboriginal and white ethnicity, miliary TB, meningeal TB and presence of co‐morbidities (such as malnutrition and malignancy) distinguished patients who were diagnosed with TB at post‐mortem from patients who were diagnosed with TB ante‐mortem. Moreover, in Canada the incidence of TB declined steadily in recent decades which particularly made the disease less prevalent in the community. Due to lack to lack of experience in diagnosing TB, physicians are less likely to consider TB, especially in patients with atypical presentations, which further enhance the possibility of a missed diagnosis. A recent Canadian study demonstrated a positive relationship between physician experience with TB and survival of TB patients (35). Education and increased awareness about presence of post‐mortem TB is therefore of particular importance in this setting.  6.4.3 Miliary TB The presence of miliary TB was not only associated with mortality, but also emerged as the most powerful predictor of TB‐caused mortality in this study (among patients who died from TB, 44% had a miliary TB diagnosis). A similar association was also shown in earlier studies. The Australian study reported a strong association (between TB‐related death and disseminated TB (15). This might be related to over‐representation of miliary TB in post‐mortem cases. Other reasons could be due to an increased prevalence of miliary TB in certain groups such as the elderly and Aboriginal people as well as in patients with history of poor prognostic factors such as renal failure, malnutrition, HIV/AIDS, drug abuse and alcoholism. Overall, 4.4% of TB patients in this cohort had a diagnosis miliary TB, which was relatively higher, compared to 209  Canadian average (2% in 2004) (36). Prevalence of miliary TB was 8% among Aboriginals, but was much higher in patients with certain co‐morbidities (HIV/AIDS: 27%, drug abuse: 19%, renal failure: 15%, malnutrition: 15%, immune‐suppressive medication: 15% and alcoholism: 10%).  Miliary TB is an indicator of the severity of diseases and has been associated with higher morbidity and mortality in other reports. In this study, CFR was 29% for TB‐caused death and 23% for TB‐contributed death. Due to atypical presentation, miliary TB is often difficult to diagnose, thus increasing the possibility of missed or delayed diagnosis (36). A high index of suspicion and initiation of therapy before the confirmed diagnosis is critical in order to prevent morbidity and mortality from this devastating form of TB. Awareness regarding the atypical presentation and the increased mortality related to miliary TB in the community as well as in health care workers is required.  6.4.4 Aboriginal People This study demonstrated that Aboriginal people were at an increased risk of death due to TB. This finding is unique in Canadian setting. The north‐American clinical trial did not find an increased risk of death from tuberculosis while on treatment among Native Americans (9). Another study in Australia reported no elevated risk of dying from tuberculosis (TB‐related) among indigenous population (15). Previous BC studies (14) did not observe a significantly increased risk of TB‐caused mortality among Aboriginals compared to the White TB patients. This increased risk in Aboriginal people is partly attributable to their life styles (such as substance abuse and alcoholism) and increased presence of certain co‐morbidities including HIV. In the current study, the frequency of drug abuse, alcoholism, malnutrition, HIV/AIDS and renal failure was significantly higher among Aboriginal TB patients than that of Canadian‐born non Aboriginal as well as foreign‐born. All these factors have been associated with increased morbidly and mortality among TB patients in TB‐mortality studies. The prevalence of HIV/AIDS, diabetes mellitus, heavy alcoholism, substance abuse and other co‐morbidities is also higher in the Aboriginal communities compared the general population (37‐41). 210  Other potential reasons for increased risk of dying was due to increased presence of severe forms of TB (such as miliary TB and meningeal TB) as well as post‐mortem TB among aboriginal population. The current study demonstrated a disproportionate burden of post‐mortem cases among Aboriginal people. This might be related to cultural barrier such as patient’s decreased desire to seek care and the isolated nature of some communities. In addition, aboriginal people represents socially vulnerable group, which are often difficult to reach for early diagnosis and treatment. Culturally safe intervention is urgently required for this high risk population.  6.4.5 Caucasian This study demonstrated an increased risk of mortality from TB among Caucasians. This finding is consistent with other studies (mainly all cause‐mortality) conducted in similar setting (42,43). However, the relationship between White and TB‐caused mortality was not significant in the previous BC study (14). A study in Birmingham, UK that specially investigated the effect of ethnicity observed a significantly higher CFR (death on treatment) among Caucasians (9.4% vs. 1.5%) than in Asians (42). The North American clinical trial (all cause mortality) found a significantly increased mortality among White TB patients (UOR: 1.81) in univariate analysis (9). The increased CFR among this ethnic group might be related to increased presence of elderly TB patients and more severe forms of TB among Caucasians. In addition, a relatively higher proportion of Caucasian TB patients were suffering from a concomitant illness ‐could be another potential reason for increased fatality among Caucasians.  6.4.6 Far Advanced PTB and CNS/Meningeal TB Far advanced PTB and meningeal TB were other important predictors of TB‐caused mortality in the current study. Severe forms of TB have been associated with mortality in TB patients in other studies (18,27,31,34,44,45). Far advanced PTB represents a severe form of TB and could increase the risk of death in TB patients due to its association with respiratory insufficiency. Tuberculosis meningitis is another life threatening form of TB and was associated with higher 211  morbidity and mortality in other studies, even after the initiation of anti‐TB medication (46‐49). Like miliary TB, meningeal TB is often difficult to diagnose, thus increase the possibility of delayed diagnosis. Advanced pulmonary TB was found to be a significant risk factor for dying from TB in the previous BC study (14).  6.4.7 Drug Resistance MDR was strongly associated with TB‐caused mortality in this study in bi‐variate association. However, this association was not found significant in the multivariable model (data not shown). A number of previous studies, especially from Russia, have reported an increased risk of mortality among patients with MDR (8,31), but others have not confirmed this relationship (15,27). No significant association between mortality and drug‐resistant diseases was found in the study in the USA (10). MDR‐TB did not have a significant impact on mortality in another USA study (18). Even studies in a high burden setting like South Africa showed no increased risk associated with drug resistance, especially with MDR (30,50). The low number of cases might lead to the non‐significant association in our study.  6.4.8 HIV HIV was significantly associated with TB‐caused mortality in bi‐variate analysis, but was not significant in multivariable model. However, it was one of most important risk factor for mortality among patients whose death was indirectly related to TB (TB was a contributing factor) in a multivariate analysis. The deleterious effect of HIV on TB‐related mortality has been well shown previously (30,45). Due to impaired immunity, it is not surprising HIV could accelerate deaths among TB patients. Majority of those reports were all‐cause mortality studies and did not distinguish between TB‐caused death and deaths caused by other illness. The effect of HIV was most evident in African studies (30,45). The Australian study could not examine the effect of HIV on TB‐related mortality due to unknown HIV status for most of the 212  patients (15). Another Russian study reported no HIV positive case among patients who died of TB (8).  The Dutch study (all‐cause mortality, AOR: 9.9) and the North‐America TB trial (all‐cause mortality, AHR: 3.9) reported a very strong association between HIV infection and mortality (9,11). On the other hand, several studies in the post‐HIV era did not observe an elevated risk of dying from TB (all cause, during treatment) among HIV‐positive patients (10,32). Several USA studies conducted in hospital settings also confirmed this finding. For examples, the all‐cause mortality study in USA among patients hospitalized for TB patients observed no significant association between HIV and mortality (10). Another hospital based USA study (51) showed similar results. Even some studies conducted in settings where HIV‐TB co‐infection was more prevalent has shown that HIV‐infection was not related to in‐hospital mortality among TB patients (27).  The stronger effect of HIV on TB‐contributed mortality than that of TB‐caused mortality is not surprising. Patients whose deaths were not directly attributed to tuberculosis might die from other co‐morbidity. HIV /AIDS could be the predominant cause of death for these patients. The South African gold‐miners study showed a significantly higher proportion of deaths due to TB among HIV‐negative patients than in HIV‐positive persons (30). The current study has shown similar evidence: a higher proportion of all deaths due to TB (13% vs. 21%) among HIV‐negative TB patients than in HIV‐positive TB persons (Table 6.11).  6.4.9 Co­morbidities This study has demonstrated that several co‐morbid diseases were significantly associated with TB‐related mortality in multivariate model (malignancy, renal failure and malnutrition for TB‐ contributed mortality, but only chronic renal failure and malnutrition for TB‐caused mortality). A similar relationship has been reported in other studies. The presence of concomitant illness at the time of diagnosis among PTB patients was significantly associated with TB‐caused mortality 213  in the Mexican study (13). In another Australian study an elevated risk of dying from TB (TB‐ related mortality) was shown among patients with coexisting malnutrition, liver disease and renal disease (15). However, in the same study, other co‐morbidities such as diabetes, malignancy and steroid treatment were not found to be significant risk factors for TB‐related mortality (15).  An all‐cause mortality study in USA among hospitalized TB patients observed that several co‐ morbid conditions such as end‐stage renal disease, malnutrition and immune‐suppression were the largest contributors to mortality over a 14‐month follow up period (10). Another recent USA study reported that TB patients who died during hospitalization had a greater likelihood of co‐morbid illness (51). An all‐cause mortality study in Bolivia among hospitalized PTB patients reported that co‐existing pathology (such as diabetes, silicosis, malignancy and renal insufficiency) was the largest predictor of mortality (2). However, this study showed that effect of co‐morbidity including HIV was much higher on TB‐contributed mortality than that of TB‐ caused mortality. Since patients with TB‐contributed deaths mainly died due to other causes other than TB, reflecting that co‐morbidity such as HIV/AIDS, renal failure and malignancy were the major contributors of deaths for those patients. A previous BC study has shown a significantly increased presence of chronic renal failure and malignant disease in patients with TB‐contributed mortality than in patients whose deaths were caused by TB or unrelated to TB (14). Malnutrition (10,15,52) and related factors such as low Body Mass Index (53,54), low weight (27,55) and weight loss (56) were shown to increase the risk of death in several studies.  Behavioral factors such as alcoholism and drug abuse were significantly associated with TB‐ related mortality, although the effects of drug abuse disappeared in a multivariate model for TB‐caused mortality. The effect of alcoholism was not significant in the Australian study (15) as well as in the Mexican study (13). An increased presence of alcoholism among TB cases diagnosed at autopsy was reported in other studies (20,34). In a USA (all cause) study (10), alcoholism was significant in a univariate model but not in the multivariate analysis. Use of IV drugs was not a significant risk in the same study. Several Russian studies have shown an 214  elevated risk of dying (all‐cause) associated in the presence of substance abuse and alcoholism (8,31). Other studies also reported a significant association between alcoholism and mortality (9,11,57).  According to the national addiction survey in 2004, the prevalence of lifetime and past year illicit drug use (other than cannabis) in BC was 23% and 4% respectively (58). BC has the highest prevalence of illicit drug use compared to other provinces in Canada (58). The frequency of drug use was relatively low in the TB cohort due to over‐representation of foreign‐born people who use illicit drugs very infrequently. Drug addiction has been associated with increased mortality and morbidity in many studies (59‐61). Both alcoholism and drug use are subject to modifications. Interventions targeting these high risk populations are urgently required. Establishment of a systematic approach in order to identify TB patients with alcoholism and substance abuse, then putting them in the appropriate drug or alcoholism treatment program should be prioritized in the current TB care. In addition, the association of malnutrition with an increased mortality risk suggests that more aggressive attention to improved nutrition during the treatment phase would be appropriate.  This study has several limitations. This study did not evaluate the effect of treatment as well as treatment regimens. However, the greater likelihood of death in the early period of chemotherapy was evident in this study (Table 6.4). About 86% of total TB‐caused deaths occurred within two months of diagnosis. This finding was consistent with earlier studies (62,63). Missing observations was another limitation of this study, which was more frequent for co‐morbid conditions. We considered missing observation as representativeness of no response. Therefore, the effects of co‐morbidities on TB‐related mortality were evaluated in comparison to patients with no or unknown co‐morbid status. However, the sensitivity analysis restricted to patients with valid responses showed similar patterns for significant associations between TB‐related mortality and comorbidity (Table A.11).  215  6.5 CONCLUSIONS In this study despite a high crude death, death specifically related to TB was low. In addition, we identified a number of important risk factors for TB related death in a population‐based TB cohort. Failure of diagnosis or failure to seek care continues to be one of the largest contributors to TB‐related deaths. Aboriginal people were at the greatest risk of dying from TB. Culturally safe interventions are urgently required for this high risk population. In addition, miliary, meningeal and advanced PTB were the strongest predictors of TB‐caused mortality. Among co‐morbidities, alcoholism, malnutrition and renal failure were strongly associated with increased risk of dying specifically from TB. Several factors including malnutrition, alcoholism and drug abuse are subject to modifications. Effective Interventions (such as nutritional supplement program and active surveillance for TB patients with alcoholism and drug abuse) should be prioritized in the current TB control program. The effects of co‐morbidity including HIV were much higher among patients whose deaths were partly attributed by TB. These findings underscore the need to intensify current efforts in order to prevent future deaths from TB in high risk populations.  216  6.6 REFERENCES (1) Raviglione MC, Snider DE,Jr, Kochi A. Global epidemiology of tuberculosis. Morbidity and mortality of a worldwide epidemic. JAMA 1995 Jan 18;273:220‐226. (2) Abos‐Hernandez R, Olle‐Goig JE. Patients hospitalised in Bolivia with pulmonary tuberculosis: risk factors for dying. International Journal of Tuberculosis & Lung Disease 2002 Jun;6(6):470‐474. (3) Olle‐Goig JE. Patients with tuberculosis in Bolivia: why do they die?. Pan American Journal of Public Health 2000 Sep;8(3):151‐155. (4) World Health Organization. Tuberculosis Facts 2009 Update. 2009; Available at: http://www.who.int/tb/publications/2009/factsheet_tb_2009update_dec09.pdf. Accessed April/02, 2010. (5) Kochi A. The global tuberculosis situation and the new control strategy of the World Health Organization. Tubercle 1991 Mar;72(1):1‐6. (6) Dye C, Scheele S, Dolin P, Pathania V, Raviglione MC. Consensus statement. Global burden of tuberculosis: estimated incidence, prevalence, and mortality by country. WHO Global Surveillance and Monitoring Project. JAMA 1999 Aug 18;282(7):677‐686. (7) World Health Organization. WHO Tuberculosis Programme: Framework for Effective Tuberculosis Control. 1994; Available at: http://whqlibdoc.who.int/hq/1994/WHO_TB_94.179.pdf. Accessed March/31, 2010. (8) Mathew TA, Ovsyanikova TN, Shin SS, Gelmanova I, Balbuena DA, Atwood S, et al. Causes of death during tuberculosis treatment in Tomsk Oblast, Russia. International Journal of Tuberculosis and Lung Disease 2006 Aug;10(8):857‐863.  217  (9) Sterling TR, Zhao Z, Khan A, Chaisson RE, Schluger N, Mangura B, et al. Mortality in a large tuberculosis treatment trial: Modifiable and non‐modifiable risk factors. International Journal of Tuberculosis and Lung Disease 2006 May;10(5):542‐549. (10) Rao VK, Iademarco EP, Fraser VJ, Kollef MH, Rao VK, Iademarco EP, et al. The impact of comorbidity on mortality following in‐hospital diagnosis of tuberculosis. Chest 1998 Nov;114(5):1244‐52. (11) Borgdorff MW, Veen J, Kalisvaart NA, Nagelkerke N. Mortality among tuberculosis patients in the netherlands in the period 1993‐1995. European Respiratory Journal 1998 Apr;11(4):816‐ 820. (12) Barker RD, Nthangeni ME, Millard FJ, Barker RD, Nthangeni ME, Millard FJC. Is the distance a patient lives from hospital a risk factor for death from tuberculosis in rural South Africa? International Journal of Tuberculosis & Lung Disease 2002 Feb;6(2):98‐103. (13) Bustamante‐Montes LP, Escobar‐Mesa A, Borja‐Aburto VH, Gomez‐Munoz A, Becerra‐ Posada F, Bustamante‐Montes LP, et al. Predictors of death from pulmonary tuberculosis: the case of Veracruz, Mexico. International Journal of Tuberculosis & Lung Disease 2000 Mar;4(3):208‐15. (14) Xie HJ, Enarson DA, Chao CW, Allen EA, Grzybowski S, Xie HJ, et al. Deaths in tuberculosis patients in British Columbia, 1980‐1984. Tubercle & Lung Disease 1992 Apr;73(2):77‐82. (15) Walpola HC, Siskind V, Patel AM, Konstantinos A, Derhy P. Tuberculosis‐related deaths in Queensland, Australia, 1989‐1998: characteristics and risk factors. International Journal of Tuberculosis & Lung Disease 2003 Aug;7(8):742‐750. (16) Fielder JF, Chaulk CP, Dalvi M, Gachuhi R, Comstock GW, Sterling TR. A high tuberculosis case‐fatality rate in a setting of effective tuberculosis control: implications for acceptable treatment success rates. International Journal of Tuberculosis & Lung Disease 2002 Dec;6(12):1114‐1117. 218  (17) de Meer G, van Geuns HA. Rising case fatality of bacteriologically proven pulmonary tuberculosis in The Netherlands. Tubercle & Lung Disease 1992 Apr;73(2):83‐86. (18) Kourbatova EV, K. LM,Jr, Romero J, Kraft C, del Rio C, Blumberg HM, et al. Risk factors for mortality among patients with extrapulmonary tuberculosis at an academic inner‐city hospital in the US. European Journal of Epidemiology 2006;21(9):715‐21. (19) Pablos‐Mendez A, Sterling TR, Frieden TR. The relationship between delayed or incomplete treatment and all‐cause mortality in patients with tuberculosis. JAMA 1996 Oct 16;276(15):1223‐1228. (20) Enarson DA, Grzybowski S, Dorken E. Failure of diagnosis as a factor in tuberculosis mortality. Can.Med.Assoc.J. 1978 Jun 24;118(12):1520‐1522. (21) Abos‐Hernandez R, Olle‐Goig JE. Patients hospitalised in Bolivia with pulmonary tuberculosis: risk factors for dying. International Journal of Tuberculosis & Lung Disease 2002 Jun;6(6):470‐474. (22) Connolly C, Davies GR, Wilkinson D. Impact of the human immunodeficiency virus epidemic on mortality among adults with tuberculosis in rural South Africa, 1991‐1995. International Journal of Tuberculosis and Lung Disease 1998;2(11):919‐925. (23) Zahar JR, Azoulay E, Klement E, De Lassence A, Lucet JC, Regnier B, et al. Delayed treatment contributes to mortality in ICU patients with severe active pulmonary tuberculosis and acute respiratory failure. Intensive Care Med. 2001 Mar;27(3):513‐520. (24) Mathur P, Sacks L, Auten G, Sall R, Levy C, Gordin F. Delayed diagnosis of pulmonary tuberculosis in city hospitals. Arch.Intern.Med. 1994 Feb 14;154(3):306‐310. (25) Katz I, Rosenthal T, Michaeli D. Undiagnosed tuberculosis in hospitalized patients. Chest 1985 Jun;87(6):770‐774.  219  (26) Penner C, Roberts D, Kunimoto D, Manfreda J, Long R. Tuberculosis as a primary cause of respiratory failure requiring mechanical ventilation. American Journal of Respiratory & Critical Care Medicine 1995 Mar;151(3 Pt 1):867‐872. (27) Sacks LV, Pendle S, Sacks LV, Pendle S. Factors related to in‐hospital deaths in patients with tuberculosis. Archives of Internal Medicine 1998 Sep 28;158(17):1916‐22. (28) Doherty MJ, Spence DP, Davies PD. Trends in mortality from tuberculosis in England and Wales: effect of age on deaths from non‐respiratory disease. Thorax 1995 Sep;50(9):976‐979. (29) Tocque K, Convrey RP, Bellis MA, Beeching NJ, Davies PD. Elevated mortality following diagnosis with a treatable disease: tuberculosis. International Journal of Tuberculosis & Lung Disease 2005 Jul;9(7):797‐802. (30) Churchyard GJ, Kleinschmidt I, Corbett EL, Murray J, Smit J, De Cock KM. Factors associated with an increased case‐fatality rate in HIV‐infected and non‐infected South African gold miners with pulmonary tuberculosis. International Journal of Tuberculosis and Lung Disease 2000;4(8):705‐712. (31) Kourbatova EV, Borodulin BE, Borodulina EA, del Rio C, Blumberg HM, K. LM,Jr, et al. Risk factors for mortality among adult patients with newly diagnosed tuberculosis in Samara, Russia. International Journal of Tuberculosis & Lung Disease 2006 Nov;10(11):1224‐30. (32) Dewan PK, Arguin PM, Kiryanova H, Kondroshova NV, Khorosheva TM, Laserson K, et al. Risk factors for death during tuberculosis treatment in Orel, Russia. International Journal of Tuberculosis and Lung Disease 2004 May;8(5):598‐602. (33) World Health Organization. TB Data: Table A‐1 Estimated epidemiological burden of TB, all forms, 1990–2008. 2009; Available at: http://www.who.int/tb/publications/global_report/2009/update/a‐1_full.pdf. Accessed June/12, 2010.  220  (34) Zafran N, Heldal E, Pavlovic S, Vuckovic D, Boe J, Zafran N, et al. Why do our patients die of active tuberculosis in the era of effective therapy? Tubercle & Lung Disease 1994 Oct;75(5):329‐ 33. (35) Khan K, Campbell A, Wallington T, Gardam M. The impact of physician training and experience on the survival of patients with active tuberculosis. CMAJ Canadian Medical Association Journal 2006 September 26;175(7):749‐753. (36) Public Health Agency of Canada and Canadian Lung Association. Canadian Tuberculosis Standards, 6th Edition. 2007; Available at: http://www.phac‐aspc.gc.ca/tbpc‐ latb/pubs/pdf/tbstand07_e.pdf. Accessed March/31, 2010. (37) Health Canada. A Statistical Profile on the Health of First Nations in Canada:Self‐rated Health and Selected Conditions, 2002 to 2005. 2009; Available at: http://www.hc‐ sc.gc.ca/fniah‐spnia/alt_formats/pdf/pubs/aborig‐autoch/2009‐stats‐profil‐vol3/2009‐stats‐ profil‐vol3‐eng.pdf. Accessed April/02, 2010. (38) Health Canada. Evaluation Strategies in Aboriginal Substance Abuse Programs: A Discussion. 1998; Available at: http://www.hc‐sc.gc.ca/fniah‐spnia/alt_formats/fnihb‐ dgspni/pdf/pubs/ads/literary_examen_review‐eng.pdf. Accessed April/02, 2010. (39) Health Canada. Heathy Canadians: A Federal Report on Comparable Health Indicators 2006. 2006; Available at: http://www.hc‐sc.gc.ca/hcs‐sss/alt_formats/hpb‐ dgps/pdf/pubs/2006‐fed‐comp‐indicat/2006‐fed‐comp‐indicat‐eng.pdf. Accessed April/02, 2010. (40) MacMillan HL, MacMillan AB, Offord DR, Dingle JL. Aboriginal health. CMAJ Canadian Medical Association Journal 1996 Dec 1;155(11):1569‐1578. (41) Public Health Agency of Canada. HIV/AIDS in Canada‐ Surveillance Report to June 30, 2006. 2006; Available at: http://www.phac‐aspc.gc.ca/publicat/aids‐sida/haic‐vsac0606/pdf/haic‐ vsac0606.pdf. Accessed March/31, 2010. 221  (42) Bakhshi SS, Hawker J, Ali S. Tuberculosis mortality in notified cases from 1989‐1995 in Birmingham. Public Health 1998 May;112(3):165‐168. (43) Humphries MJ, Byfield SP, Darbyshire JH, Davies PD, Nunn AJ, Citron KM, et al. Deaths occurring in newly notified patients with pulmonary tuberculosis in England and Wales. Br.J.Dis.Chest 1984 Apr;78(2):149‐158. (44) Cullinan P, Meredith SK, Cullinan P, Meredith SK. Deaths in adults with notified pulmonary tuberculosis 1983‐5. Thorax 1991 May;46(5):347‐50. (45) Lawn SD, Acheampong JW, Lawn SD, Acheampong JW. Pulmonary tuberculosis in adults: factors associated with mortality at a Ghanaian teaching hospital. West African Journal of Medicine 1999 Oct‐Dec;18(4):270‐4. (46) Wang J‐, Hung C‐, Sheng W‐, Wang J‐, Chang S‐, Luh K‐. Prognosis of tuberculous meningitis in adults in the era of modern antituberculous chemotherapy. Journal of Microbiology, Immunology and Infection 2002 Dec;35(4):215‐222. (47) Dube MP, Holtom PD, Larsen RA. Tuberculous meningitis in patients with and without human immunodeficiency virus infection. Am.J.Med. 1992 Nov;93(5):520‐524. (48) Kent SJ, Crowe SM, Yung A, Lucas CR, Mijch AM. Tuberculous meningitis: a 30‐year review. Clinical Infectious Diseases 1993 Dec;17(6):987‐994. (49) Hosoglu S, Geyik MF, Balik I, Aygen B, Erol S, Aygencel TG, et al. Predictors of outcome in patients with tuberculous meningitis. International Journal of Tuberculosis & Lung Disease 2002 Jan;6(1):64‐70. (50) Murray J, Sonnenberg P, Shearer SC, Godfrey‐Faussett P. Human immunodeficiency virus and the outcome of treatment for new and recurrent pulmonary tuberculosis in African patients. American Journal of Respiratory and Critical Care Medicine 1999;159(3):733‐740.  222  (51) Hansel NN, Merriman B, Haponik EF, Diette GB, Hansel NN, Merriman B, et al. Hospitalizations for tuberculosis in the United States in 2000: predictors of in‐hospital mortality. Chest 2004 Oct;126(4):1079‐86. (52) Zachariah R, Spielmann MP, Harries AD, Salaniponi FML. Moderate to severe malnutrition in patients with tuberculosis is a risk factor associated with early death. Trans.R.Soc.Trop.Med.Hyg. 2002;96(3):291‐294. (53) Gustafson P, Gomes VF, Vieira CS, Samb B, Naucler A, Aaby P, et al. Clinical predictors for death in HIV‐positive and HIV‐negative tuberculosis patients in Guinea‐Bissau. Infection 2007 Apr;35(2):69‐80. (54) Hargreaves NJ, Kadzakumanja O, Whitty CJ, Salaniponi FM, Harries AD, Squire SB. 'Smear‐ negative' pulmonary tuberculosis in a DOTS programme: poor outcomes in an area of high HIV seroprevalence. International Journal of Tuberculosis & Lung Disease 2001 Sep;5(9):847‐854. (55) Santha T, Garg R, Frieden TR, Chandrasekaran V, Subramani R, Gopi PG, et al. Risk factors associated with default, failure and death among tuberculosis patients treated in a DOTS programme in Tiruvallur District, South India, 2000. International Journal of Tuberculosis and Lung Disease 2002 Sep;6(9):780‐788. (56) Garcia‐Garcia Mde L, Ponce‐De‐Leon A, Garcia‐Sancho MC, Ferreyra‐Reyes L, Palacios‐ Martinez M, Fuentes J, et al. Tuberculosis‐related deaths within a well‐functioning DOTS control program. Emerging Infectious Diseases 2002 Nov;8(11):1327‐33. (57) Kolappan C, Subramani R, Karunakaran K, Narayanan PR. Mortality of tuberculosis patients in Chennai, India. Bull.World Health Organ. 2006;84(7):555‐560. (58) Health Canada. Canadian Addiction Survey (CAS), Detailed Report, March 2005. 2005; Available at: http://www.ccsa.ca/2005%20CCSA%20Documents/ccsa‐004028‐2005.pdf. Accessed March/31, 2010.  223  (59) Patra J, Taylor B, Rehm JT, Baliunas D, Popova S. Substance‐attributable morbidity and mortality changes to Canada's epidemiological profile: measurable differences over a ten‐year period. Canadian Journal of Public Health.Revue Canadienne de Sante Publique 2007;98(3):228‐ 234. (60) Oyefeso A, Ghodse H, Clancy C, Corkery J, Goldfinch R. Drug abuse‐related mortality: a study of teenage addicts over a 20‐year period. Social Psychiatry & Psychiatric Epidemiology 1999 Aug;34(8):437‐441. (61) Paulozzi LJ, Annest JL. US data show sharply rising drug‐induced death rates. Injury Prevention 2007 Apr;13(2):130‐132. (62) Davis CE,Jr, Carpenter JL, McAllister CK, Matthews J, Bush BA, Ognibene AJ. Tuberculosis. Cause of death in antibiotic era. Chest 1985 Nov;88(5):726‐729. (63) Ellis ME, Webb AK. Cause of death in patients admitted to hospital for pulmonary tuberculosis. Lancet 1983 Mar 26;1(8326 Pt 1):665‐667.  224  CHAPTER 7: SUMMARY, CONTRIBUTIONS, POLICY IMPLICATIONS & RECOMMENDATIONS, FUTURE RESERACH AND CONCLUSIONS  225  7.1 SUMMARY OF OBJECTIVES This PhD thesis is a population‐based research study of TB in British Columbia. Issues examined in this research project are rate, patterns and determinants of recurrence, re‐infection and mortality related to TB. The body of the thesis consists of five manuscripts, which are presented in Chapters from 2 to 6. The primary focus of Chapter 2 is the estimation of incidence of recurrence and the risk factors for the development of recurrence. Chapter 3 is an estimation of re‐infection and relapse as a mechanism of recurrent TB. Chapter 4 is a systematic review of the published literature on TB‐related mortality. Chapters five and six discuss all‐cause mortality and cause‐specific mortality among TB patients respectively. This is the concluding chapter that summarizes key findings from Chapter two to six, and outlines implications for future research and intervention efforts.  7.2 SUMMARY OF FINDINGS This research project adds to the growing body of literature indicating that morbidity and mortality related to TB remains a significant public health problem. Based on the available literature, there has been limited research on TB recurrence, re‐infection and TB‐related mortality in Canada. To our knowledge, there is no Canadian study that has estimated the incidence of recurrence and re‐infection. The recurrence study conducted in the Province of Manitoba used a case‐control design (1). In addition, there is a lack of longitudinal population‐ based data on TB recurrence and mortality in Canada and internationally. Moreover, a limited number of studies investigated specific causes of death among TB patients and identified risk factors among TB patients who died primarily due to TB (2,3). This research project therefore addressed several research gaps by analyzing data from a population based TB cohort in BC.  Chapter 2 investigated the incidence of recurrence, including its risk factors. This study of recurrence shows that the proportion of recurrent/relapse cases (prevalent cases) during the study period is relatively high and has not decreased over time. In addition, BC has a higher relapse/recurrence rate compared to the Canadian national rate as well as other major 226  provinces of Canada (4). The stable prevalence of recurrence is an issue of concern due to this being a source of ongoing increased risk of infection.  The incidence of recurrence was relatively low (370 per 100, 000 person‐years), but not as low as what found in a recent study conducted in a similar setting in Australia (5). Moreover, the current research identified several important risk factors for recurrent TB in a population‐based cohort. Place of initial diagnosis (outside of Canada vs. in BC), treatment related factors such as incomplete treatment and poor‐adherence to treatment were the strongest predictors of recurrence. Besides these factors, foreign‐birth, substance abuse and HIV/AIDS have consistently emerged as risk factors for recurrence.  Chapter 3 estimated the incidence of re‐infection as a mechanism of recurrence. The recurrence of TB among culture‐confirmed patients with successful treatment was relatively low in an area with low rates of TB disease (169 per 100,000 PYs). Re‐infection was also an infrequent cause of recurrence in this setting, which contributed 8% of the total recurrent cases (out of 24 recurrent cases, 2 cases were re‐infected with a new strain of MTB). In the current TB cohort, the incidence of re‐infection was 19 per 100,000 person‐years among culture‐positive patients with history of successful treatment. The Provincial TB incidence rate was 7.4 per 100, 000 in 2004 (6). Culture proven cases represent approximately 80 percent of the total TB cases in BC (6). The rate of TB re‐infection after successful treatment did not parallel the local TB incidence and was three‐fold higher compared to the incidence of TB in the population.  Chapter 4 presented the findings from a systematic review focusing on mortality among TB patients. Globally, mortality among TB patients was high, ranging from 2.2% (7) to 65.2% (8). The overall crude mortality (proportion of patients dying from TB) among TB‐patients was 14.5% during treatment (Minimum: 2.2%, Median: 10.3%, Maximum: 39.7%). Few studies investigated the specific cause of death among TB patients. The lowest TB‐related death (proportion of patients dying from TB) was 0.1% (9) and the highest was 15.6% (10). A wide discrepancy in the proportion of subjects with TB‐related death (which ranged from 1.4% to 227  75.3%) was found across studies, reflecting the differences in methodology (such as definition of TB‐caused death) and study populations. Mortality was two‐fold higher in high‐burden versus low‐burden countries. The higher mortality among HIV positive TB patients compared to that of HIV non‐infected patients was observed in the majority of studies. Besides HIV, old age was also identified as a significant risk factor. Other important risk factors were unemployment, homelessness, co‐morbid conditions (such as malignancy, renal failure and malnutrition), extra pulmonary TB, smear negative TB and MDR TB.  Chapter 5 provided the rates and patterns of mortality among TB patients and factors associated with mortality (all‐cause) among TB patients. This study demonstrated that a relatively higher excess mortality exists in the TB cohort than the general population of BC. Overall, the age standardized mortality ratio (ASMR) showed an increasing trend from 1990 to 2006 (Range: 9‐29, Median: 16). In addition, TB patients died at a younger age compared to the general population. The overall occurrence of death was 29 per 1000 PYs. The Kaplan Meier (KM) survival curve demonstrated that the cumulative mortality at the first 6 (393 deaths) and 12 months (466 deaths) was 8% and 10% respectively.  This all‐cause mortality study also identified a number of important risk factors for mortality among TB patients in BC. Increasing age, male gender, Canadian‐born non‐aboriginal status, Aboriginal ethnicity, Caucasian ethnicity, culture confirmed TB, miliary/disseminated TB, far advanced PTB, HIV/AIDS, malignancy, renal failure, alcoholism and substance abuse were significantly associated with mortality in multivariable setting. Aboriginal people had not only one of the highest risks of mortality but were also more likely to die at a younger age. Deaths in younger TB patients were largely attributable to behavioral factors such as substance abuse and alcoholism, and HIV/AIDS. In contrast deaths in older patients were more likely to be due to associated co‐morbid‐conditions including malignancy and renal failure.  Chapter 6 investigated the rates and determinants of death caused by TB (TB was the underlying cause or a contributing factor). Despite a high crude death rate among TB‐patients 228  (1069/5408=20%), death specifically related to TB was considerably lower. Overall, the percentage of TB‐caused deaths (TB was the underlying cause) and TB‐contributed deaths (TB was a contributing factor) was 2.0% (109/5408) and 3.3% (177/5408) respectively. This study identified a number of important risk factors for dying from TB in resource‐rich setting. Failure of diagnosis continues to be one of the biggest contributors to TB‐related deaths. Aboriginal people were at greatest risk of dying from TB. In addition, miliary TB, meningeal TB and advanced PTB were other important predictors of TB‐caused mortality. Among co‐morbidities, alcoholism, malnutrition and renal failure were strongly associated with increased risk of dying specifically from TB. However, the effects of co‐morbidity including HIV were much higher among patients whose deaths were partly attributed by TB (TB was a contributing factor only).  7.3 UNIQUE CONTRIBUTIONS TO RESEARCH The methodology and analyses of this thesis provide an important body of evidence that makes significant additions to the epidemiological literature, specifically related to TB recurrence and mortality. Overall, this research project underscores the public‐health importance of conducting longitudinal analyses with population‐based data. Globally, a limited number of countries maintain a central registry of TB patients that contains a broad range of information including demographic, clinical, behavioral and radiological data. In Canada, not all provinces have a centralized TB control system.  Using this population level information (rather than those obtained in institutional or selected hospital settings) provides more precise risk estimates and a high level of external validity (generalizability) for study findings. In addition, combining clinical, radiological and demographic information into a single model also facilitates obtaining unbiased estimates of hazard ratios for potential predictors. To the best of our knowledge, data presented in these analyses are one of the largest studies and in addition by observing patients for a period of 17 years, this research project represents one of longest follow‐up studies on TB‐related recurrence and mortality. 229  This is the first Canadian study that provides a population‐based estimate on incidence of TB recurrence and re‐infection. We followed a systematic approach to investigate the recurrence of TB. The quantification of re‐infection demonstrate that community transmission still exists and provides important public health messages for TB control programs including vaccine design, and evaluation of treatment regimens.  The systematic review evaluating TB mortality, is the first to address this question, and as such is another original and important addition to the current epidemiological literature on TB‐ related mortality. This review identifies gaps in the existing literature and identifies variations in mortality rate and factors associated with mortality that exists between high and low‐burden setting. This review also highlights the profound effect of HIV infection on TB‐related mortality throughout the world.  The increased risk of mortality from TB among patients of Aboriginal ancestry is a unique finding that makes an important contribution to the literature. This highlights the increased social and environmental vulnerabilities in this population.  7.4 LIMITATIONS There are several limitations in this research project that warrants consideration in interpreting of some of the findings. Although the administrative databases provide information that is not subject to recall biases, there are some inherent limitations associated with administrative databases. Some of the demographic and clinical information for patients was collected through self‐administered questionnaire. Therefore, self‐reporting bias and incomplete observations are potential limitations of the current TB databases. The proportion of subjects with incomplete information was greatest for variables that documented co‐morbidities. We attempted to reduce missing data on related outcomes such as TB recurrence and TB‐related death through the review of physician notes associated each individual TB‐patient.  230  We considered that patients with unknown co‐morbidity status were more representative of patients without co‐morbid status (no response) due to nature of collection of information followed in the TB control division. Therefore, the effects of co‐morbidity on outcome (recurrence or mortality) were evaluated in compared to patients without comorbidity or unknown status (combined into a single category). However, we conducted separate multivariable models (sensitivity analysis) for co‐morbidities restricting to patients with valid responses to comorbidity status. Patterns of significant variables were similar in both scenarios (primary analysis and sensitivity analysis). It is noteworthy to mention that due to the retrospective nature of data collection, an administrative database does not allow us to accurately investigate all risk factors.  Absence of active follow could be another limitation for both recurrent and mortality studies. Due to passive nature follow up, this study might not account for a few cases which have moved from BC. Some of these cases could have outcome that might lead to an under‐ estimation of incidence of recurrence or mortality. However, a study conducted by Onofre et al. 2004 in BC among TB contacts using the BC linked health databases showed that migration of TB patients was low (11).  The possibility of re‐infection by the same strain could be a potential limitation of the re‐ infection study. However, we do not think this would significantly affect the re‐infection estimates due to the high level of heterogeneity that exists among the TB strains currently transmitted within the community and the high discriminatory power of MIRU fingerprinting method (12‐15).  7.5 POLICY IMPLICATIONS AND RECOMMENDATIONS FOR TB CONTROL PROGRAM This thesis offers an important body of evidences regarding the TB recurrence and TB‐related mortality which have important implications for control and prevention of TB both in Canada 231  and internationally. Each chapter (Chapters 2 to 6) provides recommendations that are specific to that chapter’s findings, but it is worthwhile to highlight several recommendations in this concluding chapter due to the cumulative evidence across studies and have a direct relationship to health policy. Several factors were significantly associated with TB recurrence and mortality; some of these factors are potentially modifiable.  This research project provides empiric evidence to suggest that excessive mortality occurs among TB patients compared to the general population. In addition TB patients are dying at a much younger age. The excess burden of recurrence and mortality among TB patients both at an individual level and on the health care system highlights aspects of TB control program that could be improved. Moreover, these findings supports the need for additional funding for research, prevention, treatment and effective interventions to reduce the burden of TB as well as morbidity and mortality associated with TB, especially in certain ethnic groups (such as Aboriginals).  Place of diagnosis (outside of Canada) for the initial episode was one of the strongest predictors of TB‐related recurrence. This was more reflective of patients who were born outside of Canada with a prior history of TB. Vancouver is a place of immigrants, especially from Asian countries (such as China, Vietnam, Hong Kong, Philippines and India). Resources and infrastructure for TB‐control programs are severely limited in these countries. Moreover, in foreign‐born TB patients, about 45% of recurrence occurred within three years after their arrival in Canada. These findings have practical implications to local TB control program and provide an empiric evidence to prioritize the development and implementation of effective screening and treatment protocols for this high risk group (foreign‐born patients with prior history of TB at their host country) after their arrival in Canada. Moreover, these findings highlight that local TB control activities are not adequate to combat TB‐related morbidity locally unless there are initiatives in immigrant’s origin countries to reduce the burden of TB there. For foreign‐born patients with a history of TB, interventions such as aggressive monitoring, post‐arrival active surveillance and prophylaxis should be prioritized in the current TB control program. 232  The further recommendations for the prevention of recurrence stem from the increased likelihood of recurrence among TB patients who had a history of incomplete treatment as well as non‐compliance to treatment. Despite increased efforts, non‐completion of treatment as well as poor‐adherence to treatment continues to occur in resource‐rich countries like Canada. Explanations for incomplete treatment fall into two categories: one is intrinsic to treatment (such as longer duration of treatment and side effects) and other is related to individual/behavioral characteristics (such as substance abuse, alcoholism and poor‐compliance with the recommended regimens). Factors related to treatment duration cannot currently be modified or reversed, although the development of more effective therapies in the future could make this a modifiable process. However, interventions emphasizing behavioral characteristics (such as poor‐adherence) could have greater potential to improve the treatment outcome and eventually decrease the recurrence rate.  Enhanced surveillance activities and aggressive monitoring targeting high‐risk patients (who default or did not comply the initial therapy) should be prioritized in TB control programs. Moreover, efforts including involvement of well‐trained and committed outreach workers, offering of incentives (such as bus tickets and meal voucher) and forceful confinement for defaulters are needed in order to improve adherence to treatment to prevent the recurrence of TB. Increased awareness/heath education among patients and other family members, and identifying patients at risk of non‐adherence at an early stage could have beneficial impact on treatment outcome by increasing adherence rate among potential defaulters.  This research project identifies an elevated risk (four‐fold increased risk compared to reference category) of mortality from TB among Aboriginal persons, even after accounting the effects of important co‐morbid conditions. Another disturbing finding was the decreased life‐expectancy for Aboriginal subjects with TB. The average age of death for Aboriginal subjects with TB was 54 years, about twenty‐years younger than foreign‐born TB‐patients. Explanations for this increased risk of dying among Aboriginal TB patients are multifaceted and could be outcome of a complex interaction between social, environmental and individual characteristics. 233  One potential explanation for the disproportionate mortality among Aboriginal people might be attributable to life style that increases their vulnerability to death. In the current TB cohort, the prevalence of drug abuse and alcoholism among aboriginal people was 15% and 23% respectively, which were higher compared to Canadian‐born non‐aboriginal (10% and 10%) and much higher compared to the foreign‐born (1% and <1%). There is cumulative evidence about the over‐representation of risky behaviors (such as drug addictions and alcoholism), and several co‐morbid conditions (such as Diabetes, HIV and Hepatitis C) among Aboriginal communities compared to the general population (16‐20), eventually leading to a worse prognosis. Although socio‐economic context (such as poverty, homelessness, crowding, traumatic life experience and lack of self‐esteem) of Aboriginal subjects are not assessed directly in the thesis, the detrimental effects of these factors towards their decreased life expectancy cannot be ruled out.  Other potential reasons for the increased risk of dying was due to the greater likelihood of severe forms of TB (such as miliary TB and meningeal TB) as well as post‐mortem TB among the Aboriginal population. The current study demonstrated a disproportionate burden of post‐ mortem cases among Aboriginal people (Chapter 6, Table 6.10). Aboriginal people were also over‐represented in marginalized populations that were also hard to reach. As a result, the early detection of diseases sometimes becomes more challenging in Aboriginal population, which eventually increases the likelihood of a missed diagnosis. These might be related to patient’s decreased desire to seek care and barriers to access especially given the remote location of many of the Aboriginal communities (21). Despite universal health care, failure to seek care and barriers to access remain issues of concern. Although, this study did not directly evaluate the influence of cultural barriers, but the potential effects of cultural factors on decreased desire to seek care and subsequent mortality could not be ruled out in this setting. With respect to health care and delivery system, all these explanations for increased mortality among aboriginal TB patients have differential implications to the TB control program.  234  Alleviation of poverty and other socio‐cultural factors are not the direct responsibilities of a TB control program. However, due to complex interaction between social and environmental factors, there is an immediate urgency for a close collaboration between TB program and other related service providers targeting Aboriginal communities. Strengthening of TB control initiatives targeting Aboriginal population should be employed immediately in order to reduce the risk of death. Culturally safe and effective interventions would be valuable in this setting. In addition, there is an increased need to develop and respond with Aboriginal‐led prevention strategies to reduce the burden of HIV, substance abuse and other high risk behaviors among Aboriginal persons.  This research project clearly demonstrates that HIV co‐infection substantially increases the risk of recurrence and mortality among TB‐patients. The systematic review on risk factors of TB‐ related mortality also confirmed this association. Resolution to this problem is multifaceted. Prevention of HIV would be ideal solution. People with HIV are many‐ times (20‐ 40) more likely to develop active TB disease and subsequently to die from TB (22). Therefore, another essential strategy would be decrease the burden of TB among HIV‐infected persons. The WHO recommends several strategies (such as intensified case finding, INH prophylaxis and infection control) for people living with HIV in order to decrease morbidity and mortality related to this fatal interaction (22). The penultimate strategy would be improved survival of dually infected persons. In the context of health care delivery, all these strategies provide different implications to TB control program locally and internationally.  Harm‐reduction strategies would be critical for targeted prevention of HIV infection (since IDU is the well‐known cause of HIV transmission in North America), but this is very challenging and requires an integrated/collaborative approach between several service providers. Providing INH prophylaxis for HIV‐infected persons is straight‐forward and is implemented in many TB control programs, mostly in resource‐rich setting. In addition, there is cumulative evidence that proper management of HIV–infected persons including aggressive monitoring of patients, long‐term‐ follow up and introduction of anti‐viral therapy, especially HARRT have greater potentials to 235  improve the survival among HIV‐TB co‐infected persons (23,24). This offers an opportunity for improved management of dually infected persons.  The recurrent study has also shown the increased frequency of non‐completion (two‐fold than the overall non‐completion rate) and non‐adherence to treatment (two‐fold than the overall non‐compliance rate) in HIV co‐infected patients. Therefore, interventions (such as establishment of a longitudinal surveillance system and co‐ordination with HIV care providers) targeting HIV infected persons should be employed so that this high‐risk population complete the treatment regimens within an acceptable time‐frame and with good adherence.  Drug abuse and alcoholism, which were associated with higher incompletion rates and poor adherence to treatment, were more frequent among patients with a recurrence. These behavioral factors are subject to modification. Patients who use drugs (injection or non‐ injection) or alcohol are more likely not to comply with recommended treatment regimens, which eventually increase the likelihood of recurrence. In addition, the risk of mortality was significantly increased among TB patients who had a history of substance abuse and alcoholism. In this regard, a systematic approach including the identification and monitoring TB patients with substance abuse and/or alcoholism should be prioritized in the current TB control program. Moreover, these patients also offer an opportunity for harm‐reduction strategies addressing their addictions.  Malnutrition was associated with increased mortality in this TB‐cohort. Effective interventions (such as aggressive monitoring and nutritional supplement program during treatment phase) targeting this high risk population are urgently required.  236  7.6 FUTURE RESEARCH The findings of current research project demonstrate several key areas for future research. The thesis highlights the need for interventional research that targets the high‐risk populations identified here. The efficacy and cost‐effectiveness of such interventions need to be evaluated.  Although an increased risk of dying was observed in certain ethnic groups, we could not distinguish whether this was related to biological susceptibility or related to social and environmental factors and their interaction. This uncertainty highlights the need for additional research in TB patients in evaluating the impact of place and environment on the prevention of TB‐related morbidity and mortality. The identification of social determinants using GIS mapping will be particularly helpful in this regard.  Despite the presence of universal health care access and effective chemotherapy for TB treatment, Aboriginal people are at increased risk of dying from TB. We could not evaluate if any cultural barrier directly attribute to this increased mortality. Qualitative research evaluating potential cultural barriers among Aboriginal persons may improve our understanding the potential cause of the increased mortality among TB‐patients.  Foreign‐born persons account for the majority of TB‐cases both in BC and in the rest of Canada. There was an increased rate of recurrence in this population. However, the risk of dying was disproportionately low (compared to Canadian‐born non‐Aboriginal and Aboriginal) in this group. Future investigation of risk factors stratified by country of origin as well as ethnic status might be crucial in providing evidence‐based guidelines in controlling and preventing tuberculosis locally and internationally.  There was evidence to suggest the negative effect of malnutrition among TB‐patients in resource‐rich countries. We could not distinguish whether malnutrition precedes the disease or malnutrition was the outcome of the disease itself. Moreover, the causal mechanism between increased mortality and malnutrition in TB‐patients is unknown. Further research with 237  better documentation of malnutrition and its long‐term effect on outcomes in TB‐as well as any associated morbidity and mortality would be helpful. In addition, the potential benefit of interventions such as nutritional supplement program for patients with malnutrition should be formally evaluated in prospective trials.  Failure of diagnosis was a risk factor in this study. Despite universal health care policy in Canada, the occurrence of delay in diagnosis and initiation of therapy of TB patients warrant immediate consideration. Future research to evaluate the effect of delay in treatment, in particular health system delay (diagnostic delay) and patients delay in seeking care in a Canadian setting would be valuable.  7.7 CONCLUSIONS Overall, this research project offers an important body of evidence regarding the recurrence of TB and TB‐related mortality in a population‐based TB‐cohort. The findings have practical implications for the control and prevention of TB locally and internationally. Despite the availability of an effective chemotherapy and extensive efforts to reduce the burden of TB, community transmission and mortality from TB is not infrequent even in resource‐rich settings. Patients whose initial diagnosis took place outside of BC had at least five times higher risk for the development of recurrence compared to patients with a diagnosis in BC. Treatment related factors such as incomplete treatment and poor‐adherence to treatment were other important predictors of recurrence.  Not surprisingly there is an excess mortality among TB patients compared to the general population. An issue of concern is the variation in mortality risk especially being very high in Aboriginal persons. The risk of recurrence and mortality are significantly elevated among TB patients with HIV‐co‐infection. Failure of diagnosis or missed diagnosis continues to be a major contributor to TB‐caused deaths. Among co‐morbidities, alcoholism, malnutrition and renal failure were strongly associated with increased risk of dying specifically from TB. However, the 238  effects of co‐morbidity including HIV were much higher among patients whose deaths were partly attributed by TB (TB was a contributing factor only). Some of these factors are potentially modifiable. Effective interventions targeting these high‐risk populations are urgently required in order to prevent recurrence and mortality related to TB.  239  7.8 REFERENCES (1) Johnson IL, Thomson M, Manfreda J, Hershfield ES. Risk factors for reactivation of tuberculosis in Manitoba. CMAJ Canadian Medical Association Journal 1985 Dec 15;133(12):1221‐1224. (2) Walpola HC, Siskind V, Patel AM, Konstantinos A, Derhy P. Tuberculosis‐related deaths in Queensland, Australia, 1989‐1998: characteristics and risk factors. International Journal of Tuberculosis & Lung Disease 2003 Aug;7(8):742‐750. (3) Bustamante‐Montes LP, Escobar‐Mesa A, Borja‐Aburto VH, Gomez‐Munoz A, Becerra‐ Posada F, Bustamante‐Montes LP, et al. Predictors of death from pulmonary tuberculosis: the case of Veracruz, Mexico. International Journal of Tuberculosis & Lung Disease 2000 Mar;4(3):208‐15. (4) Public Health Agency of Canada. Tuberculosis in Canada 2007, Pre‐Release. 2007; Available at: http://www.phac‐aspc.gc.ca/publicat/2008/tbcanpre07/pdf/tbpre2007‐eng.pdf. Accessed March/31, 2010. (5) Dobler CC, Crawford AB, Jelfs PJ, Gilbert GL, Marks GB. Recurrence of tuberculosis in a low‐ incidence setting. European Respiratory Journal 2009 Jan;33(1):160‐167. (6) BC Center for Disease Control. TB Annual Report 2004. 2004; Available at: http://www.bccdc.ca/NR/rdonlyres/F80BA3E8‐4E46‐4AFC‐9B05‐ CE3643C2947E/0/2004_TB_Annual.pdf. Accessed March/31, 2010. (7) El‐Sony AI, Khamis AH, Enarson DA, Baraka O, Mustafa SA, Bjune G. Treatment results of DOTS in 1797 Sudanese tuberculosis patients with or without HIV co‐infection. International Journal of Tuberculosis & Lung Disease 2002 Dec;6(12):1058‐1066. (8) Kang'ombe CT, Harries AD, Ito K, Clark T, Nyirenda TE, Aldis W, et al. Long‐term outcome in patients registered with tuberculosis in Zomba, Malawi: mortality at 7 years according to initial  240  HIV status and type of TB. International Journal of Tuberculosis & Lung Disease 2004 Jul;8(7):829‐836. (9) Sterling TR, Zhao Z, Khan A, Chaisson RE, Schluger N, Mangura B, et al. Mortality in a large tuberculosis treatment trial: Modifiable and non‐modifiable risk factors. International Journal of Tuberculosis and Lung Disease 2006 May;10(5):542‐549. (10) L'Ecuyer PB, Woeltje KF, Seiler SM, Fraser VJ. Management and outcome of tuberculosis in two St Louis hospitals, 1988 to 1994. Infection Control & Hospital Epidemiology 1998 Nov;19(11):836‐841. (11) A. Morán Mendoza. The value of the tuberculin skin test size in predicting the development of tuberculosis in contacts of active cases. University of British Columbia: University of British Columbia; 2004. (12) Cowan LS, Mosher L, Diem L, Massey JP, Crawford JT. Variable‐number tandem repeat typing of Mycobacterium tuberculosis isolates with low copy numbers of IS6110 by using mycobacterial interspersed repetitive units. J.Clin.Microbiol. 2002 May;40(5):1592‐1602. (13) Centers for Disease Control and Prevention. Notice to Readers: New CDC Program for Rapid Genotyping of Mycobacterium tuberculosis Isolates . MMWR 2005;54(02):47. (14) Savine E, Warren RM, van der Spuy GD, Beyers N, van Helden PD, Locht C, et al. Stability of variable‐number tandem repeats of mycobacterial interspersed repetitive units from 12 loci in serial isolates of Mycobacterium tuberculosis. J.Clin.Microbiol. 2002 Dec;40(12):4561‐4566. (15) Supply P, Lesjean S, Savine E, Kremer K, van Soolingen D, Locht C. Automated high‐ throughput genotyping for study of global epidemiology of Mycobacterium tuberculosis based on mycobacterial interspersed repetitive units. J.Clin.Microbiol. 2001 Oct;39(10):3563‐3571. (16) Health Canada. A Statistical Profile on the Health of First Nations in Canada:Self‐rated Health and Selected Conditions, 2002 to 2005. 2009; Available at: http://www.hc‐ 241  sc.gc.ca/fniah‐spnia/alt_formats/pdf/pubs/aborig‐autoch/2009‐stats‐profil‐vol3/2009‐stats‐ profil‐vol3‐eng.pdf. Accessed April/02, 2010. (17) Health Canada. Heathy Canadians: A Federal Report on Comparable Health Indicators 2006. 2006; Available at: http://www.hc‐sc.gc.ca/hcs‐sss/alt_formats/hpb‐ dgps/pdf/pubs/2006‐fed‐comp‐indicat/2006‐fed‐comp‐indicat‐eng.pdf. Accessed April/02, 2010. (18) MacMillan HL, MacMillan AB, Offord DR, Dingle JL. Aboriginal health. CMAJ Canadian Medical Association Journal 1996 Dec 1;155(11):1569‐1578. (19) Health Canada. Evaluation Strategies in Aboriginal Substance Abuse Programs: A Discussion. 1998; Available at: http://www.hc‐sc.gc.ca/fniah‐spnia/alt_formats/fnihb‐ dgspni/pdf/pubs/ads/literary_examen_review‐eng.pdf. Accessed April/02, 2010. (20) Public Health Agency of Canada. HIV/AIDS in Canada‐ Surveillance Report to June 30, 2006. 2006; Available at: http://www.phac‐aspc.gc.ca/publicat/aids‐sida/haic‐vsac0606/pdf/haic‐ vsac0606.pdf. Accessed March/31, 2010. (21) Public Health Agency of Canada and Canadian Lung Association. Canadian Tuberculosis Standards, 6th Edition. 2007; Available at: http://www.phac‐aspc.gc.ca/tbpc‐ latb/pubs/pdf/tbstand07_e.pdf. Accessed March/31, 2010. (22) World Health Organization. Tuberculosis Facts 2009 Update. 2009; Available at: http://www.who.int/tb/publications/2009/factsheet_tb_2009update_dec09.pdf. Accessed April/02, 2010. (23) Burman W, Benator D, Vernon A, Khan A, Jones B, Silva C, et al. Acquired rifamycin resistance with twice‐weekly treatment of HIV‐related tuberculosis. American Journal of Respiratory & Critical Care Medicine 2006 Feb 1;173(3):350‐356.  242  (24) Dheda K, Lampe FC, Johnson MA, Lipman MC. Outcome of HIV‐associated tuberculosis in the era of highly active antiretroviral therapy. J.Infect.Dis. 2004 Nov 1;190(9):1670‐1676.  243  APPENDIX A  244  Table A.1: Socio‐demographic and clinical characteristics of study cohort (n=5403)  Variable  N (%)  Age at diagnosis of primary episode Mean (SD) Median (Range) Biological gender Male Female Birth place Canadian‐ born non‐Aboriginal Canadian‐ born Aboriginal Foreign‐born Unknown Ethnicity Caucasian Aboriginals Chinese South‐East Asian Filipino Vietnamese Others Missing Marital status Single Not single Other Unknown/missing Diabetes Mellitus Yes No Unknown/missing Malnutrition Yes No Unknown/missing Alcoholism Yes No Unknown/missing Drug abuse Yes No Unknown/missing Either HIV or AIDS Yes No Unknown/missing  Years 46.3 (21.9) 44 (1‐104) 2950 (55) 2453 (45) 876 (16) 708 (13) 3605 (67) 214 (4) 937 (17) 708 (13) 1293 (24) 830 (15) 428 (8) 306 (6) 741 (14) 160 (3) 1110 (20) 1564 (29) 252 (5) 2477 (46) 353 (7) 2432 (45) 2618 (48) 74 (1) 2524 (47) 2805 (52) 314 (6) 2476 (46) 2613 (48) 219 (4) 2480 (46) 2689 (50) 296 (5) 2571 (48) 2536 (47)  245  Variable  N (%)  Any Malignancy Yes No Unknown/missing Immunosuppressive medication Yes No Unknown/missing Chronic Renal failure Yes No Unknown/missing  159 (3) 2477 (46) 2767 (51) 123 (2) 2524 (47) 2756 (51) 92 (2) 2526 (47) 2785 (51)  246  Table A.2: Comparisons of socio‐demographic characteristics between TB patients who had a history of recurrence (n=490) and who had not (n=4913) Variable  Age at enrolment visit Mean (SD) Biological gender Male Female Ethnicity Caucasian Aboriginals Chinese South‐East Asian Filipino Vietnamese Others/unknown Marital status Single Not single Birth place Canadian born non‐Aboriginal Canadian born Aboriginal Foreign born Birth Country regions PAHO African Region Eastern Mediterranean European Region Southeast Asia Region Western Pacific Region Place of initial diagnosis In BC Outside of BC  Recurrent (Prevalent) cases n (%)  New cases  Years 32.4 (17.9)  Years 47.7 (21.8)  0.96 (0.96, 0.97)  307 (10) 183 (7)  2643 (90) 2270 (93)  1.44 (1.19, 1.75) Reference  79 (8) 86 (12) 182 (14) 43 (5) 31 (7) 18 (6) 46 (6)  858 (92) 622 (88) 1111 (86) 787 (95) 397 (93) 288 (94) 695 (94)  Reference 1.50 (1.09, 2.07) 1.78 (1.35, 2.35) 0.59 (0.40, 0.87) 0.85 (0.55, 1.31) 0.68 (0.40, 1.15) 0.72 (0.49, 1.05)  90 (8) 213 (14)  1020 (92) 1351 (86)  Reference 1.79 (1.38, 2.32)  58 (7) 86 (12) 342 (9)  818 (93) 622 (88) 3263 (91)  Reference 1.95 (1.38, 2.76) 1.48 (1.11, 1.97)  149 (9) 7 (7) 6 (4) 25 (9) 38 (5) 258 (12)  1536 (91) 90 (93) 132 (96) 245 (91) 724 (95) 1923 (88)  Reference 0.80 (0.37, 1.76) 0.47 (0.20, 1.08) 1.05 (0.67, 1.64) 0.54 (0.38, 0.78) 1.38 (1.12, 1.71)  189 (5) 280 (66)  3843 (95) 142 (34)  Reference 40.09 (31.25, 51.44)  247  Unadjusted Odds Ratio (95% CI)  n (%)  Table A.3: Comparisons of treatment and related behaviors between TB patients who had a history of recurrence (n=490) and who had not (n=4913) Variable  Duration of Treatment Mean (SD) Rx compliance of primary episode Yes No Rx completion of primary episode Complete Incomplete Major mode of Treatment Supervised Self‐administered Culture results Negative Positive Drug resistance status Sensitive Mono/poly resistance MDR Disease type Pulmonary Extra‐pulmonary Both  Recurrent (Prevalent) cases n (%)  New cases  Years 0.89 (0.72)  Years 0.68 (0.32)  3.05 (2.39, 3.89)  95 (2) 35 (7)  3861 (98) 453 (93)  Reference 3.14 (2.11, 4.68)  149 (4) 100 (26)  3741 (96) 281 (76)  Reference 8.94 (6.75, 11.83)  42 (5) 82 (3)  771 (95) 3041 (97)  Reference 0.50 (0.34, 0.72)  8 (1) 95 (2)  634 (99) 3813 (98)  Reference 1.97 (0.96, 4.08)  84 (2) 10 (3) 1 (4)  3464 (98) 304 (97) 22 (96)  Reference 1.36 (0.70, 2.64) 1.87 (0.25, 14.07)  382 (10) 72 (5) 17 (6)  3259 (90) 1381 (95) 273 (94)  2.25 (1.73, 2.92) Reference 1.19 (0.69, 2.06)  248  Unadjusted Odds Ratio (95% CI)  n (%)  Table A.4: Comparisons of co‐morbidity and other clinical characteristics between TB patients who developed a recurrence (n=108) and who did not (n=4356)53  Variable  Diabetes Mellitus Yes No Malnutrition Yes No Alcoholism Yes No Drug abuse Yes No Either HIV or AIDS Yes No Any Malignancy Yes No Immunosuppressive medication Yes No Chronic Renal failure Yes No  53  Number of Incident cases/Total person years  Incidence per 100, 000 person years  Unadjusted Hazard Ratio (95% CI)  5/1714 103/10894  292 946  0.31 (0.12, 0.75) Reference  2/239 106/11471  835 924  0.86 (0.21, 3.49) Reference  7/1390 101/11341  504 891  0.59 (0.27, 1.26) Reference  12/731 96/11332  1641 847  1.80 (0.99, 3.28) Reference  14/838 94/12000  1670 783  1.86 (1.06, 3.26) Reference  2/378 106/11407  529 929  0.56 (0.14, 2.26) Reference  4/479 104/11542  835 908  0.89 (0.33, 2.40) Reference  2/237 106/11407  845 920  0.91 (0.23, 3.69) Reference  ‐ This analysis was restricted to TB patients with valid responses (yes or no) for comorbidity status.  249  Table A.5: Multivariable cox regression analysis to evaluate the effect of co‐morbid conditions on TB recurrence54 Variable  Controlled for age, gender, birth place, culture status, diagnosis place  Controlled for age, gender, birth country regions, culture status, diagnosis place  Controlled for age, gender, birth‐country regions, diagnosis place, treatment completion and compliance  Diabetes mellitus  0.33 (0.13, 0.82)  0.34 (0.13, 0.84)  0.29 (0.09, 0.93)  Alcoholism  0.77 (0.33, 1.80)  0.67 (0.29, 1.56)  0.69 (0.28, 1.72)  Drug abuse  2.67 (1.30, 5.48)  3.30 (1.60, 6.82)  2.71 (1.19, 6.15)  Either HIV or AIDS  2.77 (1.49, 5.16)  3.55 (1.73, 7.29)  3.18 (1.56, 6.49)  Any Malignancy  0.61 (0.15, 2.52)  0.73 (0.18, 3.04)  0.99 (0.23, 4.18)  Renal Failure  1.33 (0.33, 5.47)  1.36 (0.33, 5.56)  1.67 (0.41, 6.89)  Malnutrition  1.19 (0.28, 5.11)  1.40 (0.33, 5.92)  1.12 (0.26, 4.75)  Immunosuppressive medication  1.32 (0.48, 3.64)  1.31 (0.48, 3.60)  1.06 (0.33, 3.41)  54  ‐ This analysis was restricted to TB patients with valid responses (yes or no) for comorbidity status.  250  Table A.6: Risk factors for recurrence among patients with more than 6 months of inactivity  Variable  Unadjusted Hazard Ratio (95% CI)  Age at diagnosis (primary episode) per year Mean (SD)  1.00 (0.99, 1.01)  Female Birth place Canadian born non‐Aboriginal Canadian born Aboriginal Foreign born Birth Country regions Pan‐American Health Region African Region East. Mediterranean/Europe Southeast Asia Region Western Pacific Region Place of Diagnosis Outside of BC In BC Culture positivity Positive Negative/unknown Rx compliance of primary episode Yes No Rx completion of primary episode Complete Incomplete Major mode of Treatment Supervised Self‐administered Disease type Pulmonary Extra‐pulmonary Both Presence of Cavitations Diabetes Mellitus Malnutrition Alcoholism Substance abuse Either HIV or AIDS Any Malignancy Immunosuppressive medication Chronic Renal failure  1.11 (0.74, 1.65) Reference 2.60 (0.98, 6.93) 3.34 (1.45, 7.66) Reference 3.71 (1.26, 10.91) 0.45 (0.11, 1.93) 2.42 (1.29, 4.54) 2.08 (1.29, 3.54) 16.80 (11.16, 25.30) Reference 0.42 (0.28, 0.63) Reference Reference 6.94 (3.97, 12.10) Reference 6.11 (3.76, 9.94) Reference 0.96 (0.52, 1.76) Reference 1.24 (0.81, 1.91) 1.22 (0.49, 3.05) 0.76 (0.19, 3.11) 0.86 (0.35, 2.12) 1.24 (0.17, 8.92) 1.60 (0.74, 3.45) 3.63 (1.83, 7.23) 3.60 (1.91, 6.76) 1.63 (0.40, 6.60) 2.45 (0.90, 6.66) 1.23 (0.17, 8.80)  251  Table A.7: Distribution of treatment incompletion and non‐compliance to treatment across several risk factors Incompletion N (%)  P value  Non‐ compliant N (%)  P value  Overall  275 (7)  446 (11)  Male  153 (8)  Female  122 (7)  Canadian‐born non‐Aboriginal  67 (10)  Canadian‐born Aboriginal  65 (13)  58 (11)  Foreign‐born  133 (5)  273 (10)  Pan‐ American Health Region  144 (12)  African Region  6 (9)  11 (15)  Eastern Mediterranean and European Region Southeast Asia Region  23 (8)  33 (11)  20 (4)  49 (8)  Western Pacific Region  72 (5)  151 (9)  Supervised  61 (9)  Self‐administered  181 (7)  Diabetes mellitus yes  12 (5)  0.180  29 (11)  0.573  Alcoholism yes  28 (15)  <0.001  28 (13)  0.638  HIV/AIDS positive  25 (14)  <0.001  35 (16)  0.022  Drug abuse yes  28 (22)  <0.001  26 (17)  0.056  Malignancy yes  8 (11)  0.220  27 (29)  <0.001  0.194  252 (11)  0.120  194 (10) <0.001  <0.001  0.036  73 (10)  145 (11)  86 (11)  0.701  0.105  0.979  337 (11)  252  Table A.8: Lists of variables that were included in multivariable cox regression model for risk factors of all‐cause mortality among TB patients  Model 3  Model 4  Model 5  Model 6  Model 7  Lists of variables  Adjusted Hazard Ratio (AHR) reported  Age Gender Birthplace Recurrence of TB Culture status Millary TB CNS TB Far advanced TB Moderately advanced TB HIV/AIDS status Age Gender Ethnicity Recurrence of TB Culture status Millary TB CNS TB Far advanced TB Moderately advanced TB HIV/AIDS status Any Malignancy Age Gender Ethnicity Recurrence of TB Culture status Millary TB CNS TB Far advanced TB Moderately advanced TB HIV/AIDS status Chronic Renal Failure Age Gender Ethnicity Recurrence of TB Culture status Millary TB CNS TB Far advanced TB Moderately advanced TB HIV/AIDS status Diabetes Mellitus Age Gender  All variables  253  All variables  Any malignancy  Chronic renal failure  Diabetes mellitus  Lists of variables  Model 8  Model 9  Model 10  Model 11  Ethnicity Recurrence of TB Culture status Millary TB CNS TB Far advanced TB Moderately advanced TB HIV/AIDS status Immunosuppressive medication Age groups Gender Ethnicity Recurrence of TB Culture status Millary TB CNS TB Far advanced TB Moderately advanced TB HIV/AIDS status Malnutrition Age groups Gender Ethnicity Recurrence of TB Culture status Millary TB CNS TB Far advanced TB Moderately advanced TB HIV/AIDS status Alcoholism Age groups Gender Ethnicity Recurrence of TB Culture status Millary TB CNS TB Far advanced TB Moderately advanced TB HIV/AIDS status Drug abuse Age groups Gender Ethnicity Recurrence of TB Culture status Millary TB  254  Adjusted Hazard Ratio (AHR) reported  Immunosuppressive medication  Malnutrition  Alcoholism  Drug abuse  Lists of variables  CNS TB Far advanced TB Moderately advanced TB HIV/AIDS status  255  Adjusted Hazard Ratio (AHR) reported  Table A.9 Comparisons of co‐morbidity between TB patients who died during follow up period (n=953) and who survived (n=4068)55 Variable  Number of death  Total Person Years (PYs)  Death/100 0 PYs  Unadjusted Hazard Ratio (95% CI)  Overall  953  33035  29  ‐  410 92  12980 1956  32 47  Reference 1.51 (1.20, 1.89)  427 33  13591 282  31 117  Reference 3.19 (2.24, 4.54)  395 105  13442 1621  29 65  Reference 2.30 (1.85, 2.85)  397 86  13419 916  30 94  Reference 2.72 (2.15, 3.43)  399 425 129  14116 17815 1103  28 24 117  Reference 0.94 (0.82, 1.08) 3.31 (2.71, 4.04)  398 80  13554 460  29 174  Reference 4.52 (3.55, 5.75)  434 37  13537 588  32 63  Reference 1.89 (1.35, 2.64)  415 39  13621 303  30 129  Reference 3.36 (2.42, 4.67)  Diabetes Mellitus No Yes Malnutrition No Yes Alcoholism No Yes Substance abuse No Yes Either HIV or AIDS No Unknown Yes Any Malignancy No Yes Immunosuppressive medication No Yes Chronic Renal failure No Yes  55  ‐ This analysis was restricted to TB patients with valid responses (yes or no) for comorbidity status  256  Table A.10: Multivariable cox regression analysis for risk factors of mortality among TB patients5657  Variable  Model 3 (n=4811) Adjusted HR (95% CI)  Model 4 (n=4684) Adjusted HR (95% CI)  Age at diagnosis (per decade)  1.06 (1.05, 1.06)  1.06 (1.05, 1.06)  Male gender Birth place Foreign‐born Canadian‐born non‐Aboriginal Canadian‐born Aboriginal Ethnicity Caucasian Aboriginals Southeast Asian Others Chinese/Filipino/Vietnamese Recurrent TB Culture confirmed TB Millary/disseminated TB CNS TB Far advanced pulmonary TB Moderately advanced pulmonary TB Either HIV or AIDS No Yes Unknown Any Malignancy Chronic renal failure Diabetes mellitus Immunosuppressive medication Malnutrition Alcoholism Drug abuse  1.47 (1.27, 1.70)  1.49 (1.28, 1.73)  56 57  Model 5‐11 Adjusted HR (95% CI)  Reference 1. 91 (1.60, 2.28) 3.02 (2.49, 3.66)  0.74 (0.44, 1.23) 1.52 (1.22, 1.90) 3.37 (2.65, 4.30) 1.56 (0.88, 2.79) 1.86 (1.33, 2.60) 1.16 (0.93, 1.45)  1.83 (1.52, 2.20) 2.93 (2.36, 3.63) 0.82 (0.65, 1.04) 0.95 (0.68, 1.34) Reference 0.64 (0.36, 1.14) 1.40 (1.13, 1.73) 3.61 (2.84, 4.58) 1.76 (1.03, 3.02) 2.10 (1.52, 2.90) 1.19 (0.96, 1.48)  Reference 4.00 (3.10, 5.15) 0.85 (0.73, 1.00)  Reference 4.00 (3.10, 5.16) 0.84 (0.73, 0.98) 2.39 (1.72, 3.34) 1.59 (0.98, 2.58) 1.20 (0.90, 1.60) 1.24 (0.84, 1.84) 1.32 (0.86, 2.00) 1.35 (1.02, 1.80) 1.87 (1.26, 2.77)  ‐ This analysis was restricted to TB patients with valid responses (yes or no) for comorbidity status. For model description, please see Table A.8  257  Table A.11: Comparisons of risk factors among canadian‐born non‐aboriginals, aboriginals, and foreign‐born people (n=5191)  Age at diagnosis (last episode) Mean (SD) Age groups (at diagnosis) 0‐40 years 41‐60 years 61‐80 years 80 plus years Age at Death Mean (SD) Age groups (at death) 0‐40 years 41‐60 years 61‐80 years 80 plus years Male gender Recurrence Culture confirmed TB Millary/disseminated TB CNS/Meningeal TB Far advanced PTB Moderately advanced PTB Cavitary diseases Drug resistance Sensitive Mono/Poly resistance MDR Diabetes mellitus Malnutrition Alcoholism Drug abuse Either HIV or AIDS Any Malignancy Immunosuppressive Medication Chronic Renal failure  Canadian‐born non‐ aboriginals (876) N (%)  Aboriginals (n=709) N (%)  Foreign‐born (n=3606) N (%)  46.1 (22.7)  40.5 (20.1)  50.0 (21.3)  355 (41) 275 (31) 207 (24) 39 (4)  360 (51) 223 (31) 114 (16) 12 (2)  1473 (41) 877 (24) 955 (27) 301 (8)  62.8 (18.1)  55.4 (16.6)  74.5 (15.0)  32 (14) 74 (33) 76 (33) 45 (20) 591 (68) 58 (7) 625 (71) 41 (5) 10 (1) 25 (3) 85 (10) 64 (7)  41 (20) 83 (41) 6 (32) 15 (7) 387 (55) 86 (12) 549 (77) 57 (8) 15 (2) 18 (2) 79 (11) 51 (7)  24 (5) 48 (10) 218 (46) 186 (39) 1832 (51) 342 (9) 2798 (78) 110 (3) 42 (1) 56 (2) 223 (6) 195 (5)  593 (95) 30 (5) 3 (<1)  529 (96) 20 (4) 0 (0)  2488 (88) 299 (11) 32 (1)  45 (5) 22 (2) 91 (10) 89 (10) 110 (13) 46 (5) 19 (2) 7 (1)  27 (4) 28 (4) 162 (23) 104 (15) 98 (14) 9 (1) 15 (2) 15 (2)  271 (7) 16 (<1) 47 (1) 19 (<1) 72 (2) 91 (2) 80 (2) 54 (1)  258  Table A.12: Univariate multinomial logistic regression analysis for co‐morbid risk factors of TB‐related mortality58  Variable Diabetes Mellitus No Yes Malnutrition No Yes Alcoholism No Yes Substance abuse No Yes Any Malignancy No Yes Immunosuppressive medication No Yes Chronic Renal failure No Yes Either HIV or AIDS No Yes  58  TB was the underlying cause Unadjusted OR (95% CI)  TB was a contributing factor Unadjusted OR (95% CI)  Reference 0.27 (0.09, 0.68)  Reference 0.60 (0.35, 1.03)  Reference 3.31 (1.54, 7.15)  Reference 3.04 (1.60, 5.80)  Reference 2.02 (1.23, 3.30)  Reference 1.58 (1.04, 2.43)  Reference 1.56 (0.84, 2.90)  Reference 2.58 (1.69, 3.93)  Reference 0.16 (0.02, 1.16)  Reference 3.54 (2.30, 5.46)  Reference 0.60 (0.19, 1.91)  Reference 1.67 (0.92, 3.04)  Reference 2.56 (1.14, 5.73)  Reference 5.84 (3.54, 9.63)  Reference 1.81 (1.06, 3.09)  Reference 2.86 (1.96, 4.17)  ‐This analysis was restricted to TB patients with valid responses (yes or no) for co‐morbidity status  259  APPENDIX B: HUMAN ETHICS APPROVAL CERTIFICATE  260  The University of British Columbia Office of Research Services Clinical Research Ethics Board – Room 210, 828 West 10th Avenue, Vancouver, BC V5Z 1L8  ETHICS CERTIFICATE OF EXPEDITED APPROVAL: RENEWAL PRINCIPAL INVESTIGATOR:  DEPARTMENT: UBC/Medicine, Faculty of/Medicine, J. Mark FitzGerald Department of INSTITUTION(S) WHERE RESEARCH WILL BE CARRIED OUT:  UBC CREB NUMBER: H06-03435  Institution  UBC BC Centre for Disease Control  Other locations where the research will be conducted:  Site  Vancouver (excludes UBC Hospital) BC Centre for Disease Control  N/A  CO-INVESTIGATOR(S): Arminee Kazanjian Richard K. Elwood Hubert Wong AKM Moniruzzaman Victoria J. Cook  SPONSORING AGENCIES: N/A  PROJECT TITLE: Morbidity and Mortality related to Tuberculosis in British Columbia, Canada EXPIRY DATE OF THIS APPROVAL:  January 28, 2011 APPROVAL DATE:  January 28, 2010 CERTIFICATION: In respect of clinical trials: 1. The membership of this Research Ethics Board complies with the membership requirements for Research Ethics Boards defined in Division 5 of the Food and Drug Regulations. 2. The Research Ethics Board carries out its functions in a manner consistent with Good Clinical Practices. 3. This Research Ethics Board has reviewed and approved the clinical trial protocol and informed consent form for the trial which is to be conducted by the qualified investigator named above at the specified clinical trial site. This approval and the views of this Research Ethics Board have been documented in writing. The Chair of the UBC Clinical Research Ethics Board has reviewed the documentation for the above named project. The research study, as presented in the documentation, was found to be acceptable on ethical grounds for research involving human subjects and was approved for renewal by the UBC Clinical Research Ethics Board.  Approval of the Clinical Research Ethics Board by one of:  Dr. Peter Loewen, Chair Dr. James McCormack, Associate Chair  https://rise.ubc.ca/rise/Doc/0/G3AK1UE2FPFKJ7LHFVLPDPT94B/fromString.html[6/28/2010 9:14:02 PM]  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            data-media="{[{embed.selectedMedia}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0071046/manifest

Comment

Related Items