UBC Faculty Research and Publications

The neglected burden of tuberculosis disease among health workers: a decade-long cohort study in South… O’Hara, Lyndsay M; Yassi, Annalee; Zungu, Muzimkhulu; Malotle, Molebogeng; Bryce, Elizabeth A; Barker, Stephen J; Darwin, Lincoln; Mark FitzGerald, J. Aug 7, 2017

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

Item Metadata

Download

Media
52383-12879_2017_Article_2659.pdf [ 916.87kB ]
Metadata
JSON: 52383-1.0351994.json
JSON-LD: 52383-1.0351994-ld.json
RDF/XML (Pretty): 52383-1.0351994-rdf.xml
RDF/JSON: 52383-1.0351994-rdf.json
Turtle: 52383-1.0351994-turtle.txt
N-Triples: 52383-1.0351994-rdf-ntriples.txt
Original Record: 52383-1.0351994-source.json
Full Text
52383-1.0351994-fulltext.txt
Citation
52383-1.0351994.ris

Full Text

RESEARCH ARTICLE Open AccessThe neglected burden of tuberculosisdisease among health workers: a decade-long cohort study in South AfricaLyndsay M. O’Hara1*, Annalee Yassi1, Muzimkhulu Zungu2,3, Molebogeng Malotle2, Elizabeth A. Bryce4,Stephen J. Barker1, Lincoln Darwin2 and J. Mark FitzGerald5AbstractBackground: Health workers (HWs) in resource-limited settings are at high-risk of exposure to tuberculosis (TB) atwork. The aim of this study was to estimate the rate of TB disease among HWs in the Free State Province of SouthAfrica between 2002 and 2012 and to compare demographic and clinical characteristics between HWs and thegeneral population with TB. This study also explores the effect of occupational variables on risk of TB among HWs.Methods: Probabilistic record linkage was utilized to identify HWs who were also registered as TB patients. Thishistorical prospective cohort study calculated incidence rate ratios (IRR) for TB disease among HWs in Free Statefrom 2002 to 2012. Generalized linear mixed-effects regression was used to model the association between sex,race, facility type, occupation, duration of employment, and the rate of TB.Results: There were 2677 cases of TB diagnosed among HWs from 2002 to 2012 and 1280 cases were expected.The overall TB incidence rate in HWs during the study period was 1496·32 per 100,000 compared to an incidencerate of 719·37 per 100,000 in the general population during the same time period. IRR ranged from 1·14 in 2012 to3·12 in 2005. HWs who were male, black, coloured and employed less than 20 years had higher risk of TB. Facilitytype and occupation were not associated with increased risk of TB when adjusted for other covariates.Conclusion: HWs in South Africa have higher rates of TB than the general population. Improved infectionprevention and control measures are necessary in all high-burden TB healthcare settings.BackgroundThe 2015 Ebola outbreak in West Africa was an infec-tious disease tragedy of epic proportions that drew at-tention to the daily occupational risks faced by healthworkers (HWs). One study estimated that during theoutbreak in Liberia, 0·11% of the general population diedfrom Ebola compared to 8·07% of the country’s doctors,nurses and midwives [1]. While Ebola dominated theheadlines, HWs continue to quietly die from tubercu-losis (TB) in numbers far greater than those seen withless common communicable diseases such as Ebola. TBhas become a “burgeoning global health crisis” with theemergence of drug-resistant tuberculosis [2] and whencoupled with the ongoing struggle to control human im-munodeficiency virus (HIV) [3]. Further compoundingthe TB and HIV epidemics is the critical shortage ofHWs globally and especially in Africa [4]. Recent atten-tion to high rates of TB among HWs [5, 6] as well ashospital-based outbreaks of multidrug and extensivelydrug-resistant TB among patients and workers [3, 7]have led to increased concern about the risk of Mycobac-terium tuberculosis transmission in healthcare settings.Several studies have confirmed that TB is a significantoccupational risk among HWs in low-and middle-income countries [8] and it is estimated that the inci-dence of TB among HWs in high burden countries(>100 cases/100,000 population) is 8·4% greater (95% CI2·7%-14·0%) than the general population [9], yet thishigh-risk population has not been the focus of system-atic research. The issue of TB in HWs in low-income* Correspondence: lohara@epi.umaryland.edu1School of Population and Public Health, University of British Columbia,Vancouver, CanadaFull list of author information is available at the end of the article© The Author(s). 2017 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, andreproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link tothe Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.O’Hara et al. BMC Infectious Diseases  (2017) 17:547 DOI 10.1186/s12879-017-2659-3countries was highlighted by articles published in theBulletin of the World Health Organization almost20 years ago [10, 11], but little has been done to obtainrigorous estimates of the true burden of disease amongHWs in regions where TB continues to flourish.Most estimates of TB disease among HWs in high inci-dence regions are based on the results of occupationalhealth record reviews thereby excluding HWs diagnosedand treated outside their workplace [5, 6]. Some studiesrely on self-reporting of TB status in a climate of HIV andTB associated stigma that predisposes to non-disclosure[7, 12]. A recent study in the province of KwaZulu-Natalof South Africa conducted a retrospective review of TBregisters at occupational health clinics in 11 hospitals andfour community health centres. The authors concludedthat under-reporting of TB among HWs likely masked thetrue high incidence in this group [13].Although it is well established that HWs in high bur-den countries are at high-risk of exposure to TB at work[8, 9, 14], the true incidence rate and burden of TB dis-ease among HWs in South Africa and other low andmiddle income countries remains unclear. Furthermore,previous methodologies utilized to generate current esti-mates of TB among HWs suffer from important limita-tions. To our knowledge, this is the first study in a low/middle-income, high TB burden country to link con-firmed cases of TB disease to healthcare human resourcerecords thereby addressing the limitations of self-reporting associated with previous estimates. Other de-terminants affecting exposure and outcomes among thispopulation are also poorly understood. This is particu-larly problematic as the lack of good data precludes theprioritization for resource allocation and evaluation ofprevention strategies. This study presents incidence rateratios (IRR) of TB disease among HWs in the Free Stateprovince of South Africa from 2002 to 2012. Demo-graphic and clinical characteristics were compared be-tween HWs and the general population with TB. Thisstudy also explores the effect of occupational variableson risk of TB among HWs.MethodsStudy design and participantsThis is a historical prospective cohort study and prob-abilistic record linkage between the South African na-tional human resource (HR) database called PERSALand the national TB registry called ETR.Net. Theseregistries do not share a unique identification number.HWs were defined as “all people engaged in the pro-motion, protection or improvement of the health of thepopulation” [15]. This definition was not limited to thosewho provide direct patient care, but was also extendedto all who work in a healthcare facility such as cleaners,porters, security personnel, etc. All employees of theFree State Department of Health from 2002 to 2012 whowere employed by the health department for at least onemonth were eligible for inclusion. HWs at all facilitylevels were included (local clinics, primary, secondaryand tertiary hospitals and non-clinical settings).OutcomesWorkers with laboratory-confirmed M. tuberculosis(including pulmonary, extra-pulmonary, disseminated,miliary and TB meningitis) [16] as identified in ETR.-Net were eligible for inclusion in the linkage. HWs withconfirmed and documented reactivation of TB were in-cluded in the linkage, but were only included in the calcu-lations of the incidence of TB, if the date of diagnosis wasafter the date of employment. Age, sex, race, HIV status,occupation, facility type, duration of employment, diagno-sis type (new, relapse/re-treatment), disease classification(pulmonary, extra-pulmonary, both), outcome (cured/completed, defaulted/failed, transferred/moved, died, un-known) and TB drug sensitivity (multidrug-resistant tu-berculosis (MDR-TB) yes, MDR-TB no) were included ascovariates. The total number of TB patients in the FreeState (general population and HWs) was calculated foreach year from ETR.Net. The total number of HWsemployed in Free State was recorded from PERSAL andaverage Full-Time Equivalent (FTE) were calculated foreach year (2002–2012).ProceduresA probabilistic record linkage was performed to estimatethe probability that a PERSAL and ETR.Net record refersto the same person. Raw data were acquired and importedinto Microsoft SQL Server 2008 using a custom applica-tion. A matching algorithm was written as a custom appli-cation using the programming language C#. Following thetheories presented by Newcombe [17], variables wereassigned a linkage weight according to their reliability anddiscriminatory power. Based on these parameters, thetotal weight (or “percentage score”) was derived by sum-ming the separate field comparisons across all fields. Atotal score was calculated as the sum of surname (40),given name (30), age/date of birth (30), and gender (30)scores where the maximum possible total score was 130(Fig. 1 and also see Additional file 1 for further details).Any final percentage scores less than 70% (91 out of130) were filtered out and were not included in the finaldataset. Scores greater than 90% (117 out of 130) wereincluded without manual review. Scores between 80 and90% were reviewed manually (Fig. 1). Decision rules formanual matching, as described in detail in Table 1, weredeveloped and were employed by two reviewers using acustomized web-based tool. Finally, all accepted possiblematches were re-assessed by a second reviewer using thesame decision rules.O’Hara et al. BMC Infectious Diseases  (2017) 17:547 Page 2 of 11Statistical analysisThe total number of TB cases among HWs in Free Statewas tabulated for each year (2002–2012). Descriptivestatistics were utilized to show demographic and clinicalcharacteristics of HWs and the general population withTB in the province.Person-years at risk for TB for HWs were estimated byassigning a full-time equivalent (FTE) score to each HW.For example, if a HW worked part-time in 2008, theywould contribute 0.5 person-years for that year. For eachindividual, their FTE were summed over the 10 years ofthe study to generate their individual person-time at risk.Average FTE was then calculated for each year to generatedenominators for subsequent calculations.The number of observed cases of TB among HWs andperson years at risk were identified (from HR databasePERSAL) for each year (2009–2012). Expected numbersof cases for each year were calculated by multiplying thenumber of person years at risk each year by the corre-sponding national TB incidence rate in the general adultpopulation. To calculate the IRR, observed numbers ofcases of TB among HWs were divided by the expectednumbers in the general adult population for each year.Poisson regression was used to model the associationbetween facility type, occupation and duration of employ-ment and the rate of TB, with the relative risk being ameasure of this association. Birth year, race and sex wereentered as independent variables in the multivariate re-gression to obtain adjusted effects. A random effect forhospital was included producing a generalized mixed-effects regression to account for the fact that HWs arenaturally clustered by facility.1Fig. 1 Overview of record linkage weighting and score cut-offsTable 1 Manual matching decision rulesIncluded Not IncludedIf difference between birthdates greater than 1with all other variables an exact matchIf only first initial (full first/given name not available) (even if other variables match)If month and day of birthdate were reversed If difference between birthdates/year was greater than 2If one digit of day or month of birth were reversed If difference between birthdates/year was greater than 1 with at least 1 other variablethat was not an exact matchIf there was an obvious typo or spelling mistake If surname was very common (e.g. Mofokeng) and other variables were not perfectmatchesIf the surnames differed but one could be assumedto be a nickname of the otherIf date of birth was different by one or two daysIf the given names differed but one could be assumedto be a nickname of the otherIf the Surname, Given Name/s and year of birth match perfectly but month and day ofbirth are differentIf all other variables matched perfectly but sex differedIf the Surname, Given Name/s and year of birth matchbut month/day in birthdate are missingIf surname sounds the same but is spelled differentlyand at least one of the given names and year of birth matchO’Hara et al. BMC Infectious Diseases  (2017) 17:547 Page 3 of 11To explore the impact of utilizing alternate cut-offscores, IRR were calculated using the number of observedcases from cut-off scores of 80%, 85% and 95% in com-parison the 90% presented here. Finally, a sub-set of 390possible matches that scored within the 90–100% rangeand a sub-set of 411 possible matches that scored lessthan 70% were examined manually to validate the com-puter algorithm and the selection of 70% and 90% as thelower and upper cut-off points.ResultsA flow chart of the 23,924 partial ETR.Net-PERSALmatches and the procedures employed to obtain the finalstudy population is presented in Fig. 2. Overall, therewere 231,834 people diagnosed with TB in Free Statefrom 2002 to 2012. There were 32,039 HWs employedby 258 facilities during this timeframe. During these11 years, 2677 cases of TB were diagnosed among HWsbut only 1280 TB cases were expected. The overall TBincidence rate in HWs during the study period was1496·32 per 100,000 compared to an incidence rate of719·37 per 100,000 in the general population during thesame time period (Table 2).The number of observed cases of TB among HWswas greater than the number of expected cases forevery year during the study period (Table 2). Thenumber of observed cases among HWs ranged from80 in 2002 to 371 in 2007. The number of TB casesdiagnosed among HWs between 2002 and 2012followed a similar trend over time when compared tothe general population (Fig. 3) however, the numberof TB cases among HWs was much higher than thegeneral population between 2005 and 2008 and thendecreased drastically in 2009. IRR ranged from 1·14Fig. 2 Free State Province ETR.Net- PERSAL record linkage processO’Hara et al. BMC Infectious Diseases  (2017) 17:547 Page 4 of 11(95% CI: 0·98 to 1·32) in 2012 to 3·12 in 2005 (95%CI: 2·81 to 3·45).As shown in Table 3, most (n = 1989, 74·3%) HWsthat were diagnosed with TB from 2002 to 2012 wereaged between 30 and 49 years old at the time of diagno-sis. There were slightly more females (n = 1574, 58·8%)than males and the majority were African/Black(n = 2546, 95·1%). About half (n = 1551, 57·9%) wereemployed in a hospital, while 882 (32·9%) worked in aclinic and 244 (9·2%) were employed in “other” settingssuch as the provincial department of health or centrallaundry facilities. Most HWs with TB were nurses(n = 1113, 41·6%), 767 (28·7%) were support staff suchas maintenance workers, laundry workers, food serviceworkers, security personnel, cleaners and porters, 282(10·5%) were physicians and surgeons, and 407 (15·2%)were administrative staff. There were 108 (4·2%) alliedhealth professionals (physical therapists, audiologists,Table 2 Incidence rate ratios for TB among HWs by year (2002–2012)Year HCW PersonYears(FTE) aObservedCasesExpectedCasesIncidence in GeneralPopulation(per 100,000) bIncidence in HealthcareWorkers(per 100,000)Incidence RateRatio(95% CI)P-Value c2002 13,473 80 34 248·5 593·8 2·35 (1·88–2·91) <0·0012003 13,964 198 77 552·2 1417·9 2·57 (2·23–2·95) <0·0012004 15,101 255 98 648·2 1688·6 2·60 (2·30–2·94) <0·0012005 16,448 365 117 708·1 2219·1 3·12 (2·81–3·45) <0·0012006 16,391 362 129 787·6 2208·5 2·81 (2·53–3·11) <0·0012007 16,641 371 136 815·1 2229·4 2·73 (2·46–3·02) <0·0012008 16,722 326 149 891·6 1949·5 2·19 (1·96–2·44) <0·0012009 16,404 157 138 842·1 957·1 1·39 (0·97–1·33) 0·1062010 16,298 203 137 840·0 1245·6 1·48 (1·29–1·70) <0·0012011 17,973 191 148 822·7 1062·7 1·29 (1·12–1·49) <0·0012012 19,491 169 148 757.0 867·1 1·14 (0·98–1·32) 0·084Overall TB incidence among HWs for study period 1496·32aFTE = Full Time EquivalentbAs reported by the South African National Department of Health (from ETR.Net) and Statistics South AfricacChi-Square TestFig. 3 TB incidence rate and TB cases among HWs and the general population (2002–2012)O’Hara et al. BMC Infectious Diseases  (2017) 17:547 Page 5 of 11technologist/technicians, pharmacists, social workersand dieticians) with TB during the study period. MostHWs with TB did not know their HIV status or did notdisclose it (n = 2035, 75·6%) but 498 were known to beHIV positive (18·6%). Pulmonary TB was diagnosed in2039 (76·2%) cases. There were 560 cases (20·9%) ofextra-pulmonary TB and 78 (2·9%) were classified ashaving both pulmonary and extra-pulmonary disease.Most (n = 2149, 80·3%) were newly diagnosed cases withthe rest being relapses or re-treatment cases. There wereonly 18 documented cases in ten years that were classi-fied as being multiple-drug resistant TB (MDR-TB)among HWs. The majority (n = 1742, 65·1%) of HWscompleted their course of treatment and were classifiedas “cured” while 306 (11·3%) died. One hundred andthirty-six (5.1%) defaulted or failed treatment, 405(15·1%) transferred or moved out of province and theoutcome was unknown for 90 HWs (3·4%).Table 4 shows the unadjusted and adjusted relative riskestimates from the mixed-effects Poisson regressionmodel. The final adjusted model included birth year,race, sex, occupation and duration of employment. Therisk of TB disease was greatest among HWs born from1960 to 1969 (RR = 7·29, 95% CI: 5·48 to 9·72) whencompared to those who were born in the 1980’s. Black/African HWs had a greater than 5-fold increased risk ofTB when compared to their white colleagues (RR = 5·30,95% CI: 3·90 to 7·20). Similarly, coloured HWs (peopleof mixed ethnic origin) had an almost 3-fold increasedrisk of TB (RR = 2·90, 95% CI: 1·87 to 4·50). The risk ofTB was greater among male compared to female HWs(RR = 1·51, 95% CI: 1·38 to 1·64). In the unadjusted ana-lysis, TB risk was greater among support staff (RR = 2·36,95% CI: 1·79 to 3·11), nursing staff (RR = 2·07, 95% CI:Table 3 Demographic and clinical characteristics of HWs andthe general population with TB in Free State, South AfricaVariable Health Workers(N = 2677)Frequency (%)General Population(N = 229,157)Frequency (%)Age (at diagnosis)< 19 0 (0·0) 4714 (2·1)20–29 205 (7·7) 51,409 (22·4)30–39 992 (37·1) 76,463 (33·3)40–49 997 (37·2) 58,223 (25·4)50–59 429 (16·0) 27,598 (12·0)60+ 54 (2·0) 10,750 (4·7)SexMale 1103 (41·2) 98,872 (43·1)Female 1574 (58·8) 130,285 (56·0)RaceAfrican 2546 (95·1) −White 70 (2·6) −Coloured 59 (2·2) −Indian 2 (0·08) −HIV StatusHIV+ 498 (18·6) 71,778 (31·3)HIV- 154 (5·8) 22,420 (9·8)Unknown 2025 (75·6) 98,959 (43·2)Facility TypeHospital 1551 (57·9) −Clinic 88 (32·9) −Other 244 (9·2) −OccupationDoctor/Surgeon 282 (10·5) −Nurse 1113 (41·6) −Allied Health Professional 108 (4·2) −Administrative/Clerical 407 (15·2) −Support Services 767 (28·7) −Duration of Employment1–5 years 529 (20·0) −6–10 years 359 (13·4) −11–15 years 536 (20·0) −16–20 years 498 (18·6) −20+ years 461 (17·2) −Disease ClassificationPulmonary 2039 (76·2) 179,579 (78·4)Extra-pulmonary 560 (20·9) 44,952 (19·6)Both 78 (2·9) 4626 (2·0)Diagnosis TypeNew 2149 (80·3) 186,679 (81·5)Relapse/Re-treatment 528 (19·7) 42,478 (18·5)Table 3 Demographic and clinical characteristics of HWs andthe general population with TB in Free State, South Africa(Continued)MDR-TB (pre-treatment)Yes 18 (0·7) 1430 (0·6)No 2659 (99·3) 227,727 (99·4)OutcomeCured/Completed 1742 (65·1) 134,857 (58·8)Defaulted/Failed 136 (5·1) 14,783 (6·5)Transferred/Moved 405 (15·1) 38,585 (16·8)Died 306 (11·3) 29,156 (12·7)Unknown 90 (3·4) 11,776 (5·1)*Physicians and Surgeon = Surgeon, Radiologist, Anaesthesiologist, OtherPhysicians, Medical Registrar; Nurse = Professional nurse, Assistant or auxiliaryNurse, Staff Nurse; Allied Health Professional = Therapist (i.e. audiologist),Technologist/technician, Pharmacist, Social Worker; Administrative/Clerical = Manager/administrator, Clerk, General Assistant; SupportStaff = Maintenance Worker, Laundry Worker, Food Service Worker, Security,Cleaner, PorterO’Hara et al. BMC Infectious Diseases  (2017) 17:547 Page 6 of 111·58 to 2·71) and administrative staff (RR = 1·74, 95% CI:1·29 to 2·32) when compared to allied health profes-sionals. These estimates attenuated in the adjustedmodel and were no longer statistically significant. Therisk of TB was also slightly higher among doctors andsurgeons in the unadjusted analysis, but this finding wasnot statistically significant (RR = 0·85, 95% CI: 0·62 to1·15). The RR estimates for all occupation categories at-tenuated in the adjusted model and were no longer sta-tistically significant. HWs who had worked in thehealthcare sector for less than 20 years had a greater riskof TB compared to those who had been employed formore than 20 years. In particular, HWs who wereemployed for 11–15 years had a more than 3-fold in-creased risk of TB (RR = 3·60, 95% CI: 2·97 to 4·37). Fa-cility type was not associated with increased risk of TB.Figure 4 provides a visual depiction of the IRR overtime for cut-off scores of 80%, 85%, 90% and 95%. Withan 80% cut-off point, IRR ranged from 12·62 in 2005 to2·00 in 2012. With an 85% cut-off point, the range wasfrom 7·23 in 2006 to 1·52 in 2011. With a 95% cut-offpoint, IRR ranged from 1·99 in 2006 to 0·64 in 2012.Manual matching of the 411 records that scored <70%,resulted in only 2 that were deemed to be false negativesTable 4 Relative risk estimates from mixed-effects poisson regression model of HWs with TB and those without TB (N = 32,039)Unadjusted P-Value Adjusted P-ValueRR (95% CI) RR (95% CI)Birth Year1980–1989 1·00 1·001970–1979 4·28 (3·24 to 5·65) <0·0001 3·84 (2·89 to 5·09) <0·00011960–1969 7·95 (6·06 to 10·42) <0·0001 7·29 (5·48 to 9·72) <0·00011950–1959 6·10 (4·59 to 8·10) <0·0001 7·11 (5·22 to 9·69) <0·00011940–1949 3·43 (2·43 to 4·85) <0·0001 4·03 (2·78 to 5·83) <0·00011930–1939 1·79 (0·16 to 20·84) 0·99 1·39 (0·13 to 15·13) 0·99RaceWhite 1·00 1·00African 6·74 (4·98 to 9·11) <0·0001 5·30 (3·90 to 7·20) <0·0001Coloured 3·59 (2·32 to 5·56) <0·0001 2·90 (1·87 to 4·50) <0·0001Asian 1·15 (0·18 to 7·02) 0·99 1·05 (0·18 to 6·07) 0·99SexFemale 1·00 1·00Male 1·41 (1·31 to 1·53) <0·0001 1·51 (1·38 to 1·64) <0·0001Facility TypeNon-Clinical 1·00 − −Hospital 2·04 (0·79 to 5·26) 0·19 − −Clinic 2·47 (0·67 to 9·12) 0·24 − −OccupationAllied Health Professional 1·00 1·00Doctor/Surgeon 1·13 (0·83 to 1·53) 0·84 0·85 (0·62 to 1·15) 0·60Nurse 2·07 (1·58 to 2·71) <0·0001 1·24 (0·93 to 1·63) 0·25Administrative/Clerical 1·74 (1·29 to 2·32) <0·0001 1·13 (0·84 to 1·51) 0·82Support Staff 2·36 (1·79 to 3·11) <0·0001 1·28 (0·96 to 1·70) 0·14Duration of Employment (yrs)20+ 1·00 1·0016–20 2·05 (1·71 to 2·47) <0·0001 1·97 (1·63 to 2·39) <0·000111–15 3·36 (2·80 to 4·04) <0·0001 3·60 (2·97 to 4·37) <0·00016–10 1·21 (0·99 to 1·48) 0·09 1·72 (1·38 to 2·14) <0·00011–5 1·06 (0·89 to 1·28) 0·97 1·92 (1·56 to 2·37) <0·0001< 1 2·74 (2·24 to 3·53) 0·03 1·60 (1·44 to 1·82) <0·0001O’Hara et al. BMC Infectious Diseases  (2017) 17:547 Page 7 of 11that should have been included in the linked dataset and409 that were appropriately discarded. Similarly, manualmatching of the 390 records that scored >90%, 383 weretrue positive matches and that only 7 were false positivesthat should have been excluded.DiscussionThese findings confirm that HWs in Free State, SouthAfrica have higher rates of TB than the general popula-tion. Although the rates were higher than the generalpopulation in all study years, the excess of cases wasparticularly high from 2002 to 2008 and highest in 2005.For this year, there was an alarming 312% more cases ofTB among HWs than expected meaning that the inci-dence of TB was more than 3-fold greater among HWsthan the general population in this year.We observed a dramatic drop in HW TB rates around2009. It is possible that this could in part be explainedby the implementation of ‘The Draft National InfectionPrevention And Control Policy For TB, MDRTB AndXDRTB’ and the ‘Tuberculosis Strategic Plan For SouthAfrica, 2007-2011’ across the country in 2007. Thesetwo policy documents had implications for TB infectioncontrol in health care settings. We were not able toidentify any formal changes to the reporting systemsduring the study period. It is also possible that the dropin TB rates could be explained by the aggressive role outof a free antiretroviral treatment program in the countryin 2004. It is estimated that there were 919,923 HIVpatients enrolled in the public program by November2009- a drastic increase from only 32,895 in January2005 [18]. It is also possible that the case definitionsused in the ETR.Net system were changed. For example,if they changed the way they entered re-infections forthe same person, there would have been a drop in inci-dence. Further investigations are necessary to fully ex-plore the cause of the drop in incidence rates in 2009.Our estimates of TB among HWs are consistent withother reports from the region including a study by O’Don-nell and colleagues from South Africa estimated rates ofMDR and extensively drug-resistant tuberculosis (XDR-TB)related hospital admissions [7] and a retrospective recordreview in one hospital in Kenya to document TB case noti-fication rates among hospital staff [19]. The results pre-sented here show that almost 30,000 people died from TBin Free State during the study period. More than 300 ofthose who died were HWs. This loss of skilled personnel isa huge detriment to a health system that is over-burdenedby the TB/HIV syndemic and where health human resourceshortages are common. We found that there were more TBpatients in the 60+ age category and in the age categories<29 in the general population group than in the HW group.This is likely due to the fact that many HWs retire in theirsixties and may still be completing their education andtraining in their twenties and therefore are not yetemployed. WHO estimates that 61% of TB patients inSouth Africa are co-infected with HIV [20]. These findingssuggest that only 31·3% of the general population (non-Fig. 4 Incidence rate ratio by year and percentage cut-off scoreO’Hara et al. BMC Infectious Diseases  (2017) 17:547 Page 8 of 11HWs) were known to be HIV positive. It was also interest-ing to note that the rate of TB patients who were known tobe HIV positive was still much higher in the general popu-lation group when compared to the HW group. This is be-cause the HIV status was unknown for the majority ofHWs with TB (75·6%) suggesting that HWs in Free Stateare either not receiving adequate access to HIV counsellingand testing or that they are afraid to disclose their status.Similar to the study from Kenya [19], HWs in FreeState had sub-optimal cure rates. HWs must thereforereceive early diagnosis and treatment for TB in additionto improved infection prevention and control efforts[21] in accordance with international guidelines [22]. Asurvey administered to medical doctors diagnosed withTB in South Africa found that a prompt diagnosis within7 days was only made in 20% of participants and 95% ofrespondents expressed concerns regarding a lack of ICin the workplace and negative attitudes of senior admin-istrators and colleagues [23]. HWs should also bescreened regularly for TB by programs that are free, con-fidential and available in the workplace [24]. These re-sults also show that occupation and facility type are notas strongly associated with increased incidence of TBamong HWs as expected. This suggests that all HWswho work in hospitals, clinics and even administrativesettings are at risk of exposure to TB in the workplaceand that there should be greater effort to include non-clinical personnel in TB infection control education andtraining. These findings also show that health workerswith less healthcare sector work experience (employedless than 20 years) were more likely to have TB thanthose who were employed more than 20 years. Interest-ingly, health workers who were employed for 11–15 yearshad the highest risk of TB. This could perhaps be due towhat is known as the “healthy worker survivor effect.”This is the tendency for the least healthy workers toleave the active workforce. Furthermore, health workerswho worked more than 20 years may have also been ableto better protect themselves, may have had work taskswith less exposure, and may have had fewer concomitantrisk factors.Although this probabilistic record linkage study is thefirst in the region to objectively estimate TB incidenceamong HWs, it does have several limitations. First, thequality of the data in ETR.Net is variable as the systemrelies on input from paper forms collected by nurses ateach facility. Furthermore, the information contained inETR.Net did not allow us to distinguish between relapseand retreatment cases. We recognize that the major riskfactors for relapse include inadequate therapy due to irregu-larity, high disease burden in the population, inadequateduration of therapy and underlying drug resistance. Recur-rence of disease due to true relapse would ideally be distin-guished from reinfection. Occupational cohort studies arevulnerable to several biases such as misclassification bias.Misclassification bias on exposure is not likely in this studyhowever misclassification of the outcome (TB status) ispossible. The ETR.Net registry does not necessarily containall records of patients diagnosed with TB and HWs in par-ticular may be less likely to report their disease. It is there-fore possible that the estimates of TB among HWs areunder-reported here. Despite these limitations, the resultsof the sensitivity analysis shown in Fig. 2 illustrate that 90%was a reasonable cut-point to accept all matches. The qual-ity of the matches decreased dramatically at 85% as evi-denced by the large jump in IRR. With all cut-off scores(80%, 85%, 90% and 95%) there is a noticeable drop in IRRin 2008–2009 as discussed previously.South Africa has adequate policies in place for the pro-tection of HWs from TB and other workplace conditions[25–28] yet this study illustrates the urgent need for the im-plementation of these policies, in particular TB infectionprevention and control measures and occupational healthand safety practices [29]. There is a need for better work-place as well as workforce surveillance, with promptfollow-up of cases of HWs with TB to ensure that all infec-tion control measures are being followed in areas in whichstaff that contracted TB worked. HWs work in stressful en-vironments where they are at high risk of exposure to infec-tious diseases such as hepatitis, HIV, TB and even Ebola.Many HWs who are diagnosed with TB, report feeling stig-matized [30] and unsupported in their journey back tohealth [30]. The findings presented here re-affirm the ur-gent call for action to protect the healthcare workforce.ConclusionsHWs in Free State, South Africa have higher rates of TBthan the general population. HWs are the backbone ofhealth systems worldwide and this study reinforces that weare not doing enough to protect them. Additional effortsmust be made to protect this high-risk, high-value popula-tion by implementing effective infection control measuresand providing timely TB screening, diagnosis, treatmentand support.What is already known about this topicSeveral small studies, based on occupational health clinicrecords or hospital admission data, have suggested thathealthcare workers are at increased risk for acquiringtuberculosis.What new knowledge this study contributesThis historical prospective cohort study is the first rec-ord linkage which documents that healthcare workers inthe Free State province of South Africa have an up tothree-fold risk of TB disease compared to the generalpopulation.O’Hara et al. BMC Infectious Diseases  (2017) 17:547 Page 9 of 11Additional fileAdditional file 1: Record linkage technical appendix. (DOCX 118 kb)AbbreviationsCI: confidence interval; FTE: full time equivalent; HIV: humanimmunodeficiency virus; HR: human resources; HWs: health workers;IRR: incidence rate ratio; MDR-TB: multidrug-resistant tuberculosis; RR: relativerisk; TB: tuberculosis; XDR-TB: extensively drug-resistant tuberculosisAcknowledgementsThe authors thank the Free State Department of Health for facilitating thedata linkage.FundingThis study was funded by the Canadian Institutes for Health Research (grantnumber ROH-115212). The funder played no role in in the study design; inthe collection, analysis, and interpretation of data; in the writing of the re-port; and in the decision to submit the article for publication.Availability of data and materialsData will be shared upon request.Authors’ contributionsLMO: conceptualized the study; formulated the overarching research goalsand aims; developed the methodology; applied statistical, mathematical,computational, or other formal techniques to analyze or synthesize studydata; wrote the initial manuscript draft; critically reviewed and revisedmanuscript drafts. AY: conceptualized the study; formulated the overarchingresearch goals and aims; developed the methodology; provided oversightand mentorship to the core team; critically reviewed and providedcommentary of manuscript drafts; acquired the financial support for theproject leading to this publication. MZ: formulated the overarching researchgoals and aims; developed the methodology; applied statistical,mathematical, computational, or other formal techniques to analyze orsynthesize study data; critically reviewed and revised manuscript drafts. MM:formulated the overarching research goals and aims; developed themethodology; applied statistical, mathematical, computational, or otherformal techniques to analyze or synthesize study data; critically reviewed andrevised manuscript drafts. EAB: formulated the overarching research goalsand aims; developed the methodology; provided oversight and mentorshipto the core team; critically reviewed and provided commentary ofmanuscript drafts. SJB: provided programming and software development;designed computer programs; implemented the computer code andsupporting algorithms; applied statistical, mathematical, computational, orother formal techniques to analyze or synthesize study data. LD: providedprogramming and software development; designed computer programs;implemented the computer code and supporting algorithms. JMF:formulated the overarching research goals and aims; developed themethodology; provided oversight and mentorship to the core team; criticallyreviewed and provided commentary of manuscript drafts. All authors readand approved the final manuscript.Ethics approval and consent to participateAll components and associated methods were approved by the University ofBritish Columbia’s Behavioural Research Ethics Board [certificate # H12–02489] and the Free State Department of Health.Consent for publicationNot Applicable.Competing interestsThe authors declare that they have no competing interests.Publisher’s NoteSpringer Nature remains neutral with regard to jurisdictional claims inpublished maps and institutional affiliations.Author details1School of Population and Public Health, University of British Columbia,Vancouver, Canada. 2National Institute for Occupational Health, NationalHealth Laboratory Service, Johannesburg, South Africa. 3School of HealthSystems and Public Health, University of Pretoria, Pretoria, Gauteng, SouthAfrica. 4Department of Pathology and Laboratory Medicine, University ofBritish Columbia, Vancouver, Canada. 5Division of Respiratory Medicine,University of British Columbia, Vancouver, Canada.Received: 11 January 2017 Accepted: 1 August 2017References1. Evans DK, Goldstein M, Popova A. Health-care worker mortality and thelegacy of the Ebola epidemic. The Lancet Glob Health. 2015;3(8):e439–40.2. Dheda K, Gumbo T, Gandhi NR, Murray M, Theron G, Uwadia Z, Migliori GB,Warren R. Global control of tuberculosis: from extensively drug-resistant tountreatable tuberculosis. Lancet Respir Med. 2014;2(4):321–38.3. Gandhi NR, Moll A, Sturm AW, Pawinski R, Govender T, Lalloo U, Zeller K,Andrews J, Friedland G. Extensively drug-resistant tuberculosis as a case ofdeath in patients co-infected with tuberculosis and HIV in a rural area ofSouth Africa. Lancet. 2006;368:1575–80.4. World Health Organization. World Health Report: Working together forhealth Available: http://www.who.int/whr/2006/en/. 2006; AccessedSeptember 20155. Wilkinson D, Gilks CF. Increasing frequency of tuberculosis among staff in asouth African district hospital: impact of the HIV epidemic on the supplyside of health care. Trans R Soc Trop Med Hyg. 1998;92:500–2.6. Naidoo S, Jinabhai CC. TB in health care workers in KwaZulu-Natal, SouthAfrica. Int J Tuberc Lung Dis. 2006;10(6):676–82.7. O’Donnell MR, et al. High incidence of hospital admissions with multidrug-resistant and extensively drug-resistant tuberculosis among south Africanhealthcare workers. Ann Intern Med. 2010;153(8):516–22.8. Joshi R, Reingold AL, Menzies D, Pai M. Tuberculosis among health-careworkers in low-and middle-income countries: a systematic review. PLoSMed. 2006;2(12):2376–84.9. Baussano I, Nunn P, Williams B, Pivetta E, Bugiani M, Scano F. Tuberculosisamong health care workers. Emerg Infect Dis. 2011;17(3):488–94.10. Harries AD, Maher D, Nunn P. Practical and affordable measures for theprotection of health care workers from tuberculosis in low-incomecountries. Bull World Health Organ. 1997;75(5):477.11. Harries AD, Hargreaves NJ, Gausi F, Kwanjana JH, Salaniponi FM. Preventingtuberculosis among health workers in Malawi. Bull World Health Organ.2002 Jul;80(7):526–31.12. Siegel J, Yassi A, Rau A, et al. Workplace Interventions to Reduce HIV and TBStigma among Healthcare Workers -Where do we go from here? GlobPublic Health 2015;Mar 13:1–13.13. Mahomed S, Khilawan DR, Knight S. Tuberculosis among public sectorhealthcare workers in eThekwini district, KwaZulu-Natal: original research.Occ Health Southern Afr. 2016 Sep 1;22(5):8–12.14. Grobler L, Mehtar S, Dheda K, et al. The epidemiology of tuberculosis inhealthcare workers in South Africa: a systematic review. BMC Health ServRes. 2016;16(1):416.15. World Health Organization. Counting health workers: Definitions, data,methods and global results. 2007. Available: http://www.who.int/hrh/documents/counting_health_workers.pdf\16. World Health Organization: International Union Against Tuberculosis andlung Disease; Royal Netherlands Tuberculosis Association Revisedinternational definitions in tuberculosis control Int J Tuberc Lung Dis 2001;5:213–215.17. Newcombe HB. Handbook of record linkage: methods for healthstatistical studies, administration and business. Oxford, UK: OxfordUniversity Press; 1988.18. Wouters E, et al. The long road to universal antiretroviral treatmentcoverage in South Africa. Future Virol. 2011;6(7):801–12.19. Wahome E, Makori L, Gikera M, Wafula J, Chakaya J, Edginton ME, KumarAMV. Tuberculosis treatment outcomes among hospital workers at a publicteaching and national referral hospital in Kenya. Public Health Action. 2013;3(4):323–7.20. World Health Organization. Global TB Report. 2015. Available: http://www.who.int/tb/publications/global_report/gtbr15_annex02.pdf?ua=1O’Hara et al. BMC Infectious Diseases  (2017) 17:547 Page 10 of 1121. Zungu M, Malotle M. Do we know enough to prevent occupationally acquiredtuberculosis in healthcare workers? Occ Health Southern Afr. 2011:17–2.22. World Health Organization, International Labour Office, UNAIDS the JointUnited Nations Programme on HIV/AIDS. The joint WHO-UNAIDS policyguidelines on improving health workers’ access to HIV and TB prevention,treatment, care and support services. Geneva WHO, 2011. Available: http://www.who.int/occupational_health/publications/hiv_tb_guidelines/en/23. Naidoo A, Naidoo S, Gathiram P, Lalloo U. Tuberculosis in medical doctors- astudy of personal experiences and attitudes. S Afr Med J. 2013;103(3):176–80.24. O’Hara NN, Roy L, O’Hara LM, Spiegel J, Lynd L, Fitzgerald JM, Yassi A,Nophale LE, Marra C. Active tuberculosis screening preferences ofhealthcare workers in South Africa: a best-worst scaling experiment. PLoSOne. 2015;10(7):e0133304.25. Constitution of the Republic of South Africa, 1996 (Act No. 108 of 1996).26. National Department of Health South Africa. Draft national infectionprevention and control policy for TB, MDR-TB and XDR-TB. Pretoria, SouthAfrica: National Department of Health, South Africa, 2007.27. National Department of Health. National Tuberculosis ManagementGuidelines. Pretoria. South Africa: National Department of Health, SouthAfrica; 2014.28. Department of Labour. Occupational Health and Safety Act, 1993 (Act No.85 of 1993) and its Hazardous Biological Agents Regulations.29. McCarthy KM, et al. High incidence of latent tuberculous infection amongsouth African health workers: an urgent call for action. Int J Tuberc LungDis. 2015;19(6):647–53.30. von Delft A, Dramowski A, et al. Why healthcare workers are sick of TB. Int JTuberc Lung Dis. 2015;32:147–51.•  We accept pre-submission inquiries •  Our selector tool helps you to find the most relevant journal•  We provide round the clock customer support •  Convenient online submission•  Thorough peer review•  Inclusion in PubMed and all major indexing services •  Maximum visibility for your researchSubmit your manuscript atwww.biomedcentral.com/submitSubmit your next manuscript to BioMed Central and we will help you at every step:O’Hara et al. BMC Infectious Diseases  (2017) 17:547 Page 11 of 11

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}]}"
                            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:
http://iiif.library.ubc.ca/presentation/dsp.52383.1-0351994/manifest

Comment

Related Items