UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Health economic studies : a focus on health related quality of life, health resource utilization and… Davis, Jennifer Colleen 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_davis_jennifer.pdf [ 2.54MB ]
Metadata
JSON: 24-1.0070986.json
JSON-LD: 24-1.0070986-ld.json
RDF/XML (Pretty): 24-1.0070986-rdf.xml
RDF/JSON: 24-1.0070986-rdf.json
Turtle: 24-1.0070986-turtle.txt
N-Triples: 24-1.0070986-rdf-ntriples.txt
Original Record: 24-1.0070986-source.json
Full Text
24-1.0070986-fulltext.txt
Citation
24-1.0070986.ris

Full Text

HEALTH ECONOMIC STUDIES: A FOCUS ON HEALTH RELATED QUALITY OF LIFE, HEALTH RESOURCE UTILIZATION AND FALLS PREVENTION IN VULNERABLE COMMUNITY DWELLING SENIORS  by Jennifer Colleen Davis BSc, The University of British Columbia, 2003 MSc, The University of British Columbia, 2006   A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY   in  The Faculty of Graduate Studies (Experimental Medicine)   THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  June, 2010   © Jennifer Colleen Davis, 2010  ii Abstract Introduction: This thesis comprises six studies that address important economic issues related to falls prevention in community living seniors. Aims: 1) To ascertain the economic burden of falls in different countries and  examine why these costs differ. 2) To determine which falls prevention strategies provide the best value for money.  3) and 4) To estimate the cost-effectiveness and cost-utili ty of once and twice weekly resistance training compared with balance and tone classes for one year and two years in terms of falls prevented and quality adjusted life years (QALYs) gained. 5) To: a) quantify the difference in the incremental cost-effectiveness ratio (ICER) when the QALYs are generated from the EuroQol 5D (EQ-5D) and from the Short Form 6D (SF-6D) over the 12-month intervention period of the Brain Power study, b) determine key predictors of changes in health related quality of life and health resource utilization.  6) To determine whether executive functions are independently associated with health related quality of life in older women. Methods: I conducted two systematic reviews, two comprehensive cost-effectiveness and cost-utili ty analyses, a comparison of two generic preference based utility instruments and an exploratory study of the association between QALYs and cognition. Results: 1) The mean cost of falls ranged from US $3476 per faller to US $26 483 per fall requiring hospitalization. 2) The best value for money in falls prevention comes from interventions targeting high-risk groups. 3) Once weekly and twice weekly resistance training were cost saving compared with balance and tone classes (comparator). 4) The benefits in the year after participating were not sustained for both resistance training groups. 5) ICERs estimated from the SF-6D and EQ-5D may result in different conclusions. 6) Executive function is an independent predictor of QALYs in older women. Summary: Falls prevention strategies can be cost saving and are related to quality of life and executive function in specific groups of community dwelling seniors. Establishing consistency in economic evaluation methods is a priority for comparison of costs, QALYs and value for money between countries and between  iii effective falls prevention strategies.  iv Table of Contents  Abstract .............................................................................................................................................................................................. ii Table of Contents ............................................................................................................................................................................. iv List of Tables .................................................................................................................................................................................. viii List of Figures................................................................................................................................................................................... ix Glossary of Terms and Abbreviations............................................................................................................................................ x Acknowledgements  ......................................................................................................................................................................... xi Co-authorship Statement ............................................................................................................................................................... xii 1 Introduction ................................................................................................................................................................................ 1 1.1 Overview of thesis ............................................................................................................................................................... 1 1.2 Literature review .................................................................................................................................................................. 2 1.2.1 Incidence of fa lls in o lder adults.................................................................................................................................. 3 1.2.2 Costs of fa lls................................................................................................................................................................ 9 1.2.3 Fracture – a common and costly consequence of falls  ............................................................................................ 17 1.2.4 Defining quality adjusted life years: direct and indirect measurement ..................................................................... 24 1.2.5 Quality of life implications.......................................................................................................................................... 33 1.2.6 Defining categories of economic evaluations ........................................................................................................... 34 1.2.7 Interventions reduce falls, but at what cost? ............................................................................................................. 39 1.2.8 Proposed thesis manuscripts .................................................................................................................................... 50 1.3 References........................................................................................................................................................................ 58 2 International comparison of cost of falls in older adults living in the community: a systematic review  ...................... 76 2.1 Introduction ........................................................................................................................................................................ 76 2.2 Materials and methods ...................................................................................................................................................... 77 2.2.1 Literature searches ................................................................................................................................................... 77 2.2.2 Selection of studies ................................................................................................................................................... 77 2.2.3 Data ex traction and data synthesis........................................................................................................................... 78 2.2.4 Standardized cost outcomes..................................................................................................................................... 85 2.2.5 Quality assessment ................................................................................................................................................... 88 2.3 Results ............................................................................................................................................................................... 90 2.3.1 Literature searches – overview of studies identified ................................................................................................. 90 2.3.2 United States – national cost of falls and other settings – five studies .................................................................... 90 2.3.3 United States – hospitalization costs of fa ll-related injuries – four studies............................................................... 91 2.3.4 United Kingdom – population based estimate of falls in older adults – two studies................................................. 91 2.3.5 Australia – prospective cost of fa lls– one study........................................................................................................ 92 2.3.6 Australia – population based study assessment of health system costs of fa lls – one study .................................. 92 2.3.7 Europe – prospective cohor t study in older adults hospi talized after a fall – one study  .......................................... 92 2.3.8 Europe – retrospective assessment of costs and health resource use – one study ................................................ 93 2.3.9 Europe – population based study – one study  ......................................................................................................... 93 2.3.10 Europe – population based assessment of cost to health care system from injur ies – one study  .......................... 93 2.3.11 Quality of included studies ........................................................................................................................................ 93 2.4 Discussion ......................................................................................................................................................................... 94 2.4.1 Range between population based studies................................................................................................................ 94 2.4.2 Range in prospective studies .................................................................................................................................... 95  v 2.4.3 Challenges in comparing costs across countries ..................................................................................................... 95 2.4.4 Incidence and prevalence based costing approaches.............................................................................................. 95 2.4.5 Quality of included studies ........................................................................................................................................ 96 2.4.6 Conclusion................................................................................................................................................................. 96 2.5 References ........................................................................................................................................................................ 98 3 Does a home based strength and balance program in people aged ≥ 80 years provide the best value for money to prevent falls? A systematic review of economic evaluations of falls prevention interventions ......................................... 102 3.1 Introduction ...................................................................................................................................................................... 102 3.2 Methods ........................................................................................................................................................................... 104 3.2.1 Literature search strategy ....................................................................................................................................... 104 3.2.2 Selection of studies ................................................................................................................................................. 104 3.2.3 Abstraction of data  .................................................................................................................................................. 105 3.2.4 Data synthesis and analysis ................................................................................................................................... 105 3.2.5 Standardized cost outcomes................................................................................................................................... 105 3.2.6 Quality assessment ................................................................................................................................................. 106 3.3 Results ............................................................................................................................................................................. 106 3.3.1 Overview of studies identified  ................................................................................................................................. 106 3.3.2 Individually customized multifactorial interventions – one cost-effectiveness study .............................................. 118 3.3.3 Multip le intervention at a community level  .............................................................................................................. 118 3.3.4 Single factor interventions – five cost-effectiveness studies and one cost-utility study  ......................................... 118 3.4 Discussion ....................................................................................................................................................................... 120 3.4.1 Quality of included studies ...................................................................................................................................... 121 3.4.2 How do economic evaluations aid in decision making?  ......................................................................................... 122 3.4.3 Implications for future research ............................................................................................................................... 123 3.4.4 Implications for policy-makers and clinical practice ................................................................................................ 123 3.4.5 Conclusion............................................................................................................................................................... 124 3.5 References ...................................................................................................................................................................... 125 4 Economic evaluation of dose-response resistance training in older women: a cost-effectiveness and cost-utility analysis .......................................................................................................................................................................................... 130 4.1 Introduction ...................................................................................................................................................................... 130 4.2 Methods ........................................................................................................................................................................... 131 4.2.1 Overview of economic evaluation  ........................................................................................................................... 131 4.2.2 Costs ....................................................................................................................................................................... 132 4.2.3 Effectiveness outcomes .......................................................................................................................................... 132 4.2.4 Adverse events and mor tality.................................................................................................................................. 133 4.2.5 Handling missing data ............................................................................................................................................. 133 4.2.6 Cost-effectiveness analysis .................................................................................................................................... 134 4.2.7 Cost-utility analysis ................................................................................................................................................. 134 4.2.8 Sensitivity analysis .................................................................................................................................................. 134 4.3 Results ............................................................................................................................................................................. 135 4.3.1 Health care use and costs....................................................................................................................................... 136 4.3.2 Health outcomes ..................................................................................................................................................... 137 4.3.3 Adjusting QALYs for baseline utility in each group ................................................................................................. 138 4.3.4 Cost-effectiveness analysis .................................................................................................................................... 138 4.3.5 Cost-utility analysis ................................................................................................................................................. 138 4.3.6 Sensitivity analysis .................................................................................................................................................. 139 4.4 Discussion ....................................................................................................................................................................... 141 4.5 References ...................................................................................................................................................................... 144 5 Sustainability of a 12-month resistance training intervention in older community dwelling women: a cost - effectiveness and cost-utility analysis ....................................................................................................................................... 147  vi 5.1 Introduction ...................................................................................................................................................................... 147 5.2 Methods ........................................................................................................................................................................... 148 5.2.1 Overview of economic evaluation  ........................................................................................................................... 148 5.2.2 Participant recruitment and sample  ........................................................................................................................ 148 5.2.3 Costs ....................................................................................................................................................................... 149 5.2.4 Effectiveness outcomes .......................................................................................................................................... 150 5.2.5 Adverse events and mor tality.................................................................................................................................. 150 5.2.6 Handling missing data ............................................................................................................................................. 150 5.2.7 Cost-effectiveness analysis .................................................................................................................................... 151 5.2.8 Cost-utility analysis ................................................................................................................................................. 152 5.2.9 Sensitivity analysis .................................................................................................................................................. 152 5.3 Results ............................................................................................................................................................................. 152 5.3.1 Health care use and costs....................................................................................................................................... 153 5.3.2 Health outcomes ..................................................................................................................................................... 155 5.3.3 Adjusting QALYs for baseline utility in each group ................................................................................................. 155 5.3.4 Cost-effectiveness analysis .................................................................................................................................... 156 5.3.5 Cost-utility analysis ................................................................................................................................................. 156 5.3.6 Sensitivity analysis .................................................................................................................................................. 159 5.4 Discussion ....................................................................................................................................................................... 160 5.4.1 Comparison with other studies................................................................................................................................ 163 5.4.2 Uncertainty in findings............................................................................................................................................. 163 5.4.3 Time horizon............................................................................................................................................................ 164 5.4.4 Limitations ............................................................................................................................................................... 164 5.4.5 Strengths ................................................................................................................................................................. 165 5.4.6 Conclusions and future directions........................................................................................................................... 165 5.5 References ...................................................................................................................................................................... 166 6 A prospective comparison of generic preference based utility instruments (SF -6D and EQ-5D) and predictors of health care resource utilization in older women ....................................................................................................................... 169 6.1 Introduction ...................................................................................................................................................................... 169 6.2 Methods ........................................................................................................................................................................... 170 6.2.1 Sample .................................................................................................................................................................... 170 6.2.2 Measures................................................................................................................................................................. 171 6.2.3 Data analysis........................................................................................................................................................... 173 6.3 Results ............................................................................................................................................................................. 174 6.3.1 Sample .................................................................................................................................................................... 174 6.3.2 Incremental cost-effectiveness ratios for the Brain Power study using the EQ-5D and SF-6D ............................. 175 6.3.3 Descriptive classification of health state vectors between groups for the EQ-5D.................................................. 177 6.3.4 Description of domain specific frequencies and percentages - EQ-5D.................................................................. 178 6.3.5 Model to determine independent predictors of QALYs........................................................................................... 181 6.3.6 Correlation coefficients............................................................................................................................................ 181 6.3.7 Multivariate linear regression results for QALYs calculated from the EQ-5D and SF-6D ...................................... 182 6.3.8 Model to determine independent predictors of health resource utilization  ............................................................. 185 6.4 Discussion ....................................................................................................................................................................... 186 6.4.1 Predicting health resource utilization among older community dwelling women  ................................................... 187 6.4.2 Clinical versus statistical significance – placing the results in contex t ................................................................... 188 6.4.3 Limitations ............................................................................................................................................................... 189 6.4.4 Conclusions............................................................................................................................................................. 189 6.5 References ...................................................................................................................................................................... 190 7 The independent contribution of executive functions to health related quality of life in older women  ...................... 193 7.1 Introduction ...................................................................................................................................................................... 193 7.2 Methods ........................................................................................................................................................................... 195  vii 7.2.1 Study design and participants................................................................................................................................. 195 7.2.2 Functional comorbidity index  .................................................................................................................................. 195 7.2.3 Global cognition measures – Mini Mental State Examination  ................................................................................ 195 7.2.4 Central executive functions—set shifting, updating and response inhibition  ......................................................... 196 7.2.5 Preference based measures – HSUV instrument................................................................................................... 197 7.2.6 Timed up and go ..................................................................................................................................................... 198 7.2.7 Data analysis........................................................................................................................................................... 198 7.3 Results ............................................................................................................................................................................. 199 7.3.1 Sample .................................................................................................................................................................... 199 7.3.2 Correlation coefficients............................................................................................................................................ 200 7.3.3 Multivariate linear regression results for QALYs calculated from the EQ-5D ........................................................ 201 7.4 Discussion ....................................................................................................................................................................... 202 7.4.1 Relationship between executive functions and QALYs – HRQL ............................................................................ 202 7.4.2 Relating working memory and health related quality of life  .................................................................................... 203 7.4.3 Response inhibition and health related quality of life – comparison with another study ........................................ 203 7.4.4 Contrasting the imputed and complete case analyses  ........................................................................................... 204 7.4.5 Timed up and go was a key explanatory variable in our model  ............................................................................. 204 7.4.6 Conclusions............................................................................................................................................................. 205 7.5 References ...................................................................................................................................................................... 206 8 Integrated discussion............................................................................................................................................................ 212 8.1 Contemporary economic methods and methodological advances  ................................................................................. 212 8.1.1 Development of cost items for cost-effectiveness and cost-utility analyses .......................................................... 213 8.1.2 Valuation of cost items and assessment of study quality  ....................................................................................... 213 8.1.3 Multip le imputation of missing data for economic evaluations alongside clin ical trials  .......................................... 214 8.1.4 Probabilistic sensitivity analyses to ascer tain the uncer tainty around the point estimates of cos ts and health benefits 215 8.1.5 In depth comparison of the EQ-5D and the SF-6D among older women............................................................... 216 8.2 New information and knowledge transfer for health policy  ............................................................................................. 217 8.3 Limitations........................................................................................................................................................................ 219 8.4 Future directions for economic research among older adults ......................................................................................... 220 8.4.1 The current limitations of fall-related economic evaluations – new potentia l for methodological innova tions for future studies ........................................................................................................................................................................ 220 8.4.2 Development of disease specific tool to assess QALYs for faller s  ........................................................................ 223 8.5 Conclusions ..................................................................................................................................................................... 224 8.6 References ...................................................................................................................................................................... 226 Appendix A: Ethics ....................................................................................................................................................................... 231 Appendix B: Consent form and letter of initial contact ............................................................................................................ 234 Appendix C: Data collection forms ............................................................................................................................................. 255 Appendix D: List of publications................................................................................................................................................. 263  viii List of Tables  Table 1-1. Categories of cost items for inclusion in cost of illness studies  ...................................................................... 13 Table 2-1. Fall related cost items repor ted in cost of falls studies  ................................................................................... 80 Table 2-2. Study population and outcome measures ....................................................................................................... 82 Table 2-3. Study methodology .......................................................................................................................................... 86 Table 2-4. Modified version of Drummond‘s checklist to assess quality of economic studies  ........................................ 89 Table 3-1. Characteristics of studies  .............................................................................................................................. 109 Table 3-2. Outcome measures  ....................................................................................................................................... 112 Table 3-3. Cost items measured in economic evaluations  ............................................................................................ 114 Table 3-4. Quality of Health Economic Studies scores  .................................................................................................. 115 Table 3-5. Checklist for economic evaluations by Drummond and colleagues  ............................................................. 117 Table 4-1. Characteristics of par ticipants at baseline  .................................................................................................... 135 Table 4-2. Unit costs for each component of resource utilization  .................................................................................. 136 Table 4-3. Results of base case analysis ....................................................................................................................... 137 Table 4-4. Results of one way sensitivity analyses ........................................................................................................ 140 Table 5-1. Characteristics of par ticipants at the end of the 12-month intervention period (Complete Case Analysis; N = 98)  ........................................................................................................................................... 153 Table 5-2. Unit costs for each component of resource utilization  .................................................................................. 154 Table 5-3. Results of base case analysis. ...................................................................................................................... 155 Table 5-4. Results of one-way sensitivity analyses ........................................................................................................ 160 Table 6-1. Characteristics of study sample at baseline entry into the trial..................................................................... 175 Table 6-2. Global utility scores from the EQ-5D and the SF-6D .................................................................................... 176 Table 6-3. Health state vectors for the EQ-5D  ............................................................................................................... 178 Table 6-4. Domain specific frequencies and percentages for the EQ-5D...................................................................... 179 Table 6-5. Domain specific frequencies and percentages for the EQ-5D...................................................................... 180 Table 6-6. Correlation coefficient matrix‡ ....................................................................................................................... 182 Table 6-7. Multiple linear regression summary for quality adjusted life years in older women calculated from EQ-5D and SF-6D ..................................................................................................................................................... 184 Table 6-8. Multiple linear regression summary for health resource utilization ............................................................... 186 Table 7-1. Characteristics of the Brain Power cohort at baseline  .................................................................................. 200 Table 7-2. Correlation coefficient matrix ......................................................................................................................... 201 Table 7-3. Multiple linear regression summary for predicting QALYs in older women as calculated from EQ-5D HSUVs............................................................................................................................................... 202  ix List of Figures  Figure 1-1. Measurement iterative loop  ........................................................................................................................................... 10 Figure 1-2. Recent h ip fracture trends by age and sex .................................................................................................................... 19 Figure 1-3. Quality adjusted life years  ............................................................................................................................................. 24 Figure 1-4. Standard gamble example for a health state i preferred to death ................................................................................. 27 Figure 1-5. Time trade-off method for a chronic health state preferred to death  ............................................................................ 28 Figure 1-6. Cost-effectiveness plane  ............................................................................................................................................... 38 Figure 2-1. QUOROM flow diagram of selection of studies ............................................................................................................. 78 Figure 3-1. QUOROM flow diagram of selection of studies........................................................................................................... 108 Figure 4-1a. Cost effective plane depicting the 95%  confidence ellipses of incremental cost and effectiveness for comparison between tw ice weekly resistance training (2RT) and tw ice weekly balance and tone (2BT, comparator); b. Cost effective plane depicting the 95%  confidence ellipses of incremental cost and effectiveness for comparison between once weekly resistance training (1RT) and twice weekly balance and tone (2BT, comparator). QALY estimates are based on utility scores from the EQ-5D. ........................................................................................................................................................ 139 Figure 5-1a. Cost effective plane depicting the 95%  confidence ellipses of incremental cost and effectiveness for comparison between once weekly resistance training (1RT) and balance and tone (2BT, comparator) with QALYs estimated from the EQ-5D; b. Cost effective plane depicting the 95%  confidence ellipses of incremental cost and effectiveness for comparison between once weekly resistance training (1RT) and balance and tone (2BT, comparator ) with QALYs estimated from the EQ-5D. 1c. Cost effective plane depicting the 95%  confidence ellipses of incremental cost and effectiveness for comparison between once weekly resistance training (1RT) and balance and tone (2BT, comparator) with QALYs estimated from the SF-6D; d. Cost effective plane depicting the 95%  confidence ellipses of incremental cost and effectiveness for comparison between once weekly resistance training (1RT) and balance and tone (2BT, comparator) with QALYs estimated from the SF-6D........................................................................................................................................ 158 Figure 5-2a. Cost effective acceptability curve for once weekly resistance training (1RT) versus balance and tone w ith QALYs estimated from the EQ-5D; b. Cost effective acceptability curve for once weekly resistance training (1RT) versus balance and tone w ith QALYs estimated from the SF-6D; 2c. Cost effective acceptability curve for once weekly resistance training (2RT) versus balance and tone with QALYs estimated from the EQ-5D and; d. Cost effective acceptability curve for once weekly resistance training (2RT) versus balance and tone with QALYs estimated from the SF -6D. ................................. 162 Figure 6-1.  Box plot of g lobal utility score for the EQ-5D  ............................................................................................................. 177  x Glossary of Terms and Abbreviations  Abbreviation           Definition  ED Emergency department EQ-5D EuroQol-5D HRQL Health related quality of life  HRU Health resource utilization HSUV Health state utility value HUI3 Health utilities index mark 3 ICECAP ICEpop CAPability measure for older people  ICER Incremental cost-effectiveness ratio IRR  Incidence rate ratio MMSE Mini mental state examination OEP  Otago exercise programme QALY Quality adjusted life year QUOROM  Quality of reporting of meta-analyses RCT  Randomized controlled trial RT Resistance training SF-6D Short form-6D TUG  Timed up and go   xi Acknowledgements  During the past four years, I have enjoyed the privilege of working with two amazing thesis supervisors – Dr Carlo Marra and Professor Karim Khan. Carlo, you were an instrumental part of my decision to train in applying health economics to the field of falls prevention research. Thank you for providing such a rich learning environment. Your thoughtful contributions and your constructive feedback o n the papers in my thesis have provided me with a steep learning curve during the past four years. Karim, thank you for your strong support during my PhD. You always took the time to engage in every conversation as a learning opportunity. Your mentorship can be counted in so many ways, but most importantly, I appreciate you empowering your students with the tools to become successful researchers. Karim and Carlo, you both have fostered the research environment as a challenging, but fun and exciting place to  be.  I was extremely fortunate to have some ‗hands on‘ committee members for my PhD – Drs.Teresa Liu- Ambrose, Clare Robertson, Maureen Ashe and Meghan Donaldson. Teresa, first and foremost thank you for providing the Brain Power study as a vehicle for me to learn about economic evaluations alongside clinical trials. You are a wonderful role model and I appreciate your generosity with time during the past four years. I have learned many important research methods regarding clinical trials and team management from you and I will be forever grateful. Clare, working with you during my PhD has truly been a gift and your mentorship to help me attain the skills I was seeking during my PhD was nothing short of extraordinary. Thank you for your yearly two-week visits where we worked through many economic evaluations together and of course, we will not forget Paris! Maureen, you were my first mentor in the research world starting with my MSc. Thank you for the support and encouragement over the years. Meghan, I greatl y appreciate your friendship and encouragement during my PhD. I always knew you were only a phone call away. Thank you for sharing your PhD journey with me to provide mentorship and guidance to my experience.  Of course, this thesis would not be completed without the Brain Power participants. I extend my warmest gratitude to all of the participants in this study who so generously contributed their time. The meticulously filled out health resource utilization forms were much appreciated. Thank you to the Brain Power research assistants, who made the data collection periods possible. A big thank you to the trainees in the Collaboration for Outcomes Research and Evaluation (CORE) and the Centre for Hip Health and Mobility (CHHM), who helped create a rich learning environment. I would like to extend a special thank you to Dr. Penny Brasher who took the time to meet with me weekly for a year to teach me the fundamentals of regression techniques. Penny you are truly a dynamic teacher who makes statistics come alive . Dr Heather McKay, thank you for taking the time out of your busy schedule to meet with me and to listen to my thesis presentations where you offered valuable feedback. I appreciate your commitment to the trainees at CHHM.  Thank you to my friends and family for your love and encouragement over the years – Mom, Dad and my husband Steve. I always knew you were only a phone call away. This experience would not have been possible with out you!   xii Co-authorship Statement  Sections of this thesis have been published as multi-authored manuscripts in peer-reviewed journals and are indicated with below. I provide details of the authors‘ contributions for Chapters 1 through 8.  Chapter 1: Author Contribution: Jennifer C. Davis was responsible for design, literature search, collation and summary of papers, retrieval of articles, review of studies and writing of Chapter 1.  Chapter 2: Author Contributions: Jennifer C. Davis, M. Clare Robertson, Maureen C. Ashe were responsible for study concept and design, extraction of data, analysis and interpretation of data, preparation of manuscript, and critical review of manuscript. Teresa Liu-Ambrose, Carlo A. Marra and Karim M. Khan were responsible for study concept and design, analysis and interpretation of data, preparation of manuscript and critical review of manuscript.  Chapter 3: Author Contributions: Jennifer C. Davis was principal investigator, was responsible for design, quality assessment, literature search, collation and summary of papers, retrieval of articles, revie w of studies and writing of manuscript. M. Clare Robertson, Maureen C. Ashe, Teresa Liu-Ambrose, Karim M. Khan and Carlo A. Marra were also responsible for design, quality assessment, review of studies and writing the manuscript. Maureen C. Ashe, Jennifer C. Davis and M. Clare Robertson were responsible for quality assessment.  Chapter 4: Author Contributions: Jennifer C. Davis was principal investigator for the economic evaluation, was responsible for design, data analysis, data interpretation and writing of manuscript. Teresa Liu- Ambrose was principal investigator for the Brain Power study and was responsible for study concept and design, acquisition of data, data analysis and data interpretation and reviewing of the manuscript. M. Clare  xiii Robertson, Mehdi Najafzadeh and Carlo A. Marra were responsible for study design, data interpretation and critical review of manuscript. Maureen C. Ashe and Karim M. Khan were responsible for study design, data acquisition and critical review of manuscript.  Chapter 5: Author Contributions: Jennifer C. Davis was principal investigator for the economic evaluation, was responsible for design, data analysis, data interpretation and writing of manuscript. Teresa Liu- Ambrose was principal investigator for the Brain Power study and was responsible for study concept and design, acquisition of data, data analysis and data interpretation and reviewing of the manuscript. M. Clare Robertson, Mehdi Najafzadeh and Carlo A. Marra were responsible for study design, data interpretation and critical review of manuscript. Maureen C. Ashe and Karim M. Khan were responsible for study design, data acquisition, and critical review of manuscript.  Chapter 6: Author Contributions: Jennifer C. Davis was principal investigator for the evaluation of HRQL and healthcare resource use and, was responsible for design, data analysis, data interpretation, writing of manuscript. Teresa Liu-Ambrose was principal investigator for the Brain Power study and was responsible for study concept and design, acquisition of data, data analysis, data interpretation, and reviewing of the manuscript. Carlo A. Marra and Mehdi Najafzadeh were responsible for design, data interpretation and critical review of manuscript.  Chapter 7: Authors Contributions: Jennifer C. Davis was principal investigator for the evaluation of HRQL and healthcare resource use and, was responsible for design, data analysis, data interpretation, writing of manuscript. Teresa Liu-Ambrose was principal investigator for the Brain Power study and was respons ible for study concept and design, acquisition of data, data analysis, data interpretation and reviewing of the manuscript. Carlo A. Marra and M. Clare Robertson were responsible for design, data interpretation and critical review of manuscript.  xiv Chapter 8: Author Contribution: Jennifer C. Davis was responsible for design, literature search, collation and summary of manuscript-based chapters within the thesis, retrieval of articles, review of studies and writing of Chapter 8.  1 1 Introduction  In Chapter 1, I provide: 1) an overview of this thesis, 2) a literature review and 3) a description of the objectives of the six studies that comprise this thesis. 1.1 Overview of thesis  Falls are a substantial public health problem in older adults in terms of incidence, health burden and health related costs. Cohort studies have consistently demonstrated that 30% of those aged 65 years and older experience at least one fall each year and half of those fall recurrently. 1,2 There is compelling evidence that falls are preventable;3 individually customized multifactorial interventions,4 5 multiple interventions6 and single factor interventions,7-9 have all prevented falls in community dwelling seniors. Non-fatal fall injuries are associated with increased morbidity,10 decreased functioning11 and increased health care resource utilization.12 Falls and fall-related injuries including fracture account for 10-15% of emergency department presentations of those aged 65 years and older 13,14. Based on the evidence that falls are costly and they are preventable, I focus my thesis on applying economic concepts to the field of falls.  The importance of economics applied to the field of falls and falls prevention has largely been overlooked. I undertook a comprehensive literature review to 1) identify the cost of falls, 2) determine the falls prevention strategies that provide the best value for money, 3) ascertain the costs and consequences of hip fracture resulting from falls and 4) examine economic evaluation methodology currently used in falls prevention research. I present a synthesis of the evidence in Chapter 1 (Section 1.2). At the end of Chapter 1, I formally outline my Rationale, Objectives and Contribution for the six manuscripts in my thesis.  Chapters 2 to 7 contain the manuscripts describing the six studies in this thesis. In Chapter 2 the systematic review of the international cost of falls studies is presented (Study 1), and in Chapter 3, the systematic review that investigates which effective falls prevention strategies provide the best value for  2 money (Study 2).  My third study, described in Chapter 4, consisted of an economic evaluation of the interventions tested in the first year of the Brain Power randomized controlled trial (RCT). 15 I was responsible for study concept, design, data acquisition and analysis for the economic evaluation o f this trial. Briefly, the Brain Power study was a RCT aimed at improving cognition and preventing falls using two doses of resistance training compared with balance and tone classes (comparator) among community dwelling older women. A strength of the Brain Power study is that participants were followed for a further one year after the initial one year clinical trial to determine if any longer terms benefits were sustained. In Chapter 5, I detail the results of the second year followup study (Study 4).  Given that different health utility instruments can be used to calculate quality adjusted life years (QALYs), I took the innovative step of comparing the incremental cost-effectiveness ratios (ICERs) from QALYs estimated using the SF-6D and the EQ-5D (Study 5, Chapter 6). Further, I utilized two multiple linear regression models to determine the key independent contributors to both health resource utilization (HRU) and health related quality of life (HRQL). In Chapter 7, I specifically ascertain the contributi on of executive function to HRQL as assessed using the EQ-5D (Study 6).  Finally, in Chapter 8, I provide an integrated discussion of the six studies and suggest directions for future health economics research applied to the area of falls prevention. 1.2 Literature review  In Section 1.2, I provide a comprehensive literature review that places falls prevention research in the context of economic evaluations. I introduce economic evaluation concepts relevant to this thesis and review the international economic burden of falls and the value for money of falls prevention strategies.  3 1.2.1 Incidence of falls in older adults  Falls and injuries resulting from falls in elderly people represent a significant health burden. 16-18 The risk of falls increases with age as demonstrated by a cohort study where adults 70 -74 years experienced 47 falls per 100 person years and adults aged 80 years and older experienced 121 falls per 100 person years.19 Falls are the leading cause of death from injury in older adults with more than 40% of hospital related admissions being secondary to falls.20 Further, approximately one half of older adults admitted to hospital for a fall related injury are subsequently discharged to a nursing facility. 20  Population-based studies have examined the incidence of injurious falls.3,21,22 A population based study in the US estimated that 3.7 million individuals sustained a single fall in 2002 and 3.1 million individuals suffered recurrent falls.23 Two million of these individuals had an injurious fall. These estimates were based on a population estimate of 44 million enrollees in US Medicare.  An understanding of falls at a population level and at an individual level is crucial from the individual, health care system and health policy perspectives. Careful examination and tracking of the incidence of falls and subsequent fall related injuries could lead to changes in care that may improve healthy functioning of older adults. These health outcomes have economic implications given that injurious falls result in he alth resource utilization (HRU). 1.2.1.1 International estimates of incidence rates of falls  The incidence of falls in community dwelling older adults has been documented internationally in several prospective cohort studies. I provide specific evidence from four cohort studies regarded as pivotal.1,2,24,25 Of the tested methods for prospective collection of falls incidence data, monthly fall diaries that are completed and returned by participants is the current gold standard. 26 Only two1,2 of the four prospective cohort studies had used this method to document the incidence  of falls. The other two24,25 studies required  4 their participants to return the falls diaries less frequently – such as bi-monthly or tri-monthly. Hence, there may be potential for under-reporting of falls in these two prospective cohort studies due to recall bias.  A nine-year prospective cohort study detailed the number of falls among seniors and demonstrated that on average, the mean number of falls experienced per participant was 1 (SD: 1.6, n=300) during a 1 year period that fall diaries were tracked.24 Approximately 50% of the women enrolled in this study reported at least one fall (range: 1 to 11); 86 experienced only one fall and 65 (21.7%) experienced at least two fa lls. These proportions are comparable with other prospective studies. 2 One limitation of this study is the generalizability of the results as the study sample included only community dwelling women.  In a US study (1988), 336 older adults aged 75 years and older were enrolled and monitored for one year to study risk factors for falling.2 Participants reported the occurrence of falls in their monthly fall diaries and study personnel phoned participants semi-monthly to remind participants to fill out and mail in their calendars. Of the 334 participants enrolled, 108 (32%, 95%CI: 27 to 37%) participants reported at least one fall during the year, 29% fell twice and 27% fell three times. In summary, 272 falls were reported during the one year period for 334 individuals.2 Forty-four percent of falls occurred in the presence of an environmental hazard, 10% during acute illness and 5% during hazardous activity. This study, conducted approximately 20 years prior to the 9-year prospective study24 highlights that the incidence of falls is a problem for community dwelling older men and women. One year later than the US study, another prospective cohort study of 761 older adults aged 70 years and older was conducted in New Zealand.1 Participants reported falls in monthly fall diaries and were telephoned monthly. To mitigate recall bias, families of participants with known memory impairment were also contacted monthly by telephone. This was the only study that attempted to address recall bias for individuals with known impairments. Of the 507 falls reported, 39.6% of women and 28.4% of men experienced at least one fall. Of note, a greater proportion of women (30%) fell compared with men (25%)  5 in this study.1 Key strengths of this study were the inclusion and direct comparison of men and women. The proportion of individuals that fell at least once was greater than that observed in the European and US studies. One possible explanation for this could be the vigilant efforts of contacting families of participants to collect falls information. Thus, this study may provide the most accurate depiction of the incidence of falls in community dwelling older adults.  In 1994, another one year prospective cohort study of 341 older adults aged 65 years and older was launched in Australia.25 Participants reported the occurrence of falls monthly by filling out questionnaires that were mailed every two months. A key strength of this study was the details of the falls provided. The falls details included: 1) number of falls in the past 2 months, 2) location, 3) causes and 4) injuries sustained. Participants who did not return these questionnaires were telephoned. Of the 287 falls reported, 39.3% of women reported one or more falls. This study was similar to the New Zealand study in two ways: 1) the inclusion of men and women; and 2) the incidence of falls reported.  From these four pivotal prospective cohort studies,1,2,24,25 older adults at least aged 65 years have a 28% to 39% probability of having at least one fall and 9% to 21% probability of at least two falls in a 12 -month period. The estimates of incidence were greater for the studies based in New Zealand and Australia. It is possible that this observed difference is due to: 1) time trend (i.e., reduction in incidence of falls over time), 2) geography or most importantly 3) methodology designed to minimize recall bias. 1.2.1.2 International incidence rates of falls in high risk older adults  Three pivotal prospective cohort studies provided the first data on incidence rates of falls in older, high risk adults.27,5,28 High risk adults were defined in these studies as individuals: 1) with recurrent falls, 2) prone to injurious falls or 3) persons presenting after a fall.29,30 A 1989 US prospective cohort study examined risk factors for recurrent nonsyncopal falls in 325 community dwelling women and men aged 60 years and older  6 who reported a history of at least one fall.27 Participants were required to report their falls weekly. This is the only study that required such frequent reporting of falls over a 12-month followup period. Of the 593 falls reported, 57% of participants had at least one nonsyncopal fall, 25% fell just one time, 31% experienced two or more falls and 19% suffered three or more falls in 12 months.27 The odds of two or more falls increased for individuals who had (at baseline) difficulty standing up from a chair, difficulty performing a tandem walk, arthritis or had Parkinson's disease. The odds of three or more falls during the previous year increased for individuals who had a fall with injury during the previous year (Odds Ratio: 3.1, 95%CI: 1.5-6.4) and for Caucasians (Odds Ratio: 2.4, 95%CI: 1.3-4.4).27 Further, the proportion of individuals with at least two falls per year increased from 0.10 for those with one or fewer risk factors for falling to 0.69 for those with at least four risk factors for falling. Whereas in the US trial discussed above 2 32% of participants fell at least once during a one year followup, the proportion who fell in a given year in this trial dramatically increased as the number of risk factors for falling increased. 27  Further, a UK cohort study was conducted of 397 adults aged 65 years and older who had attended the accident and emergency department because of a fall.5 Results of this study I highlight in this paragraph were specific to the control group of the Prevention of Falls in the Elderly Trial (PROFET). Participants fall records were tracked for one year by a questionnaire delivered at 4 months, 8 months and 12 months. Hence, a limitation of this study is that falls data were only collected at three time points over a 12-month period predisposing these data to recall bias and under-reporting of falls. Despite this limitation, 52% of participants fell at least once and 26% fell at least three times in the 12-month followup period. This high falls rate is likely reflective of the high risk frail sample of fallers that were targeted. These individuals falls were severe enough to warrant a visit to the accident and emergency department. Hence, this population has generalizability only to a specific population of adults aged 65 years and older.   7 In 2006, a Finnish study examined falls incidence among 555 high risk adults aged 85 years and older over a two-year followup period.28 Participants‘ falls were tracked via a telephone interview bimonthly. The circumstances surrounding the falls including timing, related activity, type of fall and injury sustained were recorded during the telephone interviews. Seventy four percent of participants fell at least once and 47% of participants fell at least two times over the two-year study followup. The notable larger incidence of falls in this sample is likely due in part to the older age and subsequent presence of a greater number of risk factors for falling of participants included in this study. A strength of this study is the two-year followup of these older adults and the comparison of men and women – in particular – the finding that falls incidence among men increases to a greater extent than women with ageing.  These three key studies of higher risk cohorts27,5,28 of older adults highlighted that the proportion of participants who fell was greater compared with lower risk cohorts and ranged from 52% to 58%. Specifically, the incidence of falls was greatest for cohorts at least 85 years old.  1.2.1.3 Canadian data – estimates of incidence and prevalence  Estimates of the incidence of falls and injurious falls among older adults in Canada are sparse. Only one such study prospectively assessed the incidence of falls in community dwelling older adults. 31 Included in this 48-week prospective cohort study were 409 community dwelling adults aged 65 years and older. An interviewer telephoned participants every four weeks to obtain a self-report account of their falls since the last interview. A strength of this study is that participants were provided with a monthly calendar with adhesive stickers that they could use to track their falls and facilitate memory recall. This was the only study that employed this methodology. Twenty-nine percent of study participants reported a fall during followup. Of these, 17.6% fell once and 11.5% fell two or more times. The observed incidence of falls in this study was less than that observed in US, European, Australian and New Zealand prospective cohorts and  8 this may reflect a geographical difference. These differences are likely explained by the short time horizon (i.e., only 4 months) that may not be truly representative of fall rates in the longer term. Given the lack of other prospective cohort studies of falls incidence in Canada, it is not possible to accurately estimate to date the incidence of falls in Canada. 1.2.1.4 British Columbia and Vancouver data – pilot data of incidence  To date, no prospective studies among older adults have been done in British Columbia. However, there are recent population based falls related data from Smart Risk. Founded in 1992, Smart Risk is a Canadian national non-profit organization that focuses on injury prevention.32 In 2004 in BC, falls accounted for 333 deaths, 15 496 hospitalizations, 115 508 hospital related visits, 4254 permanent disabilities and 358 permanent total disabilities. In 2004, the BC population consisted of approximately 4 196 383 individuals of all ages and of these, 258 929 (6.1%) were aged 65 years and older.  However, these data are not specific to low-trauma falls given the limitations of how the database was originally characterized. In this database, a low trauma fall is defined as a fall sustained from a standing height or less and not due to an overwhelming force.33 These findings may not be generalizable to all older adults who sustain a low-trauma fall because these data include a broader definition of falls such as falls from a ladder.  Specific to the Vancouver General Hospital Emergency Department setting in Vancouver, British Columbia, I report some pilot data below between December 1 2006 and March 31 2007. 34 I was part of a team that identified 390 falls by the 381 individuals aged 70 years and older who presented to the Emergency Department secondary to a fall. The mean (SD) age of fallers was 83.6 (7.4) years, 69% were women and 134 (34%) falls resulted in hospital admission with an average (SD) length of stay of 31.6 (41.6) days. One hundred and eighty three (47%) fallers sustained fractures. Not surprisingly, common diagnoses were fractures to the hip or contusions/lacerations and syncope.  9 1.2.2 Costs of falls  In Section 1.2.2, I introduce cost of illness studies and review the current cost of falls literature. I highlight the importance, specific use and future directions for cost of illness (falls) studies. 1.2.2.1 Introducing cost of illness studies  To date, cost of falls studies have examined the impact of falls and fall related injuries from a health care system, health service or societal prospective either at the population level or per person based estimates. In general, cost of illness studies assess the impact of mortality and morbidity on society with respect to resource availability and welfare.35 By definition, all the effects quantified in cost of illness studies are tabulated as costs; these may include resource utilization, resource losses or welfare losses. 35 I will detail cost of falls studies by country of study because each country has a unique health care system and different population density; therefore, I would not expect the cost of falls to be directly comparable . 1.2.2.2 How to quantify burden of disease?  Rice 35,36 formalized cost of illness approaches to include three broad cost categories. These include: 1) direct costs defined as health care and health related costs, 2) indirect costs relating to loss of work productivity including death and early retirement, and 3) intangible costs represented by pain and suffering. For example, in cost of falls studies, direct health care costs refer to total or fall related costs incurred; these may include health professionals, admissions to hospital, laboratory investigations or procedures or medication changes. Specific to falls, indirect costs may include lost work productivity of the individual who sustained the fall or the caregiver of the faller. Intangible costs have not been included in peer reviewed cost of falls studies to date because they are difficult to measure accurately. Cost of illness studies remain an important step in the ‗Measurement Iterative Loop‘ described by Tugwell37 because of their strong use to policy-makers in health care decision-making (Figure 1-1).38-42   10              Figure 1-1. Measurement iterative loop37   Given that cohort studies have consistently demonstrated that 30% of those aged 65 years and older adults experience at least one fall each year and half of those fall recurrently, 1,2,25,31 quantifying the economic burden of falls in older people is of key importance to health care practitioners, planners and society. Specifically, 33%23 of falls in older adults are injurious and half of these lead to a visit with a health care professional. Because falls among older adults are associated with a numbe r of complications that include fracture and increased functional disability,17 it is necessary to estimate the economic burden that the injuries and other associated morbidity from falls have on health care systems and society. 1.2.2.3 Importance of cost of illness studies  Although cost of illness studies are useful for establishing different diseases as a research priority, there remains considerable discussion over the usefulness and purpose of cost of illness studies at the health policy level. I briefly focus on the relative importance for cost of illness studies in terms of establishing evidence of specific diseases or health states as a research platform. 38,43 A key concept is that each decision has an alternative choice and each choice is associated with opportunity costs. 44 Opportunity  11 costs are the net value of the next-best choice available when picking between several mutually exclusive choices.44 Within health economics, ‗allocative efficiency‘ is the primary goal. The underlying assumption is that maximizing welfare equates with maximizing health.45 Further, cost of illness studies provide decision- makers with the necessary tools for health care planning on all levels of the health care system. For example, cost of illness studies can be used for modeling cost of falls specifically in older adults. 46 Despite criticisms of cost of illness studies, they remain a relevant area of research because they inform health care decision makers as they wrestle with the competing interests demanding health care resource allocation.38- 40 1.2.2.4 Specific use for cost of illness studies  Within the economics literature, there is some debate about what cost of illness studies actually measure and how this relates to their use for health policy decisions. One practical use of cost of illness studies is that they provide a platform for establishing specific health conditions, diseases or illnesses as a priority in terms of the economic burden. Given that falls in older adults make up the single largest burden of injury, it is essential to quantify the cost of falls using a cost of illness study framework. 35,47  Critics cite that cost of illness estimates might be interpreted as the savings that c ould be averted if prevention efforts are undertaken.38,43 The risk of this type of interpretation is that it creates the potential for a disproportionate amount of resource allocation toward a disease where the health care expenditure is already high.38,39,42,43,48 This is defined as circularity where past decisions inform and predict future allocation and together this decreases efficiency.43 Therefore, it is essential to recognize that cost of illness studies inherently include past and current treatment and prevention efforts. For example, population based cost of falls studies, by definition, include the costs of falls and related injuries for cohorts of individuals. These individuals may or may not have been treated or taken part in an intervention following previous falls. Thus, cost of falls studies incorporate past and current prevention and treatment efforts.  12 1.2.2.5 Future direction for cost of falls studies  Prior to reviewing the cost of falls studies from each country, I highlight key components and criteria that I used in evaluating cost of falls studies (Table 1-1). Further, I provide a list of recommendations for the future direction of cost of falls studies. From the above estimates of costs, it is clear that there is a need for a standardized approach to collecting health care costs associated with falls. I address such a framework in Table 1-1 with a list of recommended cost items that all cost of falls studies should include and I discuss this issue specifically in Chapter 2 of my thesis. Briefly, elements of analysis that need to be considered in such a standard table would include the following cost items: inpatient, health care professional, home health, emergency department, ambulance, medication, and personal. Further, it may be important to investigate the intangible costs to ascertain whether these costs are significant drivers of total cost of falls estimates. Cost of illness studies should always state the time horizon of the analysis, the analytic perspective, population denominator and the costing based approach that was used (i.e., incidence versus prevalence based costing). Further, there is a need  for emphasis on consistent classification of falls. For example, some studies define falls as fatal and non-fatal, others classify falls into categories of injurious falls and some studies do not attempt to classify the type of fall into categories (i.e., injurious, fatal). In order to accurately cost falls, we need to reach a consensus on the categories of falls that are important to cost (i.e., injurious, fatal) and what cost items are key parameters in obtaining an accurate estimate. These recommendations coupled with rigorous reporting of the population or study sample, analysis perspective and time horizon will facilitate future comparison for the cost of falls across countries.      13 Table 1-1. Categories of cost items for inclusion in cost of illness studies Direct health care resource use costs Type of costing approach Inpatient hospital Health care professional Home health care Emergency departments or ambulance costs Medications Admission to a long term care facility Personal Out of pocket expenses Family/caregiver time Absenteeism/Presenteeism  1.2.2.5.1 Cost of falls in the United Kingdom  In the United Kingdom, two studies12,49 provided estimates for the cost of falls in older people. In 2000, the cost of falls for adults aged 60 years and older was estimated at UK £981 million; 59.2% of these costs incurred by the National Health Service and the remainder by the Personal Social Service for long term care.12 Costs included in this analysis were direct medical costs such as ambulance, physician, hospital admission and discharge costs. Another study ascertaining costs to the ambulance service due to falls in the community was UK £145.83 per fall or UK £376 018 per year based on 41 338 individuals.49 The cost items for the North East Ambulance service included ambulance and emergency costs only 49 while Scuffham12 included emergency costs, inpatient costs and health care professional fees. 12 Estimating ambulance service costs only for falls is a limitation of this study due to the narrow scope. However, it does  14 enable a detailed overview of the cost item components specific to an ambulance service that contribute to health care resource use of potentially high risk falle rs and thus provides a detailed description of this one component. 1.2.2.5.2 Cost of falls in Australia  Two studies estimated the cost of falls in Australia: one prospective cohort study ascertained the costs of falls in older adults living in the community50 and one population based study quantified health system costs of falls in older people in Western Australia.51 In the prospective study, the total estimated falls related costs over a short three month period ranged from AUD $316 155 to AUD $333 648 based on 79 individuals. 50 Of these costs, hospital costs accounted for 80%, community costs accounted for 16% and personal costs accounted for 4%. Community costs were defined as services received from the community such as informal care or meal services. Sixty percent of all costs were incurred within the first month post fall. Extrapolation of these three-month cost data for the 79 adults included in the study to a one year estimate indicated that post-hospital costs were AUD $24.12 million per year mean extrapolating to the Western Australia population. The second population-based study in Western Australia in 2001-2002 (one year time horizon for costing falls) demonstrated there were 18 706 ED presentations and 6222 hospital admissions for fall-related injuries.51 These falls related events were projected to cost the health system AUD $86.4 million, with more than half of this figure attributable to hospital inpatient treatment. This population-based estimate of AUD $86.4 million is over three times greater that the extrapolated estimate from the prospective cohort study. Three potential reasons for the difference in these estimates may be: 1) the small sample size of the prospective cohort study which may indicate the sample is not representative of the target population, 2) the short followup time of the cohort study and/or 3) the inability of participants to identify and track all fall related HRU (i.e., recall bias) for the population based study only.    15 1.2.2.5.3 Cost of falls in the United States  A recent US study found that fallers incurred 80% greater HRU costs than non-fallers.23 For example, the total mean health care costs in 2002 were US $12647 for fallers who fell at least two times and US $7049 for non-fallers over a one year period.23 The cost of falls is dependent on the definition and classification of falls as injurious, fatal or unclassified. For example nine United States studies 52,53,54,55,56,57,58,59,60 detail the costs of falls; these studies vary in their inclusion criteria from non-injurious falls57 to falls resulting in death.55 In 2000, estimates of total costs from fatal and non-fatal injuries from a fall were US $0.2 billion for fatal and US $19 billion for non-fatal injuries.55 Of the non-fatal injuries, 63% of the costs were for hospitalizations, 21% for emergency department visits and 16% for outpatient treatments. 55 Further, one study estimated US $406 billion (for 34 028 786 individuals) for the total fall and HRU related lifetime costs that accounted for lost productivity.47 The direct health care costs of a fall injury event were also estimated based on a fall prevention trial. 61 It was estimated based on 272 community dwelling women aged 70 years and older, that each injurious fall related event ranged from US $63 to US $85 984 with a median cost of US $658.61 There were a total of 55 injurious falls.61 Femur fracture was the most expensive type of injury resulting from a fal l and costs US $18 638 (SD: 19 990).58 According to 1989 hospitalization records examined in Washington state, fall related injuries account for 5.3% of all hospitalizations of older adults.57  From the US estimates of fall related costs, it is apparent that the variation is much greater compared with other countries; the annual estimates are in the order of billions of dollars, rather than millions. A limitation of a number of these studies55,60 was the failure to include the population denominator (i.e., the number of individuals that were included in the cost of falls estimate) that the cost estimate was based upon. Given this limitation, it is only possible to speculate that the dramatic variation in costs in the US publications is greater given the larger population size.  16 1.2.2.5.4 Cost of falls in Canada  To my knowledge, there are no peer reviewed studies on the total economic burden of falls in Canada. I conducted a comprehensive literature review in accordance with MEDLINE, PUBMED, EMBASE and, NHS EED databases to identify cost and economic burden studies based on falls specifically published in the English language (from 1945 through May 2008). Included in my search terms and medical subject headings were fall, economic evaluation, cost, cost analysis.  Smart Risk32 did produce a report on the economic burden of injury in 2009 based on 2004 prices. However, I cannot obtain an accurate estimate of low-trauma falls from these data given the a priori classification of falls. For example, falls were not classified as low-trauma, injurious or fatal; falls were either unclassified due to missing data or classified according to age. This highlights the need for future cost of illness studies in Canada to obtain an accurate estimate of the cost of injurious and other falls in older adults. 1.2.2.5.4.1 Cost of falls in British Columbia  Previous estimates of the costs of falls have likely underestimated the economic burden due to the underestimation of falls incidence. Administrative data often have incomplete records leading to underestimated falls incidence rates. I was part of a team that collected falls data directly within an Emergency Department (ED). Each individual who presented with a fall was tracked using the ED census, to estimate the direct health costs of falls resulting in a presentation to the ED. 34 I collected information on patients aged 70 years and older that presented to the ED of a tertiary care teaching hospital. From hospital reports and the ED database, our team collected information on patient demographics, diagnoses, admission status, Canadian Triage Acuity Score of injury severity (CTAS, 1-5 where 1 is the most acute), and hospital length of stay. We used the third party payer perspective to estimate total hospital costs that included the following unit costs: ED visits, hospitalizations were taken from the fully allocated Vancouver General Hospital Cost Model valued in 2006 Canadian dollars. Between December 1 2006 and March 31  17 2007, we identified 390 falls by 381 individuals costing CAD $2 520 641, with the majority of fallers being women aged 80 years and older. Thirty four percent of falls resulted in hospital admission with a mean length of stay of 31.6 days and mean cost of hospitalization CAD $18 375. 1.2.3 Fracture – a common and costly consequence of falls  Fall related fractures are a significant cause of hospitalizations, thus adding to the financial burden on our health care system18 with approximately 90% of hip fractures resulting from a fall.62 Hip fractures have a multifactorial origin and may occur because of an increased fall frequency, reduced bone strength and/or a loss of protective reflexes. 1.2.3.1 Hip fracture  In 1990, the worldwide incidence of hip fracture was estimated at 1.3 million and was projected to increase to 2.3 million by 2008; a key component of hip fracture prevention is falls prevention. 63 More recently, estimates indicated approximately 9 million new fall-related fractures sustained each year out of the total 56 million individuals worldwide that have sustained a fall-related fracture).64 Hip fracture was the most frequent fall-related fracture (i.e., 16% of worldwide fractures were sustained as the result of a fall). 64 I focus this section on hip fractures, the most common and costly 18 consequence of a fall which also have the greatest impact on mortality, even 10 years after the incident fracture.65-69  In terms of consequences, approximately 30% of individuals who sustain a hip fracture die within one year70-72 with at least 40% of individuals experience significant functional decline.73-76 Hip fractures are a common cause of admission or transfer to institutional care, permanent disability or death and are among the most damaging of all fall-related fractures in older adults.77 The frequency of hip fractures, which varies among countries, was globally predicted to increase due to the longer life expectancy among individuals  18 aged 60 years and older.78,79 In subsequent sections, I detail the previous trends and highlight current trends for hip fracture incidence in women and men. 1.2.3.2 Hip fracture incidence  Hip fractures are a major public health problem and a costly18 consequence of falls.80 Multiple factors affect the incidence of hip fracture including, most importantly, the individual‘s risk of falling. In 1990, the total estimated number of hip fractures worldwide was estimated at 1.7 million. 81 For this literature review, I included research that provided an estimate of incidence using a cumulative incidence approach because this analysis takes into account competing risks such as death. The incidence of hip fracture among individuals aged 50 years and older increases exponentially among individuals aged 75 years and older82 with the greatest number of hip fractures observed among individuals aged 80-89 years.83,84 Estimates indicate that 93% of all hip fractures occur among Caucasians and 79% of all hip fractures occur among women.85 Of these, one of six women will sustain a hip fracture in her lifetime.86  Individuals who sustain a hip fracture have greater comorbidities than those who do not. 80 Therefore, these individuals may be predisposed to incurring the largest health care system costs. Specifically, fractures of the hip, wrist and spine all cause enormous burdens both in terms of costs and consequences, to the health care system, families, caregivers and of course patients themselves. 80 The associated deaths, morbidity and economic burden warrant prevention efforts. A focus of my thesis is to examine falls prevention strategies that provide the best value for money. By identifying these strategies, researchers may be able to reduce the burden of fall-related hip fractures most economically.  In subsequent sections I detail hip fracture incidence and cost by country because each country has its own health care system, patient demographics, geography, economic climate and population size. Hence, it is most appropriate to compare studies within countries firs t and then across countries.  19 1.2.3.2.1 Time trends in hip fracture incidence  Most recent evidence indicates that the incidence of hip fracture is declining80 or at least stabilizing.87-90 I include a recently published hip fracture trend figure from a US study that is representative of the decline in hip fracture incidence (Figure 1-2). Of note, this decline in hip fracture among women is greater than that observed in men during the 1995 through 2004 period (i.e., 24% in women > 85 years; 13% in men > 85 years).80 Figure 1-2. Recent hip fracture trends by age and sex80  To date, the reasons why these hip fracture trends have changed remains unclear. 80 One potential reason is that the decline in incidence of hip fracture observed after 1995 corresponds temporally with the market production and release of bisphosphonates; however, no causal association has been demonstrated to date.91 However, studies documenting medications have confirmed increases in the use of Copyright restrictions may apply. Brauer, C. A. et al. JAMA 2009;302:1573-1579. Temporal Trends in Hip Fracture Incidence by Age for Men and Women  20 bisphosphonates after 1995.80,91 Other potential reasons for the decline in hip fracture incidence may be lifestyle changes that include calcium and vitamin D supplements, engagement in resistance training or cessation of smoking; however, most large epidemiologic studies do not have access to these specific data.92,93 Last, it was suggested that an awareness of falls may be one contributing factor to the decrease in hip fracture incidence.91 1.2.3.2.1.1 Hip fracture incidence in the United States  The incidence of hip fracture is greater in the US compared with other countries and this may in part be due to the greater aging population density.80 In the United States, it is estimated that a 50 year old white woman has a 17% chance of a hip fracture in her lifetime and a 50 year old man has a 6% lifetime chance of a hip fracture.94  Previously the global estimate of the incidence of hip fractures was estimated to increase by 100% for women and 135% for men; however, recent data demonstrate that from 1985 to 2005, the incidence of hip fracture was declining.63 80 Specifically, Brauer and colleagues80 have identified two distinct trends based on two periods (1986 through 1995 and 1995 though 2004). From 1986 through 1995, the incidence of hip fracture was increasing while mortality after hip fracture was declining.80 In contrast, from 1995 through 2004, the incidence of hip fracture was declining while mortality after hip fracture remained unchanged.80 This notable decline in incidence was previously corroborated.95 Specifically, research indicated that hospitalizations for hip fractures are not increasing.95 Lengths of stay are declining and thus has reduced the demand on hospital resources.95 However, the average charges per hospitalization rising and a greater number of individuals are discharged to long term care facilities indicating that the economic consequences of hip fracture should continue to increase.95    21 1.2.3.2.1.2 Hip fracture incidence in Europe, Asia compared with US  Internationally, research demonstrates the incidence of fall-related hip fractures has reached a plateau.96 Previous research had estimated the annual incidence of hip fractures was between 0.42-3.33 cases per 1 000 in those 60 to 100 years old,97 and that the incidence of hip fracture was increasing.97 These predictions were consistent with other European countries98 and the United States.99 However, we know now that this is not the case. The reported incidence of hip fracture estimated that by 2010, the number of hip fractures will decrease by 11% compared with 1996 data.96 This decrease was expected to be greater in women (19%) than in men (7%).96 Of note, the incidence of hip fracture in the US is greater than estimates from Europe and Asia, although the trend of the stabilization or decline in incidence is the same.80 1.2.3.2.1.3 Hip fracture incidence in Canada  Previously, the aging population in Canada led epidemiologists to predict an epidemic of hip fracture due to the increasing number of older adults who are at risk for fracture .63,100-102 In adults aged 70 years and older, the risk of hip fracture increases exponentially which led to the projections in Canada, that the numbers of hip fractures would quadruple from 1990 to 2040.100,102  Despite this projection, epidemiologic studies now indicate that the incidence of hip fracture in men and women in Canada are decreasing; this decline is greater for women than men. 103 Specifically, the rate of hip fracture in women was 266.42 per 100000 and the rate of hip fracture in men was 124.93 per 100 000 in 2003. In 2005 the rate of hip fracture in women decreased to 242 per 100 000 and in men decreased to 120.77 per 100000.103 These rates are specific to a population within Ontario, Canada. These Canadian data are consistent with international evidence cited almost 10 years prior where the incidence of hip fracture in Finland was on the decline.104   22 Most recently, a notable decline in hip fracture rates in Canada from 1985 through 2005 was demonstrated.105 Specifically, over this 21-year period, the age specific hip fracture rates decreased and age adjusted hip fracture rates decreased by 32% in females and 25% in males. 105 Further, the authors found an overall percentage decrease in annual hip fracture rate of 1.2% (95%CI: 1.0% -1.3%) from 1985 through 1996 and 2.4% (95%CI: 2.1%-2.6%) from 1996 through 2005.105 In summary, hip fracture rates have decreased; the rate of decrease is greater in females than males. 105  Yet, I highlight that regardless of the decline in rate of hip fracture, a substantial number of older adults in Canada will sustain hip fracture and incur the morbidity, mortality and economic burde n associated with hip fracture. 1.2.3.2.1.3.1 Hip fracture incidence in British Columbia  BC contains the third largest population of older adults aged 65 years and older compared with other provinces in Canada.106,107 Fall prevention is a high priority in BC and given 95% of hip fractures are  the result of a fall, hip fractures are also a logical area of high priority. 106,107 There are limited data on the epidemiology of hip fractures and to my knowledge, no published study has estimated the rate of hip fractures in BC. 1.2.3.2.2 Cost of hip fracture  Hip fractures result in substantial burden to the patient, the health care system and the community and have a large direct medical costs18,108 when compared with other low-trauma fractures that are a common result of a fall, such as wrist and vertebral fractures.71 Hip fractures are the leading cause of health care resource utilization in the form of hospitalization for older adults (mean (SD) age 78.4 (8.8) years). 109,110 The mean age noted in these studies were lower than the mean age noted for hip fracture;62,101 therefore, these estimates may be an underestimate of the economic burden imposed by hip fracture.  23 1.2.3.2.2.1 UK estimates of the cost of hip fracture  In 1995, hip fractures incurred costs of UK £942 million in the United Kingdom based on 51 863 hip fractures.109,110 The acute inpatient cost of osteoporosis related hip fractures were estimated at UK £4 808 per person and the social care costs and long stay hospital costs were estimated at UK £7 152 per person at 1995/1996 prices.109 Johansen and Stone111 report that Dolan and Torgerson‘s109 fracture costs are an underestimate of total costs in the UK because they extrapolated data from a study done 20 years ago to estimate their costs and thus, under estimated the incidence of hip fracture. 1.2.3.2.2.2 US estimates of the cost of hip fracture  In 1995, hip fractures consumed 8.8 billion dollars of the health care budget in the United States. 109,110 I highlight below the most comprehensive cost of illness study in the hip fracture literature by Braithwaite and colleagues.108 The lifetime attributable cost in 2001 prices for hip fracture are the sum of all direct and indirect costs that occur subsequent to the hip fracture in an individual‘s lifetime. Using these methods, a hip fracture was estimated to cost US $81 300. Specifically, US $8900 resulted from the initial hospitalization, US $3900 resulted from subsequent hospitalizations, US $2300 was due to rehabilitation, US $35 400 was due to nursing facilities and US $30 800 was due to home care.108 Of these total costs, 33% were incurred within the first six months after sustaining a hip fracture and 56% o f these costs occurred within the first year. 1.2.3.2.2.3 Canadian estimates of cost of hip fracture detailing direct and indirect costs  In Canada, the one year direct and indirect costs of hip fracture at 1997 prices, ranged from CAD $24 000 to CAD $30 000.18 These costs were less for community dwelling residents (CAD $21 385 in Canada) than long-term care residents (CAD $33 729). Direct costs are defined as resources used within the health care sector. For example, these may include inpatient and operating room resources. Indirect (productivity) costs are defined as a patient‘s out of pocket expenses that includes  time of patients and families consumed or freed up by the hip fracture. With an approximate 5% increase per year in medical costs, the  24 resultant cost of hip fractures each year will only increase.112 Delays in hip fracture surgery for medical reasons or delays because of insufficient resources (staff, operating rooms) further add to inpatient costs and health care resource utilization. In summary, data indicate that hip fractures are an economic burden in Canada and worldwide. By targeting individuals at high risk of hip fracture (for example seniors reporting a fall) there is large potential for cost savings from cost-effective falls prevention strategies. 1.2.4 Defining quality adjusted life years: direct and indirect measurement  In Section 1.2.4, I introduce QALYs, how they are measured and their use in economic evaluations. 1.2.4.1 Introduction to quality adjusted life years  QALYs are defined as the benefit of a health intervention in terms of time in a series of quality -weighted health states, in which the quality weights reflect the desirability of living in the state, typically anchored at ―perfect‖ health (weighted 1.0) to dead (we ighted 0.0).113 The quality weights spent in each state are multiplied by the time spent in each state. The sum of all these products is the total number of QALYs. I present this graphically in Figure 1-3  below. Specifically, QALYs are calculated from the area under the curve of a graph of health state utility values (HSUVs) versus time spent in each health state.  Figure 1-3. Quality adjusted life years QUANTITY OF LIFE  (Years) 1.0 0.0 Death Death Without Program With Program Quality- Adjusted Life Years-Gained Q U A L IT Y  O F  L IF E  ( W e ig h t s )  25 The QALY is one of the preferred measures of health status for econo mic evaluation because it integrates years of life (i.e., quantity) and quality of life into one outcome measure. QALYs are commonly estimated for assessment of health benefit in clinical trials, economic evaluations and in patient outcomes after receiving health care treatments or procedures. In some countries such as the UK, QALYs are a preferred outcome to use in decision making by the National Institute for Health and Clinical Excellence (NICE). 114  Briefly, the QALY is a useful measure of health benefit because it simultaneously captures both quantity and quality gains or losses.44 A key benefit of the QALY is that it enables direct comparison of patient outcomes across diseases and diverse health interventions.44 Also, it accounts for changes in both morbidity and mortality under a common metric. There are a number of developed instruments that measure HRQL; I describe these below.115-117 For my thesis work, I compared the QALYs gained between interventions. 1.2.4.1.1 Quality weights and assumptions that satisfy the QALY concept  A key challenge in developing a QALY concept is the need for valid quality weights for each different possible health state to answer the question: how can different health conditions be compared? A number of techniques are used to construct valid quality weights. There are direct and indirect techniques used to construct quality weights. Direct techniques construct quality weig hts using methods asking participants to value health states by imagining a given health state and considering the amount they would be willing to sacrifice to avoid a defined health state.44,113 Indirect utility measures are often used in observational or clinical trials because they generate scores that can be utilized to calculate QALYs. 44 The most common direct techniques are: 1) the standard gamble; and 2) the time trade -off; 3) rating scale or VAS (commonly used). These techniques have assumptions that need to be satisfied prior to estimating quality weights.44,113 These include 1) quality weights should be based on individuals‘ preferences for health states and not using psychometric approaches, 2) QALYs must be measured on an interval scale and 3) quality  26 weights must be anchored on measures of full health (1.0) and death (0.0). I will describe both techniques in sections below. 1.2.4.1.2 Direct method of estimation of preferences – two classic techniques  By way of background, preference is a general term used to describe the desirability that individuals have for a set of outcomes; preferences can be categorized as a set of values or utilities. Values are preferences measured under conditions of certainty (i.e., time trade-off technique, rating scale) and utilities are preferences measured under conditions of uncertainty (i.e., standard gamble technique). 44,113  Both the standard gamble and time trade off techniques consist of methods that ask participants to value health states by imagining a given health state and considering the amount they would be willing to sacrifice to avoid a defined health state.44,113 Alternative elicitation techniques are rating scales and ratio scales, but I will not focus on these for this thesis because in general, choice based techniques are recommended over scaling methods such as rating scales.44,113 1.2.4.1.2.1 Standard gamble  The standard gamble is one of the classic direct methods of measuring individual preferences for health states based on the third axiom of von Neumann and Morgenstern‘s 118 utility theory.44 It is based on the following premise: if lottery p is preferred or equivalent to q, then the utility from lottery p is greater than utility from lottery q and vice-versa. The administration of the standard gamble depends on whether the health state in question is considered worse than death (i.e., the HSUV is less than zero). The key concepts underlying standard gamble technique is illustrated in Figure 1-4.44      27       Figure 1-4. Standard gamble example for a health state i preferred to death44  Participants respond to standard gamble questions by rating their preferences for intermediate cases in comparison with lotteries that involve both better and worse outcomes. 44 For example, if you had to imagine living the remainder of your life with blindness, what is the percent risk that you would willingly accept for a therapy option ‗Choice A‘ (Figure 1-4) that would result in full health (‗p‘) and with the percent risk that you would accept of immediate death (1-p) compared with remaining in your existing health state ‗i‘ (Choice B)?  Valuation of an individual‘s preferences is achieved by varying the probability (p). First, a participant is offered two choices: A and B and each choice has a possible consequence. If the participant chooses choice A, he or she will either: 1) remain perfectly healthy to ‗ t‘ years with a probability (p) or 2) die immediately with a probability (1-p). The outcome of choice B is certain for t years. The probability ‗p‘ is varied like a lottery until the participant is indifferent between choice A and choice B. It is at this point o f indifference that the participant‘s preference score for health state i for t years is p. Perfect health is defined as 1.0 and death is defined as 0.0. The standard gamble is traditionally administered in face -to-face scripted interviews with visual aids. More recent modes of administrations include interactive computer approaches, paper based and group interview approaches.119,120  There are noted limitations to the standard gamble technique. Participants often d o not give sensible or consistent answers to the questions.121-123 For example, McNeil and colleagues121 found that among three different populations who were asked to imagine they had lung cancer, surgery was more attractive relative Healthy Dead State i Choice A Choice B p 1-p  28 to radiation therapy when: 1) treatments were identified rather than not, 2) participants were informed of life expectancy rather than cumulative probability of survival and 3) the problem was framed as a prob ability of living instead of the probability of dying. Similarly, another study demonstrated that framing a given decision problem in various ways led to changes in individuals preferences.123 Consequently, this suggests the need for awareness of these potential interviewer biases and guidelines for how to correctly frame a standard gamble questionnaire when interpreting the findings. 1.2.4.1.2.2 Time trade-off  Another direct method of determining an individual‘s preferences for health states is the time trade -off. Originally developed for use in health care, the time trade-off was considered an easy to use tool compared with the standard gamble with comparable methodological quality. To administer the time trade -off, the participant is offered two alternatives: 1) health state ―I‖ for time ―t‖ followed by death or 2) perfect health for time x<t followed by death. The time x is varied repeatedly until the participant is indifferent between the two choices. At the point of indifference, the preference score for health state I is x/t (hi=x/t).44 Figure 1-5 depicts the time trade-off method.44           Figure 1-5. Time trade-off method for a chronic health state preferred to death44       Healthy Dead State i x t Choice B hi TIME  29 1.2.4.1.2.3 Visual-analogue scale  The visual-analog scale is the simplest approach to measuring preferences.44 Subjects are asked to rank their most preferred, then their least preferred health outcomes. Subsequently, they place their outcome rankings on a scale where the intervals between the rankings correspond to the subject‘s preferences.44 1.2.4.1.2.4 Comparison of the directly elicited preference scores  When comparing the standard gamble with the time trade-off we expect to obtain different preference scores because they are different techniques.113 In general, the standard gamble technique generates larger HSUVs than those from the time trade-off technique; both are larger than those from using the visual-analog scale.44,113,124-129 This is partially explained by the assumption that individuals are risk averse, thus they are less willing to accept the gamble outcome in the standard gamble and they are more willing to accept the outcome that has certainty. Individuals also have positive time preference as they value life years in the immediate future more than life years in the longer-term future. Thus, individuals are generally more willing to give up life years at the end of their life rather than in the present as in the time trade -off technique.130 1.2.4.1.2.4.1 Validity and reliability of standard gamble & time trade off  Much of the focus of previous research has tested the validity and reliability of techniques used to elicit preferences.131-133 The time trade-off technique proved valid compared to the standard gamble technique as a gold standard.134 Participants found the time trade-off technique to be the easiest to follow, the standard gamble technique to be marginally more difficult and the visual analogue scale to be substantially more difficult; this was measured by participants willingness to go through the interview and from participant feedback.134 Further, both SG and time trade-off telephone interviews gave comparable results with face-to-face interviews indicating that telephone interviews may be able to replace face -to-face interviews.135 The mean time-tradeoff values were 0.84 (SD: 0.20) versus 0.86 (SD: 0.17) for the telephone interviews compared with the face-to-face interviews.135 The mean standard-gamble utilities for the  30 telephone interviews compared with the face-to-face interviews were 0.93 (SD: 0.16) vs. 0.92 (SD: 0.17).135 This could reduce the cost in terms of time and resources for administration of these techniques. 135 Importantly, the different methods of eliciting preferences are a key factor for the preference we ights attained. 1.2.4.1.3 Indirect method of estimation – preference-based utility instruments  Preferences based utility instruments can be disease specific or generic. 136 There are currently no disease specific instruments for assessing health status post fall. Therefore, I focused my research on the use of generic instruments. Generic instruments have the advantage of facilitating comparison across different diseases and health conditions rather than only within one specific population. In contrast, disease specific instruments are often more responsive to changes in acute health status but they do not enable direct comparisons between outcomes for patients with different conditions. 137 A number of generic instruments exist in the form of questionnaires that are an indirect method of estimating individuals‘ preferences. In sections below, I outline two widely used generic preference-based utility instruments and one utility instrument that originated in Canada: the Euro-Qol-5D (EQ-5D), the Short Form-6D (SF-6D) and the Health Utili ties Index Mark 3 (HUI3). 1.2.4.1.3.1 Multi-attribute health status classification systems using the multi -attribute utility theory  A popular alternative for measuring individuals‘ preferences for health outcomes is the use of a prescored instrument that follows the multi-attribute health status classification system.44 One example of such an instrument is the HUI3. Three assumptions comprise the multi-attribute utility theory.44 They include: 1) first- order utility independence, 2) mutual utility independence and 3) additive utility independence. First order utility independence implies that an individual‘s score on any of the defined attributes/domains (e.g. mobility, pain) has no interaction with any levels of another attribute/domain (e.g. anxiety/depression). Mutual utility independence is a stronger assumption that requires there to be no interaction between any of  31 the defined attributes/domains. Additive utility independence implies that overall, an individuals‘ preferences depends only on the individuals levels of the attributes/domains and does not depend on the way that the levels of the different attributes or domains are combined. 1.2.4.1.3.2 EuroQol EQ-5D  The EQ-5D is a generic preference-based utility instrument developed by the EuroQol Group. The EQ-5D captures 243 unique health s tates and captures the following domains using a short five -item questionnaire: 1) mobility, 2) self-care, 3) usual activities, 4) pain and 5) anxiety/depression.138 Individuals‘ preferences for the scoring of the EQ-5D were estimated using the time trade off technique on a random sample of adults taken from the population living in York (UK-region) (N=3395).139 The EQ-5D is the most widely used generic instrument that uses a utility-based scoring approach, yielding a single summary score on a common scale to facilitate comparison across different health conditions and patient populations.132,140 The single summary score, defined as a HSUV is anchored at zero – a health state equivalent to death and 1.0 – a state of ―full health.‖  HSUVs less than zero define health states worse than death. HSUVs are used to calculate QALYs to account for the quality of life of a patient (measured using health utili ties from a generic preference-based utility instrument such as the EQ-5D) in a given health state and the time spent in that health state. 1.2.4.1.3.3 Short Form-6D  The SF-6D is also a generic preference-based utility instrument that is derived from a widely used HRQL questionnaire, the Short Form 36 (SF-36).116 The SF-6D contains six domains that include: 1) physical functioning, 2) role limitations, 3) social functioning, 4) pain, 5) mental health and 6) vitality. Each domain contains four to six levels that enable the SF-6D to identify 18 000 unique health states. Unlike the EQ-5D, the scoring method for the SF-6D is based on the standard gamble. A random sample (N=836) of a general adult population from the UK was used to estimate the utilities for 249 different health states. Each participant was required to provide utilities for six states. Developers then used econometric modeling on  32 these data to assign utilities for all the remaining health states so that all 18 000 unique HSUVs were defined. The worst HSUV is zero (dead). The next worse HSUV in the SF-6D system is 0.30, and the best is 1.0 defined as ―full‖ health. The SF-36 can be used to obtain HSUVs using analogous questions that comprise the SF-6D. These HSUVs can subsequently be used to calculate QALYs. 1.2.4.1.3.4 Health Utilities Index Mark 3  The HUI3 is a generic preference-based utility instrument that described 972 000 unique health states.141 Similar to the EQ-5D and SF-6D, the health classification system used in the HUI3 was designed to directly link multi-attribute health status classification describing preference-based, multi-attribute utility functions.141 Specifically, the scoring of the HUI3 convert the health classifications into specific values for each attribute.142 A single value is calculated by combining for the combining the attributes used for an overall estimate of HRQL.142 Two other versions of the HUI exist and each of these versions share some common attributes. As a result, some studies138,143 combine the Health Utili ties Index Mark 2 and HUI3 instruments to enable description of well over 1 000 000 unique health states. Preferences for the HUI scoring function were based on a random sample of school children and parents residing in Hamilton, Canada using both the standard gamble technique and the visual analogue technique. Health states worse than zero (dead) were identified and measured as negative values on the zero (dead) to 1.0 (full health) scale. One key difference between the HUI3 and the EQ-5D and SF-6D is that only the HUI3 used the multi-attribute utili ty theory for the estimation of the utility formula. The EQ-5D and SF-6D use econometric modeling. 1.2.4.1.3.4.1 Comparing the EQ-5D, the SF-6D  I limit this section to a comparison of the instruments used in the Brain Power study (i.e., EQ-5D and SF- 6D). Each of these preference-based utility measures provides weightings for QALYs. The SF-6D may be more sensitive to the health s tates experienced by individuals with falls as it captures physical functioning, role limitations, social functions mental health, bodily pain and vitality. In addition it describes 18 000 discrete health states and will likely capture a greater number of changes in health status because it  33 describes a greater number of health status, thus is it more l ikely to detect smaller changes.116 The EQ-5D captures fewer (i.e., 243 health states) compared with the SF-6D.138 My research provides a further comparison of the QALYs and incremental cost effective ratios (ICERs) estimated by these generic instruments specifically in older engaged in a dose-response resistance training intervention to improve cognition and prevent falls. To my knowledge, no such comparisons have been made among this group of individuals and specific to falls prevention research. 1.2.5 Quality of life implications  Given that 90% of hip fractures are the result of a fall,62 I will briefly outline literature for hip fracture regarding quality of life. Hip fractures are among the most frequently investigated fragility fractures. 144 1.2.5.1 Quality of life implications for hip fractures  Although many studies cite or define quality of life in general terms, few studies quantify quality of l ife using a utility score. A systematic review of utility values associated with osteoporotic fracture in 2008144 initially retrieved 152 articles. However, only ten of the studies specifically provided utility values among those with hip fractures.144 Of the studies that reported utility scores, most reported those obtained with the EQ-5D and few used the Health Utilities Index Mark 2.144 For the 2008 systematic review, the frequent use of the EQ-5D facilitated comparison across studies on a common health benefit metric – namely the EQ-5D overall utility score. Overall, the utility was lower in patients following a hip or vertebral fracture and even worse for patients who had combined or multiple fractures. At one year post hip fracture, the estimated reduction in QALYs due to a hip fracture ranged from 0.17 to 0.23. 145-148 Further, the HSUV of the first-year after a hip fracture was 0.80 [95%CI: 0.77,0.83] and increased to 0.90 [95%CI: 0.89,0.91] in subsequent years after this initial year after the hip fracture.144,149    34 1.2.5.2 Gap in literature: health related quality of life resulting from falls  To my knowledge, only one economic evaluation of a falls prevention study, a randomized controlled trial of expedited (four weeks) first cataract surgery (N=148) versus usual care (routine 9-13 month wait list) (n=140),150 reported QALYs as an outcome. Although QALYs were used for the economic evaluation, a specific description of QALYs for these participants was not detailed. QALYs were used in their cost-utili ty analyses to determine the value for money of expedited cataract surgery. For this cost-utili ty analysis of expedited first cataract surgery, effectiveness was estimated from QALYs determined from the EQ-5D scores at baseline, 6 months and 12 months. To estimate the number of QALYs for the entire 12-month trial period, the area under the time versus health utility curve was determined for each participant as displayed in Figure 1-3. Over the trial duration, no significant differences in mean QALYs between the surgery and no surgery group were observed.150 I highlight that this trial was for a specific population of fallers who required cataract surgery demonstrating the need for future studies that characterize HRQL using QALYs estimated from generic instruments, such as the EQ-5D. This would provide a platform for comparison with different health care interventions. Further, it is essential to characterize our population‘s health status using common metrics to ascertain benefits of health care interventions, programs or procedures in relation to competing alternatives. Comparative studies show that the same patient groups can score quite differently depending on the instrument used.151 1.2.6 Defining categories of economic evaluations  Economic evaluations provide a structure where the value for money in terms of costs and health benefits can be evaluated among competing alternatives. 1.2.6.1 Four types of economic evaluations  Given that health care resources are limited, decisions about implementing health care interventions should ideally be based on efficiency. Efficiency – the cornerstone of economic theory – may be considered a ‗guiding principle‘ for decision makers as they try to decide among competing alternatives against a  35 background of health care resource scarcity.152 Economic evaluations comparing different treatments and different interventions would provide essential data to facilitate efficient medical decision-making. Health economic evaluation quantifies the incremental costs and benefits of an intervention compared with an appropriate comparator of interest in order to determine the intervention that offers the best value for money. The focus of my research is on cost-effectiveness analysis and cost-utility analysis conducted alongside a resistance training, dose-response randomized controlled trial in community dwelling older women. There are four different types of economic evaluations: 1) cost-minimization, 2) cost-effectiveness analysis, 3) cost-utility analysis and 4) cost-benefit analysis. I highlight that while I provide a description of cost-minimization and cost-benefit analysis – it is not a component of my thesis work. 1.2.6.1.1 Cost-minimization  Cost-minimization examines only the comparative costs of alternative health care interventions. For example, if two blood pressure treatments controlled hypertension equally well, a cost-minimization would be ideal because it would only examine costs incurred as a result of each blood pressure pill rather than also assessing any differences in health status. Briefly, the three main issues of contention for the costs components of all economic evaluations are: 1) techniques for allocating overhead costs shared between departments to individual project, 2) discounting techniques that adjust for differential timing and 3) techniques for estimating productivity costs.44 These areas of contention are not a focus of this thesis. 1.2.6.1.2 Cost-effectiveness analysis  In health care, cost-effectiveness analysis (CEA) has emerged as one of the favored techniques for economic evaluation. The key distinction between CEA and cost-benefit analysis (see Section 1.2.6.1.4) is that in CEA health outcomes are quantified in terms of health benefits (i.e., falls prevented) rather than in monetary units. As defined by Neumann, ―cost-effectiveness analyses show the relationship between the net resources used (costs) and the net health benefits achieved (effects) for a specific intervention compared with a specific alternative strategy.‖113 In my research, the focus of my cost-effectiveness  36 analyses is to compare the implementation and absence of an intervention aimed at preventing falls. The primary outcome of a CEA is the ICER. By definition, an ICER is the difference in mean costs required by the intervention compared with an alternative (e.g. usual care or a ‗do nothing‘ alternative) divided by the difference in mean health benefit gained from the intervention compared with an alternative. For example, often the ICER is the additional cost of a more expensive but more effective intervention above that of the less expensive but less effective intervention divided by the difference in effectiveness (Equation 1). 44 For this thesis, I estimated the incremental cost per falls prevented. ICER Cost Effect The major limitation of CEAs is that the units used to express the health benefits may limit comparability across disease states if an outcome specific to the intervention or disease treatment is used. 113 In an attempt to mitigate this problem, CEAs can be standardized by reporting health effects of interventions by life years gained. Although this approach sounds appealing, the endpoints of clinical trials are often shorter than what would be needed for the economic evaluation. However, not all interventions have an impact on mortality, so in some cases, the endpoint of the clinical trial may be appropriate. In economic evaluations, the most comprehensive time horizon is ‗lifetime‘ which encompasses at least 25 years. The limitations of the CEAs lead into the widely recommended approach that incorporates  both quantity and quality of gains or losses – the QALY is an example of one such measure.113 1.2.6.1.3 Cost-utility analysis  The prominent difference between CEA and cost-utility analysis (CUA) is that in CEA, the incremental cost of a program is compared with the incremental effects of a program where the health effects are measured in naturalistic units (i.e., clinically relevant units such as falls). In CUA, the health effects are measured in QALYs gained, a measure that includes quality adjusted life years gained. Of note, there are competing alternatives to the QALY that include disability-adjusted life years gained (DALYs) and healthy year Equation 1  37 equivalents (HYEs). Briefly, DALYs were developed by the World Health Organization for a study of Global Burden of Disease and Injury and they were subsequently recommended for their generalized use in CEAs. The four key differences between DALYs and QALYs are: 1) DALYs use age weights giving lower weight to the young and elderly, 2) DALYs use a constant life expectancy set at the greates t reported national life expectancy, 3) DALY weights do not reflect preferences for health states, but represent person trade -off scores from a panel of health care workers, rather than the general population and 4) DALY weights can take only one of seven discrete values in addition to dead and healthy.44 HYEs essentially have two main differences to the conventional QALY approach.153-155 First, they measure preferences for health states over an entire path that individuals would pass rather than each state alone.153-155 Second, preferences are measured using a two-stage standard gamble procedure that first measures conventional utility for the path and then assess the number of health years that would give this same utility.153-155  In summary, CEA and CUA are identical on the cost side and only differ on the outcomes side. CUA is often considered advantageous because of its broader applicability and use to decision makers. CUA unlike CEA does not use disease specific outcomes (e.g., falls prevented), it requires generic effectiveness data as defined above (i.e., HSUVs from EQ-5D, SF-6D or HUI3 and duration of time spent in health states). Given this is the only major difference, many do not differentiate between CEA and CUA. CUA should be used when: 1) HRQL is an important outcome, 2) comparator programs have a wide range of different outcomes, 3) comparing with other programs that have been evaluated using CUA. This is usually the case when you expect a program or intervention to affect both morbidity and mortality.  The results of a cost-effectiveness or cost-utili ty analysis may be plotted along what is termed the cost- effectiveness plane. When contrasting an intervention with a comparator, the horizontal axis represents the incremental effect (i.e., difference in effect between one or more comparators) and the vertical axis represents the incremental cost (i.e., different in mean cost between one or more comparators). The slope  38 of the line between Point A and C (Comparator – because it is incremental, the comparator has been placed at the origin or at coordinates 0,0) in Figure 1-6 represents the incremental cost effectiveness ratio of the intervention with the comparator. When point A lies in the northwest or southeast quadrants, the decision is clear. When in the northwest quadrant, the new intervention is less effective and more costly than the comparator; therefore, decision makers would not fund the new intervention for obvious reasons. In the southeast quadrant, the opposite is true and the new intervention dominates the comparator (i.e., more effective and less costly). The northeast and southwest quadrants depend on the max imum cost- effectiveness ratio society is willing to pay – i.e. the willingness to pay for the outcome.   Figure 1-6. Cost-effectiveness plane44     C New treatment more costly New treatment  more effective New treatment less effective New treatment less costly NENW SW SE Existing treatment dominates New treatment dominates New treatment less costly but less effective New treatment more effective but more costly Maximum acceptable ICER A  39 1.2.6.1.4 Cost-benefit analysis  I have not conducted cost-benefit analysis (CBA) in my thesis work; therefore, I provide only a brief explanation of what it is and why it is difficult to conduct cost-benefit analysis in falls research. The distinguishing feature of CBA is that it requires program consequences to be valued in monetary units.44 The overall goal of CBA is to determine if the incremental program benefits exceed the incremental program costs. Because CBA values benefits and costs in monetary units, it is not limited to only making decisions within health care, but it can span within and across sectors.44 CBA is particularly difficult in the area of falls prevention, because it is difficult to attach a cost to a fall event that accounts for all quality and quantity conditions imposed by a fall. These are difficult to accurately define; therefore, it is difficult to conduct an accurate cost-benefit analysis that appropriately quantifies benefits in monetary units. However, it is not impossible, and future research could focus on estimated the willingness to pay to avoid a fall. Willingness to pay values are defined as the maximum the decision maker is willing to pay for an additional gain of one QALY. 1.2.7 Interventions reduce falls, but at what cost?  In section 1.2.7, I review multifactorial, multiple and single factor interventions that reduce falls and the economic data available for these categories of falls prevention strategies. 1.2.7.1 Descriptive summary of major falls interventions  The recently updated Cochrane Review titled ―Interventions for preventing falls in older people living in the community‖3 included 111 randomized controlled trials aimed at preventing falls among community dwelling older adults.  Falls prevention strategies can be classified into three categories of interventions: 1) individually customized multifactorial interventions,156 2) regimens targeting the same multip le factors to all participants6 and 3) single factor interventions. Tinetti and colleagues4 conducted an individually  40 customized multifactorial intervention targeting up to eight different risk factors for falls depending on the participant‘s individual risk profile; it reduced falls by 31% in one year. In multifactorial interventions, reduction in the risk of falling ranges from not being significantly different from usual care 157 to a 61% reduction in the risk of falling.5 I distinguish individually customized multifactorial intervention from the strategy of providing a multiple intervention at a community level. In a multiple intervention, all individuals are exposed to the same intervention irrespective of the person‘s risk profile (i.e., everyone received the same intervention).6 Lastly, single factor interventions consist of only one type of individually customized intervention that is delivered to all participants. Some examples of single factor interventions include: strength and balance training, withdrawing use of psychotropic medication, home safety modification, and cataract surgery.7,8 Overall, in four trials of community living seniors, a progressive home based strength and balance retraining program reduced both falls and injuries by 35%. 158 Briefly, I do not highlight strategies that were not effective in preventing falls because economic evaluations should be conducted upon established evidence of effectiveness.37,44,159 1.2.7.1.1 Effective multifactorial interventions  Of the 111 RCTs included in the Cochrane Review,3 31 studies were multifactorial interventions as defined in section 2.7.1. I will limit this chapter to a discussion of Tinetti‘s4 multifactorial intervention conducted in 1994 as one example of a successful multifactorial intervention. 4 This study enrolled 301 community dwelling men and women aged ≥70 years who had a least one of the following falls risk -factors: 1) postural hypotension, 2) use of at least four prescription medications, 3) use of sedatives, and 4) impairments in 5) arm or leg strength or range of motion, 6) gait, 7) balance or ability to move safely from bed to chair or tub, and 8) environmental hazards for falls or tripping. The intervention group received a baseline assessment that was conducted by the study nurse practitioner and physical therapist who were blind to group allocation. The physical therapist visited the participants within one week after the nurse practitioner had assessed them. Two hundred and fifty of the participants were followed up at a median of 4.5 months after  41 the baseline visit although the actual intervention was intended to last 3 months due to issues with followup scheduling. Participants in the intervention group received interventions based on their baseline assessment. For each falls risk factor identified, decision rules and priority lists were created to ensure standardized intervention protocols. During the one year followup, 35% of the intervention group fell compared with 47 % of the control group. This corresponded to an incidence rate ratio (IRR) for falls of 0.69 (95% CI: 0.52 to 0.90).  Overall, the multifactorial intervention resulted in a statistically significant reduction in both the rate and risk of falls among community dwelling older adults.  Tinetti‘s4 study is merely one of many cited effective multifactorial interventions5,160 but was the only multifactorial intervention that included an economic evaluation. Briefly, the mean cost of delivering an individually targeted intervention was UK £880 (at 2008 prices) per participant.161 For all participants the targeted intervention was cost saving compared with usual care and social visits when the ICER was calculated using mean total health care costs or mean total cost for falls requiring medical care. The multifactorial intervention was cost saving in those who had at least four of the eight targeted risk factors for falling. 1.2.7.1.2 Effective multiple interventions  Of the 111 falls prevention RCTs reported in the Cochrane Review, 3 10 were classified as multiple interventions – a combination of two or more major interventions where all participants receive the same intervention. I will focus on a description of one such intervention, not an RCT, but a good example o f a multiple intervention and relevant to my research because it included a concurrently conducted economic evaluation. Briefly, The ―Stay on your Feet‖6,162 intervention targeted older adults with an intervention addressing falls related attitudes (e.g., awareness), knowledge, behaviours (e.g., risk taking), and with a risk factor awareness campaign. The economic evaluation of the ‗Stay on Your Fee t‘ program was a CBA that estimated that this program required a large budget in the order of millions; the net monetary benefit to  42 cost ratio for the intervention was 20.6 to 1.162 This ratio means that the benefits outweighed the costs incurred by 20.6 times. No attempt in this study was made to value money in terms of health consequences of the ―Stay on your Feet‖ intervention. Benefits in this case were only defined as cost savings. Please refer to the systematic review in chapter 3 for a detailed description of the economic evaluation of the ―Stay on Your Feet‖ intervention.6  In the North Coast of New South Wales (Australia), 80 000 older adults aged 60 years and older were targeted to received the ‗Stay on Your Feet‘ multiple intervention. 6 This was a prospective cohort study, taking place from 1991 to 1995, that used a geographic comparator where the multiple intervention was not taking place. Participants in each region were randomly enrolled via telephone to form the intervention and control region cohorts. This multiple intervention addressed vision, footwear, balance and gait, physical activity, chronic conditions, medication use, and home and public environmental hazards. Multiple strategies were implemented that included: community education, falls prevention policy developments, increased falls awareness, home hazard reduction, media campaigns and communication with health care professionals. In total, the intervention cost AUD $600 000. I emphasize that in a multiple intervention, each individual receives the same interventions regardless of their individual fall risk profile. Results indicated that there was a 22% reduction in the incidence of self-reported falls in one year in the intervention compared with the control cohort; however, this was not statistically significant (p=0.17). However, from a CEA perspective, this lack of statistical significance does not negate the 20.6 to 1 ratio of net benefit. 163 Further, participants in the intervention cohort had a 20% lower fall -related hospitalization rate in target group residents. Although this multiple intervention was costly to implement, it did result in 20.6 dollars benefit for every 1 dollar spent and it targeted a greater number of individuals that an RCT. 6    43 1.2.7.1.3 Effective single factor interventions  In the Cochrane review,3 single factor interventions were further categorized by type of intervention: exercises, medication, surgery, fluid or nutrition therapy, psychological, environment/assistive technology and knowledge/education interventions. Reviewing all o f these studies is beyond the scope of this thesis and I refer you to the Cochrane Review titled ―Interventions for preventing falls in older people living in the community.‖3 For this section, I highlight one of the most widely cited and tested exercise programs – the Otago Exercise Programme (OEP) – because four of the nine falls prevention studies that include an economic evaluation utilize the OEP. Importantly, single factor interventions, similar to multiple or multifactorial interventions, can also result in a broad range of health benefits.  The Cochrane systematic review demonstrated the effectiveness of the OEP in over 1000 participants.9,160,164-166 The OEP is a progressive home based strength and balance retraining program initiated by a health care professional such as a nurse or physiotherapist. Together, Campbell, Robertson and colleagues from the University of Otago have published five controlled trials of the OEP from 1997 through 2005.9,160,164-166 Of these, three studies demonstrated a statistically significant reduction in the risk of future falls in the intervention compared with the control with IRRs ranging from 0.47 to 0.70. 9,164,166 The beneficial effect of the OEP was maintained over a further one  year followup.9,166,167 Participants enrolled in the original 12-month study were invited to continue with their exercises if they were in the intervention group and were invited to continue receiving usual care if they were enrolled in the control group. With minimal input from the trained physiotherapist, the OEP remained effective during this one year followup.9,166 Most recently, a meta-analysis of the OEP indicated that the OEP was most effective among the following subgroups: 1) older adults aged 80 years and older who experienced at least one fall in the prior year (54.0 falls prevented per 100 person years) or 2) individuals ≥65 years who experienced at least one fall in the prior year (44.3 falls prevented per 100 person years). 158   44 Although three OEP trials9,166,167 have demonstrated effectiveness of the intervention, two RCTs160,165 using 2x2 factorial designs did not. These two trials were conducted in different populations than the RCTs that showed effectiveness and I will discuss each study in detail.  Campbell and colleagues used a 2x2 factorial design of the OEP delivered by a physiotherapist and a home safety intervention delivered by an occupational therapist.165 The four groups in the factorial design were 1) usual care, 2) home safety 3) OEP and 4) OEP and home safety. Three hundred and ninety -one adults aged 75 years and older who had visual impairment (defined as visual acuity of 6/24 or worse)  were followed for one year. The home safety intervention demonstrated effectiveness by a 41% reduction in falls (IRR 0.59, 95%CI: 0.42-0.83) while the OEP did not prove effective (IRR 1.15, 95%CI: 0.82 -1.61). Campbell and colleagues suggest the reason for the OEP not reducing falls may have been an issue of compliance, only 18% of participants were 100% compliant compared with previous trials that demonstrated 45% of participants were 100% compliant. Compliance was evaluated by a study nurse blind to group allocation at six months after entry into the study by telephone interview. Another major contributing factor is that this population of individuals with visual impairment differed from previously tested populations. Individuals with visual impairments may have been frailer and thus at higher risk for falls.  The other OEP study that did not demonstrate effectiveness of the OEP in reducing falls also used a 2x2 factorial design to determine the effect of the OEP program and withdrawal of psychotropic medic ation.160 The four groups consisted of: 1) usual care, 2) OEP, 3) psychotropic medication withdrawal and 4) OEP and psychotropic medication withdrawal. This study demonstrated that the OEP had no synergistic or unique effect of reducing falls. Noncompliance with the OEP was not an issue in this study as 68% of participants were 100% compliant. Three reasons have been suggested for the lack of effectiveness: 1) insufficient sample size (n=93), 2) insufficient followup time (44 weeks) and 3) a younger age group (aged 65 years and older), because the OEP was subsequently shown to be more effective in ≥80 years. 168  45 Briefly, I highlight the economic evaluations of the OEP below. Further, for a comprehensive review of economic evaluations of all falls preventions strategies, please refer to my systematic rev iew in Chapter 3. Three cost-effectiveness analyses of the OEP compared the program with usual care for a one year period.9,166,167 These include two RCTs and one controlled trial. The incremental cost of the OEP ranged from NZ $314 (1995 prices) in a research setting to NZ $1803 (1998 prices) per fall prevented in a community health care setting. 1.2.7.2 Descriptive summary of major falls interventions that included economic analysis  To my knowledge, nine9,150,161,162,165-167,169,170 economic evaluations of falls prevention strategies have been published. Of these, five9,150,161,166,169 economic evaluations have been published from falls prevention randomized controlled trials among community dwelling older adults aged 60 years and older. These nine studies can be divided into three types of interventions: 1) multifactorial, 2) multiple and 3) single factor interventions. From these three types of interventions, effective single factor interventions are further divided into the following intervention categories: 1) exercise training, 2) home safety and 3) expedited cataract surgery. Please refer to Chapter 3 for a full description of these studies in the systematic review titled “Does a home based strength and balance program in people aged 80 years provide the best value for money to prevent falls? A systematic review of economic analyses of falls prevention interventions .”  In summary, seven of the nine studies reported an incremental cost per fall prevented for strength and balance retraining, cataract surgery and home safety interventions – seven of the nine studies were single factor interventions. The studies of strength and balance retraining were the most methodologically sound and prevented the greatest number of falls at the least cost. Further, a multifactorial program was cost saving in a narrow range of individuals with four or more of the eight specified risk factors for falls. The most favorable and widely applicable ICER was US $295167 (at 2006/2007 prices) per fall prevented for the OEP, although this trial was in a research rather than a community based setting. 167 The costs of program  46 implementation were higher in a community based setting. For example, a number of factors make ICERs determined from CEAs alongside RCTs different from real world ICERs. Some of these  include: 1) the comparator used in the trial may not represent a real world comparator, 2) a non representative participant sample that does not reflect the target population sample and 3) insufficient collection of data on health services.  The ICERs for home safety interventions varied from US $481 per fall prevented for delivering the pro gram to people with severe visual impairment,158 to US $5030 (incorporating all health care costs) for those recently discharged from hospital and who had reported a fall in the previous year. 163 Of note, what is often referred to as an ‗ICER‘ for program delivery is, technically speaking, not a true ICER because it does not take into account the cost of health resource utilization. Therefore it is not surprising that CEAs and CUAs that include HRU costs are larger than those that include only the cost prog ram delivery. Cataract surgery was the least favorable with an incremental cost (incorporating all health care, personal and care provider costs) per fall averted of US $7757.150 Of note, although the estimated ICERs for the expedited cataract surgery per QALY saved for one year were above the currently accepted willingness to pay values of CAD $30 000 (US $30 000), the estimated cost-utility ratios for the expedited cataract surgery per QALY saved over a participants expected lifetime were within this limit (US $23 273). 1.2.7.3 Summary of gaps: why the limited QALY data for fallers?  Policy makers are faced with difficult resource allocation choices among different medical conditions (i.e., there is competition for resources for different health technologies and procedures). Cost-benefit or cost-utili ty analyses can provide vital information to direct policy makers on which diseases and chronic conditions to allocate resources. To do this, there is a need to establish the presence of QALY data in falls prevention research. Presently, economists have used utility instruments for other chronic diseases such as rheumatoid arthritis,138 cardiovascular disease171 and human immunodeficiency virus, but there is little published for falls.  47  To highlight the gap in the research cycle (Figure 1-1) described by Tugwell,37 regarding the use of generic preference-based utility instruments to calculate QALYs in falls, I will use rheumatoid arthritis as an example. Marra and colleagues138 conducted a comparison of generic, preference-based utility instruments (HUI2, HUI3, SF-6D and EQ-5D) as well as disease specific measures in rheumatoid arthritis. They assessed the construct validity for indirect measures of health utility.138 Construct validity is the extent to which the score of an instrument such as the EQ-5D correlates with other hypothesized measures such as the SF-6D as well as other clinical measures of pain.172 This was a useful and novel contribution to the field of rheumatoid arthritis research because we now have estimates of minimal important differences, correlations between the global utility scores and correlations for overall scores with the main indicators of rheumatoid arthritis severity. 138 I use the above example to detail what has been done in other fields and what can be done for older adults. Ideally, prior to using these instruments among older adults, we need to assess whether these instruments pick up the health consequences of a fall.  Some limitations for using QALYs as a measure of benefit in older people and in conditions such as falls have been noted173 – preference-based utility instruments such as the EQ-5D and SF-6D may lack sensitivity in older adults who fall. The inherent problem with using QALYs as an outcome for a complex intervention may be due to the potential of such an intervention to produce multiple benefits for older people. This makes it difficult to determine the component of the intervention that proved most useful. Robertson and colleagues158 have also not found quality of life measures sensitive to change (i.e., trial participants‘ HSUVs were too high to see a change) in published falls prevention  studies despite the beneficial outcomes of the trials. For example, the instruments used to estimate HSUVs that are used to calculated QALYs may have a ceiling effect among some populations of participants that are too well functioning. Further, data indicate that participants who agree to partake in randomized controlled trials are higher functioning than those who do not.174  48  Of the nine economic evaluations of falls preventions strategies, only two such studies were cost-benefit162 or cost-utili ty150 studies. As mentioned previously, Beard and colleagues162 conducted a cost-benefit study estimating both costs and benefits in dollars. The authors performed a cost-benefit analysis using people in two different geographic areas as the controls. Although the program required a large budget, both comparisons yielded a net monetary benefit to cost ratio for the intervention of 20.6 to 1. Sach and colleagues150 conducted a cost-utility analysis that provided estimates for QALYs using the EQ-5D. For this study evaluating expedited cataract surgery, incremental cost per QALY gained over the participants‘ lifetime after surgery was estimated at US $23 273; this is within range of the maximum willingness to pay threshold of the cost per QALY. To provide a basis for comparison with other chronic conditions, the ICER reported for aspirin in primary prevention of age-related cardiovascular disease was US $14 355.171 Thus, providing surgery is only an addition US $9000 per QALY. Further falls prevention studies that report QALYs are needed for comparison across health care interventions and health services. 1.2.7.4 Why are QALYs important?  Despite the acknowledged limitations for using QALYs as a measure of benefit in older adults at high risk of falls and fracture, comparisons between generic utility instruments have not been evaluated at all in the falls literature. The EQ-5D is currently the most widely used instrument in the fracture literature, but a systematic review reported that this measure was only used to calculate QALYs in 11 studies that used generic preference-based, utility instruments.144 The authors did not give specific details on the instruments that each of these studies used, but they did highlight that the EQ-5D was one of the most commonly used instruments.144 Given that the SF-6D measures 18 000 health states compared with only 243 health states measured by the EQ-5D,138 it is essential that we evaluate both of the instruments among the specific population of older adults at high risk of falls to assess the impact of instrument choice on decision making in health care. The EQ-5D is advantageous because it covers a wider range (-0.49 to 1.0) whereas the SF-  49 6D has a potential floor effect and has a smaller range in HSUVs (scores range from 0.30 to 1.0). Therefore, the SF-6D has the benefit of covering a greater number of health states, but the EQ-5D covers fewer health s tates over a broader range of values. The distribution of the SF-6D follows a normal distribution, whereas the distribution of the EQ-5D is often skewed and may follow a bimodal distribution. Therefore, often SF-6D data may be analysed using parametric analysis while EQ-5D data may be analysed using non-parametric techniques. 1.2.7.5 Gaps in current knowledge that motivated my thesis studies  While undertaking my literature review examining studies of falls and economics, I noticed the following gaps: 1) Although the incidence of falls is well described both in Canada and internationally, the number of high quality studies that quantify the cost of falls is few. Therefore, this led to my research question for my first systematic review in Chapter 2: ―What is the economic burden of falls in different countries and why do these costs differ?‖ 2) Despite the recently published Cochrane Review of all falls prevention randomized controlled trials, only nine peer reviewed economic evaluations of effective falls prevention strategies have been published. This instigated my second research question: ―Which falls prevention strategy provides the best value for money?‖ 3) Given resistance training is a necessary component of effective exercise interventions, what is the value for money of the two doses of resistance training in community dwelling older wome n in the Brain Power study? 4) From my economic evaluation of the Brain Power study, I also sought to examine the following: 1) What are the one year benefits post intervention of participating in an exercise intervention? 2) What are the differences in the incremental cost per QALY using the SF-6D compared with the EQ- 5D? and 3) What factors are independent predictors of HRQL?  50 I address each of these proposed research questions in Section 1.2.8. 1.2.8 Proposed thesis manuscripts  I include the rationale, objectives and potential contribution of the six manuscript-based studies that comprise this thesis. 1.2.8.1 Study 1: International comparison of cost of falls in older adults living in the community: a systematic review  1.2.8.1.1 Rationale  In Canada, the direct health care costs of falls was CAD $2.4 billion in 1998, which represented 55% of the total cost to treat unintentional injuries in Canada.175 The direct cost of falls in those 60 years and older was approximately CAD $980 million and the estimated direct and indirect costs for treating falls in Canada were CAD $3.6 billion. In British Columbia (BC), estimates indicate falls are responsible for 85% of the CAD $211 million annual direct cost of unintentional injuries;176 falls cost Canada well over 1 billion dollars annually (combining direct and indirect costs).177 These conservative estimates ignore the costs of other fall-related injuries (including other fractures and head injuries) as well as the loss of independence associated with them. In BC in 2001, the estimated direct cost for treating falls in persons aged 65 years and older was CAD $180 million, that represented approximately 9% of the direct and indirect costs for treating all injuries.178  To my knowledge, there was no previous effort to quantify the existing peer reviewed literature on economic outcomes that include cost of fall and/or fall related injuries in a format useful for future research studies and policy makers. Therefore, to help inform health care managers and professionals, I conducted a systematic review aimed at quantifying the international burden of falls.   51 1.2.8.1.2 Objective  To help policy makers, health care professionals and government recognize the global financial burden of falls, I systematically reviewed all peer-reviewed literature to answer the question: ―What is the economic burden of falls in different countries and why do these costs differ?‖ Specifically, my aim was to quantify the economic burden of falls based on a comprehensive evaluation of the current li terature and to ascertain methodological reasons for different estimates for the cost of falls. 1.2.8.1.3 Potential contribution  This is the first systematic review to describe the cost of falls for community living older people in the falls literature. This s tudy highlighted the glaring need for descriptive data to characterize the economic burden of falls and fall-related injuries. It also highlights the need for consistency in the reporting and defining of falls with the aim of comparing the cost of falls at an international level. One of the major contributions of this systematic review is the identification of the key factors that influence the range of costs reported for cost of falls studies. These include: 1) variations in the definition of falls and fall -related injuries used, 2) clinical outcomes reported (for example hip fracture), 3) the cost items collected and the units reported, 4) the population denominator that cost estimates were based on, 5) variation in time horizon and 6) the perspective of the analysis. Lastly, this systematic review will help quantify the economic burden of falls across the globe.         52 1.2.8.2 Study 2: Does a home based strength and balance program in people aged 80 years provide the best value for money to prevent falls? A systematic review of econ omic analyses of falls prevention interventions  1.2.8.2.1 Rationale  Falls and injuries resulting from falls in older people represent a significant health burden. 17,18 Given that different strategies to prevent falls exist and are categorized into 1) individually customized multifactorial interventions,156 2) regimens targeting the same multiple factors to all participants and 3) single factor interventions, there is a need to evaluate and compare the costs and consequences of these interventions. To our knowledge, no systematic review of economic evaluations has been published for falls prevention interventions. To inform health care professionals on how to allocate resources that are scarce, I systematically reviewed economic evaluations of falls prevention studies to answer the question ―Which falls prevention strategy provides the best value for money?‖ 1.2.8.2.2 Objective  To help health professionals and health managers to make better decisions, I systematically reviewed economic evaluations of falls prevention studies to answer the question ―Which falls prevention strategy provides the best value for money?‖ 1.2.8.2.3 Potential contribution  This is the first systematic review of studies in community dwelling older adults that includes a thorough review of economic evaluations of falls prevention strategies. Results from this systematic review will empower decision makers on which falls prevention intervention strategies may provide the best value for money. Further, it provides detailed information on which subgroups to target with effective falls prevention strategies that will generate savings in health care costs.  53 1.2.8.3 Study 3: Economic evaluation of dose-response resistance training in older women: a cost-effectiveness and cost-utility analysis  For studies 3, 4 and 5, I highlight that my role was to undertake an economic evaluation of a dose response exercise intervention among older adults aged 65-75 years. Dr Teresa Liu-Ambrose is the Principal Investigator of the Brain Power study. For further description and details of the clinical trial, please refer to the referenced publication.15 1.2.8.3.1 Rationale  Falls and injuries resulting from falls in older adults represent a costly and significant public health burden.16-18 Progressive resistance training is an essential component of effective exercise programs for falls prevention.3,179 Liu-Ambrose and colleagues previously demonstrated that progressive, high intensity resistance training significantly reduced falls risk score by 57%.180 What is not known is whether there is a threshold dose of resistance training aimed at combating sarcopenia and reducing falls risk in women aged 65-75 years that will reduce total health care resource utilization and thus provide the best value for money. Therefore, I included in the study design a concurrent, prospective economic analysis with individual level data on cost and effectiveness outcomes collected during the Brain Power trial. 1.2.8.3.2 Objective  My primary objective was to determine the ICER (cost per fall prevented) of once weekly or twice weekly resistance training compared with twice weekly balance and tone classes (comparator). I modeled the comparator program on the trial control group activity, a popular provincial -wide exercise program designed to reduce falls risk among seniors with low bone mass (the Osteofit program). 181     54 1.2.8.3.3 Hypothesis  The once weekly and twice weekly resistance training interventions will result in an acceptable ICER as measured by QALYs and an acceptable incremental cost per fall averted, making it economically attractive compared with twice weekly balance and tone (comparator). 1.2.8.3.4 Potential contribution  The Brain Power study is a novel intervention with the first economic evaluation to examine the value for money of two doses of resistance training compared with standard of care. My results suggest that both once weekly resistance training and twice weekly resistance training are economically efficient programs and provide better value for money than current standard of care practice. Technically, I have an ideal opportunity to test the cost-effectiveness of an intervention that may have a low ‗number needed to treat‘ (NNT) and this is unusual in a prevention setting. The proposed study will differentiate the economics o f falls from the economics of osteoporosis and hip fracture that has been addressed by Eastern Canadian research groups.18 1.2.8.4 Study 4: The long term effects of a 12-month of resistance training intervention in older community dwelling women: a cost-effectiveness and cost-utility analysis  1.2.8.4.1 Rationale  Impaired executive functions in older adults are associated with falls, an increased risk of fall-related injury and a poorer score on the physiological profile assessment to ascertain falls risk. Given that executive functions can predict functional status and risk of falling,182 interventions that maintain or improve executive functions may have a subsequent impact on HRU, health status and HRQL among older adults. For example, executive functions are cognitive skills used to modify falls risk through: medication management, dietary and lifestyle improvements, self-monitoring and followup. Given that falls and their associated injuries are a major health care problem that is highlighted as a top research priority and cognitive  55 impairment is a major risk factor for falls, it is timely that a RCT investigating the dose-response of resistance training be conducted with a built in economic evaluation. Specifically, it is essential to detail the long term effects of a dose-response resistance training intervention to ascertain any health improvements , maintenance or decline in the year following participation in the 12-month intervention. 1.2.8.4.2 Objective  To ascertain the ICER and the incremental cost-utili ty ratio after the one year followup post 12-month intervention that included delivering once weekly and twice weekly resistance training compared with twice weekly balance and tone classes (comparator). 1.2.8.4.3 Potential contribution  I have monthly HRQL data that were measured by the EQ5D (self-report) as well as a monthly record of falls. Few studies have conducted economic evaluations beyond a one year followup and to our knowledge, no other falls prevention RCT has collected monthly HRQL measures beyond the intervention duration. 1.2.8.5 Study 5: A prospective comparison of generic preference-based utility instruments (SF-6D and EQ-5D) and predictors of health care resource utilization in older women  1.2.8.5.1 Rationale  Although preference-based utility instruments have been compared and validated among patients with rheumatoid arthritis, no such comparisons exist in the falls  literature. Therefore, to fill this gap, I will evaluate HRQL via the SF-6D and the EQ-5D at baseline and then at monthly intervals for 12 months. Each of these preference-based utility measures will provide weightings for QALYs. The SF-6D captures physical functioning, role limitations, social functions mental health, bodily pain and vitality and describes 18,000 discrete health states and should capture small changes in health s tatus. The EQ-5D captures the fewest (i.e., 243) health states but the range of possible values cover a wider range than the SF-6D.  I will  56 use the EQ-5D to measure an individual‘s HRQL and health states according to the following attributes: mobility, self-care, usual activities, pain and anxiety/depression. I will calculate QALYs using the weightings from each of the two instruments and then determine if there is a statistically significant difference in the two incremental cost per QALY change ratios as calculated using each measure. 1.2.8.5.2 Objective  My aims were to: 1) quantify the difference in the ICER from the QALYs generated from the EQ-5D compared with the QALYs generated from the SF-6D, 2) determine key predictors of statistically significant changes in HRQL and HRU and 3) identify key domain specific changes and correlations between the EQ- 5D and the SF-6D. 1.2.8.5.3 Hypothesis  I hypothesize that: 1) The changes in the EQ-5D over time will result in a larger change in the ICER compared with changes in the SF-6D over time; 2) A key predictor of HRQL will be HRU; and 3) The EQ-5D and SF-6D will demonstrate a correlation of at least 30%. 1.2.8.5.4 Potential contribution  These comparisons will inform falls researchers as to whether generic preference -based utility instruments are sensitive among older adults at high risk of falls. These comparisons have not been investigated in the falls literature and are therefore novel and will provide a basis for comparison of these generic preference - based utility instruments among vulnerable seniors. These analyses will also inform future comparisons of economic evaluations of falls prevention strategies.      57 1.2.8.6 Study 6: The independent contribution of executive functions to health related quality of life in older women  1.2.8.6.1 Rationale  Health related quality of life is an important construct that describes an individual‘s  overall health status. It is commonly used in economic evaluations183 as a measure of health benefit, and may be more responsive among populations with conditions associated with high morbidity.150 Cognition is a multidimensional construct and to our knowledge, no previous studies have examined the independent contribution of specific domains of cognition to HRQL. I hypothesize that executive functions may be of particular importance to HRQL. Executive functions are higher-order cognitive processes that control planning, initiation, sequencing and monitoring of complex goal directed behavior. 184,185 These cognitive processes are essential to the person‘s ability to carry out health-promoting behaviours,186 such as medication management, dietary and lifestyle changes, self-monitoring of responses, and follow-up with health care professionals. 1.2.8.6.2 Objective  To determine whether executive functions are independently associated with HRQL assessed using QALYs calculated from the EQ-5D in older women after adjusting for measures of global cognition and known covariates. 1.2.8.6.3 Potential contribution  My research highlights that the specific executive processes of set shifting and working memory were independently associated with QALYs, a measure of HRQL. Given executive functions explain variability in QALYs, clinicians may need to consider assessing executive functions when measuring HRQL. Further, the EQ-5D may be used to track changes in health status over time and thus serve as a screening tool for clinicians in the future.  58 1.3 References  1.  Campbell AJ, Borrie MJ, Spears GF. Risk factors for falls in a community -based prospective study of people 70 years and older. J Gerontol 1989;44(4):M112-7. 2.  Tinetti ME, Speechley M, Ginter SF. Risk factors for falls among elderly persons living in the community. N Engl J Med 1988;319(26):1701-7. 3.  Gillespie LD, Robertson MC, Gillespie WJ, Lamb SE, Gates S, Cumming RG, et al. Interventions for preventing falls in older people living in the community. Cochrane Database Syst Rev 2009(2):CD007146. 4.  Tinetti ME, Baker DI, McAvay G, Claus EB, Garrett P, Gottschalk M, et al. A multifactorial intervention to reduce the risk of falling among elderly people living in the community. N Engl J Med 1994;331(13):821-7. 5.  Close J, Ellis M, Hooper R, Glucksman E, Jackson S, Swift C. Prevention of falls in the elderly trial (PROFET): a randomised controlled trial. Lancet 1999;353(9147):93-7. 6.  Kempton A, Van Beruden E, Sladden T, Garner E, Beard J. Older peop le can stay on their feet: final results of a community-based falls prevention programme. Health Promotion International 2000;15(1):27-33. 7.  Cumming RG. Intervention strategies and risk-factor modification for falls prevention. A review of recent intervention studies. Clin Geriatr Med 2002;18(2):175-89. 8.  Harwood RH, Foss AJ, Osborn F, Gregson RM, Zaman A, Masud T. Falls and health status in elderly women following first eye cataract surgery: a randomised controlled trial. Br J Ophthalmol 2005;89(1):53-9. 9.  Robertson MC, Devlin N, Gardner MM, Campbell AJ. Effectiveness and economic evaluation of a nurse delivered home exercise programme to prevent falls. 1: Randomised controlled trial. BMJ 2001;322(7288):697-701.  59 10.  Wong RY, Miller WC. Adverse outcomes following hospitalization in acutely ill older patients. BMC Geriatr 2008;8:10. 11.  Haines TP, Russell T, Brauer SG, Erwin S, Lane P, Urry S, et al. Effectiveness of a video -based exercise programme to reduce falls and improve health-related quality of life among older adults discharged from hospital: a pilot randomized controlled trial. Clin Rehabil 2009;23(11):973-85. 12.  Scuffham P, Chaplin S, Legood R. Incidence and costs of unintentional falls in older people in the United Kingdom. J Epidemiol Community Health 2003;57(9):740-4. 13.  National Center for Injury Prevention and Control: statistics and activities. Int J Trauma Nurs 1998;4(1):18-22. 14.  Sattin RW, Lambert Huber DA, DeVito CA, Rodriguez JG, Ros A, Bacchelli S, et al. The incidence of fall injury events among the elderly in a defined population. Am J Epidemiol 1990;131(6):1028- 37. 15.  Liu-Ambrose T, Nagamatsu LH, Graf P, Beattie L, Ashe MC, Handy T. The effect of resistance training on brain function among community-dwelling senior women: a randomized controlled trial. Archives of Internal Medicine 2010;170(2):170-8. 16.  Burge R, Dawson-Hughes B, Solomon DH, Wong JB, King A, Tosteson A. Incidence and economic burden of osteoporosis-related fractures in the United States, 2005-2025. J Bone Miner Res 2007;22(3):465-75. 17.  Tinetti ME, Williams CS. Falls, injuries due to falls, and the risk of admission to a nursing home. N Engl J Med 1997;337(18):1279-84. 18.  Wiktorowicz ME, Goeree R, Papaioannou A, Adachi JD, Papadimitropoulos E. Economic implications of hip fracture: health service use, institutional care and cost in Canada. Osteoporos Int 2001;12(4):271-8.  60 19.  Campbell AJ, Borrie MJ, Spears GF, Jackson SL, Brown JS, Fitzgerald JL. Circumstances and consequences of falls experienced by a community population 70 years and over during a prospective study. Age Ageing 1990;19(2):136-41. 20.  Hornbrook MC, Stevens VJ, Wingfield DJ, Hollis JF, Greenlick MR, Ory MG. Preventing falls among community-dwelling older persons: results from a randomized trial. Gerontologist 1994;34(1):16-23. 21.  Chang JT, Morton SC, Rubenstein LZ, Mojica WA, Maglione M, Suttorp MJ, et al. Interventions for the prevention of falls in older adults: systematic review and meta-analysis of randomised clinical trials. BMJ 2004;328(7441):680. 22.  Rubenstein LZ. Falls in older people: epidemiology, risk factors and strategies for prevention. Age Ageing 2006;35 Suppl 2:ii37-ii41. 23.  Shumway-Cook A, Ciol MA, Hoffman J, Dudgeon BJ, Yorkston K, Chan L. Falls in the Medicare population: incidence, associated factors, and impact on health care. Phys Ther 2009;89(4):324- 32. 24.  Sylliaas H, Idland G, Sandvik L, Forsen L, Bergland A. Does mortality of the aged increase with the number of falls? Results from a nine-year follow-up study. Eur J Epidemiol 2009;24(7):351-5. 25.  Lord SR, Ward JA, Williams P, Anstey KJ. Physiological factors associated with falls in older community-dwelling women. J Am Geriatr Soc 1994;42(10):1110-7. 26.  Ganz DA, Higashi T, Rubenstein LZ. Monitoring falls in cohort studies of community-dwelling older people: effect of the recall interval. J Am Geriatr Soc 2005;53(12):2190-4. 27.  Nevitt MC, Cummings SR, Kidd S, Black D. Risk factors for recurrent nonsyncopal falls. A prospective study. JAMA 1989;261(18):2663-8. 28.  Lehtola S, Koistinen P, Luukinen H. Falls and injurious falls late in home -dwelling life. Arch Gerontol Geriatr 2006;42(2):217-24.  61 29.  Guideline for the prevention of falls in older persons. American Geriatrics Society, British Geriatrics Society, and American Academy of Orthopaedic Surgeons Panel on Falls Prevention. J Am Geriatr Soc 2001;49(5):664-72. 30.  McMurdo ME. 'Guideline for the prevention of falls in older persons': essential reading. Age Ageing 2002;31(1):13-4. 31.  O'Loughlin JL, Robitaille Y, Boivin JF, Suissa S. Incidence of and risk factors for falls and injurious falls among the community-dwelling elderly. Am J Epidemiol 1993;137(3):342-54. 32. SmartRisk. Injury Burden Near $20 billion per year. www.smartrisk.ca. Vancouver, 2009. 33.  Zecevic AA, Salmoni AW, Speechley M, Vandervoort AA. Defining a fall and reasons for falling: comparisons among the views of seniors, health care providers, and the research literature. Gerontologist 2006;46(3):367-76. 34.  Woolcott JC, Davis JC, Buchanan J, Abu-Laban R, Khan KM, Marra CA. The direct costs of inurious falls in seniors. International Society for Pharmacoeconomics and Outcomes Research Thirteenth Annual International Meeting May 3-7, 2008. 35.  Rice DP. Estimating the cost of illness. Am J Public Health Nations Health 1967;57(3):424-40. 36.  Cooper BS, Rice DP. The economic cost of illness revisited. Soc Secur Bull 1976;39(2):21-36. 37.  Tugwell P, Bennett KJ, Sackett DL, Haynes RB. The measurement iterative loop: a framework for the critical appraisal of need, benefits and costs of health interventions. J Chronic Dis 1985;38(4):339-51. 38.  Byford S, Torgerson DJ, Raftery J. Economic note: cost of illness studies. BMJ 2000;320(7245):1335. 39.  Currie G, Kerfoot KD, Donaldson C, Macarthur C. Are cost of injury studies useful? Inj Prev 2000;6(3):175-6. 40.  Mitton CR. Priority setting for decision makers: using health economics in practice. Eur J Health Econ 2002;3(4):240-3.  62 41.  Mooney G, Wiseman V. Burden of disease and priority setting. Health Econ 2000;9(5):369-72. 42.  Wiseman V, Mooney G. Burden of illness estimates for priority setting: a debate revisited. Health Policy 1998;43(3):243-51. 43.  Shiell A, Gerard K, Donaldson C. Cost of illness studies: no aid to decision making? Health Policy 1987;8:317-323. 44.  Drummond MF, Sculpher MJ, Torrance GW, O'Brien B, Stoddart GL. Methods for the economic evaluation fo health care programmes. Third edition. New York. United States of America: Oxford University Press, 2005. 45.  Cohen D. Health promotion and cost-effectiveness. Health Promotion International 1994;9(4):281- 287. 46.  Heinrich S, Rapp K, Rissmann U, Becker C, Konig HH. Cost of falls in old age: a systematic review. Osteoporos Int 2009 Epub ahead of print. 47.  Corso P, Finkelstein E, Miller T, Fiebelkorn I, Zaloshnja E. Incidence and lifetime costs of injuries in the United States. Inj Prev 2006;12(4):212-8. 48.  Mitton C, Donaldson C. Tools of the trade: a comparative analysis of approaches to priority setting in healthcare. Health Serv Manage Res 2003;16(2):96-105. 49.  Newton JL, Kyle P, Liversidge P, Robinson G, Wilton K, Reeve P. The costs of falls in the community to the North East Ambulance Service. Emerg Med J 2006;23(6):479-81. 50.  Hall SE, Hendrie DV. A prospective study of the costs of falls in older adults living in the community. Aust N Z J Public Health 2003;27(3):343-51. 51.  Hendrie D, Hall SE, Arena G, Legge M. Health system costs of falls of older adults in Western Australia. Aust Health Rev 2004;28(3):363-73. 52.  Englander F, Hodson TJ, Terregrossa RA. Economic dimensions of slip and fall injuries. J Forensic Sci 1996;41(5):733-46.  63 53.  Carroll NV, Slattum PW, Cox FM. The cost of falls among the community -dwelling elderly. J Manag Care Pharm 2005;11(4):307-16. 54.  Rizzo JA, Friedkin R, Williams CS, Nabors J, Acampora D, Tinetti ME. Health care utilization and costs in a Medicare population by fall status. Med Care 1998;36(8):1174-88. 55.  Stevens JA, Corso PS, Finkelstein EA, Miller TR. The costs of fatal and non-fatal falls among older adults. Inj Prev 2006;12(5):290-5. 56.  Finkelstein EA, Chen H, Miller TR, Corso PS, Stevens JA. A comparison of the case -control and case-crossover designs for estimating medical costs of nonfatal fall -related injuries among older Americans. Med Care 2005;43(11):1087-91. 57.  Alexander BH, Rivara FP, Wolf ME. The cost and frequency of hospitalization for fall -related injuries in older adults. Am J Public Health 1992;82(7):1020-3. 58.  Roudsari BS, Ebel BE, Corso PS, Molinari NA, Koepsell TD. The acute medical care costs of fall - related injuries among the U.S. older adults. Injury 2005;36(11):1316-22. 59.  Ellis AA, Trent RB. Do the risks and consequences of hospitalized fall injuries among olde r adults in California vary by type of fall? J Gerontol A Biol Sci Med Sci 2001;56(11):M686-92. 60.  Mahoney JE, Glysch RL, Guilfoyle SM, Hale LJ, Katcher ML. Trends, risk factors, and prevention of falls in older adults in Wisconsin. WMJ 2005;104(1):22-8. 61.  Findorff MJ, Wyman JF, Nyman JA, Croghan CF. Measuring the direct healthcare costs of a fall injury event. Nurs Res 2007;56(4):283-7. 62.  Hedlund R, Lindgren U. Trauma type, age, and gender as determinants of hip fracture. J Orthop Res 1987;5(2):242-6. 63.  Gullberg B, Johnell O, Kanis JA. World-wide projections for hip fracture. Osteoporos Int 1997;7(5):407-13. 64.  Johnell O, Kanis JA. An estimate of the worldwide prevalence and disability associated with osteoporotic fractures. Osteoporos Int 2006;17(12):1726-33.  64 65.  Cauley JA, Thompson DE, Ensrud KC, Scott JC, Black D. Risk of mortality following clinical fractures. Osteoporos Int 2000;11(7):556-61. 66.  Empana JP, Dargent-Molina P, Breart G. Effect of hip fracture on mortality in elderly women: the EPIDOS prospective study. J Am Geriatr Soc 2004;52(5):685-90. 67.  Forsen L, Sogaard AJ, Meyer HE, Edna T, Kopjar B. Survival after hip fracture: short- and long- term excess mortality according to age and gender. Osteoporos Int 1999;10(1):73-8. 68.  Magaziner J, Lydick E, Hawkes W, Fox KM, Zimmerman SI, Epstein RS, et al. Excess mortality attributable to hip fracture in white women aged 70 years and older. Am J Public Health 1997;87(10):1630-6. 69.  Vestergaard P, Rejnmark L, Mosekilde L. Increased mortality in patients with a hip fracture-effect of pre-morbid conditions and post-fracture complications. Osteoporos Int 2007;18(12):1583-93. 70.  Goldacre MJ, Roberts SE, Yeates D. Mortality after admission to hospital with fractured neck of femur: database study. BMJ 2002;325(7369):868-9. 71.  Melton LJ, 3rd, Gabriel SE, Crowson CS, Tosteson AN, Johnell O, Kanis JA. Cost-equivalence of different osteoporotic fractures. Osteoporos Int 2003;14(5):383-8. 72.  Roberts SE, Goldacre MJ. Time trends and demography of mortality after fractured neck of femur in an English population, 1968-98: database study. BMJ 2003;327(7418):771-5. 73.  Boonen S, Autier P, Barette M, Vanderschueren D, Lips P, Haentjens P. Functional outcome and quality of life following hip fracture in elderly women: a prospective controlled study. Osteoporos Int 2004;15(2):87-94. 74.  Cooper C. The crippling consequences of fractures and their impact on quality of life. Am J Med 1997;103(2A):12S-17S; discussion 17S-19S. 75.  Hall SE, Williams JA, Senior JA, Goldswain PR, Criddle RA. Hip fracture outcomes: quality of life and functional status in older adults living in the community. Aust N Z J Med 2000;30(3):327-32.  65 76.  Randell AG, Nguyen TV, Bhalerao N, Silverman SL, Sambrook PN, Eisman JA. Deterio ration in quality of life following hip fracture: a prospective study. Osteoporos Int 2000;11(5):460-6. 77.  Cummings SR, Kelsey JL, Nevitt MC, O'Dowd KJ. Epidemiology of osteoporosis and osteoporotic fractures. Epidemiol Rev 1985;7:178-208. 78.  Dennison E, Mohamed MA, Cooper C. Epidemiology of osteoporosis. Rheum Dis Clin North Am 2006;32(4):617-29. 79.  Johnell O, Gullberg B, Allander E, Kanis JA. The apparent incidence of hip fracture in Europe: a study of national register sources. MEDOS Study Group. Osteoporos Int 1992;2(6):298-302. 80.  Brauer C, Coca-Perraillon M, Cutler D. Incidence and Mortality of Hip Fractures in the United States. JAMA 2009;302(14):1573-1579. 81.  Cooper C, Campion G, Melton LJ, 3rd. Hip fractures in the elderly: a world -wide projection. Osteoporos Int 1992;2(6):285-9. 82.  Donaldson LJ, Cook A, Thomson RG. Incidence of fractures in a geographically defined population. J Epidemiol Community Health 1990;44(3):241-5. 83.  Porter RW, Miller CG, Grainger D, Palmer SB. Prediction of hip fracture in elderly women: a prospective study. BMJ 1990;301(6753):638-41. 84.  Rockwood PR, Horne JG, Cryer C. Hip fractures: a future epidemic? J Orthop Trauma 1990;4(4):388-93. 85.  Kauffman TL, Albright L, Wagner C. Rehabilitation outcomes after hip  fracture in persons 90 years old and older. Arch Phys Med Rehabil 1987;68(6):369-71. 86.  Cummings SR, Black DM, Rubin SM. Lifetime risks of hip, Colles', or vertebral fracture and coronary heart disease among white postmenopausal women. Arch Intern Med 1989;149(11):2445-8. 87.  Chevalley T, Guilley E, Herrmann FR, Hoffmeyer P, Rapin CH, Rizzoli R. Incidence of hip fracture over a 10-year period (1991-2000): reversal of a secular trend. Bone 2007;40(5):1284-9.  66 88.  Jaglal SB, Weller I, Mamdani M, Hawker G, Kreder H, Jaakkimainen L, et al. Population trends in BMD testing, treatment, and hip and wrist fracture rates: are the hip fracture projections wrong? J Bone Miner Res 2005;20(6):898-905. 89.  Lofman O, Berglund K, Larsson L, Toss G. Changes in hip fracture epidemiology: redistribution between ages, genders and fracture types. Osteoporos Int 2002;13(1):18-25. 90.  Melton LJ, 3rd, Kearns AE, Atkinson EJ, Bolander ME, Achenbach SJ, Huddleston JM, et al. Secular trends in hip fracture incidence and recurrence. Osteoporos Int 2009;20(5):687-94. 91.  Stafford RS, Drieling RL, Hersh AL. National trends in osteoporosis visits and osteoporosis treatment, 1988-2003. Arch Intern Med 2004;164(14):1525-30. 92.  Chevalley T, Rizzoli R, Nydegger V, Slosman D, Rapin CH, Michel JP, et al. Effects of calcium supplements on femoral bone mineral density and vertebral fracture rate in vitamin-D-replete elderly patients. Osteoporos Int 1994;4(5):245-52. 93.  Schurch MA, Rizzoli R, Slosman D, Vadas L, Vergnaud P, Bonjour JP. Pro tein supplements increase serum insulin-like growth factor-I levels and attenuate proximal femur bone loss in patients with recent hip fracture. A randomized, double -blind, placebo-controlled trial. Ann Intern Med 1998;128(10):801-9. 94.  Osteoporosis: review of the evidence for prevention, diagnosis and treatment and cost- effectiveness analysis. Introduction. Osteoporos Int 1998;8 Suppl 4:S7-80. 95.  Gehlbach SH, Avrunin JS, Puleo E. Trends in hospital care for hip fractures. Osteoporos Int 2007;18(5):585-91. 96.  Lofthus CM, Osnes EK, Falch JA, Kaastad TS, Kristiansen IS, Nordsletten L, et al. Epidemiology of hip fractures in Oslo, Norway. Bone 2001;29(5):413-8. 97.  Herrera A, Martinez AA, Ferrandez L, Gil E, Moreno A. Epidemiology of osteoporotic hip fractures in Spain. Int Orthop 2006;30(1):11-4.  67 98.  Schwartz AV, Kelsey JL, Maggi S, Tuttleman M, Ho SC, Jonsson PV, et al. International variation in the incidence of hip fractures: cross-national project on osteoporosis for the World Health Organization Program for Research on Aging. Osteoporos Int 1999;9(3):242-53. 99.  Riggs BL, Melton LJ, 3rd. The worldwide problem of osteoporosis: insights afforded by epidemiology. Bone 1995;17(5 Suppl):505S-511S. 100.  Jaglal SB, Sherry PG, Schatzker J. The impact and consequences of hip fracture in Ontario. Can J Surg 1996;39(2):105-11. 101.  Kannus P, Parkkari J, Sievanen H, Heinonen A, Vuori I, Jarvinen M. Epidemiology of hip fractures. Bone 1996;18(1 Suppl):57S-63S. 102.  Papadimitropoulos EA, Coyte PC, Josse RG, Greenwood CE. Current and projected rates of hip fracture in Canada. CMAJ 1997;157(10):1357-63. 103.  Jaglal S. Falling hip fracture rates. J Bone Miner Res 2007;22(7):1098; author reply 1099. 104.  Kannus P, Niemi S, Parkkari J, Palvanen M, Vuori I, Jarvinen M. Nationwide decline in incidence of hip fracture. J Bone Miner Res 2006;21(12):1836-8. 105.  Leslie WD, O'Donnell S, Jean S, Lagace C, Walsh P, Bancej C, et al. Trends in hip fracture rates in Canada. JAMA 2009;302(8):883-9. 106.  Census. Census Canada, 2007. 107.  Provincial health officer. Prevention of falls and Injuries Among the Elderly: A Special Report from the Office of the Provincial Health Officer, 2004. 108.  Braithwaite RS, Col NF, Wong JB. Estimating hip fracture morbidity, mortality and costs. J Am Geriatr Soc 2003;51(3):364-70. 109.  Dolan P, Torgerson DJ. The cost of treating osteoporotic fractures in the United Kingdom female population. Osteoporos Int 1998;8(6):611-7.  68 110.  Ray NF, Chan JK, Thamer M, Melton LJ, 3rd. Medical expenditures  for the treatment of osteoporotic fractures in the United States in 1995: report from the National Osteoporosis Foundation. J Bone Miner Res 1997;12(1):24-35. 111.  Johansen A, Stone M. The cost of treating osteoporotic fractures in the United Kingdom fem ale population. Osteoporos Int 2000;11(6):551-2. 112.  Norris RJ. Medical costs of osteoporosis. Bone 1992;13 Suppl 2:S11-6. 113.  Neumann PJ, Goldie SJ, Weinstein MC. Preference-based measures in economic evaluation in health care. Annu Rev Public Health 2000;21:587-611. 114.  Earnshaw J, Lewis G. NICE Guide to the Methods of Technology Appraisal: pharmaceutical industry perspective. Pharmacoeconomics 2008;26(9):725-7. 115.  Brooks R. EuroQol: the current state of play. Health Policy 1996;37(1):53-72. 116.  Brazier J, Roberts J, Deverill M. The estimation of a preference -based measure of health from the SF-36. J Health Econ 2002;21(2):271-92. 117.  Feeny D, Furlong W, Boyle M, Torrance GW. Multi -attribute health s tatus classification systems. Health Utilities Index. Pharmacoeconomics 1995;7(6):490-502. 118.  J. von Neumann., O. Morgenstern. Theory of games and economic behaviour. Princeton, New Jersey, 1944. 119.  Dawson NV, Singer ME, Lenert L, Patterson MB, Sami SA, Gonsenhouser I, et al. Health state valuation in mild to moderate cognitive impairment: feasibility of computer-based, direct patient utility assessment. Med Decis Making 2008;28(2):220-32. 120.  Littenberg B, Partilo S, Licata A, Kattan MW. Paper Standard Gamble: the reliability of a paper questionnaire to assess utility. Med Decis Making 2003;23(6):480-8. 121.  McNeil BJ, Pauker SG, Sox HC, Jr., Tversky A. On the elicitation of preferences for alternative therapies. N Engl J Med 1982;306(21):1259-62.  69 122.  Thompson MS. Willingness to pay and accept risks to cure chronic disease. Am J Public Health 1986;76(4):392-6. 123.  Tversky A, Kahneman D. The framing of decisions and the psychology of choice. Science 1981;211(4481):453-8. 124.  Tengs TO, Wallace A. One thousand health-related quality-of-life estimates. Med Care 2000;38(6):583-637. 125.  Bleichrodt H, Johannesson M. An experimental test of a theoretical foundation for rating -scale valuations. Med Decis Making 1997;17(2):208-16. 126.  Hornberger JC, Redelmeier DA, Petersen J. Variability among  methods to assess patients' well- being and consequent effect on a cost-effectiveness analysis. J Clin Epidemiol 1992;45(5):505-12. 127.  Nease RF, Jr., Brooks WB. Patient desire for information and decision making in health care decisions: the Autonomy Preference Index and the Health Opinion Survey. J Gen Intern Med 1995;10(11):593-600. 128.  O'Leary JF, Fairclough DL, Jankowski MK, Weeks JC. Comparison of time -tradeoff utili ties and rating scale values of cancer patients and their relatives: evidence for a possible plateau relationship. Med Decis Making 1995;15(2):132-7. 129.  Read JL, Quinn RJ, Berwick DM, Fineberg HV, Weinstein MC. Preferences for health outcomes. Comparison of assessment methods. Med Decis Making 1984;4(3):315-29. 130.  P. Dolan. The measurement of health-related quality of life for use in resource alocation decisions in health care Chapter 32. In: AJ Culyer JN, editor. Handbook of Health Economics. London, UK, 2000. 131.  Gold MR, Siegel JE, Russel LB, Weinstein MC. Cost-effectiveness in Health and Medicine. Oxford, UK: Oxford University Press, 1996. 132.  Torrance GW. Measurement of health state utilities for economic appraisal. J Health Econ 1986;5(1):1-30.  70 133.  Torrance GW, Boyle MH, Horwood SP. Application of multi-attribute utility theory to measure social preferences for health states. Oper Res 1982;30(6):1043-69. 134.  Torrance GW. Social preferences for health states: an empirical evaluations of three measurement techniques. Socio-Econ Plan Sci 1976;10:128-36. 135.  van Wijck EE, Bosch JL, Hunink MG. Time-tradeoff values and standard-gamble utilities assessed during telephone interviews versus face-to-face interviews. Med Decis Making 1998;18(4):400-5. 136.  Guyatt GH, Feeny DH, Patrick DL. Measuring health-related quality of life. Ann Intern Med 1993;118(8):622-9. 137.  Fitzpatrick R, Davey C, Buxton MJ, Jones DR. Evaluating patient-based outcome measures for use in clinical trials. Health Technol Assess 1998;2(14):i-iv, 1-74. 138.  Marra CA, Woolcott JC, Kopec JA, Shojania K, Offer R, Brazier JE, et al. A comparison of generic, indirect utility measures (the HUI2, HUI3, SF-6D, and the EQ-5D) and disease-specific instruments (the RAQoL and the HAQ) in rheumatoid arthritis. Soc Sci Med 2005;60(7):1571-82. 139.  Dolan P. Modeling valuations for EuroQol health states. Med Care 1997;35(11):1095-108. 140.  Torrance GW. Utility approach to measuring health-related quality of life. J Chronic Dis 1987;40(6):593-603. 141.  Horsman J, Furlong W, Feeny D, Torrance G. The Health Utili ties Index  (HUI): concepts, measurement properties and applications. Health Qual Life Outcomes 2003;1:54. 142.  Feeny D, Furlong W, Torrance GW, Goldsmith CH, Zhu Z, DePauw S, et al. Multiattribute and single-attribute utility functions for the health utili ties index mark 3 system. Med Care 2002;40(2):113-28. 143.  Le Gales C, Costet N, Gentet JC, Kalifa C, Frappaz D, Edan C, et al. Cross -cultural adaptation of a health status classification system in children with cancer. First results of the French adaptation of the Health Utilities Index Marks 2 and 3. Int J Cancer Suppl 1999;12:112-8.  71 144.  Hiligsmann M, Ethgen O, Richy F, Reginster JY. Utility values associated with osteoporotic fracture: a systematic review of the literature. Calcif Tissue Int 2008;82(4):288-92. 145.  Borgstrom F, Zethraeus N, Johnell O, Lidgren L, Ponzer S, Svensson O, et al. Costs and quality of life associated with osteoporosis-related fractures in Sweden. Osteoporos Int 2006;17(5):637-50. 146.  A prospective study of the health related quality of life impact of hip fracture. ScHARR; 2000; University of Sheffield, Sheffield. 147.  Murray CWS, Brazier JE. Utility following a fracture in a group of eldelry women. Quality of LIfe Research 2002;11:642. 148.  Tosteson AN, Gabriel SE, Grove MR, Moncur MM, Kneeland TS, Melton LJ, 3rd. Impact of hip and vertebral fractures on quality-adjusted life years. Osteoporos Int 2001;12(12):1042-9. 149.  Brazier JE, Green C, Kanis JA. A systematic review of health state utility values for osteoporosis - related conditions. Osteoporos Int 2002;13(10):768-76. 150.  Sach TH, Foss AJ, Gregson RM, Zaman A, Osborn F, Masud T, et al. Falls and health status in elderly women following first eye cataract surgery: an economic evaluation conducted alongside a randomised controlled trial. Br J Ophthalmol 2007;91(12):1675-9. 151.  Conner-Spady B, Suarez-Almazor ME. Variation in the estimation of quality-adjusted life-years by different preference-based instruments. Med Care 2003;41(7):791-801. 152.  Robbins L. An Essay on the Nature and Significance of Economic Science. 2nd Edition ed. London: MacMillan and Co. Ltd, 1932. 153.  Mehrez A, Gafni A. Quality-adjusted life years, utility theory, and healthy-years equivalents. Med Decis Making 1989;9(2):142-9. 154.  Mehrez A, Gafni A. The healthy-years equivalents: how to measure them using the standard gamble approach. Med Decis Making 1991;11(2):140-6.  72 155.  Mehrez A, Gafni A. Preference based outcome measures for economic evaluation of drug interventions: quality adjusted life years (QALYs) versus healthy years equivalents (HYEs). Pharmacoeconomics 1992;1(5):338-45. 156.  Gates S, Fisher JD, Cooke MW, Carter YH, Lamb SE. Multifactorial assessment and targeted intervention for preventing falls and injuries among older people in community and emergency care settings: systematic review and meta-analysis. BMJ 2008;336(7636):130-3. 157.  Kamide N, Shiba Y, Shibata H. Effects on balance, falls, and bone mineral density of a home - based exercise program without home visits in community -dwelling elderly women: a randomized controlled trial. J Physiol Anthropol 2009;28(3):115-22. 158.  Robertson MC, Campbell AJ, Gardner MM, Devlin N. Preventing injuries in older people by preventing falls: a meta-analysis of individual-level data. J Am Geriatr Soc 2002;50(5):905-11. 159.  Glick HA, Doshi JA, Sonnad SA, Polsky D. Economic Evaluation in Clinical Trials. New York USA: Oxford University Press, 2007. 160.  Campbell AJ, Robertson MC, Gardner MM, Norton RN, Buchner DM. Psychotropic medication withdrawal and a home-based exercise program to prevent falls: a randomized, controlled trial. J Am Geriatr Soc 1999;47(7):850-3. 161.  Rizzo JA, Baker DI, McAvay G, Tinetti ME. The cost-effectiveness of a multifactorial targeted prevention program for falls among community elderly persons. Med Care 1996;34(9):954-69. 162.  Beard J, Rowell D, Scott D, van Beurden E, Barnett L, Hughes K, et al. Economic analysis of a community-based falls prevention program. Public Health 2006;120(8):742-51. 163.  Claxton K. The irrelevance of inference: a decision-making approach to the stochastic evaluation of health care technologies. J Health Econ 1999;18(3):341-64. 164.  Campbell AJ, Robertson MC, Gardner MM, Norton RN, Tilyard MW, Buchner DM. Randomised controlled trial of a general practice programme of home based exercise to prevent falls in elderly women. BMJ 1997;315(7115):1065-9.  73 165.  Campbell AJ, Robertson MC, La Grow SJ, Kerse NM, Sanderson GF, Jacobs RJ, et al. Randomised controlled trial of prevention of falls in people aged >=75 with severe visual impairment: the VIP trial. BMJ 2005;331(7520):817. 166.  Robertson MC, Gardner MM, Devlin N, McGee R, Campbell AJ. Effectiveness and economic evaluation of a nurse delivered home exercise programme to prevent falls. 2: Controlled trial in multiple centres. BMJ 2001;322(7288):701-4. 167.  Robertson MC, Devlin N, Scuffham P, Gardner MM, Buchner DM, Campbell AJ. Economic evaluation of a community based exercise programme to prevent falls. J Epidemiol Community Health 2001;55(8):600-6. 168.  Donaldson MG. Falls risk in frail seniors: clinical and methodological studies. University of British Columbia, 2007. 169.  Salkeld G, Cumming RG, O'Neill E, Thomas M, Szonyi G, Westbury C. The cost effectiveness of a home hazard reduction program to reduce falls among older persons. Aust N Z J Public Health 2000;24(3):265-71. 170.  Smith RD, Widiatmoko D. The cost-effectiveness of home assessment and modification to reduce falls in the elderly. Aust N Z J Public Health 1998;22(4):436-40. 171.  Pignone M, Earnshaw S, Pletcher MJ, Tice JA. Aspirin for the primary prevention of cardiovascular disease in women: a cost-utility analysis. Arch Intern Med 2007;167(3):290-5. 172.  Streiner DL, Norman GR. Health Measurement Scales: a practical guide to their development and use Third Edition Oxford University Press. 2003:130-135. 173.  Harwood RH. Economic evaluations of complex services for older people. Age Ageing 2008;37(5):491-3. 174.  Vind AB, Andersen HE, Pedersen KD, Jorgensen T, Schwarz P. Baseline and  follow-up characteristics of participants and nonparticipants in a randomized clinical trial of multifactorial fall prevention in Denmark. J Am Geriatr Soc 2009;57(10):1844-9.  74 175.  SmartRisk. The economic burden of unitentional injury in Canada p. 59. Ottawa, 1998. 176.  SmartRisk. Economic burden of unintentional injury in British Columbia. British Columbia Injury Research and Prevention Unit: Vancouver 2001:30. 177.  Provincial Health Offier. Prevention of falls and inuries in the elderly. Office of the Provincial Health Officer: Victoria 2003. 178.  SmartRisk. Economic burden of unintentional injury in British Columbia p.36. British Columbia Injury Research and Prevention Unit. Vancouver, 2001. 179.  Orr R, Raymond J, Fiatarone Singh M. Efficacy of prog ressive resistance training on balance performance in older adults : a systematic review of randomized controlled trials. Sports Med 2008;38(4):317-43. 180.  Liu-Ambrose T, Khan KM, Eng JJ, Janssen PA, Lord SR, McKay HA. Resistance and agility training reduce fall risk in women aged 75 to 85 with low bone mass: a 6-month randomized, controlled trial. J Am Geriatr Soc 2004;52(5):657-65. 181.  BC Women's Osteofit Program. www.osteofit.org, Accessed January 1 2010. 182.  Liu-Ambrose TY, Ashe MC, Graf P, Beattie BL, Khan KM. Increased risk of falling in older community-dwelling women with mild cognitive impairment. Phys Ther 2008;88(12):1482-91. 183.  Sadatsafavi M, Marra CA, Ayas NT, Stradling J, Fleetham J. Cost-effectiveness of oral appliances in the treatment of obstructive sleep apnoea-hypopnoea. Sleep Breath 2009;13(3):241-52. 184.  Royall DR, Lauterbach EC, Cummings JL, Reeve A, Rummans TA, Kaufer DI, et al. Executive control function: a review of its promise and challenges for clinical research. A report from the Committee on Research of the American Neuropsychiatric Association. J Neuropsychiatry Clin Neurosci 2002;14(4):377-405. 185.  Stuss DT, Alexander MP. Executive functions and the frontal lobes: a conceptual view. Psychol Res 2000;63(3-4):289-98.  75 186.  Kuo HK, Lipsitz LA. Cerebral white matter changes and geriatric syndromes: is there a link? J Gerontol A Biol Sci Med Sci 2004;59(8):818-26.    aA version of this chapter has been accepted for publication. DAVIS JC, Robertson MC, Ashe MC, Khan KM, Marra CA 2009 International comparison of cost of falls in older adults living in the community: a systematic review Osteoporosis International (accepted, December 7, 2009)  76 2 International comparison of cost of falls in older adults living in the community: a systematic reviewa 2.1 Introduction  Falls and injuries resulting from falls in older adults represent a significant health burden1 since approximately 30% of those aged 65 years and older experience at least one fall each year and half of those fall recurrently.2,3 Non-fatal fall injuries are associated with increased morbidity, decreased functioning and increased health care resource utilization. Falls and fall related injuries including fracture account for 10-15% of emergency department presentations of those aged 65 years and older. 4,5  Policy makers are continually challenged as to how, where and to whom to allocate scarce resources. By quantifying the economic burden of falls on the health system this geriatric syndrome can be compared with other important medical conditions. By evaluating data at an international level, investigating reasons for variation in cost of falls estimates and standardizing the economic methods and time horizon over which costs are collected, researchers can begin to seek potential for cost saving and track changes over time.  To my knowledge, there is no systematic review cost of illness study that includes cost of fall and/or fall related injuries. There is one systematic review of economic evaluations of falls prevention strategies. 6 Therefore, to help policy makers, health care professionals and government recognize the global financial burden of falls, I systematically reviewed all peer-reviewed literature to quantify the question: ―What is the economic burden of falls in different countries and why do these costs differ?    77 2.2 Materials and methods 2.2.1 Literature searches  In accordance with QUOROM7 guidelines, I conducted a comprehensive search of MEDLINE, EMBASE, CINAHL, the Cochrane Database of Systematic Reviews, the Cochrane Database of Abstracts of Reviews of Effects and NHS EED databases to identify cost of illness (falls) studies published in the English language from 1945 through December 2008. Included in my search terms and medical subject headings were fall$, cost$ and older adults. I searched the references of retrieved articles to trace potentially relevant papers. 2.2.2 Selection of studies  I selected peer reviewed published studies that included an estimate of the cost of falls in  older adults aged 60 years and older living in the community. I excluded studies if the costs were from falls prevention intervention trials that were not designed a priori to estimate cost of falls, if they did not display results separately for older people, or if they were not a cost of illness (falls) study. After I critically reviewed titles and abstracts, seventeen full text manuscripts met my inclusion criteria (Figure 2-1).   78                                     Figure 2-1. QUOROM flow diagram of selection of studies 2.2.3 Data extraction and data synthesis  I developed standardized forms to extract information from the published manuscripts included in my systematic review. Extracted information included: publication, journal, impact factor, study design, study sample, cost items, analysis perspective, country, reported currency, year of currency, time period costs were measured, sensitivity analysis, main economic outcomes, falls definition, main clinical outcomes and Excluded studies (title and abstract) n=397  Duplicate paper = 47  Review papers = 91  Wrong age = 4  Not relevant = 169  Intervention costs = 54  No costs reported = 16  Not community dwelling = 16 26 duplicates  Retrieved full text for more detail n=33 Excluded studies (full text) n=16  Not specific to falls = 2  Not relevant = 12  Not age specific = 1  Intervention trial = 1   Total number of studies included in systematic review n=17 Potentially relevant studies identified    n=430  Reference list of studies reviewed (full text)     n=33   79 quality assessment based on Drummond‘s checklist.8 I also extracted information on specific methodology used in each cost of illness (falls) study included in my systematic review (Table 2-1). This information included: 1) the source of cost item information (i.e., database), 2) the population size where stated (i.e.., population denominator), 3) the costing approach – these included incidence based costs or prevalence based costing and 4) the time horizon over which costs were  collected. Further, I identified and classified falls related cost items for all cost of illness (falls) studies included in my systematic review (Table 2 -2). Cost items identified I categorized as follows: inpatient hospital, health care professional or outpatient, home health care, emergency department or ambulance, medications, personal out of pocket and non- injurious fall related costs. I obtained any missing information from the papers that the article of interest cited. Two authors performed data abstraction (JCD, MCR) and a third author (one of MCA, TLA or CAM) independently checked the data abstracted. Any discrepancies were discussed and checked by all authors and resolved by consensus.   Table 2-1. Fall related cost items reported in cost of falls studies  Reference Inpatient hospitala Health care professional Home health care Emergency dept or Ambulance costs Medications Personal  Non injurious fall related Funder reported United States: population based studies Englander 199611  + + + (rehabilitation)     + Carrollb 200512 + + + + +  + + Rizzo 199813  +  + +   + + Stevensc 200614  + + + +    + Finkelsteind 200515 + +  +    + Alexander 199216 + Roudsari 200517  + +  + +   + Ellise 200118  + Mahoneye 200519 +   + United Kingdom: population based studies Newton 200620     + Scuffham 200321 + +  +    + Australia: population based study Hendrie 200423  + + + + + Australia: prospective cohort study Hallb 200322  + + + + + + Europe: population based studies Pannemane 200326 +    + Sjogren 199127  + + Europe: retrospective or prospective cohort studies Cotterf 200625  + + Seematter- Bagnoude 200624       +* a  Included hospital inpatient, acute care, rehabil itation and long term care costs b  Cost data were self reported c  Costs were reported according to falls classified as fatal or non-fatal falls 80   d  Fall injuries were stratified into three groups (hospitalization, emergency department with no hospitalization and outpatient) e  Data reported were charges not costs f  Hip fracture cost estimates were prov ided for five randomly selected patients 81   Table 2-2. Study population and outcome measures Reference  Population size (i.e., denominator for cost estimate) Study sample  Results: reported currency Results: converted to US dollars (2008 prices) United States: population based studies Englander 199611  For adults ≥65 years, 33.2 million, 53.3 million ≥65 yrs in 2020 US resident population stratified by age group Average US $7399 per injurious fall, total cost US $20.2 billion (1994 prices) Total cost US $32.3 billion (2020 prices) Average US $10 749 per injur ious fall, total cost US $29.4 billion  Carroll 200512  34 028 786 Data for non institutionalized elderly population of the United States were taken from the 1997 Medical Expenditure Panel Survey (n=4025) Total costs and utilization of fa ll related medical US $7.8 billion Mean cost per fa ller US $2591 Total costs and utilization of fa ll related medical US $9.3 billion Mean cost per fa ller US $3476 Rizzo 199813  1017 Aged ≥72 years, probability sample of residents of New Haven, Connecticut (Project Safety)  ≥1 injur ious fall compared with non fa llers (hospital US $11 042, nursing home US $5325, ER US $253, Home health US $2820, total US $19 440) ≥1 injur ious fall compared with non fallers (hospital US $15 152, nursing home US $7307, ER US $347, Home health US $3870, total US $26 676) Stevens 200614  10 300 fatal and 2.6 million non fatal fa ll related injuries  Aged ≥65 years Individual data for fatal and non fatal fa ll injuries obtained from National Center for Health Statistics National Vital Statistics System (death), Medical Expenditure Panel Survey (office, outpatient), Medicare Standard Analytical Files (non- fatal) Direct medical costs include US $0.2 billion for fatal and US $19 billion for non-fatal in juries Direct medical costs include US $0.3 billion for fatal and US $23 billion for non- fatal in juries Finkelstein 200515  For case control 22 514, For case cross-over 51 861 fallers and 102 755 non fallers Aged ≥65 years who were beneficiaries of Medicare from 1998-1999 who had a non-fatal fall Case control design (overall average hospitalization fall injury costs US $22 260) Case-crossover design (overall average hospitalization fall in jury costs US $20 920) Case control design (overall average hospitalization fall injury costs US $27 832) Case-crossover design (overall average hospitalization fall injury costs US $28 657) Alexander 199216  149 504 Adults ≥65 years (excluding Veteran Administration and military hospitals) in Washington State Commission Hospital Abstract Repor ting System Hospitalization for fa ll related trauma US $53 346 191 Females, average cost per hospitalization US $6559, males, US $7379 Cost of falls US $92 per person per year Hospitalization for fa ll related trauma  US $92.6 million Females, average cost per hospitalization US $11 388, males, US $12 812 Cost of falls US $160 per person per year Roudsari 200517  550 fall related Aged ≥65 years, Market Scan, Medicare Mean (SD) hospitalization  Mean (SD) hospitalization 82   Reference  Population size (i.e., denominator for cost estimate) Study sample  Results: reported currency Results: converted to US dollars (2008 prices)  hospitalizations from 153 000 hospital admission records Supplemental database, fa ll related hospital admission, emergency department visit, or outpatient visit, ICD-9 and E880-888.9 codes US $17483 (22 426), ED visits US $236 (388), outpatient visit US $412 (1146)  US $19 927 (25 561), ED visits US $269 (442), outpatient visit US $470 (1306)  Ellis 200118  For adults ≥65 years, 174 436 Adults aged ≥ 20 years hospitalized with a fall injury, computerized hospital discharge summaries from the California Office of Statewide Health Planning and Development Mean hospital charges: For adults aged ≥65 years, mean hospital charges for hospitalized falls US $17 299  Mean hospital charges: For adults aged ≥65 years, mean hospital charges for hospitalized falls US $26 483  Mahoney 200519  Not stated Aged ≥65 years, fa lls-related deaths and injuries from WISH (Web-based interactive query system). ICD-0 (E880- 886, E888) and ICD-10 (W00-19) codes Statewide charges: US $43 703 516 (1995) and US $81.6 million (2002) Average charge per hospital admission for fa ll related in jury US $12 741 (2002) Average ED charge per visit US $780 (2002) Statewide charges: US $52.3 million and US $98 million Average charge per hospital admission for fall related injury US $15 248 Average ED charge per visit US $934 United Kingdom: population based studies Newton 200620  41 338 Aged ≥65 year, N=1504 who fell and reported to the Nor th East Ambulance Service Falls cost UK £115 per ambulance call out Total cost was UK £172 960 Falls cost US $191 per ambulance call out Total cost was US $286 512 Scuffham 200321  10 000 randomly selected cases Aged ≥60 years, attending an Accident and Emergency or admitted to hospital after a fall UK £300 000 (60-64 years), UK £1.5 million (≥75 years), Total cost to UK government UK £981 million US $497 000 (60-64 years), US $1.7 million (≥75 years), Total cost to UK government US $1.6  billion Australia: population based study Hendrie 200423  18706 emergency department presentations, 6222 hospital admissions Aged≥65 years who fell and presented to the emergency department Overall health system costs of accidental fa lls in o lder people across all services was AUD $86 million Overall health system costs of accidental fa lls in o lder people across all services was US $72 million Australia: prospective cohort study Hall 200322   Aged ≥65, N=79 who presented to the emergency department after a fall Mean cost of falls per patient was AUD $4343 (without informal care) AUD $4642 (with informal care) Total fa ll re lated costs for a three month period are AUD $318 993 (without informal care) AUD $333 648 (with informal care). Mean cost of falls per patient was US $3767 (without informal care) US $4026 (with informal care) Total fa ll re lated costs for a three month period are US $276 680 (without informal care) US $289 391 (with informal care). Europe: population based studies Panneman 200326  2497 Aged ≥55 years, residents from 24 Benzodiazepine fall related Benzodiazepine fall related 83   Reference  Population size (i.e., denominator for cost estimate) Study sample  Results: reported currency Results: converted to US dollars (2008 prices)  geographical areas in The Netherlands, Benzodiazepine associated hospitalization as a result of a fall, Cases defined by E-codes hospitalization costs  €48.5 million (range: 39.0-58.8), The Netherlands, €1.8 billion (95% CI: 1.5-2.2), European Union hospitalization costs US $71 million (range: 49-74), The Netherlands, US $2.7 billion (95% CI: 2.2-3.3), European Union Sjogren 199127  1311 Aged ≥60 years, treated at Umea Regional Hospital for an intentional in jury Total cost of medical care for a fall was SEKa 11.1 million in the older group, mean SEK 12400 Total cost of medical care for a fall was US $3 million in the older group, mean US $3376 Europe: retrospective or prospective cohort studies Cotter 200625  810 fall related admissions All patients aged ≥ 65 years from hospital in-patient enquiry (HIPE) system, supplemented by review of hospital case notes and emergency department records Total costs of one year fall related admissions to acute hospital  €10.8 million Average cost of h ip fracture admission €14339 Total costs of one year fall related admissions to acute hospital  US $15.2 million Average cost of h ip fracture admission US $20 189 Seematter-Bagnoud 200624  690 Aged ≥75 years, hospitalized if ≥24 hours after presenting to the Emergency Department  (in jurious falls excluded) Enrolled in two prospective cohor t studies Fallers versus non faller s, average costs per day $138.5 vs $148.7 (P=0.66), acute care costs $83.0 vs $124.5 (P=0.06), long term care $55.5 vs $24.1 (P<0.001) Fallers versus non faller s, average costs per day US $175.3 vs  US $158.8 (P=0.66), acute care costs $83.0 vs $133.0 (P=0.06), long term care $64.7 vs $25.7 (P<0.001) Assumed US dollars a Swedish kronas 84    85 2.2.4 Standardized cost outcomes  The studies reported costs in different currencies and different years. To attenuate this variation, I reported costs in two ways (Table 2-2): 1) the value documented by year and currency from the manuscript and 2) converted to 2008 US dollars by inflating the values to 2008 prices using the Consumer Price Index for that country and then using Purchasing Power Parity values in 2008 to convert the currency to US dollars. 9,10 To enable comparisons I have reported costs in US dollars at 2008 prices in the Results and Discussion sections. I note that the population for which the cost outcomes are reported it also important to quantify, therefore in Table 2-2 and Table 2-3, I report the population denominator for which each estimate is based, where available.  Table 2-3. Study methodology  Reference, Journal, Impact Factor Type of costing approach – Incidence (I) vs. Prevalence (P) study Time horizon Perspective  Currency Year of costs United States: population based studies Englander 199611 Journal of Forensic Science 1.037 I Lifetime projections Not stated US dollars 1994, costs projected to 2020 Carroll 200512 Journal of Managed Care Pharmacy Not available P 1 year Not stated US dollars 1997 prices inflated to 2002 Rizzo 199813 Medical Care 3.554 I 1 year Not stated  US dollars 1996 Stevens 200614a Injury Prevention 1.401 I 1 year Not stated US dollars 2000 Finkelstein 200515 Medical Care 3.554 P 1 year, 2 years Not stated  US dollars 2000 Alexander 199216 Amer ican Journal of Public Health 3.612 P 1 year Not stated  US dollars 1989 Roudsari 200517 Injury 1.509 P 1 year Not stated US dollars 2004 Ellis 200118 Journal of Gerontology: Medical Sciences 3.455 P 3 years Not stated  US dollars Not stated Mahoney 200519a Wisconsin Medical Journal Not available P 8 years Not stated  US dollars 2002 United Kingdom: population based studies Newton 200620 Emergency Medicine I 7 months Cost to North East Ambulance Service Pounds sterling 2004-2005 86   Reference, Journal, Impact Factor Type of costing approach – Incidence (I) vs. Prevalence (P) study Time horizon Perspective  Currency Year of costs Journal 0.929 Scuffham 200321b Journal of Epidemiology & Community Health 2.956 I 1 year National Health Service and Personal Social Services Pounds sterling 2000 Australia: population based study Hendrie 200423 Australian Health Review Not stated P 1 year Health system Australian dollars 2001-2002 Australia: prospective cohort study Hall 200322c Australian and New Zealand Journal of Public Health 1.335 I 3 months Not stated Australian dollars 1999 Europe: population based studies Panneman 200326 Drugs & Aging 2.140 P 16 years Not stated Euros 2000 Sjogren 199127 Scandinavian Journal of Primary Health care 1.908 P 1 year Health care system Swedish kronas  1985 Europe: retrospective or prospective cohort studies Cotter 200625 Irish Journal of Medical Science 0.290 P 1 year Not stated Euros Not stated Seematter-Bagnoud 200624c Journal of the American Geriatrics Society 3.539 I 6 months Not stated  Not stated Not stated a The size of the study population used (denominator) was not prov ided b A  one-way selective sensitiv ity  analysis was reported for estimating cost of falls c Questionnaires were used for estimating costs; all other studies used a database  87    88 2.2.5 Quality assessment  There are no validated tools to assess the quality of studies of the cost of illness, specifically cost of falls without a comparator. Therefore, I assessed the quality of each study using seven questions from Drummond‘s checklist8 and one question I developed which was specific to falls: ―Were fall events clearly defined for the purposes of the study?‖ Two authors (JCD,  MCR) independently evaluated each study and any discrepancies were discussed and reviewed by a third party (one of CAM, MCA, TLA) and resolved by consensus (Table 2-4). The authors have reported further detailed examples of quality assessments for economic evaluations of falls prevention strategies.6  Table 2-4. Modified version of Drummond‘s checklist to assess quality of economic studies* Reference 1 2 3 4 5 6 7 8 Alexander16 + - + - + - + - Carroll12 + - + + + + + - Cotter25  + - + + + - + - Ellis18  + - + + + - + - Englander11  + - - - + + - - Finklestein15  + - + + + - + - Hall22  + - - + + + - - Hendrie23 + + + + + + + - Mahoney19 + - + - - - + - Newton20  + + + + + + - - Panneman26  + - + + + + + - Rizzo13  + - + + + + + - Roudsari17  + - + + + + + - Scuffham21  + + + + + + + + Seematter- Bagnoud24 + - + - - - - - Sjogren27  + + - - + + + - Stevens14 + - + + + + + - + Study met the cr iterion - All or part of the criterion was not met *7 criteria of Drummond‘s checklist plus our item 3 (specific to falls): Was the study objective well defined? Was the viewpoint of the analysis stated? Were fall events clearly defined for the purposes of the study? Were all impor tant and relevant costs identified (based on study objective)? Were costs measured accurately in appropriate physical units? Were costs valued credibly? Were costs tracked for one year or more? Was a sensitivity analysis per formed (i.e., a llowance for uncertainty in the estimates of costs)? 89    90 2.3 Results 2.3.1 Literature searches – overview of studies identified  After critical review of the 430 titles and abstracts and 33 full text manuscripts, 17 studies met the inclusion criteria (Figure 2-1). These included 9 studies from the United States, two from Australia, four from Europe and two from the United Kingdom (Table 2-1). Studies differed in the categorization and definition of falls and the cost items included. I report results using ‗location‘ as the main subheading, rather than ‗study design‘ to provide the reader easier access to the relevant costs. 2.3.2 United States – national cost of falls and other settings – five studies Englander and colleagues reviewed economic dimensions of slip and fall injuries.11 The total cost of fall injuries for all age groups in 1994 was projected at US $64.4 billion and in 2020 was projected at US $85.4 billion based on costs in 1985. For a population of 18.7 million adults aged 65-74 years, the average cost per fall was US $10 749 (at 2008 prices) and the total cost of falls alone (not their associated injuries) was US $13.1 billion (at 2008 prices). One study reported the total medical cost of falls in community dwelling elderly people to be US $9.3 billion for a population of 34 million.12 Of these costs, 65% were inpatient hospitalization costs. One study focused on broader health care utilization and reported the greater costs of people with injurious falls compared with non fallers for hospital, nursing home, emergency department and home health services (total US $26 676 greater cost per fall).13 Another US study classified cost of fall related injuries into fatal and non-fatal falls.14 Overall, the direct medical costs for the US totaled US $0.3 billion for fatal falls and US $23 billion for non-fatal fall related injuries. Of the non-fatal fall injuries, 63% of the costs were from hospitalizations and 21% were from emergency department visits. Lastly, one US study compared case-control and case-crossover designs for estimating medical costs of non-fatal fall related injuries and found that estimates of the cost of falls using a case control approach was between 6% and 17% greater than estimates from the case-crossover approach (51 861 fallers and 102 755 non fallers).15   91 2.3.3 United States – hospitalization costs of fall-related injuries – four studies  One study in adults aged 65 years and older estimated that the cost of fall related hospitalization for the State of Washington (based on Washington State Census projections for 1980 and 1990, aged >65 years) was US $92.6 million.16 These costs were based on hospital charges only; professional fees were excluded. Further, the cost of falls was estimated to be US $160 per person per year. Another study estimated the mean hospitalization cost of unintentional falls at US $19 927 (SD US $25 561) using 550 fall related hospitalizations from 153 000 hospital admission records.17 Hip and femur fracture were the most costly consequence of a fall (mean (SD) US $20 244 (US $14 577)). 17 One study classified hospitalized falls and mean hospital charges by type of fall and age among residents of California (184 798 fall events in 1995 to 1997, people aged >60 years).18 In all age cohorts 65 years and older, the most costly fall was from one level to another and this was followed by slipping, tripping or stumbling on the same level. The mean hospital charges of all falls was US $24 424 (aged 65-74) and this ranged from US $22 205 (55-64 years) to US $22 948 (≥85 years). In Wisconsin, the average length of stay for hospital admission for a fall related injury was 5.5 days at a charge of US $15 248.19 2.3.4 United Kingdom – population based estimate of falls in older adults – two studies  One study estimated cost of falls of an ambulance service 20 and a second study estimated cost of falls among older adults more broadly.21 Census data for the total population (N=41 338) serviced by the North East Ambulance Service in 2001 were used to estimate the cost of falls to the community in adults aged 65 years and older.20 After accounting for ambulance attendants ‘ time, the total community costs of attending to fallers in Newcastle was US $622 883 and averaged US $242 per fall.  Another population based study quantified the incidence and costs of unintentional falls in older adults by subgroups aged 60-64, 65-69, 70-74 and ≥75 years.21 The cost per 10 000 population was US $496 957 (60-64 years) and US $2.2 million (≥75 years). The authors estimated that falls cost the UK government US   92 $1.6 billion. Inpatient admissions (49.4%) and long term care costs (41%) were the major cost drivers for the cost of falls. 2.3.5 Australia – prospective cost of falls – one study22  Seventy-nine adults aged 65 years and older were followed prospectively for 3 months after a fall. Each participant completed a diary recording all community, informal care and expenses incurred due to each fall. Assessment of costs included medical services, community nursing, allied health, community domestic services, transport and other personal costs. In total over the 3-month followup period, falls cost between US $ 461 000 to US $ 487 000 depending on model assumptions. The mean cost per fall varied between US $6260 and US $6770 and the hospital costs accounted for 80 % of the total community costs. 2.3.6 Australia – population based study assessment of health system costs of falls – one study23  The total cost of falls to the health system in adults aged 65 years and older was US $143 million (18 706 ED presentations and 6222 hospital admissions). Assuming the rate of falls remains constant by age, the projected estimate of the cost of falls in 2021 was US $151 million. 2.3.7 Europe – prospective cohort study in older adults hospitalized after a fall – one study24  One study used two prospective cohorts with a 6 -month followup that included 690 adults aged 75 years and older who were hospitalized after presenting to the Emergency Department. Fallers were more frequently discharged to rehabilitation care and were more likely to be admitted to a nursing home compared with non fallers. Of all patients 10% (N=70) were hospitalized after a non-injurious fall. Institutional costs per day of follow were US $148 and long term costs per day were US $59. The long term care costs of fallers were higher than those of non fallers (p<0.001).     93 2.3.8 Europe – retrospective assessment of costs and health resource use – one study25  This retrospective study collected data for one year at a University teaching hospital with on-site orthopedic and geriatric medical services. Included in the cost calculations were the number of inpatient bed days occupied for fall related admissions, the number of rehabilitation bed days and the number of readmission bed days. From these cost items, the total cost was estimated at US $15.2 million. A hip fracture resulting from a fall was estimated at US $20 134. 2.3.9 Europe – population based study – one study26  A population-based study of 1 million residents of The Netherlands estimated the cost of fall related injuries due to benzodiazepine use (12.4 million prescription records). Individuals who took benzodiazepines were 60% times more likely to sustain fall injuries. The risk was almost 4-fold in adults aged 85 years and older. Fall related injuries that were attributable to benzodiazepine use cost a total of US $2.7 billion. 2.3.10 Europe – population based assessment of cost to health care system from injuries – one study27  One study assessed the impact of injuries in older adults on the health care system in 1985-1986. All individuals treated at Umea hospital were included in the study. The mean medical care costs in the elderly cost almost four times more than younger adults. Falls accounted for 63% of the total cost used by the older group (≥ 60 years, n=1311) compared with the younger (<60 years) group and cost almost 2.5 times more. In total, falls cost US $29.2 million. 2.3.11 Quality of included studies  For the quality assessment, I highlight the results for each of the 8 questions defined in Table 2-4. All of the 17 studies included a well defined study objective. Only four studies stated the viewpoint of the analysis. I did not allocate a point unless the viewpoint was specifically stated. Fourteen studies included a fall event that was well defined. Based the objective for each study, I determined that 12 studies included all the   94 important and relevant costs, 15 studies measures costs in appropriate and accurate physical units, 11 studies valued costs with credible references and 13 studies tracked  costs for one year or more. Only one study included a sensitivity analysis allowing for uncertainty in the estimates of costs. In summary, eight studies scored a ‗+‘ for at least six out of the eight item checklist. Of the eight item checklist, the two questions that most often received a negative score were: 1) Was the viewpoint of the analysis stated? and 2) Was a sensitivity analysis performed (i.e., allowance for uncertainty in the estimates of costs)? 2.4 Discussion  In many areas of health care the burden of disease is reasonably well described.28,29 There is a glaring need for such descriptive data for falls and fall related injuries. As might be expected for an emerging field (geriatrics), there was considerable variability in the cost of falls depending on whether estimates arose from population databases, cohort studies or analytic models for cost of falls. For example, some studies identified acute hospitalization for fall related injuries while others focused on costs of fatal, non-fatal, or injurious falls. Importantly, I found that the cost of falls and fall related injuries is likely in the vicinity of nine billion dollars annually in the US and this exceeds the total spending on pharmacotherapy in that nati on. 2.4.1 Range between population based studies  Costs for falls were presented in three broad categories: 1) total cost of falls to the health care system among community dwelling older adults, 2) cost of falls due to injury -related hospitalization only and 3) other specific costs (ambulance, due to benzodiazepine use). Costs varied greatly even within these categories. The cost per fall based on three scenarios in the US were: costs for hospital admission (US $22 269), emergency department visit alone (US $3890), or falls requiring office/outpatient visits alone (US $5040) for a one year time horizon.15 Falls are cited as the most costly injury among older people. From this type of study, falls accumulate an estimated annual cost of US $75 to US $100 billion annually.    95 2.4.2 Range in prospective studies  The estimates for the cost of falls from the prospective studies also demonstrated a large variation. Two key factors that affect this variation were the time horizon used in the study and the units used for cost of falls (cost per fall, cost per day, cost per person, total cos t per month). One European prospective study reported that after a non-injurious fall, overall institutional costs per day of followup were US $149 and long term costs per day were US $59.24 A 3-month prospective study reported that total costs of falls over the 3- month followup period were US $486 632 to US $461 118 (US $221 570 to US $233 829 per day, 2008 prices) in Western Australia.22 Consensus among the units reported and the cost items collected will facilitate comparability between studies. I suggest ways to standardize the results for individual studies into categories of units and cost items collected and reported (Tables 2-1 and 2-3). 2.4.3 Challenges in comparing costs across countries  The variations in results of economic analyses from different countries likely reflect inconsistencies among the following: year and currency costs were collected, the cost items collected in population level databases, reporting the population denominator, time horizon and analysis viewpoint used and reported. In the present study I compared costs by converting all estimates to a co mmon currency and year (converted to 2008 US dollars, Table 2-2). However, population size estimates were not provided for all studies, therefore, it was not possible to present cost per population in each case. Drummond and colleagues highlight a list of variables that could demonstrate variability across jurisdictions and they include a set of recommendations for dealing with analysis of individual patient data. 30 I recommend that further guidelines be developed that are specific to cost of illness studies. 2.4.4 Incidence and prevalence based costing approaches  I highlight the importance of considering the different costing approaches because each approach, even within the category of incidence or prevalence based costing, can provide substantially different results. 31,32 There are important methodological differences for incidence and prevalence based costing approaches for   96 cost of illness studies.33,34 Prevalence based approaches enable quantification of the annual burden of a disease such as falls. Thus, it is the method most often used for determining expected expenditures. A limitation of prevalence based costing is that it does not enable comparison of the value for money of specific interventions. In contrast, incidence based costing approaches are specific to falls; the researcher needs to identify the clinical outcomes such as fracture attributable to falls, the number of new falls, duration of illness associated with each fall complication, survival rates and prospective collection of health care resources used. In my systematic review, seven studies reported an incidence based costing approach. However of these seven, the only fracture that was costed out as a consequence of a fall was hip fracture. 2.4.5 Quality of included studies  The quality assessment highlighted that seven of the 17 studies met six or more of the eight criteria. Additionally, it provided a reference list of components that should be included in studies aimed at costing falls (Table 2-4). A strength of Drummond‘s Checklist is that it covers the key elements that should be included in an economic evaluation. A limitation of Drummond‘s Checklist for comparing studies is that it provides a qualitative, rather than quantitative assessment. 2.4.6 Conclusion  My study provides a solid evidence base for estimates of the economic burden of fall related injuries in older adults. Given that effective and cost-effective fall prevention strategies exist,6,35 the magnitude of the costs reported here should provide impetus to policymakers to examine public health interventions or focused intervention in high risk groups. The main factors driving the range reported for cost of falls include: 1) variations in the definition of falls and fall related injuries used, 2) clinical outcomes reported (for example hip fracture), 3) the cost items collected and the units reported, 4) the population denominator that cost estimates were based on, 5) variation in time horizon and 6) perspective of analysis. I recommend that   97 sensitivity analyses to vary key cost drivers be conducted to calculate the uncertainty of the point estimates. A consensus on these factors would enable better comparisons across cost of falls studies.    98 2.5 References  1.  Tinetti ME, Williams CS. Falls, injuries due to falls, and the risk of admission to a nursing home. N Engl J Med 1997;337(18):1279-84. 2.  Campbell AJ, Borrie MJ, Spears GF. Risk factors for falls in a community -based prospective study of people 70 years and older. J Gerontol 1989;44(4):M112-7. 3.  Tinetti ME, Speechley M, Ginter SF. Risk factors for falls among elderly persons living in the community. N Engl J Med 1988;319(26):1701-7. 4.  National Center for Injury Prevention and Control: statistics and activities. Int J Trauma Nurs 1998;4(1):18-22. 5.  Sattin RW, Lambert Huber DA, DeVito CA, Rodriguez JG, Ros A, Bacchelli S, et al. The incidence of fall injury events among the elderly in a defined population. Am J Epidemiol 1990;131(6):1028- 37. 6.  Davis JC, Robertson MC, Ashe MC, Liu-Ambrose T, Khan KM, Marra CA. Does a home-based strength and balance programme in people aged ≥ 80 years provide the best value for money to prevent falls? A systematic review of economic analyses of falls prevention interventions. British Journal of Sports Medicine 2009;Accepted: Online First. 7.  Moher D, Cook DJ, Eastwood S, Olkin I, Rennie D, Stroup DF. Improving the quality of reports of meta-analyses of randomised controlled trials: the QUOROM statement. Quality of Reporting of Meta-analyses. Lancet 1999;354(9193):1896-900. 8.  Drummond MF, Sculpher MJ, Torrance GW, O'Brien B, Stoddart GL. Methods for the economic evaluation fo health care programmes. Third edition. New York. United States of America: Oxford University Press, 2005. 9.  US Department of Labour. www.bls.gov/data/, 2009.   99 10.  Welte R, Feenstra T, Jager H, Leidl R. A decision chart for assessing and improving the transferability of economic evaluation results between countries. Pharmacoeconomics 2004;22(13):857-76. 11.  Englander F, Hodson TJ, Terregrossa RA. Economic dimensions of slip and fall injuries. J Forensic Sci 1996;41(5):733-46. 12.  Carroll NV, Slattum PW, Cox FM. The cost of falls among the community-dwelling elderly. J Manag Care Pharm 2005;11(4):307-16. 13.  Rizzo JA, Friedkin R, Williams CS, Nabors J, Acampora D, Tinetti ME. Health care utilization and costs in a Medicare population by fall status. Med Care 1998;36(8):1174-88. 14.  Stevens JA, Corso PS, Finkelstein EA, Miller TR. The costs of fatal and non-fatal falls among older adults. Inj Prev 2006;12(5):290-5. 15.  Finkelstein EA, Chen H, Miller TR, Corso PS, Stevens JA. A comparison of the case -control and case-crossover designs for estimating medical costs of nonfatal fall -related injuries among older Americans. Med Care 2005;43(11):1087-91. 16.  Alexander BH, Rivara FP, Wolf ME. The cost and frequency of hospitalization for fall -related injuries in older adults. Am J Public Health 1992;82(7):1020-3. 17.  Roudsari BS, Ebel BE, Corso PS, Molinari NA, Koepsell TD. The acute medical care costs of fall - related injuries among the U.S. older adults. Injury 2005;36(11):1316-22. 18.  Ellis AA, Trent RB. Do the risks and consequences of hospitalized fall injuries among older adults in California vary by type of fall? J Gerontol A Biol Sci Med Sci 2001;56(11):M686-92. 19.  Mahoney JE, Glysch RL, Guilfoyle SM, Hale LJ, Katcher ML. Trends, risk factors, and prevention of falls in older adults in Wisconsin. WMJ 2005;104(1):22-8. 20.  Newton JL, Kyle P, Liversidge P, Robinson G, Wilton K, Reeve P. The costs of falls in the community to the North East Ambulance Service. Emerg Med J 2006;23(6):479-81.   100 21.  Scuffham P, Chaplin S, Legood R. Incidence and costs of unintentional falls in older people in the United Kingdom. J Epidemiol Community Health 2003;57(9):740-4. 22.  Hall SE, Hendrie DV. A prospective study of the costs of falls in older adults living in the community. Aust N Z J Public Health 2003;27(3):343-51. 23.  Hendrie D, Hall SE, Arena G, Legge M. Health system costs of falls of older adults in Western Australia. Aust Health Rev 2004;28(3):363-73. 24.  Seematter-Bagnoud L, Wietlisbach V, Yersin B, Bula CJ. Healthcare util ization of elderly persons hospitalized after a noninjurious fall in a Swiss academic medical center. J Am Geriatr Soc 2006;54(6):891-7. 25.  Cotter PE, Timmons S, O'Connor M, Twomey C, O'Mahony D. The financial implications of falls in older people for an acute hospital. Ir J Med Sci 2006;175(2):11-3. 26.  Panneman MJ, Goettsch WG, Kramarz P, Herings RM. The costs of benzodiazepine -associated hospital-treated fall Injuries in the EU: a Pharmo study. Drugs Aging 2003;20(11):833-9. 27.  Sjogren H, Bjornstig U. Trauma in the elderly: the impact on the health care system. Scand J Prim Health Care 1991;9(3):203-7. 28.  Bansback N, Ara R, Ward S, Anis A, Choi HK. Statin therapy in rheumatoid arthritis: a cost- effectiveness and value-of-information analysis. Pharmacoeconomics 2009;27(1):25-37. 29.  Mycek S. The dire cost of obesity. Mater Manag Health Care 2004;13(4):18-21. 30.  Drummond MF, Barbieri M, Cook J, Glick HA, Lis J, Malik F, et al. Transferability of Economic Evaluations Across Jurisdictions: ISPOR Good Research Practices Task Force Report. Value In Health 2009;12(4):409-418. 31.  Yabroff KR, Warren JL, Banthin J, Schrag D, Mariotto A, Lawrence W, et al. Comparison of approaches for estimating prevalence costs of care for cancer patients: what is the imp act of data source? Med Care 2009;47(7 Suppl 1):S64-9.   101 32.  Yabroff KR, Warren JL, Schrag D, Mariotto A, Meekins A, Topor M, et al. Comparison of approaches for estimating incidence costs of care for colorectal cancer patients. Med Care 2009;47(7 Suppl 1):S56-63. 33.  Koopmanschap MA. Cost-of-illness studies. Useful for health policy? Pharmacoeconomics 1998;14(2):143-8. 34.  Hodgson TA, Meiners MR. Cost-of-illness methodology: a guide to current practices and procedures. Milbank Mem Fund Q Health Soc 1982;60(3):429-62. 35.  Gillespie LD, Robertson MC, Gillespie WJ, Lamb SE, Gates S, Cumming RG, et al. Interventions for preventing falls in older people living in the community. Cochrane Database Syst Rev 2009(2):CD007146.  bA version of this chapter has been published. DAVIS JC, Rober tson MC, Ashe MC, Khan KM, Marra CA 2009 Does a home based strength and balance program in people aged ≥ 80 years provide the best value for money  to prevent falls?: A systematic review of economic evaluations of falls prevention interventions British Journal of Sports Medicine Online First doi:10.1136/bjsm.2009.060988 102 3 Does a home based strength and balance program in people aged ≥ 80 years provide the best value for money to prevent falls?: A systematic review of economic evaluations of falls prevention interventions b 3.1 Introduction Physical activity provides a vast and well documented range of health benefits.1,2 In recent BJSM editorials, Church and Blair asked why, given the powerful, pleiotropic  medical benefits of exercise, it is not prescribed more in clinical settings.3,4 Similarly, Sallis5 argued that ―Exercise is medicine and physicians need to prescribe it!‖ A medical condition where there is both a substantial burden of disease6,7 and high quality evidence that exercise improves patient outcomes is in falls prevention among seniors. 8 Despite all of the foregoing elements, the clinical impression is that there is not universal adoption of ex ercise as a therapy by individual clinicians,9,10 nor widespread concerted public health spending on exercise programs to prevent falls in seniors. Before addressing whether this gap between what should happen and what does happen is due to economic factors (i.e., the cost to deliver falls prevention interventions) I briefly update the reader on the burden of falls and current evidence to prevent them.  Cohort studies have consistently demonstrated that 30% of those aged 65 years and older experience at least one fall each year and half of those fall recurrently.11,12 In the United Kingdom, falls by adults aged 60 years and older cost UK £981 million (at 2000 prices) and 59% of these costs were incurred by the National Health Service (NHS).13 The burden of disease is similar in other countries for which I could find data. 14-16  Some interventions that aimed to prevent falls in seniors have proven successful. To aid analysis of the various interventions, it is useful to categorize interventions into: 1) individually customized multifactorial interventions,17 2) regimens targeting the same multiple factors to all participants and 3) single factor    103  interventions. The initial individually customized multifactorial intervention targeted up to eight different risk factors for falls depending on the participant‘s individual risk profile; it reduced falls by 31% in one year. 18 In multifactorial interventions, reduction in the risk of falling ranges from not being significantly different from usual care, to 61% reduction.19 I distinguish this individually customized multifactorial approach from the strategy of providing multiple interventions at a community level whe re all individuals are exposed to a universal intervention irrespective of the person‘s risk profile.20 The third category is single factor intervention, for example strength and balance training, withdrawing use of psychotropic medication, home safety modification, and cataract surgery.21,22 Overall, in four trials of community living seniors, a progressive home based strength and balance retraining program reduced both falls and injuries by 35%. 8  Which of these approaches to falls prevention provide the best value for money? Given the ageing population, the large financial burden imposed by falls, and the scarcity of health care system resources, economic evaluations are increasingly important to assist health care decision makers in allocating resources. There are three main types of economic evaluations used  to determine value for money when comparing falls prevention interventions with usual care or when comparing two different interventions: cost-effectiveness analysis, cost-utili ty analysis and cost-benefit analysis. The common feature of all these analyses is the comparison of monetary units between competing alternative interventions; however, they differ in the approach taken to measure the health benefits of the intervention. In cost-effectiveness analysis, the benefits are measured using a clinically relevant outcome such as life years gained or number of falls prevented. The primary outcome of a cost-effectiveness analysis is the ICER. By definition, an ICER is the difference between the costs of providing the competing interventions divided by the dif ference in effectiveness (that is, the number of falls prevented) (Equation 1).23 ICER Cost Effect Equation 1    104 In cost-utility analyses, health benefits are measured by a QALY, and for cost-benefit analyses, in monetary units.  To my knowledge, no systematic review of economic evaluations has been published for falls prevention interventions. Therefore, to help health professionals and health managers make better decisions, I systematically reviewed economic evaluations of falls prevention studies to ans wer the question ―Which falls prevention strategy provides the best value for money?‖ 3.2 Methods 3.2.1 Literature search strategy In accordance with QUOROM24 and Cochrane Collaboration guidelines,25 I searched MEDLINE, PUBMED, EMBASE and, NHS EED databases to identify cost-effectiveness, cost-utility or cost-benefit studies based on fall prevention interventions published in the English language from 1945 through July 2008. I limited my search to the English language and to studies of people aged 60 years and older. Included in my search terms were keywords: fall prevention, economic evaluation, cost-effectiveness, cost-utility, and cost-benefit analysis. 3.2.2 Selection of studies I selected peer reviewed, published studies that included a comprehensive (full) cost-effectiveness, cost- benefit, or cost-utility analysis, that is, they reported the costs and consequences of two alternatives and an incremental cost-effectiveness, cost-benefit or cost-utility ratio. I included only studies in community dwelling older adults aged 60 years and older. I excluded studies if they were set in a hospital or long term care facility; were still in progress; or the intervention was aimed primarily at fracture prevention rather than falls prevention.    105 3.2.3 Abstraction of data Two raters (JCD, MCA) independently extracted data from each study and any discrepancies were discussed and reviewed by a third party (one of MCR, CAM). I noted the country, type of economic analysis (analytic model or evaluation within a clinical trial), journal, impact factor, study sample, intervention and comparator evaluated, length of intervention phase, effectiveness in terms of falls reduction, perspectives, type of currency, year of currency, analytic time horizon, results of sensitivity analyses, discount rate, cost items measured, intervention costs, and incremental cost-effectiveness, cost-benefit and, cost-utility ratios. 3.2.4 Data synthesis and analysis I emailed authors of the papers that are included in this systematic review for additional information that was not available from the publications themselves. Due to the heterogeneity of reported cost items, I was not able to complete a meta-analysis; therefore, I provide a qualitative description of outcomes. 3.2.5 Standardized cost outcomes The studies reported costs in different currencies and from different years. To attenuate this variation I express monetary values in two ways: (i) by year and currency as reported in the manuscript and (ii) converted to 2008 pounds sterling, by inflating the values to 2008 prices using the Consumer Price Index and then using Purchasing Power Parity values in 2008 to convert the currency to pounds sterling.26,27 The Consumer Price Index (CPI), specific to each country, is an indicator of consumer prices in that country. It is obtained by calculating, on a monthly basis, the cost of a fixed list of goods and services purchased by a ‗typical‘ consumer during a given month. The CPI is a widely used indicator of inflation and deflation. Purchasing power parity assumes there is equilibrium between exchange rates when the purchasing power is equal in the comparator countries. Therefore, the ratio of the comparator countries' price levels for the goods and services of interest should equal the exchange rate between the comparator countries. I used this ratio in order to convert all monetary values into pounds sterling.    106 3.2.6 Quality assessment I assessed the quality of each study using: 1) a checklist developed by Drummond and colleagues 23 and 2) the Quality of Health Economic Studies instrument,28,29  which were designed to evaluate all common types of health economic analyses (e.g. cost-effectiveness analysis, cost-utili ty analysis). The Quality of Health Economics Studies instrument was originally developed from a co mprehensive literature search of all existing checklists and guidelines for economic evaluations. Weightings for each question were determined by conjoint analysis based on survey results from an international panel of health economists.  The Quality of Health Economic Studies instrument comprises 16 criteria that cover presentation of study objectives, description of methods, and comprehensive reporting of results. I excluded item four which was not applicable for all the included studies, therefore the weighted total for the 15 items was a score of 99. I considered studies with a score of 75% or greater as ‗good‘ quality.29  Two raters (JCD, MCA) independently evaluated each study and any discrepancies were discussed and reviewed by a third party (one of MCR, CAM). One author (MCR) did not review or rate her own studies. 30- 33 3.3 Results 3.3.1 Overview of studies identified After critical review of the 29 full text manuscripts, nine studies met my inclusion criteria (Figure 3 -1). One study reported a cost-benefit analysis, seven a cost-effectiveness analysis and one reported both a cost- effectiveness and a cost-utility analysis (Table 3-1). Five of the cost-effectiveness analyses reported the incremental cost of delivering the intervention per fall prevented, and five reported a cost-effectiveness ratio which incorporated incremental health care costs related to falls during the  study, or total health care costs (Table 3-2). The studies differed in the particular cost items included (Table 3-3) and the methods used for    107 valuing these items. The cost-utility analysis used QALYs gained as the measure of effectiveness. The cost-benefit analysis used dollars to quantify both effectiveness and costs.  There were also distinct variations across studies with respect to time horizon, interventions evaluated, methods used for sensitivity analyses, and discounting rates. The comparators evaluated for all nine studies were similar and included social visits or usual care, which in five of the studies was assigned a monetary value of zero. Six studies scored 75% or more on the Quality of Health Economic Studies instrument (Table 3-4). Two studies that fell just below 75% did not provide a comprehensive presentation of the economic model or a full explanation of all assumptions made.    108 Figure 3-1. QUOROM flow diagram of selection of studies       109   Table 3-1. Characteristics of studies  Publication, Country, Type of trial or model Study sample  Intervention, Comparator (Length of intervention phase) Measures of effectiveness Perspective, Type of currency, Time period costs were measured Sensitivity analysis Discounting, Time horizon Beard 200635 Australia Controlled tria l Approximately 90 000 people aged 60 years, resident in North Coast region of New South Wales ‗Stay on Your Feet20‘ 1992- 1996 (community education, policy development, home hazard reduction, media campaigns and working with health care professionals), interstate control region (Method 1) and the state of New South Wales (Method 2) (5 years) Falls-related hospital admissions 1989 to 1996 calculated based on Method 1: hospital admission rates, aged 60 years, total inpatient costs averted AUD $7 107 703 Method 2: falls related diagnostic related group costs, aged 65 years, total inpatient costs averted AUD $6 184 530 Societal Australian dollars 1995-1996   No 8% Intervention 1992 to 1996 Hospital admissions 1989 to 1996  Campbell 200530 New Zealand Randomised controlled tr ial (2x2 factorial) 391 women and men, aged 75 years, severe visual impairment (visual acuity 6/24 or worse), recruited through Royal New Zealand Foundation of the Blind and low vision clin ics Home safety assessment and modification (n=198), no home safety program (n=193) (1 to 2 home visits by experienced occupational therapist)  41%  reduction in falls, 99 falls prevented in 1 year Societal New Zealand dollars 2004  Selective one- way using a range of estimates of cost items for intervention Not applicable 1 year Rizzo 199634 United States Randomised controlled tr ial 301 women and men, aged 70 years, 1 or more of 8 specified risk factors for falling  Targeted multifactorial intervention (n=153) (behavioural instructions, exercise programs, adjustment to medications, home safety), home visits, usual care (n=148) (delivered by physician, at home by nurse and physiotherapist, 3 months with additional 3 months maintenance phase) 31%  reduction in falls,18 66 falls prevented in 1 year Not stated US dollars 1993 Selective one- way using a range of estimates of cost items for intervention Not applicable 1 year Robertson 200132  233 women, aged 80 Exercise intervention41 (specific 32%  reduction in falls,42 64 falls Societal Selective one- Not applicable  110  Publication, Country, Type of trial or model Study sample  Intervention, Comparator (Length of intervention phase) Measures of effectiveness Perspective, Type of currency, Time period costs were measured Sensitivity analysis Discounting, Time horizon New Zealand Randomised controlled tr ial years, recruited from 17 general practices  set of muscle strengthening and balance retraining exercises prescribed at home by physiotherapist, 4 home visits and monthly phone calls) (n=116); social visits and usual care (n=117) (1 year) prevented in 1 year  New Zealand Dollar s 1995 way using a range of estimates of cost items for intervention 1 year Robertson 200131 New Zealand Randomised controlled tr ial 240 women and men, aged 75 years, recruited from 17 general practices Exercise intervention41 (specific set of muscle strengthening and balance retraining exercises prescribed at home by trained community nurse, supervised by physiotherapist, 5 home visits and monthly phone calls (n=121), usual care (n=119) (1 year) 46%  reduction in falls,42 29 falls prevented  Societal New Zealand Dollar s 1998 Selective one- way using a range of estimates of cost items for intervention Not applicable 1 year Robertson 200133 New Zealand Controlled tria l 450 women and men, aged 80 years, recruited from 32 general practices Exercise intervention41 (specific set of muscle strengthening and balance retraining exercises prescribed at home by 3 trained nurses, supervised by physiotherapist, 5 home visits and monthly phone calls (n=330), usual care (n=120) (1 year) 30%  reduction in falls,42 90.77 (pro rata) falls prevented Societal New Zealand Dollar s 1998 Selective one- way using a range of estimates of cost items for intervention Not applicable 1 year Sach 200738 United Kingdom Randomised controlled tr ial and cost-utility model 306 women, aged 70 years, bilateral cataracts  Expedited (4 weeks) first eye cataract surgery (n=148), usual care (routine 9-13 month wait list) (n=140) (routine cataract operation) 34%  reduction in falls,22 number of fa lls prevented not repor ted National Health Service, Personal Social Services Pounds sterling 2004 Probabilistic using a range of estimates of cost items and quality of life variability for intervention Not applicable (1 year results) 3.5%  (lifetime analysis) Salkeld 200036 Australia Randomised 530 women and men, age 65 years, 444 recruited before discharge from Routine occupational therapy home safety assessment and modification (n=264), usual 14%  reduction in falls,43 98 falls prevented Societal Australian Dollars 1997 Selective one- way using a range of Not applicable 378 days  110  Publication, Country, Type of trial or model Study sample  Intervention, Comparator (Length of intervention phase) Measures of effectiveness Perspective, Type of currency, Time period costs were measured Sensitivity analysis Discounting, Time horizon controlled tr ial selected hospital wards, 26 from outpatient clinics, 60 from day centres care (n=266) (1 home visit, fo llowup phone call 2 weeks later) estimates of cost items for intervention Smith 199837 Australia Decision analytic model Not applicable  Assessment of home hazards and provision of fa ll prevention devices (hypothetical intervention), usual care (estimated total 4 hours per person by nurse or occupational therapist) Assumed 25% 18 reduction in number of fa llers  Not stated Australian Dollars 1996 Selective one- way using a range of estimates of cost items for intervention Not applicable (1 year) 5%  (10 years)    111   112  Table 3-2. Outcome measures Publication, Type of economic evaluation Cost items measured  Intervention costs Incremental cost-effectiveness/cost-benefit ratio    Reported currency Pounds sterling (2008 prices) Beard 200635 Cost-benefit analysis Costs of intervention program (staff, printing, marketing, overheads), fall related inpatient hospital costs Total d irect costs of intervention program (public and private expenditure) AUD $805 579 Average benefit to cost ratio 8.5:1 (State Government), 13.75:1 (Commonwealth Government), and 20.6:1 (Australian community)  Campbell 200530 Cost-effectiveness analysis Intervention costs (training costs, recruitment, occupational therapists‘ time and transport, administration, services and equipment installed in homes, overhead costs) Total cost  NZ $64 337, mean (SD) NZ $325 (292) per participant Incremental cost per fa ll prevented for delivering intervention NZ $650 Incremental cost per fa ll prevented for delivering intervention UK £304 Rizzo 199634 Cost-effectiveness analysis Intervention costs (developmental and training costs, recruitment costs, overheads, equipment and staff related costs), fa ll related and total health care use (hospitalization, emergency department, outpatient, home care, skilled nursing facilities) Cost per participant mean (range) US $905 (588 to 1346) Using mean costs: US $1772 per fall prevented for intervention costs only, total health care cost per fa ll prevented <US $0, total health care cost per medical fall prevented <US $0 Using mean costs: UK £1724 per fall prevented for intervention costs only, total health care cost per fall prevented < UK £0, total health care cost per ‗medical‘ fall prevented < UK £0 Robertson 200132 Cost-effectiveness analysis Intervention costs (recruitment, program delivery, overhead), health care costs resulting from falls and total health care costs during tria l (actual cost of hospital admissions and outpatient services, estimates of general practitioner and other costs) Mean cost per person for 1 year NZ $173 Incremental cost per fa ll prevented for delivering intervention NZ $314 Incremental cost per fa ll prevented for delivering intervention UK £173 Robertson 200131 Cost-effectiveness analysis Intervention costs (training course, recruitment, supervision of exercise instructor, program delivery, overhead), actual hospital costs resulting from falls during tr ial Total cost NZ $52 299, mean cost per person for 1 year NZ $432 Incremental cost per fa ll prevented NZ $1803 (delivering intervention), NZ $155 (delivering intervention and hospital costs averted), participants ≥80 years <NZ $0 (delivering intervention and hospital costs averted) Incremental cost per fa ll prevented UK £942 (delivering intervention), UK £81 (delivering intervention and hospital costs aver ted), participants ≥80 years < UK £0 (delivering intervention and hospital costs aver ted) Robertson 200133 Cost-effectiveness analysis Intervention costs (training course, recruitment, supervision of exercise instructor, program delivery, overhead) Total cost NZ $137 878, mean cost per person for 1 year NZ $418 Incremental cost per fa ll prevented for delivering intervention NZ $1519 Incremental cost per fa ll prevented for delivering intervention UK £794  112  Publication, Type of economic evaluation Cost items measured  Intervention costs Incremental cost-effectiveness/cost-benefit ratio    Reported currency Pounds sterling (2008 prices) Sach 200738 Cost-effectiveness analysis and cost- utility analysis Secondary health care (cataract operation, be days, outpatient, emergency department, lower and upper limb fractures), primary health care (general practitioner visits, practice/district nurse visits), personal social services (home care, day care centre, residential and nursing home care, meals on wheels, special equipment), patient and carers‘ costs (home care, time costs) Mean (SD) cataract operation UK £672 Incremental cost per fa ll prevented UK £4390 (1 year), UK £35 704 per QALY gained (1 year), UK £13 172 per QALY gained (over participants‘ expected lifetime)  Incremental cost per fa ll prevented UK £4732 (1 year), UK £38 482 per QALY gained (1 year), UK £14 197 per QALY gained (over participants‘ expected lifetime)  Salkeld 200036 Cost-effectiveness analysis Hospitalization, other health care costs provided in an institutional setting (e.g. outpatients), other health care costs provided in the home (e.g. home nursing), informal care costs (e.g. personal care provided by a relative or friend and help around the home), home modification costs, occupational therapist (intervention costs) in subsample of 103 in the intervention group and 109 in the control group (last 212 recruited into tria l) Not repor ted (mean occupational therapist cost AUD $116, mean home modification costs AUD $7) Incremental cost per fa ll prevented AUD $4986 (hospital costs, other health care costs, occupational therapist, home modifications, informal care costs), participants with a previous fall <AUD 0 (sensitivity analysis, outliers removed) Incremental cost per fa ll prevented UK £3040 (hospital costs, other health care costs, occupational therapist,  home modifications, informal care costs), participants with a previous fall < UK £0 (sensitivity ana lysis, outliers removed) Smith 199837 Cost-effectiveness analysis Intervention cost (installation and recommendation of aids, 1 hour visit from occupational therapist,  travel time), nursing home care, hip fracture rehabilitation, home health care, hospitalization as a result of a fall injury Incremental cost per person AUD $172 Incremental cost per fa ll prevented AUD $1721 (intervention and fall related health care costs) Incremental cost per fa ll prevented UK £1052 (intervention and fall related health care costs)        114  Table 3-3. Cost items measured in economic evaluations  Cost item Beard 200635 Campbell 200530 Rizzo 199634 Robertson 200132 Robertson 200131 Robertson 200133 Sach 200738 Salkeld 200036 Smith 199837 Costs of implementing the intervention  Program development + * + * * * * *  Training  + + + + +  Participant enrollment  + + + + +  Staff  + + + + + +  + +  Equipment  + + + + +  + +  Administration + + + + + +  Other intervention delivery  costs (e.g. cataract  surgery, travel)  + + + + + +  +  Overhead costs + + + + + +   + Control group intervention costs  +† +† +† +† +†  + Fall related health care resource use  Hospital costs +  + + + +   +  Health care professional   service costs   + +     + Total health care resource use  Hospital costs +  + +   + +  Health care professional   service costs   + +   + + Personal out-of-pocket expenses    +   + + + Costs were repor ted for this item * Intervention program was developed or available prior to the trial † Costs of control group activities were valued at zero  115   Table 3-4. Quality of Health Economic Studies scores*  Criteria Beard 200635 Campbell 200530  Rizzo 199634  Robertson 200132 Robertson 200131 Robertson 200133 Sach 200738 Salkeld 200036 Smith 199837 1 - + + + + + + + + 2 - + - + + + + + - 3 + + + + + + + + - 4 5 - + + + + + + - + 6 + + + + + + + + + 7 + + + + + + + + + 8 - + + + + + + + + 9 - + + + + + + + - 10 + + + + + + + + - 11 + + + + + + + + - 12 + + - + + - - - - 13 + + - + + + + - + 14 - + - + + + + - + 15 + + + + + + + + - 16 + + + + + + + + - Percentage (sum/total score) 59 (58/99) 100 (99/99) 75 (74/99) 100 (99/99) 100 (99/99) 92 (91/99) 92 (91/99) 73 (69/99) 47 (47/99) + Study met the cr iterion - All or part of the criterion was not met *16 criteria of the Quality of Health Economic Studies (QHES) instrument: 1  Was the study objective presented in a clear, specific, and measurable manner? (7 points) 2  Were the perspective of the analysis ( societal, th ird party payer, etc.) and reasons for its selection stated? (4 points) 3  Were variable estimates used in the analysis from the best available source (i.e., randomised controlled trial—best, expert opinion—worst)? (8 points) 4  If estimates came from a subgroup analysis, were the groups prespecified at the beginning of the study? (1 point) 5  Was uncertainty handled by: 1) statistical analysis to address random events; 2) sensitivity analysis to cover a range of assumptions? (9 points) 6  Was incremental analysis per formed between alternatives for resources and costs? (6 points) 7  Was the methodology for data abstraction (including the value of health states and other benefits) stated? (5 points) 8  Did the analytic horizon allow time for all relevant and impor tant outcomes? Were benefits and costs that went beyond one ye ar discounted (3%-5% ) and justification given for the discount rate? (7 points) 9  Was the measurement of costs appropriate and the methodology for the estimation of quantities and unit costs clearly described? (8 points) 10  Were the primary outcome measure(s) for the economic evaluation clearly stated and were the major shor t term, long term, and negative outcomes included? (6  points) 11    Were the health outcomes measures/scales valid and reliable? I f previously tested valid and reliable measures were n ot available, was justification given for the       measures/scales used? (7 points)  116   12  Were the economic model (including structure), study methods and analysis, and the components of the numerator and denominato r displayed in a clear, transparent manner? (8 points) 13  Were the choice of economic model, main assumptions, and limitations of the study stated and justified? (7 points) 14  Did the author(s) explicitly discuss direction and magnitude of potentia l biases? (6 points) 15  Were the conclusions/recommendations of the study justified and based on the study results? (8 points) 16  Was there a statement d isclosing the source of funding for the study? (3 points)   117  Table 3-5. Checklist for economic evaluations by Drummond and colleagues*  Criteria Beard 200635 Campbell 200530  Rizzo 199634  Robertson 200132 Robertson 200131 Robertson 200133 Sach 200738 Salkeld 200036 Smith 199837 1 - + + + + + + + + 2 + + + + + + + + - 3 + + + + + + + - - 4 + + + + + + + - + 5 + + + + + + + + + 6 + + + + + + + + + 7 + - - - - - + - + 8 + + + + + + + + - 9 - + + + + + + + + 10 - + - + + + + - + + Study met all items in the criterion - All items in the criterion were not met *10 criteria of the checklist for economic evaluations developed by Drummond and colleagues (each has several i tems): 1  Was a well defined question posed in answerable form? 2  Was a comprehensive description of the competing alternatives given? 3 Was the effectiveness of the programs or services established? 4 Were all the impor tant and relevant costs and consequences for each alternative identified? 5 Were costs and consequences measured accurately in appropriate physical units? 6 Were costs and consequences valued credibly? 7 Were costs and consequences adjusted for differential timing? 8 Was an incremental analysis of costs and consequences of a lternatives per formed? 9 Was the allowance made for uncertainty in the estimates of costs and consequences? 10 Did the presentation and discussion of study results include all issues of concern to users?   118 3.3.2 Individually customized multifactorial interventions – one cost-effectiveness study 3.3.2.1 Assessment, exercise, behaviour modification, medication In the only US cost-effectiveness analysis, the mean cost of delivering an individually targeted intervention was UK £880 (at 2008 prices) per participant.34 For all participants the targeted intervention was cost saving compared with usual care and social visits when the ICER was calculated using mean total health care costs or mean total cost for falls requiring medical care. In a subgroup analysis, the targeted intervention was cost saving for the 54% (156 of 288) of participants at high risk (defined as four or more of the eight specified risk factors for falls), whereas for participants at lower risk of falls (three or fewer risk factors), the ICER (total health care costs per fall prevented) was UK £2696. The high risk group represented 8% (156 of 1950) of those originally screened for eligibility for this trial; the cost of screening was incorporated in the reported cost-effectiveness ratios. 3.3.3 Multiple intervention at a community level The ―Stay on your Feet‖ intervention targeted all older adults within the No rth Coast of New South Wales, Australia with a falls related knowledge, attitudes, behaviours, and risk factor awareness campaign. 35 The authors performed a cost-benefit analysis using a different region and the whole state as the comparators. Although the programs required a large budget, both comparisons yielded a net monetary benefit to cost ratio for the intervention of 20.6 to 1. 3.3.4 Single factor interventions – five cost-effectiveness studies and one cost-utility study 3.3.4.1 Strength and balance training intervention Three cost-effectiveness analyses of a nurse- or physiotherapist-delivered home exercise program compared the program with usual care for a one year period.31-33 These two randomised controlled trials and one controlled trial in New Zealand recruited participants on the basis of age only. The mean cost of   119 delivering the exercise program for one year ranged from UK £173 in a research setting to UK £942 per fall prevented in a community health care setting. Importantly, the ICER for the intervention in the study incorporating the costs of hospital admissions averted as a result of the intervention, demonstrated that the Otago Exercise Program was not cost saving for all participants, but was cost saving for those aged 80 years and older.31 3.3.4.2 Three home safety interventions Community dwelling women and men in New Zealand aged ≥75 who were at fall risk because of severely impaired vision (acuity of 6/24 or worse) were randomised to receive a home safety assessment and modification intervention.30 The delivery of the home safety program by an experienced occupational therapist cost UK £304 per fall prevented in one year.  An Australian study compared falls prevented and costs in participants recently discharged from hospital who had a home safety intervention delivered by an occupational therapist, or no intervention.36 The ICER, which incorporated health, home, and community care costs, was UK £3040 per fall prevented. A sensitivity analysis with outliers removed demonstrated that the intervention was cost saving in the subgroup of participants (39%, 203 of 527) who had a fall in the previous year.  A decision analytic model with assumptions for fall, injury, and hospitalization rates obtained from the published literature was used to assess the cost-effectiveness of a hypothetical home assessment and modification program.37 The authors estimated the intervention compared with usual care would  cost UK £1052 per fall prevented over a one year period.37 Selective one-way sensitivity analyses indicated that the assumed probability of a fall and assumed effectiveness of the intervention were key variables that affected the magnitude of the ICER.   120 3.3.4.3 Expedited cataract surgery For the cost-utility analysis of expedited first eye cataract surgery, effectiveness was estimated from QALYs determined from the EuroQol-5D (EQ-5D) scores at baseline and after 6 months.38 The cost-utility ratios of UK £38 482 per QALY from the NHS perspective, and UK £34 911 from the personal social services perspective for one year, were above the currently accepted willingness to pay value of UK £20 000. However, when the costs and QALYs were modeled over the participants‘ expected lifetime, the incremental cost per QALY was within this limit at UK £14 197. The base case one year ICER incorporating primary and secondary health care, and social services, personal, and carer costs was UK £4732 per fall prevented. The authors also provided cost-effectiveness acceptability curves. 3.4 Discussion There have been comprehensive economic evaluations published from only nine RCTs of falls prevention strategies in community dwelling older adults. Some of the strategies included exercise while others did not. There were considerable differences in economic evaluation methodologies used across studies (Table 3-3). Before discussing the implications of my findings I review the methodological issues that must be taken into account to interpret this systematic review.  In eight studies that reported incremental cost per fall prevented, effective interventions included strength and balance retraining, cataract surgery and home safety interventions. Of these, the studies testing strength and balance retraining, which had the highest quality assessment scores, prevented the greatest number of falls at the least cost. A multifactorial program was cost saving in a narrow range of individuals with four or more of the eight specified risk factors for falls. The most favourable incremental and widely applicable cost-effectiveness ratio was UK £173 (at 2008 prices) per fall prevented for the Otago Exercise Program, although this trial was in a research setting.32    121 The ICERs for home safety interventions varied from UK £304 per fall prevented for delivering the program to people with severe visual impairment,30 to UK £3040 (incorporating all health care costs) for those recently discharged from hospital.36 Cataract surgery was the least favorable with a cost (incorporating all health care, personal and carer costs) per fall averted of UK £4732.38  There are often inherent logistical difficulties in comparing and combining the results of economic analyses because of differences in: the health system in the country of the study, the currency, cost items included, perspectives taken, the time frame for measurement of costs, and the methodology used to calculate incremental cost-effectiveness and cost-benefit ratios. The clinical trials were powered for falls and not for costs, and hospital costs particularly are very often skewed. I facilitate some comparability between studies and report cost per fall prevented for delivering the intervention where this was reported (converted to 2008 pounds sterling, Table 3-2). 3.4.1 Quality of included studies The Quality of Health Economic Studies instrument used for the quality assessment underscored five strong papers with scores of 92% and greater (Table 3-4). Additionally, it provided a reference list of components that economic evaluations should include.28,29 This instrument has three main limitations. It does not specifically examine details on study design and study population, or include an item on price adjustments for inflation or currency conversion. A further key limitation is that the scoring of all items is dichotomous. For example, a selective one-way sensitivity analysis of a small number of factors over implausible ranges is scored the same as a comprehensive one -way sensitivity analysis or a probabilistic sensitivity analyses. Given that six of the nine studies included in my systematic review were published prior to the routine use of probabilistic sensitivity analyses, I anticipated the dichotomous nature of the instrument would have minimal impact on my quality assessment. However, to mitigate this problem, I also used a checklist developed by Drummond and colleagues23 to support my findings from the Quality of   122 Health Economic Studies instrument. 3.4.2 How do economic evaluations aid in decision making? Economic analyses are important for policy makers, funders, and providers making decisions among falls prevention strategies.23 Cost-effectiveness analyses allow decision makers and clinicians to compare value for money from a multifactorial intervention with a multiple or a single factor approach. For example, the multifactorial intervention cost UK £2696 per fall prevented for delivering the one-off intervention.34 In contrast, the single factor strength and balance retraining program compared with a control group cost UK £942 per fall prevented for delivering the program in a community health care setting for one year.32 I encourage all authors of randomised controlled trials reporting effective falls prevention strategies to  report details of the cost items required to deliver the intervention (Table 3-3) and the associated ICER to allow comparison of value for money. Detailed reporting of all falls and cost items for 1) delivering the intervention, 2) fall-related health care resource utilization, and 3) total health care resource utilization, will enable comparison of the incremental cost per fall prevented for an intervention compared with the control activity. Detailed reporting of these three cost categories will provide information on value for money among different falls prevention strategies. These details are essential for decision making.  Policy makers are faced with difficult resource allocation choices among different medical conditions. Ideally these choices require cost-benefit or cost-utili ty analyses such that outcomes are on the same metric, but both these approaches are problematic for evaluating falls prevention interventions. One study using a population health approach valued the benefits in dollar terms based on the number of hospital admissions averted.35 Caution is needed when interpreting hospital costs in randomised controlled trials powered for falls and not hospital admissions. The EQ-5D was used to estimate incremental cost per QALYs gained after expedited cataract surgery, which was UK £14 197 over the participants‘ lifetime.38 However there are particular problems with using QALYs as an outcome for a complex intervention   123 potentially resulting in multiple benefits for older people.39 I have not found quality of life measures sensitive to change in the falls prevention studies despite the beneficial outcomes of the trials.8 3.4.3 Implications for future research I recommend that future economic evaluations be guided in part by the checklists available for assessing economic evaluations.23 Key components are: the establishment of effectiveness, comprehensive inclusion and reporting of all relevant costs and consequences for each comparator, accurate and reliable valuation of cost and consequences, discounting when appropriate, reporting an incremental analysis of costs and consequences, reporting a comprehensive one way and probabilistic sensitivity analysis, and a discussion of the results.23 One major problem with comparing the economic evaluations in my systematic review was that some ICERs incorporated intervention costs only, some included fall related costs and some included total health care costs. Therefore, I recommend that all studies report: 1) total health care resource utilization costs, 2) fall related health care resource utilization costs, and 3) intervention costs only to enable a base case comparison of value for money. If future economic evaluations of falls prevention strategies were to follow these recommendations, it would help reduce hete rogeneity among economic evaluations and facilitate comparison between studies. 3.4.4 Implications for policy-makers and clinical practice In practice, clinically appropriate risk reduction strategies will continue to determine the most appropriate intervention for a particular patient. However from a public health perspective, this systematic review suggests that single interventions (such as the Otago Exercise Program) targeted at high risk groups can prevent the greatest number of falls at the lowest incremental costs. This study also suggests that individual clinicians can be assured that prescribing strength and balance training programs is evidence - based – not only for effective fall prevention but also as an economically valid choice. Clinicians in the physical activity and medicine setting (i.e., various relevant physician groups, physiotherapists, clinical physiologists, athletic trainers) should advocate for this treatment to be included judiciously in patient   124 prescriptions. I acknowledge that just as pharmaceutical prescription can have adverse effects40 and is contraindicated in certain patients, clinical judgment needs to apply when prescrib ing exercise. Nevertheless, the Otago Exercise Program has proven to be both safe and effective in over 500 patient years of data in studies as well as in subsequent implementation in the community setting. 8,30-33 3.4.5 Conclusion In summary, as with every field of research, additional data will provide a more solid foundation for recommendations. Recommendations may need to be changed in time. However at this point in time, I conclude that single interventions (such as the Otago Exercise Program) targeted at high risk groups can prevent the greatest number of falls at the lowest incremental costs.     125 3.5 References 1.  Booth FW, Chakravarthy MV, Gordon SE, Spangenburg EE. Waging war on physical inactivity: using modern molecular ammunition against an ancient enemy. J Appl Physiol 2002;93(1):3-30. 2.  Booth FW, Gordon SE, Carlson CJ, Hamilton MT. Waging war on modern chronic diseases: primary prevention through exercise biology. J Appl Physiol 2000;88(2):774-87. 3.  Church TS, Blair SN. When will we treat physical activity as a legitimate medical therapy...even though it does not come in a pill? Br J Sports Med 2009;43(2):80-1. 4.  Blair SN. Physical inactivity: the biggest public health problem of the 21st ce ntury. Br J Sports Med 2009;43(1):1-2. 5.  Sallis RE. Exercise is medicine and physicians need to prescribe it! Br J Sports Med 2009;43(1):3- 4. 6.  Tinetti ME, Williams CS. Falls, injuries due to falls, and the risk of admission to a nursing home. N Engl J Med 1997;337(18):1279-84. 7.  Wiktorowicz ME, Goeree R, Papaioannou A, Adachi JD, Papadimitropoulos E. Economic implications of hip fracture: health service use, institutional care and cost in Canada. Osteoporos Int 2001;12(4):271-8. 8.  Robertson MC, Campbell AJ, Gardner MM, Devlin N. Preventing injuries in older people by preventing falls: a meta-analysis of individual-level data. J Am Geriatr Soc 2002;50(5):905-11. 9.  Donaldson MG, Khan KM, Davis JC, Salter AE, Buchanan J, McKnight D, et al. Emergency department fall-related presentations do not trigger fall risk assessment: a gap in care of high-risk outpatient fallers. Arch Gerontol Geriatr 2005;41(3):311-7. 10.  Salter AE, Khan KM, Donaldson MG, Davis JC, Buchanan J, Abu-Laban RB, et al. Community- dwelling seniors who present to the emergency department with a fall do not receive Guideline   126 care and their fall risk profile worsens significantly: a 6 -month prospective study. Osteoporos Int 2006;17(5):672-83. 11.  Campbell AJ, Borrie MJ, Spears GF. Risk factors for falls in a community-based prospective study of people 70 years and older. J Gerontol 1989;44(4):M112-7. 12.  Tinetti ME, Speechley M, Ginter SF. Risk factors for falls among elderly persons living in the community. N Engl J Med 1988;319(26):1701-7. 13.  Scuffham P, Chaplin S, Legood R. Incidence and costs of unintentional falls in older people in the United Kingdom. J Epidemiol Community Health 2003;57(9):740-4. 14.  Carroll NV, Slattum PW, Cox FM. The cost of falls among the community -dwelling elderly. J Manag Care Pharm 2005;11(4):307-16. 15.  Hall SE, Hendrie DV. A prospective study of the costs of falls in older adults living in the community. Aust N Z J Public Health 2003;27(3):343-51. 16.  Hendrie D, Hall SE, Arena G, Legge M. Health system costs of falls of older adults in Western Australia. Aust Health Rev 2004;28(3):363-73. 17.  Gates S, Fisher JD, Cooke MW, Carter YH, Lamb SE. Multifactorial assessment and targeted intervention for preventing falls and injuries among older people in community and emergency care settings: systematic review and meta-analysis. BMJ 2008;336(7636):130-3. 18.  Tinetti ME, Baker DI, McAvay G, Claus EB, Garrett P, Gottschalk M, et al. A multifactorial intervention to reduce the risk of falling among elderly peop le living in the community. N Engl J Med 1994;331(13):821-7. 19.  Close J, Ellis M, Hooper R, Glucksman E, Jackson S, Swift C. Prevention of falls in the elderly trial (PROFET): a randomised controlled trial. Lancet 1999;353(9147):93-7. 20.  Kempton A, Van Beruden E, Sladden T, Garner E, Beard J. Older people can stay on their feet: final results of a community-based falls prevention programme. Health Promotion International 2000;15(1):27-33.   127 21.  Cumming RG. Intervention strategies and risk-factor modification for falls prevention. A review of recent intervention studies. Clin Geriatr Med 2002;18(2):175-89. 22.  Harwood RH, Foss AJ, Osborn F, Gregson RM, Zaman A, Masud T. Falls and health status in elderly women following first eye cataract surgery: a randomised controlled trial. Br J Ophthalmol 2005;89(1):53-9. 23.  Drummond MF, Sculpher MJ, Torrance GW, O'Brien B, Stoddart GL. Methods for the economic evaluation fo health care programmes. Third edition. New York. United States of America: Oxford University Press, 2005. 24.  Moher D, Cook DJ, Eastwood S, Olkin I, Rennie D, Stroup DF. Improving the quality of reports of meta-analyses of randomised controlled trials: the QUOROM statement. Quality of Reporting of Meta-analyses. Lancet 1999;354(9193):1896-900. 25.  Mulrow C, Oxman A. Cochrane reviewers' handbook. September 2008 [online]. Available at http:www.cochrane.org/resources/handbook/ Accessed January 2009. 26.  US Department of Labour. www.bls.gov/data/, 2009. 27.  Welte R, Feenstra T, Jager H, Leidl R. A decision chart for assessing and improving the transferability of economic evaluation results between countries. Pharmacoeconomics 2004;22(13):857-76. 28.  Chiou CF, Hay JW, Wallace JF, Bloom BS, Neumann PJ, Sullivan SD, et al. Development and validation of a grading system for the quality of cost-effectiveness studies. Med Care 2003;41(1):32-44. 29.  Ofman JJ, Sullivan SD, Neumann PJ, Chiou CF, Henning JM, Wade SW, et al. Examining the value and quality of health economic analyses: implications of utilizing the QHES. J Manag Care Pharm 2003;9(1):53-61.   128 30.  Campbell AJ, Robertson MC, La Grow SJ, Kerse NM, Sanderson GF, Jacobs RJ, et al. Randomised controlled trial of prevention of falls in people aged >=75 with severe visual impairment: the VIP trial. BMJ 2005;331(7520):817. 31.  Robertson MC, Devlin N, Gardner MM, Campbell AJ. Effectiveness and economic evaluation of a nurse delivered home exercise programme to prevent falls. 1: Randomised controlled trial. BMJ 2001;322(7288):697-701. 32.  Robertson MC, Devlin N, Scuffham P, Gardner MM, Buchner DM, Campbell AJ. Economic evaluation of a community based exercise programme to prevent falls. J Epidemiol Community Health 2001;55(8):600-6. 33.  Robertson MC, Gardner MM, Devlin N, McGee R, Campbell AJ. Effectiveness and economic evaluation of a nurse delivered home exercise programme to prevent falls. 2: Controlled trial in multiple centres. BMJ 2001;322(7288):701-4. 34.  Rizzo JA, Baker DI, McAvay G, Tinetti ME. The cost-effectiveness of a multifactorial targeted prevention program for falls among community elderly persons. Med Care 1996;34(9):954-69. 35.  Beard J, Rowell D, Scott D, van Beurden E, Barnett L, Hughes K, et al. Economic analysis of a community-based falls prevention program. Public Health 2006;120(8):742-51. 36.  Salkeld G, Cumming RG, O'Neill E, Thomas M, Szonyi G, Westbury C. The cost effectiveness of a home hazard reduction program to reduce falls among older persons. Aust N Z J Public Health 2000;24(3):265-71. 37.  Smith RD, Widiatmoko D. The cost-effectiveness of home assessment and modification to reduce falls in the elderly. Aust N Z J Public Health 1998;22(4):436-40. 38.  Sach TH, Foss AJ, Gregson RM, Zaman A, Osborn F, Masud T, et al. Falls and health status in elderly women following first eye cataract surgery: an economic evaluation conducted alongside a randomised controlled trial. Br J Ophthalmol 2007;91(12):1675-9.   129 39.  Harwood RH. Economic evaluations of complex services for older people. Age Ageing 2008;37(5):491-3. 40.  Wu RC, Laporte A, Ungar WJ. Cost-effectiveness of an electronic medication ordering and administration system in reducing adverse drug events. J Eval Clin Pract 2007;13(3):440-8. 41.  Accident Compensation Coorporation. Otago exercise programme to prevent falls in older adults. www.acc.com/otagoexerciseprogramme (accessed 30 July 2008) August 2003. 42.  Campbell AJ, Robertson MC, Gardner MM, Norton RN, Tilyard MW, Buchner DM. Randomised controlled trial of a general practice programme of home based exercise to prevent falls in elderly women. BMJ 1997;315(7115):1065-9. 43.  Cumming RG, Thomas M, Szonyi G, Salkeld G, O'Neill E, Westbury C, et al. Home visits by an occupational therapist for assessment and modification of environmental hazards: a randomized trial of falls prevention. J Am Geriatr Soc 1999;47(12):1397-402.   cA version of th is chapter will be submitted for publication. DAVIS JC, Marra CA, Robertson MC, Khan KM, Najafzadeh M, Ashe M, Liu-Ambrose T.  Economic evaluations of a dose-response resistance training in older women: a cost-effectiveness and cost-utility analysis. 130  4 Economic evaluation of dose-response resistance training in older women: a cost-effectiveness and cost-utility analysisc 4.1 Introduction  Falls and injuries resulting from falls in older adults represent a costly and significant public health burden.1- 3 Falls are the most frequent cause of injury-related health care resource utilization, morbidity and mortality among older people.4,5 In the United Kingdom, falls result in over UK £981 million (2000 prices) in annual direct medical expenditures.6  Progressive resistance training is an essential component of effective exercise programs for falls prevention.7,8 Liu-Ambrose previously demonstrated that progressive, high intensity resistance training significantly reduced falls risk score by 57%.9 Resistance training also has benefits for balance, cognition, body composition, and cardiovascular health.7 9-13 Resistance training may save money by preventing health care resource use by combating sarcopenia. Consequences of sarcopenia include increased fa lls risk secondary to impaired strength, balance, and mobility. The Otago Exercise Programme, a home based strength and balance retraining program was cost saving for those aged 80 years and older. 14 The ICER in terms of the cost per fall prevented has been reported for a combined program o f resistance training and balance training as compared to usual care.14-16  What is not known is whether there is a threshold dose of resistance training aimed at combating sarcopenia and reducing falls risk in women aged 65-75 years that will reduce total health care resource utilization and thus provide the best value for money. Therefore, I included in my study design a concurrent, prospective economic analysis with individual level data on cost and effectiveness  outcomes collected   131 during the Brain Power trial.17 My primary objective was to determine the ICER (cost per fall prevented) of once weekly or twice weekly resistance training compared with twice weekly balance and tone classes (comparator). I modeled the comparator program on a popular provincial -wide exercise program designed to reduce falls among seniors with low bone mass (the Osteofit program). 4.2 Methods 4.2.1 Overview of economic evaluation  The data for my cost-effectiveness and cost-utility analyses were from an intention to treat analysis of the Brain Power clinical trial.17 I used a Canadian health care system perspective, and a 9-month time horizon for the economic evaluation from the 12-month the clinical trial. The main outcome for my primary (cost- effectiveness) analysis was the incremental cost per fall prevented and for my secondary (cost-utility) analysis, the main outcome was the incremental cost per QALY gained.  I previously reported study design, participant recruitment, randomization, demographics (including weight, height, average waist girth, average hip girth, waist to hip ratio, physiological profile assessment and functional comorbidity index questionnaire), methods and results of the Brain Power trial. 17 Briefly, the study sample included 155 community dwelling women aged 65 to 75 years. Participants enrolled in Brain Power were: aged 65 to 75 years, a Mini Mental State Examination (MMSE) ≥24 (i.e., cognitively intact) and visual acuity 20/40 or better with or without corrective lenses. Participants excluded: were unable to write and speak English, were partaking in resistance training in the last six months, had a current medical condition for which exercise is contraindicated, had a neurodegenerative disease, were taking cholinesterase inhibitors, being treated currently for depression or on hormone replacement therapy during the previous 12 months. The secondary objective of Brain Power that related to my economic evaluation was to determine whether once weekly resistance training versus twice weekly resistance training significantly reduced falls compared with twice weekly balance and tone classes.   132 4.2.2 Costs  I used a HRU questionnaire to track health care resource utilization prospectively for each participant for 9 months of the 12-month study period. The major resource categories were: any visits to health care professionals (including general practitioners, specialists, physiotherapists etc); all visits, admissions or procedures carried out in a hospital; and laboratory and diagnostic tests. I calculated the costs of delivering the once weekly resistance training, twice weekly resistance training and twice weekly balance and tone (comparator group) interventions. My base case analysis considered the costs of all health care resource use and my sensitivity analyses included only fall related health care resource costs. I excluded research protocol driven costs from my analysis as these do not reflect the cost of implementation in a real wo rld setting.  For each component of HRU, I assigned a unit cost. All costs for admission to hospital were based on the fully allocated cost model of a tertiary care hospital, Vancouver General Hospital. For unit costs of health care professionals, I based costs on fee for service rates from the British Columbia Medical Services Plan 2009 price list. Unit costs for specialized services such as physiotherapy, chiropractic or naturopathic medicine were taken from the BC Association website for each specialty.  I inflated or deflated (where appropriate) costs to 2008 Canadian dollars using the consumer price index reported by Statistics Canada. Discounting was not relevant given my analytic time horizon was less than 12 months. 4.2.3 Effectiveness outcomes  I used monthly fall diary calendars to track all falls for each participant during the 12-month study period. Given that I collected health care costs for 9 months, I calculated the total number of falls prevented at 9 months for once weekly resistance training compared with twice weekly balance and tone (comparator) and the total number of falls prevented at 9 months for twice weekly resistance training compared with twice weekly balance and tone (comparator). I administered the SF-36 and the EQ-5D at baseline, 6 months and   133 trial completion (12 months) and from these calculated the total QALYs lost or gained at 6 and 12 months for the three participant groups. I used linear regression to calculate the incremental QALYs for each participant adjusted for baseline utility score. All statistical analyses were carried out using STATA version 10.0. 4.2.4 Adverse events and mortality  Details on adverse events have been reported.17 Fourteen (29.8%) women in the once weekly resistance training group developed musculoskeletal complaints, five (10.9%) in the twice weekly resistance training group and four (9.5%) in the twice weekly balance and tone group. The one exercise related fall in the twice weekly balance and tone group did not result in health care resource utilization. 4.2.5 Handling missing data In the Brain Power study, 12.9% of participants had  incomplete followup during the 52-week intervention. Missing cost data in particular can introduce substantial bias into the estimation of costs because cost data are often highly skewed.18-21 I calculated the cost and effectiveness estimates for available cases (dropping observations with missing values), complete case sets and an imputed data set.  I followed recommendations for multiple imputation of missing cost and effectiveness data. 18-21 For all discrete time points, I used a combination of multiple imputation and bootstrapping to estimate uncertainty caused by missing values and I report both an available case analysis and a complete case analysis. Missing data from each followup period for each participant were determined separately for both cost and effectiveness outcomes. I imputed missing EQ-5D, SF-6D and health care resource use values at each time point. For multiple imputation, I used the ice procedure in STATA. For each missing value, I generated five possible values using multiple linear regression. Covariates included age, group, baseline utility score, and the weight and value of the missing variable in the preceding period. The final imputed value was the mean of the five data sets created.   134 4.2.6 Cost-effectiveness analysis  I calculated the ICER for both once weekly and for twice weekly resistance training compared with twice weekly balance and tone classes (comparator). I used nested imputation and nonparametric bootstrapping to model uncertainty around the estimates for costs and effectiveness. For each of the five cycles, I imputed missing values and bootstrapped the complete dataset. For each cycle of imputation and bootstrapping, I calculated the total health care resource use cost, fall related resource use cost, QALY and number of falls per participant by group allocation. I averaged results of each cycle of imputation for participants in each of the three participant groups. I evaluated the contribution of e ach cost item in relation to the total health care resource use estimated for each group. I used plots on the cost-effectiveness plane and cost-effectiveness acceptability curves generated based on 5000 iterations of nested imputation/bootstrapping using Fiellers‘ method to generate 95% confidence ellipses for the joint distribution of cost and effectiveness outcomes.22 4.2.7 Cost-utility analysis  I expressed the differences in mean costs and health outcomes in each group by reporting the incremental cost per QALY. Given that the health benefit (i.e.,  QALY) difference was close to zero, I used 5000 bootstrapped replications of mean cost and QALY differences.23 I used these to generate a cost-utili ty acceptability curve to estimate the probability that once weekly resistance training or twice weekly resistance training is considered cost effective compared with twice weekly balance and tone classes over a range of willingness to pay values.24 4.2.8 Sensitivity analysis  I applied multiple imputation, bootstrapped confidence interval estimation, adjustment for imbalances in baseline utility and bootstrapped estimates of the incremental cost-effectiveness and cost-utility ratios. Included in my sensitivity analysis were both deterministic and probabilistic assumptions. For example, I restricted my data to a complete case analysis, thus including only participants for whom I had complete   135 cost and effectiveness data to eliminate uncertainty caused by missing data. I analys ed total health care resource utilization and fall related health care resource utilization costs separately in my sensitivity analysis. 4.3 Results  I present baseline study characteristics in Table 4-1. After 12 months, compared with the balance and tone group the unadjusted incidence rate ratio (IRR) for the once weekly resistance training group indicated a 27% (IRR: 0.73, CI: 0.44-1.23) nonsignificant reduction in falls and a 12% (IRR: 0.88, CI: 0.67-1.16) nonsignificant reduction in the twice weekly resistance training group. After 9 months there were 30 falls in the once weekly resistance training group, 38 (32 excluding outlier) in the twice weekly training group and 38 in the twice weekly balance and tone group. Table 4-1. Characteristics of participants at baseline Characteristic Twice weekly balance and tone (n=49) Once weekly resistance training (n=54) Twice weekly resistance training (n=52)  Mean (SD) Mean (SD) Mean (SD) Age, years 70.0 (3.3) 69.5 (0.2) 69.4 (3.0) Weight, kg 67.0 (11.5) 69.2 (16.2) 72.1 (18.8) Height, cm 161.0 (6.9) 160.9 (7.0) 162.8 (6.5) Hip girth, cm 102.1 (9.1) 105.4 (11.6) 105.2 (13.8) Waist girth, cm 83.9 (10.1) 86.8 (14.4) 88.1 (13.7) Waist to hip ratio 0.8 (0.1) 0.8 (0.1) 0.8 (0.1) Physiological Profile Assessment 0.2 (0.9) 0.3 (1.0) 0.2 (1.0) Functional Comorbidity Index score 2.2 (1.7) 1.9 (1.7) 2.3 (1.6)   136 4.3.1 Health care use and costs  Complete health care resource utilization data were provided by 50 (92%) participants a t baseline, 51 (94%) participants at 6 months and 49 (90%) at 12 months in once weekly resistance training, 49 (94%) participants at baseline, 50 (96%) participants at 6 months and 48 (92%) participants at 12 months in twice weekly resistance training and 45 (91%) participants at baseline, 47 (95%) participants at 6 months and 45 (92%) participants at 12 months in twice weekly balance and tone respectively. There were no differences in response rates or drop outs among the three participant groups. Unit cos ts for health care cost items are provided in Table 4-2. The mean total health care costs were significantly lower for the once weekly resistance training and twice weekly resistance training groups compared with twice weekly balance and tone classes (p<0.05) (see Table 4-3). Table 4-2. Unit costs for each component of resource utilization Item Value 2008 CAD$ Unit Reference Cost of delivering twice weekly balance and tone 706.12 Cost per person year Study records Cost of delivering once weekly resistance training 353.06 Cost per person year Study records Cost of delivering twice weekly resistance training 706.12 Cost per person year Study records Health care professional visit, mean (standard deviation)  111 (124) Cost per visit 2009 Medical services plan Admission to hospital 606 Cost per day 2005 Vancouver General Hospital fully allocated cost model* Emergency Department presentations  39 Cost per hour 2005 Vancouver General Hospital fully allocated cost model* Laboratory procedures, mean (standard deviation) 51 (47) Cost per procedure 2009 Medical services plan * Taken from the fully allocated cost model at Vancouver General Hospital    137 Table 4-3. Results of base case analysis Item Twice weekly balance and tone Once weekly resistance training Twice weekly resistance training Cost of delivering program per person (2008 CAD $) 706 or 0 (usual care) 353  706  Fall related health care resource use cost (2008 CAD $) 161.80  546.61  183.55  Total health care resource use costs (2008 CAD $) 43571.23  42355.11  41408.41  Mean health care resource use cost (2008 CAD $) 1772 1379* 1684*, 1676 Number of falls per year over 9 months: 38 30 38, 32 excluding outlier Incremental cost per fall prevented based on:  Total health care resource use costs  Ý reference dominates dominates  Fall related costs Ý reference dominates dominates  Cost of program delivery Ý reference dominates dominates QALY mean (SD) based on:  SF-6D 0.697 (0.057) 0.706 (0.067) 0.698 (0.069)  EQ-5D 0.816 (0.187) 0.829 (0.211) 0.854 (0.119) Adjusted incremental QALY based on:  SF-6D ý 0 (reference) 0.003 0.003  EQ-5D ý 0 (reference) 0.084 0.179 Incremental cost per QALY based on:  SF-6D reference dominates dominates   EQ-5D reference dominates dominates * p<0.05 Ý ICER based on total HRU costs, fall related costs and cost of delivering programs ý Incremental QALYs are adjusted for the baseline utility using a linear regression model 4.3.2 Health outcomes  Complete data for the EQ-5D were provided at all three time points for 49 (90%) participants in once weekly resistance training, 47 (90%) participants in twice weekly resistance training and 45 (91%) in twice weekly balance and tone respectively. Complete data for the SF-6D were provided at all three time points by 51 (94%) in once weekly resistance training, 49 (94%) participants in twice weekly resistance training and 47 (95%) participants in twice weekly balance and tone groups respectively. There were no differences in response rates or drop outs between treatment groups. Mean QALYs calculated from the EQ-5D and SF-6D scores are provided in Table 4-3.   138 4.3.3 Adjusting QALYs for baseline utility in each group  After controlling for baseline EQ-5D levels, the incremental QALY after 12 months calculated using the EQ- 5D was 0.084 for the once weekly resistance training group and 0.179 for the twice weekly resistance training group compared with twice weekly balance and tone classes (Table 4-3).  Similarly after controlling for baseline SF-6D levels, the incremental QALY score over 12 months calculated using the SF-6D was 0.003 both for the group receiving once and the twice weekly resistance training group compared with the balance and tone group (Table 4-3). 4.3.4 Cost-effectiveness analysis  Based on the point estimates from my base case analysis I found that twice weekly resistance training was less costly and more effective than (i.e., thus, it dominates) twice weekly balance and tone classes (comparator) when I excluded one participant who was a recurrent faller. A falls histogram revealed this one outlier. This outlier who had a diagnosed hip condition experienced at least 8 falls over the 1-year period of observation. Based on the point estimates from my base case analysis I found that once weekly resistance training dominated twice weekly balance and tone classes. 4.3.5 Cost-utility analysis  Based on the point estimates from my base case analysis, I found that twice weekly resistance training dominated twice weekly balance and tone classes including and excluding one participant who was a recurrent faller. Figure 4-1a demonstrates that for twice weekly resistance training compared with twice weekly balance and tone, most of the bootstrapped cycles (>80% of the 4000 cycles) were represented in the southeast quadrant. Figure 4-1b demonstrates that for once weekly resistance training compared with twice weekly balance and tone, most of the bootstrapped cycles (>80% of the 4000 cycles) were also represented in the southeast quadrant. Given that my cost-utility acceptability curves were in the negative   139 willingness to pay zone, I do not report them here; an ICER less than zero indicate the intervention is cost saving.   Figure 4-1a. Cost effective plane depicting the 95% confidence ellipses of incremental cost and effectiveness for comparison between twice weekly resistance training (2RT) and twice weekly balance and tone (2BT, comparator); b. Cost effective plane depicting the 95% confidence ellipses of incremental cost and effectiveness for comparison between once weekly resistance training (1RT) and twice weekly balance and tone (2BT, comparator). QALY estimates are based on utility scores from the EQ-5D.  4.3.6 Sensitivity analysis  I found similar results for both my complete case analysis and my available case analysis (Table 4-4). For all scenarios in my sensitivity analysis, QALY estimates from the EQ-5D were in favour of the once weekly resistance training and twice weekly resistance training compared with twice weekly balance and tone classes. There were statistically significant mean cost savings for once weekly resistance training and twice weekly resistance training compared with twice weekly balance and tone classes (comparator) (Table 4-4). Decision makers have threshold amounts of money that they are willing to pay when comparing effective interventions with already existing community programs. Given that once weekly resistance training and twice weekly resistance training dominated twice weekly balance and tone classes (comparator), these two resistance training options are both favourable alternatives to balance and tone classes.   a b   140 Table 4-4. Results of one way sensitivity analyses Scenario Outcomes Twice weekly balance and tone Once weekly resistance training Twice weekly resistance training Available case analysis (n=103) Mean cost (2008 CAD $) 1812 1419 1599  Number of falls 24 18 20  Cost/fall prevented Reference dominates dominates Available case analysis (n=99) Mean cost (2008 CAD $) 1812 1443 1659  QALY 0.702 0.699 0.700  Adjusted QALY Reference 0.011 0.005  Cost/QALY (from SF-6D) Reference dominates dominates Available case analysis (n=84) Mean cost (2008 CAD $) 1880 1522 1665  QALY (EQ-5D) 0.816 0.814 0.855  Adjusted QALY Reference 0.009 0.018  Cost/QALY (from EQ-5D) Reference dominates dominates Complete case analysis (n=84) Mean cost (2008 CAD $) 1880 1522 1665  Number of falls 19 18 18  Cost/fall prevented Reference dominates dominates  QALY (SF-6D) 0.698 0.700 0.704  Adjusted QALY (from SF-6D) Reference 0.007 0.014  Cost/QALY (from SF-6D) Reference dominates dominates Only fall related costs included Cost per person (2008 CAD $) 709 363 710  Cost/fall prevented Reference dominates dominated, 0, excluding outlier Excluding 1 outlier in twice weekly balance and tone group Cost (2008 CAD $) 1772 1380 1676  Cost/fall prevented Reference dominates dominates   141 4.4 Discussion  From the Canadian health care system perspective, the ICER per fall prevented for both twice weekly and once weekly resistance training each dominates twice weekly balance and tone classes. I also noted a statistically significant improvement in the adjusted incremental QALY determined from the EQ-5D in the twice weekly resistance training group compared with the twice weekly balance and tone group at trial completion.  Three falls prevention strategies are cost saving in subgroups of older people. These include: (i) the home based Otago Exercise Program in people ≥80 years,14-16 (ii) an individually customized multifactorial program in those with four or more of the eight targeted fall risk factors, 25 and (iii) a home safety program in individuals who had a previous fall.26 These economic evaluation results from the Brain Power study demonstrated that both once weekly resistance training and twice weekly resistance training  dominated twice weekly balance and tone classes, a substantive finding compared with evaluations of other falls prevention strategies.  I aimed to quantify uncertainty firs t by using multiple imputation for missing values of the EQ-5D, SF-6D and health care resource utilization. I then used nonparametric bootstrapping to estimate the uncertainty around the ICERs. From both the probabilistic and selective one way sensitivity analyses, I found that the available case analysis, complete case analysis and subgroup analyses supported the conclusions that both once weekly resistance training and twice weekly resistance training dominated twice weekly balance and tone classes (comparator).  The time horizon of my study was limited to the 9 months that health care  resource utilization data were collected during the 12 month Brain Power study. I used falls data at nine months. Previous research   142 demonstrates that resistance training in older adults has long term health benefits that would be ideally captured by a longer time horizon.27  Ideally, a gold standard economic evaluation conducted alongside a clinical trial will include the following four characteristics: 1) a comparator commonly used in standard practice, 2) adequate power to assess homogeneity of economic results, 3) sufficient followup time to assess ful l health benefit and 4) appropriate time frame to aid in decision making and adoption.28 The balance and tone comparator is relevant in communities where balance and tone classes such as Osteofit, not specifically aimed a fall prevention are provided. Consequently, my evaluation may provide an overly conservative estimate of the health benefit of the once and twice weekly resistance training classes. Although once and twice weekly resistance training each dominated the twice weekly balance and tone classes, my conservative estimate should not result in any immediate change in decision making as there was no significant reduction in falls as a result of these programs.  The strength of my study is that I collected information on cost and effectiveness outcomes prospectively and thus minimized recall and response bias. Further, I collected health care resource utilization at three distinct time points. Analytic techniques consisted of multiple imputation and bootstrapping to estimate the uncertainty around the ICERs as a result of the missing values and small sample size. To my knowledge, this is the first economic evaluation to examine the best value for money of two different doses of resistance training compared with commonly available balance and tone classes in the community such as Osteofit.  Total health care resource utilization costs were significantly lower for both once weekly resistance training and twice weekly resistance training compared with twice weekly balance and tone classes. Twice weekly resistance training also resulted in a statistically significant improvement in HRQL compared with twice   143 weekly balance and tone classes. Among the three treatment options evaluated in the Brain Power study, both once weekly resistance training and twice weekly resistance training provide better value for money for fall prevention than balance and tone classes.    144 4.5 References 1.  Burge R, Dawson-Hughes B, Solomon DH, Wong JB, King A, Tosteson A. Incidence and economic burden of osteoporosis-related fractures in the United States, 2005-2025. J Bone Miner Res 2007;22(3):465-75. 2.  Tinetti ME, Williams CS. Falls, injuries due to falls, and the risk of admission to a nursing home. N Engl J Med 1997;337(18):1279-84. 3.  Wiktorowicz ME, Goeree R, Papaioannou A, Adachi JD, Papadimitropoulos E. Economic implications of hip fracture: health service use, institutional care and cost in Canada. Osteoporos Int 2001;12(4):271-8. 4.  Tinetti ME, Speechley M, Ginter SF. Risk factors for falls among elderly persons living in the community. N Engl J Med 1988;319(26):1701-7. 5.  Nevitt MC, Cummings SR, Hudes ES. Risk factors for injurious falls: a prospective study. J Gerontol 1991;46(5):M164-70. 6.  Scuffham P, Chaplin S, Legood R. Incidence and costs of unintentional falls in older people in the United Kingdom. J Epidemiol Community Health 2003;57(9):740-4. 7.  Orr R, Raymond J, Fiatarone Singh M. Efficacy of progressive resistance training on balance performance in older adults : a systematic review of randomized controlled trials. Sports Med 2008;38(4):317-43. 8.  Gillespie LD, Robertson MC, Gillespie WJ, Lamb SE, Gates S, Cumming RG, et al. Interventions for preventing falls in older people living in the community. Cochrane Database Syst Rev 2009(2):CD007146. 9.  Liu-Ambrose T, Khan KM, Eng JJ, Janssen PA, Lord SR, McKay HA. Resistance and agility training reduce fall risk in women aged 75 to 85 with low bone mass: a 6-month randomized, controlled trial. J Am Geriatr Soc 2004;52(5):657-65.   145 10.  Barrett CJ, Smerdely P. A comparison of community-based resistance exercise and flexibility exercise for seniors. Aust J Physiother 2002;48(3):215-9. 11.  Nelson ME, Fiatarone MA, Morganti CM, Trice I, Greenberg RA, Evans WJ. Effects of high- intensity strength training on multiple risk factors for osteoporotic fractures. A randomized controlled trial. JAMA 1994;272(24):1909-14. 12.  Orr R, de Vos NJ, Singh NA, Ross DA, Stavrinos TM, Fiatarone-Singh MA. Power training improves balance in healthy older adults. J Gerontol A Biol Sci Med Sci 2006;61(1):78-85. 13.  Timonen L, Rantanen T, Ryynanen OP, Taimela S, Timonen TE, Sulkava R. A randomized controlled trial of rehabilitation after hospitalization in frail older women: effects on strength, balance and mobility. Scand J Med Sci Sports 2002;12(3):186-92. 14.  Robertson MC, Devlin N, Gardner MM, Campbell AJ. Effectiveness and economic evaluatio n of a nurse delivered home exercise program to prevent falls. 1: Randomised controlled trial. BMJ 2001;322(7288):697-701. 15.  Robertson MC, Devlin N, Scuffham P, Gardner MM, Buchner DM, Campbell AJ. Economic evaluation of a community based exercise program to prevent falls. J Epidemiol Community Health 2001;55(8):600-6. 16.  Robertson MC, Gardner MM, Devlin N, McGee R, Campbell AJ. Effectiveness and economic evaluation of a nurse delivered home exercise program to prevent falls. 2: Controlled trial in multiple centres. BMJ 2001;322(7288):701-4. 17.  Liu-Ambrose T, Nagamatsu LH, Graf P, Beattie L, Ashe MC, Handy T. The effect of resistance training on brain function among community-dwelling senior women: a randomized controlled trial. Archives of Internal Medicine 2010;170(2):170-8. 18.  Briggs A, Clark T, Wolstenholme J, Clarke P. Missing... presumed at random: cost-analysis of incomplete data. Health Econ 2003;12(5):377-92.   146 19.  Manca A, Palmer S. Handling missing data in patient-level cost-effectiveness analysis alongside randomised clinical trials. Appl Health Econ Health Policy 2005;4(2):65-75. 20.  Oostenbrink JB, Al MJ. The analysis of incomplete cost data due to dropout. Health Econ 2005;14(8):763-76. 21.  Oostenbrink JB, Al MJ, Rutten-van Molken MP. Methods to analyse cost data of patients who withdraw in a clinical trial setting. Pharmacoeconomics 2003;21(15):1103-12. 22.  Laska EM, Meisner M, Siegel C. Statistical inference for cost-effectiveness ratios. Health Econ 1997;6(3):229-42. 23.  Briggs AH, Gray AM. Handling uncertainty when performing economic evaluation of health care interventions. Health Technol Assess 1999;3(2):1-134. 24.  Fenwick E, Claxton K, Sculpher M. Representing uncertainty: the role of cost-effectiveness acceptability curves. Health Econ 2001;10(8):779-87. 25.  Rizzo JA, Baker DI, McAvay G, Tinetti ME. The cost-effectiveness of a multifactorial targeted prevention program for falls among community elderly persons. Med Care 1996;34(9):954-69. 26.  Salkeld G, Cumming RG, O'Neill E, Thomas M, Szonyi G, Westbury C. The cost-effectiveness of a home hazard reduction program to reduce falls among older persons. Aust N Z J Public Health 2000;24(3):265-71. 27.  McCartney N, Hicks AL, Martin J, Webber CE. Long-term resistance training in the elderly: effects on dynamic strength, exercise capacity, muscle, and bone. J Gerontol A Biol Sci Med Sci 1995;50(2):B97-104. 28.  Glick HA, Doshi JA, Sonnad SA, Polsky D. Economic Evaluation in Clinical Trials. New York USA: Oxford University Press, 2007.  dA version of this chapter will be submitted for publication. DAVIS JC, Marra CA, Robertson MC, Khan KM, Naja fzadeh M, Liu-Ambrose T.  Sustainability of a dose-response resistance training in older women: a cost-effectiveness and cost-utility analysis. 147 5 Sustainability of a 12-month resistance training intervention in older community dwelling women: a cost-effectiveness and cost-utility analysisd  5.1 Introduction  The direct medical cost of fall related injuries in the US totaled US $0.3 billion (2000 pric es) for fatal falls and US $23 billion for non-fatal fall-related injuries.1 Falls among older adults are a major contributor to the economic burden on the health care system and are associated  with functional decline. Approximately 30% of community dwelling older adults fall at least once annually with women having almost a three -fold higher incidence of falls than men.2-4 Also, falls are a major health economic issue.5,6 Approximately 40-60% of falls among older adults result in injuries that require medical attention or hospital admissions.3,4,7,8 Therefore the burden of falls in terms of their consequences and costs is high with falls costing approximately UK £981 million annually (2000 prices).9  Although research has identified cost-effective falls prevention strategies10 – few studies have estimated health benefits and value for money beyond the time horizon of clinical trials. To date, there are nine economic evaluations published of effective falls prevention strategies among community dwelling older adults in peer reviewed journals.10 Of these, one study included a one year followup after the intervention and reported an economic evaluation at two years.11 One other study constructed a model with a 10 year time horizon to investigate the ICER of a hypothetical home safety falls prevention intervention; however, a major limitation was the assumption that the risk of falls was the same regardless of fall or fracture history.12     148 The combination of limited health care resources and the expansion of new health care interventions emphasize the need for economic evaluations of existing and new interventions within and between different health conditions to enable decision makers to better establish health care priorities. I know from my economic evaluation of the Brain Power study at 12 months (paper submitted for publication) that both once weekly and twice weekly resistance training were both cost saving compared with balance and tone classes (the comparator). What is not known is whether, over the longer term, health benefits and reductions in total health care resource utilization will be sustained. Therefore, I included in the study design a second year of followup with a concurrent, economic analysis with individual level data on cost and effectiveness outcomes.13 My primary objective was to determine the ICER (cost per falls prevented) and cost-utility ratio (cost per QALY gained) of once weekly or twice weekly resistance training compared with balance and tone classes over the two year followup period. 5.2 Methods 5.2.1 Overview of economic evaluation  The data for my cost-effectiveness and cost-utility analyses for the one year followup study were from 98 participants of the Brain Power clinical trial (135 participants completed the 12-month intervention).13 Using an intention to treat analysis, I used a Canadian health care system perspective and a two year time horizon for the followup study economic evaluation. I highlight that this study combines the data from the 12-month intervention period and the 12-month followup study. My primary outcome for the cost- effectiveness analysis was the incremental cost per fall prevented and for the cost-utili ty analysis was the incremental cost per QALY gained or lost. 5.2.2 Participant recruitment and sample  I previously reported study design, participant recruitment, randomization, demographics, methods and results for the 12-month Brain Power randomized controlled trial.13 Briefly, the original study sample included community dwelling women aged 65 to 75 years who were randomized to either: 1) once weekly   149 resistance training; 2) twice weekly resistance training; and 3) balance and tone (comparator). As previously described, I modeled the comparator program on a provincial -wide exercise program currently available to seniors in British Columbia designed to reduce falls among seniors with low bone mass (the Osteofit program). Twenty eight participants received the once weekly resistance training, 35 participants received the twice weekly resistance training and 35 participants received the balance and tone classes (the comparator) during the 12 months that they were enrolled in the study. No interventions were provided in the followup study (i.e., in year 2).  Participants enrolled in Brain Power were: aged 65 to 75 years, a MMSE ≥24 (i.e., cognitively intact) and visual acuity 20/40 or better with or without corrective lenses. I previously described exclusion criteria. For this followup study, I included the 98 participants who agreed to participate in monthly reporting of their HRQL, physical activity, falls and HRU over a second 12-month period.  The Brain Power study was approved by the Clinical Research Ethics Board of the University of British Columbia. 5.2.3 Costs  I used a questionnaire to track health care resource utilization over the second 12-months based on 12- month recall. The major resource categories included were: 1) appointments with health care professionals (including general practitioners, specialists, physiotherapists etc); 2) visits, admissions or procedures carried out in a hospital; and 3) laboratory and diagnostic tests. I included the costs of delivering the once weekly resistance training, twice weekly resistance training, and balance and tone (comparator group) interventions and HRU costs from the intervention period. For the followup study, the base case analysis considered the costs of all health care resource use over the second year of followup and the sensitivity analyses included fall related health care resource costs.  I assigned a unit cost for each component of health care resource utilization. All hospital admission related   150 costs were based on a fully allocated cost model of a tertiary care hospital. For health care professional unit service costs, I based costs on fee for service rates from the British Columbia Medical Services Plan 2009 price list. For specialized services such as physiotherapy, chiropractic or naturopathic medicine, I used unit costs from the BC Association website for each specialty. I inflated  (where appropriate) costs to 2009 Canadian dollars using the consumer price index reported by Statistics Canada. Discounting was not applied. 5.2.4 Effectiveness outcomes  Participants filled out and mailed in monthly fall diary calendars to track all falls dur ing the second year followup period. Further, participants filled out the SF-6D and the EQ-5D monthly for the followup period and mailed these forms in monthly. From these monthly assessments using these generic preference based utility instruments, I calculated the total QALYs at 2 years (i.e., one year after completing the Brain Power study) for the three experimental groups. I used multiple linear regression to calculate the incremental QALYs for each participant and adjusted for imbalances in baseline utility score. I did not use mixed effects models because there was no statistically significant differences in utility scores between the groups. I carried out all statistical analyses using STATA version 10.0. 5.2.5 Adverse events and mortality  Details on adverse events for the Brain Power study have been reported.13 There were no deaths during the followup study; however, I did not collect data on other adverse events given that there was no intervention during this time. 5.2.6 Handling missing data For the followup to the Brain Power study, 82.7%  (n=98) of participants had complete followup data for health care resource utilization,  (46) 47% of participants had complete followup data for the SF-6D at all 12 time points and 81 (83%) of participants had complete followup data for at least eight time points, and 41   151 (42%) of participants had complete followup data for the EQ-5D at all 12 time points, and 82 (84%) of participants had complete followup data for the EQ-5D for at least 8 of the 12 time points during the followup study. Given that missing cost data in particular introduces substantive bias into the cost estimation,14-17 I calculated the cost and effectiveness estimates using both an imputed data set and a complete case set.  Using the ice procedure in STATA, I followed recommendations for multiple imputation of missing cost and effectiveness data.14-17 For all discrete time points, I used a combination of multiple imputation and bootstrapping to estimate uncertainty caused by missing values. I determined the amount of missing data for the followup period at each time point and for each participant separately for both cost and effectiveness outcomes. For each time point, I imputed missing EQ-5D, SF-6D and health care resource use values and for each missing value, I generated five possible values using multiple linear regression. Covariates included intervention group, MMSE score, functional comorbid index, fall risk profile score, baseline utility score, and the weight and value of the missing variable in the preceding period.  The final imputed value was the mean of the five data sets created. 5.2.7 Cost-effectiveness analysis  I calculated the incremental cost per mean number of falls prevented for both once weekly and for twice weekly resistance training compared with balance and tone classes (comparator). I used nested imputation and nonparametric bootstrapping to model uncertainty around the estimates for costs and effectiveness. For each of the five cycles, I imputed missing values and bootstrapped the complete dataset. For each cycle of imputation and bootstrapping, I calculated the total health care resource use cost, fall related resource use cost, QALY and number of falls per participant by group allocation. I averaged results of each cycle of imputation for participants in each of the three participant groups. I evaluated the contribution of each cost item in relation to the total health care resource use estimated for each group. I used plots on the   152 cost-effectiveness plane and cost-effectiveness acceptability curves generated based on 5000 iterations of nested imputation/bootstrapping using Fiellers‘ method to generate 95% confidence ellipses for the joint distribution of cost and effectiveness outcomes.18 5.2.8 Cost-utility analysis  I expressed the differences in mean costs and health outcomes in each group by reporting the incremental cost per QALY gained or lost. Given that the health benefit (i.e., QALY) difference was close to zero, I used 5000 bootstrapped replications of mean cost and QALY differences.19 I used these to generate a cost-utili ty acceptability curve to estimate the probability that once weekly resistance training or twice weekly resistance training is considered cost effective compared with balance and tone classes over a range of willingness to pay values.20 5.2.9 Sensitivity analysis  I applied multiple imputation, bootstrapped confidence interval estimation, adjustment for imbalances in baseline utility and bootstrapped estimates of the incremental cost-effectiveness and cost-utility ratios. Included in my sensitivity analyses were both deterministic and probabilistic assumptions. For example, I restricted my data to a complete case analysis, thus including only participants for whom I had complete cost and effectiveness data to eliminate uncertainty caused by missing data. I analysed total health care resource utilization and fall related health care resource utilization costs separately in my sensitivity analyses. 5.3 Results  I present characteristics of participants after completion of the followup study (i.e., 2 years after randomization into the Brain Power study, see Table 5-1). Compared with the balance and tone group the unadjusted IRR for the once weekly resistance training group indicated a 47% (IRR: 0.53, CI: 0.17-1.70) nonsignificant reduction in falls and a 33% (IRR: 1.33, CI: 0.77-2.31) nonsignificant increase in falls in the   153 twice weekly resistance training group. During the second year of followup there were 17 falls in the once weekly resistance training group, 37 in the twice weekly training group and 20 in the balance and tone group. Table 5-1. Characteristics of participants at the end of the 12-month intervention period (Complete Case Analysis; N = 98) Characteristic Twice weekly balance and tone (n=28) Once weekly resistance training (n=35) Twice weekly resistance training (n=35)  Mean (SD) Mean (SD) Mean (SD) Age, years 71.6 (2.9) 71.7 (2.8) 71.2 (3.1) Weight, kg 67.0 (9.9) 71.3 (17.0) 71.7 (15.9) Height, cm 160.3 (7.3) 162.4 (7.3) 162.6 (6.7) Hip girth, cm 86.2 (8.3) 90.3 (14.6) 89.5 (11.4) Waist girth, cm 104.6 (7.9) 108.0 (12.7) 106.8 (11.5) Waist to hip ratio 0.8 (0.1) 0.8 (0.1) 0.8 (0.0) Physiological Profile Assessment 0.4 (0.7) 0.5 (0.9) 0.3 (0.7) Functional Comorbidity Index score 2.2 (1.6) 2.0 (1.6) 2.8 (1.9) MMSE, max score = 30 28.8 (1.5) 28.1 (1.2) 28.9 (1.2) 5.3.1 Health care use and costs  Of the 98 participants who took part in the followup study assessment, I obtained complete health care resource utilization data for 96 (98%). Unit costs for health care cost items detailed by treatment group are provided in Table 5-2. Overall, admission to hospital was the main driver of total HRU.               154  Table 5-2. Unit costs for each component of resource utilization Item Value 2009 CAD$ Unit Reference  Twice weekly balance and tone (n=28) Once weekly resistance training (n=35) Twice weekly resistance training (n=35)  Health care professional visit, mean (standard deviation)  783 (430) 622 (506) 729 (684) Cost per visit 2009 Medical services plan Admission to hospital, mean (standard deviation) 1526 (5059) 683 (2202) 588 (3257) Cost per day 2005 Vancouver General Hospital fully allocated cost model* Laboratory procedures, mean (standard deviation) 122 (134) 254 (612) 132 (207) Cost per procedure 2009 Medical services plan Total Health Resource Utilization, mean (standard deviation) 2580 (4998) 1126 (2005) 1591 (3179) Cost per 12-months 2009 Medical services plan Fall Related Costs Only Total Health care professional visit costs  496 0 1573 Cost per visit 2009 Medical services plan Total admission to hospital costs 0 0 84 Cost per day 2005 Vancouver General Hospital fully allocated cost model* Total laboratory procedure costs 0 0 221 Cost per procedure 2009 Medical services plan Total Health Resource Utilization costs 496 0 1878 Cost per procedure 2009 Medical services plan * Taken from the fully allocated cost model at Vancouver Gene ral Hospital          155 5.3.2 Health outcomes  For my followup study, there were no differences in drop outs or response rates between treatment groups. Mean QALYs and adjusted incremental QALYs calculated from the EQ-5D and SF-6D scores are summarized in Table 5-3. Table 5-3. Results of base case analysis. Item Twice weekly balance and tone Once weekly resistance training Twice weekly resistance training Number of falls per year over 9 months: 20 17 37 Incremental cost per fall prevented based on:  Total health care resource use costs  Ý reference dominates dominated  Fall related costs Ý reference dominates dominated QALY mean (SD) based on:  SF-6D 4.96 (0.41) 4.89 (0.57) 4.92 (0.44)  EQ-5D 5.45 (0.73) 5.40 (1.07) 5.49 (0.78) Adjusted incremental QALY based on:  SF-6D ý 0 (reference) -0.167 -0.207  EQ-5D ý 0 (reference) -0.051 -0.081 Incremental cost per QALY for fall related health resource utilization based on:   SF-6D reference 8707 4778   EQ-5D reference 28510 12210 Incremental cost per QALY for total health resource utilization based on:   SF-6D reference 2970 dominated   EQ-5D reference 9725 dominated * p<0.05 Ý ICER based on total HRU costs, fall related costs ý Incremental QALYs are adjusted for the baseline utility using a linear regression model 5.3.3 Adjusting QALYs for baseline utility in each group  After controlling for EQ-5D HSUVs at baseline, the incremental QALY at two years calculated using the EQ-5D was -0.051 for the once weekly resistance training group and -0.081 for the twice weekly resistance training group compared with balance and tone classes (Table 5-3). After controlling for SF-6D HSUVs at baseline, the incremental QALY at two years calculated using the SF-6D was -0.167 for the once weekly resistance training group and -0.207 for the twice weekly resistance training group compared with balance and tone classes (Table 5-3).   156 5.3.4 Cost-effectiveness analysis  I included all participants in my cost-effectiveness analysis based on an intention to treat analysis. Based on the point estimates from my base case analysis I found that twice weekly resistance training was less costly and less effective than balance and tone classes with an ICER of $CAD -58 per fall prevented (2009 prices). Based on the point estimates from my base case analysis I found that, at two years, once weekly resistance training dominated (i.e., cost less and was more effective) balance and tone classes. 5.3.5 Cost-utility analysis  Based on the point estimates for total health care resource use and QALYs calculated from the EQ-5D for my base case analysis, I found that the incremental cost-utility ratio of once weekly resistance training per QALY gained was CAD $28 510 compared with balance and tone. The incremental cos t-utility ratio of twice weekly resistance training per QALY gained was CAD $12 210 compared with balance and tone classes. For fall-related health care resource use, the incremental cost-utili ty ratio of once weekly resistance training per QALY gained was CAD $9725 compared with balance and tone classes. The twice weekly resistance training was dominated compared with balance and tone.  For Figures 5-1a and 5-1c, QALYs were estimated from the EQ-5D. Figure 5-1a demonstrates that for once weekly resistance training compared with balance and tone classes, half of the bootstrapped cycles (51% of the 5000 cycles in southeast quadrant) were represented in the southeast quadrant and half in the southwest quadrant. Figure 5-1c demonstrates that for twice weekly resistance training compared with balance and tone classes, most of the bootstrapped cycles (70% of the 5000 cycles in southwest quadrant) were also represented in the southwest quadrant. Briefly, CEA enables us to compare both costs and consequences of competing alternatives both within and across different health care interventions and health states. Hence, CEAs allow us to compare new technologies with existing technologies to determine which provide the best value for money. The primary outcome of a CEA is the ICER which can be plotted   157 on a cost-effectiveness plane to determine which strategies provide the best value for money (i.e., least costly and most effective) (Figures 5-1a-5-1d).   158    Figure 5-1a. Cost effective plane depicting the 95% confidence ellipses of incremental cost and effectiveness for comparison between once weekly resistance training (1RT) and balance and tone (2BT, comparator) with QALYs estimated from the EQ-5D; b. Cost effective plane depicting the 95% confidence ellipses of incremental cost and effectiveness for comparison between once weekly resistance training (1RT) and balance and tone (2BT, comparator) with QALYs estimated from the SF-6D. 1c. Cost effective plane depicting the 95% confidence ellipses of incremental cost and effectiveness for comparison between twice weekly resistance training (2RT) and balance and tone (2BT, comparator) with QALYs estimated from the EQ-5D; d. Cost effective plane depicting the 95% confidence ellipses of incremental cost and effectiveness for comparison between twice weekly resistance training (2RT) and balance and tone (2BT, comparator) with QALYs estimated from the SF-6D.     159 Based on the point estimates for total health care resource use and QALYs calculated from the SF-6D for my base case analysis, I found that the incremental cost-utility ratio of once weekly resistance training per QALY gained was CAD $8707 compared with balance and tone classes. For total health care resource use, the incremental cost-utili ty ratio of twice weekly resistance training per QALY gained was CAD $4778 compared with balance and tone classes. For fall-related health care resource use, the incremental cost- utility ratio of once weekly resistance training per QALY gained was CAD $2970 compared with balance and tone classes. The twice weekly resistance training strategy was dominated by the balance and tone intervention.  For Figures 5-1b and 5-1d, QALYs were estimated from the SF-6D. Figure 5-1b demonstrates that for once weekly resistance training compared with balance and tone classes, all of the bootstrapped cycles (100% of the 5000 cycles) were represented in the southwest quadrant. Figure 5-1d demonstrates that for twice weekly resistance training compared with balance and tone classes, most of the bootstrapped cycles (100% of the 5000 cycles) were also represented in the southeast quadrant (i.e., more effective and less costly). 5.3.6 Sensitivity analysis  I found similar results for both my complete case analysis compared with my imputed data set for the cost- effectiveness analysis (Table 5-4). Specifically, the incremental cost per fall prevented for the once weekly resistance training group dominated the balance and tone group, and twice weekly resistance training strategy was dominated by the balance and tone intervention. For the complete case analysis for the cost- utility analysis, I used only cases with data available for all 12 time points for the QALY calculation. For the cost-utili ty analysis using QALY estimates from the EQ-5D, twice weekly resistance training dominated balance and tone classes, and the ICER for once weekly resistance training was less costly and less effective than balance and tone classes. For the cost-utili ty analysis using QALY estimates from the SF6D,   160 the once and twice weekly resistance training strategies compared with balance and tone classes were less costly and less effective. Table 5-4. Results of one-way sensitivity analyses Scenario Outcomes Twice weekly balance and tone Once weekly resistance training Twice weekly resistance training Complete case analysis for cost- effectiveness analysis (n=90) Mean (SD) cost (2009 CAD $) 2425 (5001) 1179 (1153) 1481 (3296)  Number of falls 16 15 31  Cost/fall prevented Reference dominates dominated Complete case analysis for cost- utility analysis* (n=33) Mean (SD) cost (2009 CAD $) 4269 (7529) 1176 (1313) 938 (531)  QALY (SF-6D) 5.04 (0.44) 4.90 (0.71) 5.06 (0.51)  Adjusted QALY (from SF-6D) Reference -0.144 -0.127  Cost/QALY (from SF-6D) Reference 21 479 26 228  QALY (EQ-5D) 5.61 (0.87) 5.15 (1.50) 5.72 (0.83)  Adjusted QALY (from EQ-5D) Reference -0.0459 0.127  Cost/QALY (from EQ-5D) Reference 67 386 dominates * Complete Case Analysis for cost-utili ty analysis included only participants who had all 12 time points for the EQ-5D and the SF-6D.  5.4 Discussion  From the Canadian health care system perspective, the once weekly resistance training strategy dominated balance and tone classes, but not the twice weekly resistance training classes. Overall, I noted that health benefits obtained while participating in the 12-month intervention were not sustained for the twice weekly resistance training group, but were in part sustained for the once weekly re sistance training group. Three potential reasons for these findings include: 1) small sample size, 2) the once weekly resistance training group had the greatest level of leisure time physical activity indicating they were able to sustain a once weekly intensive program beyond the intervention period and 3) the twice weekly resistance training group   161 were not able to sustain such frequent activity post intervention and experienced the greatest decline as a result.  Decision makers have threshold amounts of money that they are willing to pay to introduce effective interventions (Figures 5-2a-d). Given that both resistance training interventions resulted in lower health care resource utilization costs, these two resistance training options may be considered economically attractive alternatives to balance and tone classes.   162    Figure 5-2a. Cost effective acceptability curve for once weekly resistance training (1RT) versus balance and tone with QALYs estimated from the EQ-5D; b. Cost effective acceptability curve for once weekly resistance training (1RT) versus balance and tone with QALYs estimated from the SF-6D; 2c. Cost effective acceptability curve for twice weekly resistance training (2RT) versus balance  and tone with QALYs estimated from the EQ-5D and; d. Cost effective acceptability curve for twice weekly resistance training (2RT) versus balance and tone with QALYs estimated from the SF-6D.   2a 2b 2c 2d   163 5.4.1 Comparison with other studies  Only one other economic evaluation has estimated the incremental cost per fall prevented beyond the time horizon of the clinical trial; this study used a lifetime time horizon. 21 In that study, the cost-effectiveness of implementing a home based strength and balance training intervention was lower after the second year compared with the first year of followup and the intervention was equally effective. 11 Specifically, the cost of implementing the exercise program for one and two years respectively was NZ $314 and NZ $265 (1995 prices) per fall prevented, and NZ $457 and NZ $426 per fall resulting in a moderate or serious injury prevented.  My economic evaluation results from the Brain Power study do not concur with this finding. Although the health benefit was sustained for the once weekly resistance training group, the benefit was not sustained for the twice weekly resistance training group in terms of falls prevented. There are three major factors that likely could explain the differences between studies: 1) Brain Power participants were high functioning and fell less often than their New Zealand counterparts, 2) the Brain Power study interventions consisted solely of resistance training in contrast to the New Zealand study that included strength and balance retraining, and 3) the cost items included in the economic evaluations differed. Most importantly, I incorporated total health care resource utilization costs in addition to the cost of p rogram implementation and delivery into the cost-effectiveness ratios. The New Zealand study did report both fall related and total health care resource utilization costs but did not incorporate these costs into their ICERs. 5.4.2 Uncertainty in findings  To quantify uncertainty, I used multiple imputation for missing values of the EQ-5D, SF-6D and health care resource utilization and nonparametric bootstrapping to estimate the uncertainty around the incremental cost-effectiveness/utility ratios. From both probabilistic and selective one way sensitivity analyses, I found that there were a few differences between the imputed data analysis, complete and available case   164 analyses. Further, I also explored uncertainty caused by the use of different preference based gene ric utility instruments to calculate QALYs. The main difference in the results between these two instruments was that using the EQ-5D, approximately 50% of the bootstrapped estimates straddled the zero axis; whereas when the SF-6D was used, the reduction in health benefit was clear with 100% of the bootstrapped estimates in the southwest quadrant (the intervention cost less but was less effective than the comparator). It is known that different instruments produce different HSUVs for the same patients and t hat the size and direction of differences between instruments can also vary between conditions.22,23 5.4.3 Time horizon  I followed participants for a total of two years upon recruitment into the Brain Power study. Altho ugh previous research suggested resistance training in older adults may have long term benefits, I did not find the benefits of resistance training to be sustained without active participation in once or twice weekly resistance training.24 5.4.4 Limitations  There are two main limitations to my economic evaluations. First, the large variability in the point estimates reflects the small sample size for the followup study. Even if a true difference existed, the Brain Power study was only powered to see a minimum 35% reduction in falls. Further, the comparator activity was based on classes commonly available to seniors in the community (i.e., Osteofit). This provides an extremely conservative estimate of effectiveness and there was no significant reduction in falls as a result of these programs at one or two years. Because only once weekly resistance training continued to dominate the balance and tone classes (in terms of falls prevented) at the two year mark, I do not recommend any immediate change in decision making.     165 5.4.5 Strengths  A unique aspect of my study is that I followed individuals beyond the duration of the clinical trial. This is the first long-term followup of a resistance training intervention in older adults. Further, I collected HSUV information monthly for the 1-year followup study using two different generic preference based utility instruments (i.e., SF-6D and EQ-5D) to provide the most comprehensive estimation of QALYs and to minimize recall bias. 5.4.6 Conclusions and future directions  There was a reduction in health care resource utilization costs and falls (nonsignificant) one year after completion of once or twice weekly resistance training compared with balance and tone classes. The fact that this reduction was not statistically significant does not undermine these results because it is  argued that rules of statistical inference are entirely irrelevant and arbitrary to the decisions that economic evaluations inform.25 Specifically, decisions should be based only on the mean net benefits (i.e., mean benefits and mean costs) regardless of the statistical significance of these differences.25 Therefore, in order for benefits of resistance training to be sustained beyond the duration of the clinical trial, I recommend that individuals continue these activities. Further, programs effective in reducing falls mostly include a balance retraining component, therefore, I would also recommend this addition.26 Although the point estimates indicate once weekly resistance training resistance training provides better value for money for fall prevention than balance and tone classes, I interpret this finding with caution given the wide variability surrounding this estimate.   166 5.5  References  1.  Stevens JA, Corso PS, Finkelstein EA, Miller TR. The costs of fatal and non-fatal falls among older adults. Inj Prev 2006;12(5):290-5. 2.  Campbell AJ, Borrie MJ, Spears GF. Risk factors for falls in a community -based prospective study of people 70 years and older. J Gerontol 1989;44(4):M112-7. 3.  Cummings SR, Nevitt MC, Browner WS, Stone K, Fox KM, Ensrud KE, et al. Risk factors for hip fracture in white women. Study of Osteoporotic Fractures Research Group. N Engl J Med 1995;332(12):767-73. 4.  Tinetti ME, Speechley M, Ginter SF. Risk factors for falls among elderly persons living in the community. N Engl J Med 1988;319(26):1701-7. 5.  Corso P, Finkelstein E, Miller T, Fiebelkorn I, Zaloshnja E. Incidence and lifetime costs of injuries in the United States. Inj Prev 2006;12(4):212-8. 6.  Rizzo JA, Friedkin R, Williams CS, Nabors J, Acampora D, Tinetti ME. Health care utilization and costs in a Medicare population by fall status. Med Care 1998;36(8):1174-88. 7.  Bergland A, Wyller TB. Risk factors for serious fall related injury  in elderly women living at home. Inj Prev 2004;10(5):308-13. 8.  Kannus P, Sievanen H, Palvanen M, Jarvinen T, Parkkari J. Prevention of falls and consequent injuries in elderly people. Lancet 2005;366(9500):1885-93. 9.  Scuffham P, Chaplin S, Legood R. Incidence and costs of unintentional falls in older people in the United Kingdom. J Epidemiol Community Health 2003;57(9):740-4. 10.  Davis JC, Robertson MC, Ashe MC, Liu-Ambrose T, Khan KM, Marra CA. Does a home-based strength and balance programme in peop le aged ≥ 80 years provide the best value for money to prevent falls? A systematic review of economic analyses of falls prevention interventions. British Journal of Sports Medicine 2009;Accepted: Online First.   167 11.  Robertson MC, Devlin N, Scuffham P, Gardner MM, Buchner DM, Campbell AJ. Economic evaluation of a community based exercise programme to prevent falls. J Epidemiol Community Health 2001;55(8):600-6. 12.  Smith RD, Widiatmoko D. The cost-effectiveness of home assessment and modification to reduce falls in the elderly. Aust N Z J Public Health 1998;22(4):436-40. 13.  Liu-Ambrose T, Nagamatsu LH, Graf P, Beattie L, Ashe MC, Handy T. The effect of resistance training on brain function among community-dwelling senior women: a randomized controlled trial. Archives of Internal Medicine 2010;170(2):170-8. 14.  Briggs A, Clark T, Wolstenholme J, Clarke P. Missing... presumed at random: cost-analysis of incomplete data. Health Econ 2003;12(5):377-92. 15.  Manca A, Palmer S. Handling missing data in patient-level cost-effectiveness analysis alongside randomised clinical trials. Appl Health Econ Health Policy 2005;4(2):65-75. 16.  Oostenbrink JB, Al MJ. The analysis of incomplete cost data due to dropout. Health Econ 2005;14(8):763-76. 17.  Oostenbrink JB, Al MJ, Rutten-van Molken MP. Methods to analyse cost data of patients who withdraw in a clinical trial setting. Pharmacoeconomics 2003;21(15):1103-12. 18.  Laska EM, Meisner M, Siegel C. Statistical inference for cost-effectiveness ratios. Health Econ 1997;6(3):229-42. 19.  Briggs AH, Gray AM. Handling uncertainty when performing economic evaluation of healthcare interventions. Health Technol Assess 1999;3(2):1-134. 20.  Fenwick E, Claxton K, Sculpher M. Representing uncertainty: the role of cost-effectiveness acceptability curves. Health Econ 2001;10(8):779-87. 21.  Sach TH, Foss AJ, Gregson RM, Zaman A, Osborn F, Masud T, et al. Falls and health status in elderly women following first eye cataract surgery: an economic evaluation conducted alongside a randomised controlled trial. Br J Ophthalmol 2007;91(12):1675-9.   168 22.  Brazier J. Valuing health States for use in cost-effectiveness analysis. Pharmacoeconomics 2008;26(9):769-79. 23.  Marra CA, Woolcott JC, Kopec JA, Shojania K, Offer R, Brazier JE, et al. A comparison of generic, indirect utility measures (the HUI2, HUI3, SF-6D, and the EQ-5D) and disease-specific instruments (the RAQoL and the HAQ) in rheumatoid arthritis. Soc Sci Med 2005;60(7):1571-82. 24.  McCartney N, Hicks AL, Martin J, Webber CE. Long-term resistance training in the elderly: effects on dynamic strength, exercise capacity, muscle, and bone. J Gerontol A Biol Sci Med Sci 1995;50(2):B97-104. 25.  Claxton K. The irrelevance of inference: a decision-making approach to the stochastic evaluation of health care technologies. J Health Econ 1999;18(3):341-64. 26.  Sherrington C, Whitney JC, Lord SR, Herbert RD, Cumming RG, Close JC. Effective exercise for the prevention of falls: a systematic review and meta-analysis. J Am Geriatr Soc 2008;56(12):2234- 43.   eA version of this chapter will be submitted for publication. DAVIS JC, Marra CA, Robertson MC, Khan KM, Najafzadeh M, Liu-Ambrose T.  A prospective comparison of generic preference based utility instruments (SF -6D and EQ-5D) and predictors of health care resource utilization in older women. 169 6 A prospective comparison of generic preference based utility instruments (SF-6D and EQ-5D) and predictors of health care resource utilization in older womene 6.1 Introduction Falls are the leading cause of chronic physical disability and are very common. Approximately 30% of community dwelling adults aged 65 years and older fall at least once per year. 1,2 Chronic physical disability is associated with decline in HRQL. Improvement of HRQL is one of the most important goals for managing a number of severe chronic conditions.3 To date HRQL and health status questionnaires such as the EQ- 5D have only been used in one economic evaluation within a falls prevention randomized controlled trial. 4 Therefore, more research that quantifies and compares observed differences among different instruments that measure HRQL is needed in falls research to better inform future economic  evaluations of falls prevention strategies.  Currently, there are no validated disease-specific measures that assess HRQL in older adults. Therefore, generic preference based utility instruments are the only current option to provide an assessment of heal th status in older adults. This is not necessarily a limitation because generic preference based utility instruments are often an essential component of clinical research. The scores they generate can be used to assess health status specifically to calculate QALYs which measure health gains or losses for economic evaluations. I focus on two examples of these generic measures of health status: the EQ-5D and the SF- 6D. Each of these two preference-based utility measures provides weightings for QALYs. The SF-6D describes 18,000 discrete health states and captures small changes in health status, 5 whereas the EQ-5D captures fewer health states (i.e., 243) but has a broader range of possible HSUVs. 6    170 In the 20th century, the incidence of chronic conditions has increased dramatically. 7 Specifically, in adults aged 65 years and older, greater than 70% live with at least one chronic condition. 7,8 In the US, total health care resource utilization and costs of fall-related medical care was estimated to be US $9.3 billion (inflated to 2008 prices) annually. Non-fatal fall injuries are associated with increased morbidity, decreased functioning and increased health care resource utilization.2,9 Given that falls are a leading cause of chronic physical disability which increases the costs of basic and long term health care, the burden on the health care system should increase. Therefore, it is necessary to determine key drivers of health care resource utilization in older adults at risk of a falls.  In falls prevention research, few studies have prospectively monitored HRQL in older women and none has specifically compared EQ-5D and SF-6D values. My aims were to: 1) quantify the difference in the incremental cost effectiveness ratio from the QALYs generated from the EQ-5D compared with the QALYs generated from the SF-6D, 2) determine key predictors of clinically and statistically significant changes in HRQL, 3) describe domain specific (i.e., mobility, self-care, usual activities, pain and anxiety/depression) changes and correlations with the EQ-5D total score and 4) determine key predictors of health care resource utilization. 6.2 Methods 6.2.1 Sample Characteristics of participants from the Brain Power study were previously reported both at baseline (cross sectional) and during the randomized controlled trial.10,11 Briefly, the study sample included 135 community dwelling older women who completed the 12- month intervention. Of these, 89 participants completed the EQ-5D at all three time points (at baseline, 6 months and one year) and 127 participants completed the SF- 36 at all three time points. I have used these the complete case set in previous chapters, but report the imputed data set to include all participants (i.e., data adjusted for missing values) for all data analyses in   171 this chapter. The design and the primary results of the Brain Power study have been reported elsewhere.10 Briefly, participants enrolled in Brain Power were: women aged 65 to 75 years, community dwelling and had a MMSE ≥ 24. A description of cos t items, health care resource utilization costs and costs of delivering the intervention are described elsewhere (see Chapters 4 and 5). 6.2.2 Measures 6.2.2.1 Descriptive variables I collected basic descriptive variables that included age, height, weight, average wais t girth and average hip girth. Specific questionnaires I collected were the Functional Comorbidity Index, Barthel Geriatric Depression Scale, Activity Specific Balance Confidence, Mini-Mental State Examination,12,13 Montreal Cognitive Assessment and Physical Activities Scale for the Elderly. 6.2.2.2 Physiological profile assessment I used the Physiological Profile Assessment, a valid and reliable tool, to assess participants‘ individual falls risk profile.14 This assessment is based on an individual‘s performance for five physiological doma ins (hand reaction time, edge contrast sensitivity, quadriceps strength, postural sway and proprioception). An individual‘s performance on each of these domains is then used to compute a fall risk score for each individual with 75% predictive accuracy in o lder adults.15 The Physiological Profile Assessment scoring system begins at zero. Scores above 3 indicate a marked risk of falls. 6.2.2.3 Health resource utilization questionnaire The health resource utilization (HRU) questionnaire asked participants to report the following visits over a specified time period: 1) to health care professionals, 2) admissions or visits to hospital and 3) laboratory work. The health resource questionnaire is described and supported in previous studies. 16,17 In total, I collected nine months of data from the 12-month Brain Power study on associated health care resource   172 use. Participants recalled their health care resource use every three months. A detailed list of cost items collected was previously reported (see Chapters 4 and 5). I estimated total health care related costs over the nine months from a Canadian health care system perspective and presented all costs in 2008 Canadian dollars. I assigned costs to all items using the BC Minis try of Health Fee Payment Schedule and the Vancouver General Hospital Fully Allocated Cost Model. 6.2.2.4 EuroQol 5D The EQ-5D is a generic preference based utility instrument developed by the EuroQol Group and is one of the most commonly used instruments.18 This five item questionnaire includes the following domains: mobility, self-care, usually activity, pain/discomfort, and anxiety/depression with each attribute having three possible options: 1) no problems, 2) some problems and 3) major problems. These options enable a total of 243 health states to be identified. Individuals‘ preferences for the scoring of the EQ-5D were measured using the time trade off technique on a random sample adult population living in the York region (n=3000). 19 Zero is defined as a health state equivalent to death, 1.0 is defined as a state of ―full health‖ and less than zero is a health state considered worse than death. Health states less than zero are possible for the EQ-5D (range: -0.54 to 1.00). Participants in the Brain Power study completed the EuroQol at baseline, 6 months and 12 months. 6.2.2.5 Short Form 6D The SF-6D is also a generic preference based utility instrument that is based on a widely used HRQL questionnaire, the Short Form 36.5 The SF-36 can be used to calculate a utility score for the SF-6D. This six-item questionnaire contained six domains that include: physical functioning, role limitations, social functioning, pain, mental health and vitality. Each attribute contains four to six levels that account for the 18 000 unique health states captured by the SF-6D. Unlike the EQ-5D, the scoring model for the SF-6D is based on the standard gamble utility measurements. A random sample (n=836) of a general adult population from the UK was used to estimate the utili ties for 249 different health states. Each participant   173 was required to provide utilities for six states. The range of values for possible health states for the SF-6D is narrower than the EQ-5D at 0.30 to 1.00. Participants in our study completed the EuroQol at baseline, 6 months and 12 months. 6.2.3 Data analysis I analysed all data using STATA version 10.0. I report descriptive data as mean (standard deviation) and median (interquartile range) for all covariates and key predictors measured unless otherwise stated. Using frequencies, I also report the domain for both the SF-6D and the EQ-5D that is most commonly reported as severe. I report the most common health state vector for best and poorest health for the SF-6D and the EQ- 5D.  For the SF-6D and EQ-5D, I also report mean (standard deviation) and median (interquartile range) at baseline, 6-months and 12-months. Based on the QALY calculations from the SF-6D and the EQ-5D, I tested if there was a statistically significant difference between the ICERs (previously reported) using a t - test and I determined whether this difference was clinically significant.  I used the Pearson product moment correlation coefficient to determine the level of correlation between QALYs or HRU (the dependent variables) and the independent variables that included key covariates (e.g. age, function comorbidities index, depression status and fall risk profile) as well as key predictors in my final model. Alpha was set at p < 0.05.  I constructed a multiple linear regression model to determine the independent association of QALYs calculated from the SF-6D or the EQ-5D and fall risk profile. I repeated all analyses described below for QALYs calculated from the SF-6D and then from the EQ-5D.    174 I constructed a multiple linear regression model to determine the independent association of HRU and fall risk profile. HRU was the dependent variable. In my model, age, group, functional comorbidity index, average waist girth and timed up and go (TUG) test were statistically controlled by forcing these four variables into the regression model (Model 1). I then entered the Physiolog ical Profile Risk score into the model. I determined independent variables from the results of the Pearson product moment coefficient of correlation analyses for both QALYs and HRU. These independent variables were determined based on the results of the Pearson product moment coefficient analyses (i.e., selected if alpha level ≤ 0.05) but those with assumed biological relevance, such as age, group and functional comorbidity index were entered into the model regardless of the results of the correlation analyses. I assessed the assumptions of normality of the residuals and heteroscedasticity. To test the assumption of normality, I used a quantile - quantile plot and assessed if there were any deviations in the linearity of the line. To test for heteroscedasticity, I used a residuals versus fit plot and assessed the presence of any funnel effect of the residuals. 6.3 Results 6.3.1 Sample Table 6-1 reports descriptive statistics for my relevant descriptor variables and my outcomes of interest (QALYs and HRU). From my analyses in Chapters 4, 5 and 7, I note that participants included in my imputed and case analysis were similar on demographic characteristics.          175 Table 6-1. Characteristics of study sample at baseline entry into the trial Characteristic Mean (SD) Median (IQR) Age, years 69.6 (3.0) 69 (67 to 72) Weight, kg 69.4 (15.1) 66.8 (59.5 to 77.5) Height, cm 161.5 (6.8) 161.4 (156.6 to 165.6) Hip girth, cm 104.2 (11.7) 102.5 (96.5 to 110.0) Waist girth, cm 86.2 (13.0) 84.8 (76.2 to 94.0) Waist to hip ratio 0.83 (0.07) 0.82 (0.78 to 0.86) PPA score 0.27 (1.02) 0.21 (-0.51 to 0.91) Functional Co morbid Index score 2.1 (1.7) 2.0 (1.0 to 3.0) Falls in past 12 months 0.32 (0.47) 0 (0 to 1) Average ABC score 88.0 (12.6) 92.2 (81.9 to 97.5) Mean Quad Strength, kg 27.4 (7.5) 26.7 (22.7 to 32.7) Timed Up and Go, sec 6.6 (1.4) 6.5 (5.7 to 7.0) Right grip strength, kg 21.7 (5.4) 22.0 (18.7 to 25.3) 6- minute walk, m 518.5 (75.8) 524.5 (479.0 to 568.0)  6.3.2 Incremental cost-effectiveness ratios for the Brain Power study using the EQ-5D and SF-6D Table 6-2 and Figure 6-1 provide a descriptive summary of the HSUVs from both the EQ-5D and SF-6D. As indicated from Table 5-3 in Chapter 5 of this thesis, I note that although the incremental QALYs calculated were different in value, this did not change the conclusion of the overall ICER. However, the EQ-5D HSUVs were substantially larger than the SF-6D HSUVs. The ICER estimated from the EQ-5D was approximately three times larger than the ICER generated from the SF-6D (i.e., CAD $9725 per QALY gained compared with CAD $2970 per QALY gained respectively). This is due to the magnitude of differences between the incremental QALYs estimated from EQ-5D HSUVs and the SF-6D HSUVs. Specifically, for once weekly resistance training compared with balance and tone classes, the incremental QALY was 3.3 times greater for the SF-6D HSUVs (-0.167 QALYs) compared with EQ-5D HSUVs (-0.051 QALYs). For twice weekly resistance training compared with balance and tone classes, the incremental QALY was 2.6 times greater for the SF-6D HSUVs  (-0.207 QALYs) compared with EQ-5D HSUVs (-0.081 QALYs).  176   Table 6-2. Global utility scores from the EQ-5D and the SF-6D  Twice weekly balance and tone (comparator) (n=49) Once weekly resistance training (n=54) Twice weekly resistance training (n=52)  EQ-5D SF-6D EQ-5D SF-6D EQ-5D SF-6D  Mean (SD) Median (IQR) Mean (SD) Median (IQR) Mean (SD) Median (IQR) Mean (SD) Median (IQR) Mean (SD) Median (IQR) Mean (SD) Median (IQR) Imputed Dataset (unadjusted) Baseline 0.803 0.193 0.800 0.270 0.704 0.060 0.696 0.074 0.821 0.193 0.800 0.270 0.697 0.071 0.696 0.099 0.836 0.194 0.800 0.231 0.688 0.069 0.681 0.101 6 months 0.792 0.208 0.800 0.270 0.692 0.068 0.696 0.099 0.820 0.237 0.800 0.240 0.701 0.078 0.696 0.099 0.818 0.191 0.800 0.270 0.696 0.086 0.697 0.099 12 months 0.796 0.192 0.800 0.275 0.687 0.068 0.696 0.099 0.826 0.219 0.800 0.240 0.687 0.068 0.696 0.110 0.847 0.177 0.821 0.200 0.700 0.067 0.696 0.099 Complete Case Analysis (unadjusted) Baseline 0.823 0.197 0.800 0.220 0.703 0.060 0.696 0.074 0.833 0.193 0.800 0.200 0.698 0.072 0.696 0.099 0.833 0.193 0.800 0.200 0.687 0.0699 0.681 0.101 6 months 0.805 0.203 0.826 0.242 0.694 0.070 0.701 0.099 0.826 0.242 0.800 0.240 0.706 0.077 0.696 0.084 0.819 0.189 0.800 0.270 0.696 0.086 0.697 0.099 12 months 0.797 0.199 0.800 0.270 0.693 0.066 0.696 0.087 0.860 0.116 0.850 0.200 0.695 0.086 0.696 0.099 0.860 0.116 0.800 0.200 0.699 0.065 0.696 0.091   177   Figure 6-1.  Box plot of global utility score for the EQ-5D 6.3.3 Descriptive classification of health state vectors between groups for the EQ-5D Table 6-3 details the health state vectors by group over time for the EQ-5D. I found that the most common health state vector for the EQ-5D for each of the three treatment groups at all three time points was ‗11111‘ indicating no deficits in mobility, self care, usual activities, pain and anxiety/depression. An almost equally common health state for participants in each of the three treatment groups at all three time points was ‗11121‘ indicating that a number of participants reported pain. There were no statistically significant differences in these health state vectors between these groups, or over time. In summary, the proportion of participants reporting ‗full health‘ defined as ‗11111‘ increased by 5% at 6 months in the once weekly resistance training group. The proportion of participants reporting ‗full health‘ increased by 2% in the twice weekly resistance training group at 12 months. This was not statistically significant; however, this improvement is clinically important.   178 Table 6-3. Health state vectors for the EQ-5D   Treatment Group Followup period Health State Vector Twice Weekly Balance and Tone Once Weekly Resistance Training Twice Weekly Resistance Training Baseline  11111 36% 40% 39%  11121 36% 35% 34% 6 months  11111 34% 45% 35%  11121 24% 25% 33% 12 months  11111 33% 38% 41%  11121 28% 23% 35%  6.3.4 Description of domain specific frequencies and percentages - EQ-5D For all individuals who were enrolled in the Brain Power study, I report the combined frequencies of responses across each dimension of the EQ-5D. Given that I did not base the primary power calculations to investigate individual domain differences, I report differences between groups only as descriptive data, rather than carrying out a statistical analysis. For mobility, usual activities and anxiety/depression, I found that the percentage of individuals reporting some problems increased over the 12-month intervention period. For self care and pain, I did not note any general trend. Descriptively, the general trend for the once weekly resistance training group was that few individuals reported problems in mobility and usual activities over the 12-months intervention period, while in the twice weekly balance and tone and resistance training groups, a greater number of individuals reported problems in mobility and usual activities over time. For further comparison, Tables 6-4 and 6-5 detail domain specific frequencies of the EQ-5D over time and by group.   179 Table 6-4. Domain specific frequencies and percentages for the EQ-5D  Baseline 6-months 12-months Mobility Frequency (%) No problems 102 (87.8) 109 (83.3) 92 (81.1) Some problems 2 (12.2) 22 (16.7) 20 (18.9) Confined to bed 0 (0) 0 (0) 0 (0) Self-care No problems 117 (98.9) 129 (97.8) 111 (98.9) Some problems 2 (1.1) 2 (2.2) 1 (1.1) Unable to wash/dress 0 (0) 0 (0) 0 (0) Usual Activities No problems 102 (87.8) 111 (85.6) 94 (81.1) Some problems 16 (12.2) 19 (14.4) 18 (18.9) Unable 1 (0) 1 (0) 0 (0) Pain/Discomfort No pain 50 (45.6) 57 (47.8) 51 (43.3) Moderate pain 65 (52.2) 67 (48.9) 57 (52.2) Extreme pain 4 (2.2) 7 (3.3) 4 (4.5) Anxiety/Depression Not anxious 103 (90.0) 107 (82.2) 92 (83.3) Moderately anxious 14 (8.9) 22 (15.6) 18 (15.6) Extremely anxious 1 (1.1) 2 (2.2) 1 (1.1)   180 Table 6-5. Domain specific frequencies and percentages for the EQ-5D Characteristics Baseline  6-months  12-months   Frequency (%) Mobility 2BT* 1RT* 2RT* 2BT 1RT 2RT 2BT 1RT 2RT No problems 31 (86) 32 (80) 39 (91) 31 (71) 37 (84) 41 (87) 27 (71) 33 (85) 36 (86) Some problems 4 (14) 8 (20) 4 (9) 9 (29) 7 (16) 6 (13) 8 (29) 6 (15) 6 (14) Confined to bed 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) Self-care No problems 36 (100) 39 (97) 42 (98) 40 (100) 44 (100) 45 (96) 35 (100) 38 (97) 38 (100) Some problems 0 (0) 1 (3) 1 (2) 0 (0) 0 (0) 2 (4) 0 (0) 1 (3) 0 (0) Unable to wash/dress 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) Usual Activities No problems 33 (89) 34 (86) 35 (81) 34 (82) 35 (80) 45 (96) 29 (79) 32 (82) 33 (87) Some problems 3 (11) 5 (13) 8 (19) 6 (18) 8 (18) 2 (4) 6 (21) 7 (18) 5 (13) Unable 0 (0) 1 (1) 0 (0) 0 (0) 1 (2) 0 (0) 0 (0) 0 (0) 0 (0) Pain/Discomfort 2BT 1RT 2RT 2BT 1RT 2RT 2BT 1RT 2RT No pain 14 (39) 17 (47) 19 (44) 17 (50) 21 (48) 19 (40) 15 (39) 18 (46) 18 (47) Moderate pain 20 (57) 22 (50) 23 (53) 21 (46) 20 (45) 26 (55) 18 (54) 19 (49) 20 (53) Extreme pain 2 (4) 1 (3) 1 (3) 2 (4) 3 (7) 2 (5) 2 (7) 2 (5) 0 (0) Anxiety/Depression Not anxious 30 (89) 36 (90) 37 (86) 33 (86) 39 (89) 35 (74) 28 (86) 32 (84) 32 (84) Moderately anxious 6 (11) 3 (10) 5 (12) 6 (11) 4 (9) 12 (26) 7 (14) 5 (13) 6 (16) Extremely anxious 0 (0) 0 (0) 1 (2) 1 (4) 1 (2) 0 (0) 0 (0) 1 (3) 0 (0) * 1RT: Once weekly resistance training, 2RT: tw ice weekly resistance training, 2BT: twice weekly balance and tone   181 6.3.5 Model to determine independent predictors of QALYs I report the results of the imputed data set. This included the 135 participants who completed the 12-month intervention. I report my available case analysis for descriptive reporting of frequencies within specific domains of the EQ-5D. 6.3.6 Correlation coefficients Table 6-6 reports the correlation coefficients between variables of interest and key predictors of QALYs estimated from both the EQ-5D and SF-6D based on the sample described in Table 6-1. I found that group, age, weight, physiological profile assessment, functional comorbidities index, falls in the past 3 months, right and left grip strength, TUG and six minute walk test were significantly associated with QALYs estimated from the EQ-5D (p < 0.05). Of these, the strongest associations for QALYs estimated using EQ- 5D HSUVs were physiological profile assessment (R2=-0.31), functional comorbidities index (R2=-0.48), right grip strength (R2=0.20), TUG (R2=-0.39) and six minute walk test (R2=0.27). The strongest associations for QALYs estimated using SF-6D HSUVs were physiological profile assessment (R2=-0.34), functional comorbidities index (R2=-0.42), right grip strength (R2=0.18), TUG (R2=-0.39) and the six minute walk test (R2=0.24). For HRU, the strongest associations were QALYs (R2=-0.44 for EQ-5D HSUVs and R2=-0.32 for SF-6D HSUVs), functional comorbidities index (R2=0.32) and right grip strength (R2=-0.25). I found that age, weight, physiological profile assessment, functional comorbidities index, falls in the past 3 months, right and left grip strength, TUG and six minute walk test were significantly asso ciated with QALYs estimated from the SF-6D (p < 0.05). The main differences for these bivariate relationships between the EQ-5D and the SF-6D were that 1) group allocation was not statistically significant using SF-6D HSUVs and was statis tically significant using EQ-5D HSUVs and 2) number of falls in the past 6 and 12 months was not statistically significant using EQ-5D HSUVs and was statistically significant using SF-6D HSUVs and 3) the direction of the correlation for number of falls in the past 12 months  was different for the EQ-5D HSUVs compared with the SF-6D HSUVs.   182  Table 6-6 reports the correlation coefficients between covariates of interest and key predictors of HRU based on the samples described in Table 6-1. I found that QALYs (from EQ-5D or SF-6D), group, weight, physiological profile assessment, functional comorbidities index, falls in the past 3 months, right and left grip strength and six minute walk test were significantly associated with HRU (p < 0.05). Of note, for all the key predictors detailed above, the direction of association for HRU was reversed compared with QALYs (from EQ-5D or SF-6D) in all cases. Table 6-6. Correlation coefficient matrix‡ Variables at baseline Health Resource Utilization QALYs (from EQ-5D) QALYs (from SF-6D) Group -0.0882* 0.0766* 0.0145 Age (baseline) 0.0464 -0.1286** -0.2189** QALYs (from SF-36) -0.3157** 0.5163** n/a QALYs (from EQ-5D) -0.4432** n/a 0.5163** Weight (baseline) 0.1031* -0.1407** -0.0710* Physiological Profile Assessment 0.0897* -0.3113** -0.3396* Functional Comorbidity Index 0.3188** -0.4849** -0.4211** Falls in past 12-months  0.0259 -0.0264 0.0199 Falls in past 6-months 0.0299 -0.0565 -0.0932* Falls in past 3-months 0.1042* -0.0851* -0.1043** Timed Up and Go 0.1110* -0.3921** -0.3859** Right Grip Strength -0.2512** 0.2023** 0.1813** Left Grip Strength -0.2835** 0.1511** 0.1352** Six-Minute Walk -0.1336** 0.2690** 0.2388** ‡  Correlations reported for HRU and QALYs (estimated from EQ-5D and SF-6D) using imputed data set * p < 0.05 ** p < 0.01  6.3.7 Multivariate linear regression results for QALYs calculated from the EQ-5D and SF- 6D HRU, Physiological Profile Assessment and TUG were all significant and independent predictors of QALYs estimated from the EQ-5D HSUVs and SF-6D HSUVs (p<0.05). Adding these key predictors together into the model resulted in a R2 change of 16% (p<0.05) for the EQ-5D HSUVs and a R2 change of 13%   183 (p<0.05) for the SF-6D HSUVs. The total variance accounted for by the final model ranged from 34% to 41% (i.e., adjusted R2=0.41 using EQ-5D HSUVs and adjusted R2=0.34 using SF-6D HSUVs)  (Table 6-7).  HRU, Physiological Profile Assessment, number of falls in the past 3 months and TUG were all significant and independent predictors of QALYs estimated from the SF-6D (p<0.05). Adding these key predictors resulted in an R2 change of 13% (p<0.05). The total variance accounted for by my final model was 34% (Table 6-7).   184 Table 6-7. Multiple linear regression summary for quality adjusted life years in older women calculated from EQ-5D and SF-6D Independent Variables R2 R2 change Unstandardized B (Standard Error) P-value Model 1a (EQ-5D) 0.2452 Age, years   -0.0027 (0.0019) 0.153 Group   0.0190 (0.0070) 0.007 Functional Comorbidities Index   -0.0512 (0.0034) 0.000 Model 1b (SF-6D) 20.13 Age, years   -0.0034 (0.0007) 0.000 Group   0.0010 (0.0026) 0.701 Functional Comorbidities Index   -0.0155 (0.0013) 0.000 Multivariate Model 2a (EQ-5D) 0.4079 0.1595 Group   0.0041 (0.0018) 0.025* Age, years   0.0155 (0.0062) 0.013* Functional Comorbidities Index   -0.0370 (0.0032) 0.000** Health Resource Utilization, 2008 CAD $   -0.000047 (0.000006) 0.000** Physiological Profile Assessment   -0.0328 (0.0054) 0.000** Timed Up and Go, sec   -0.0308 (0.0039) 0.000** Multivariate Model 2b (SF-6D) 0.3351 0.1338 Group   -0.0007 (0.0007) 0.365 Age, years   0.0008 (0.0025) 0.739 Functional Comorbidities Index   -0.0114 (0.0012) 0.000** Health Resource Utilization, 2008 CAD $   0.000010 (0.000002) 0.000** Falls in the past 3 months   -0.0153 (0.0056) 0.006 Physiological Profile Assessment   -0.0139 (0.0021) 0.000** Timed Up and Go, sec   -0.0103 (0.0016) 0.000** * p < 0.05 ** p < 0.01   185   6.3.8 Model to determine independent predictors of health resource utilization Weight, functional comorbidities index, QALYs (from EQ-5D), falls in the past 3 months, TUG and right grip strength were all significant and independent predictors of HRU (p<0.05). The addition of weight, falls in the past 3 months, TUG and Right Grip Strength resulting in a R2 change of 5% (p<0.05). The total variance accounted for by my final model was 15% (Table 6-8). Of note, the ß coefficients for the EQ-5D HSUVs in the multivariate models were large given that the incremental changes were in the 1/1000 decimal place.  Weight, functional comorbidities index, QALYs (from SF-6D) and falls in the past 3 months were all significant and independent predictors of HRU (p<0.05). The addition of weight and falls in the past 3 months resulted in a R2 change of 5% (p<0.05). The total variance accounted for by my final model was 15% (Table 6-8).   186 Table 6-8. Multiple linear regression summary for health resource utilization  Independent Variables R2 R2 change Unstandardized B (Standard Error) P-value Model 1a 0.1533 Group   -20 (39) 0.615 Age, years   -16 (11) 0.143 Functional Comorbidities Index   -64 (21) 0.003* QALYs (EQ-5D)   -1714 (200) 0.000** Model 1b 0.1053 Group   -49 (40) 0.217 Age, years   -21 (11) 0.061 Functional Comorbidities Index   107 (21) 0.000** QALYs (SF-6D)   -2902 (548) 0.000** Model 2a 0.2062 0.0529 Group   -47 (40) 0.238 Age, years   -6 (11) -.583 Functional Comorbidities Index   51 (23) 0.027* QALYs (EQ-5D)   -1707 (213) 0.000** Weight, kg   12 (3) 0.000** Falls in past 3 months   324 (89) 0.000** Timed Up and Go, sec   -117 (29) 0.000** Right Grip Strength, kg   -35 (7) 0.000** Model 2a (SF-6D) 0.1530 0.0477 Group   -55 (41) 0.176 Age, years   -20 (11) 0.077 Functional Comorbidities Index   97 (21) 0.000** QALYs (SF-6D)   -2716 (551) 0.000** Weight, kg   5 (2) 0.037* Falls in past 3 months   299 (91) 0.001* * p < 0.05 ** p < 0.01 6.4 Discussion Falls are the leading cause of chronic physical disability1,2 and this is is associated with a decline in HRQL. Individuals who experience the greatest declines in HRQL have the greatest mortality. 20 To my knowledge, this study, that included an economic evaluation based on a falls prevention strategy, is the first to 1) examine the ICERs generated from two generic preference based utility instruments, 2) provide a descriptive comparison of two generic preference based utility instruments in terms of their estimated   187 QALYs measured prospectively over one year in high functioning community dwelling senior women and 3) to ascertain significant and independent predictors of QALYs and HRU. 6.4.1 Predicting health resource utilization among older community dwelling women  Of particular interest, there were independent associations for HRU, physiological profile assessment and TUG for QALYs estimated from the EQ-5D or SF-6D in this cohort of senior women after accounting for age, trial group allocation and functional comorbidities. One notable discrepancy in these findings was that for QALYs estimated using SF-6D HSUVs, the number of falls in the past 3 months was also a significant and independent predictor. This was not the finding in the regression model for QALYs estimated using EQ-5D HSUVs. One potential reason for this is that the EQ-5D and the SF-6D are comprised of different domains. Specifically, the SF-6D has a physical function and vitality domain whereas the EQ-5D has a mobility domain. Although this seems surprising at first glance, it could be researched by having appropriate participants discuss how they rate ‗mobility‘, ‗vitality‘ and ‗physical function‘. Therefore, it is possible that individuals who sustained falls in the past three months may report deficits in the physical function or vitality domain that would not be detected in the mobility domain of the EQ-5D. Few peer- reviewed studies have compared the EQ-5D and SF-6D specifically in older adults. One recent study found that the EQ-5D and SF-6D score were highly correlated in adults aged 45 years and older. 21 Further, individuals who were healthier, had higher mean scores on the EQ-5D (p<0.001), whereas individuals who were less healthy had higher mean scores on the SF-6D (p<0.001).21 This is not surprising given that the range of possible values from the SF-6D is 0.30-1.00 compared with the EQ-5D‘s range of -0.54-1.00.22 Thus my study extends previous findings by highlighting differences between the SF-6D and the EQ-5D that could be explored further to better understand the usefulness of these instruments.    188 To my knowledge, no other study has examined the relationship between HRQL and HRU specifically among older adults. The novel finding that HRQL is inversely, independently and significantly associated with HRU does build upon previous findings in other fields.23 6.4.2 Clinical versus statistical significance – placing the results in context A key issue to consider is clinical versus statistical significance.24 Placing this comparison in the context of comparing ICERs and QALYs, I highlight that statistical significance is not of primary importance.24 Given that our health care system is always constrained by a limited budget, the purpose of economic evaluations is to help decision makers determine ‗best buys‘.24,25,26 A ‗p‘ value quantifies statistical inference but economic evaluations are unrelated to statistical inference.24 Rather, economic evaluations are designed to inform optimal allocation of health budget.27 The statistical uncertainty of the ICER is only relevant when budgeting for future research and prioritizing future evidence.24 Hence, it is the point estimate of the ICER that is of importance when making comparisons. These data indicate that the ICER estimated using EQ-5D HSUVs was three times greater than the ICER estimated using SF-6D HSUVs. This difference arose because the incremental QALY that was between 2.6 to 3.3 times greater for the SF-6D HSUVs and EQ- 5D HSUVs, respectively. Previous research has highlighted that the minimally important difference for the EQ-5D is 0.030 and for the SF-6D is 0.033.6 There are a range of values reporting minimally important differences for the EQ-5D and the SF-6D.28 For the EQ-5D, these values range from 0.011 to 0.140. For the SF-6D, these values range from 0.011 to 0.097.28 Further, the mean difference between these two instruments was 0.033 (95%CI: 0.000 to 0.066; p=0.05).28 Hence, the EQ-5D‘s minimally important difference is approximately double that of the SF-6D. This is reasonable given that the possible range of the EQ-5D summary score is approximately double that of the SF-6D. Given that the incremental QALYs in this study were greater than the minimally important differences for both HSUV instruments, I conclude these are clinical important differences, regardless of statistical significance. 24   189 6.4.3 Limitations  These findings should be interpreted with caution due to the following study limitations. My study sample consisted only of community dwelling women who were cognitively intact – I cannot generalize these results to all older women or older adults in general. Further, the primary outcome of the Brain Power study was to ascertain a change in cognitive performance. For this reaso n, I limited my comparisons of the EQ- 5D and the SF-6D to a descriptive analysis. 6.4.4 Conclusions  Thus, I recommend this study be used to highlight ideas for future prospective studies to ascertain 1) whether my present finding applies to related clinical populations and 2) the direction of the causal relationship between HRQL and HRU. The findings indicate that changes in EQ-5D or SF-6D HSUVs over time can be largely explained by baseline measures of age, weight, functional comorbidities index, HRU, Physiological Profile Assessment, number of falls in the past 3 months and TUG test results.   190 6.5 References 1.  Murray CWS, Lopez A. Global and regional descriptive epidemiology of disability: Incidence, prevalence, health expectancies, and year lived with disability. Boston: The Havard School of Public Health, 1996. 2.  Tinetti ME, Speechley M, Ginter SF. Risk factors for falls among elderly persons living in the community. N Engl J Med 1988;319(26):1701-7. 3.  American College of Rheumatology Subcommittee on Rheumatoid Arhtirits Guidelines for the Managment of Rheumatoid Arthritis: 2002 Update. Arthritis Rheum;2002(46):326-348. 4.  Sach TH, Foss AJ, Gregson RM, Zaman A, Osborn F, Masud T, et al. Falls and health status in elderly women following first eye cataract surgery: an economic evaluation conducted alongside a randomised controlled trial. Br J Ophthalmol 2007;91(12):1675-9. 5.  Brazier J, Roberts J, Deverill M. The estimation of a preference -based measure of health from the SF-36. J Health Econ 2002;21(2):271-92. 6. Marra CA, Woolcott JC, Kopec JA, Shojania K, Offer R, Brazier JE, et al. A comparison of generic, indirect utility measures (the HUI2, HUI3, SF-6D, and the EQ-5D) and disease-specific instruments (the RAQoL and the HAQ) in rheumatoid arthritis. Soc Sci Med 2005;60(7):1571-82. 7.  Booth FW, Gordon SE, Carlson CJ, Hamilton MT. Waging war on modern chronic diseases: primary prevention through exercise biology. J Appl Physiol 2000;88(2):774-87. 8.  Bean JF, Vora A, Frontera WR. Benefits of exercise for community-dwelling older adults. Arch Phys Med Rehabil 2004;85(7 Suppl 3):S31-42; quiz S43-4. 9.  Campbell AJ, Borrie MJ, Spears GF. Risk factors for falls in a community -based prospective study of people 70 years and older. J Gerontol 1989;44(4):M112-7.   191 10.  Liu-Ambrose T, Nagamatsu LH, Graf P, Beattie L, Ashe MC, Handy T. The effect of resistance training on brain function among community-dwelling senior women: a randomized controlled trial. Archives of Internal Medicine 2010;170(2):170-8. 11.  Liu-Ambrose TY, Ashe MC, Graf P, Beattie BL, Khan KM. Increased risk of falling in older community-dwelling women with mild cognitive impairment. Phys Ther 2008;88(12):1482-91. 12.  Folstein MF, Folstein SE, McHugh PR. "Mini-mental state". A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 1975;12(3):189-98. 13.  Folstein MF, Robins LN, Helzer JE. The Mini-Mental State Examination. Arch Gen Psychiatry 1983;40(7):812. 14.  Lord SR, Sherrington C, Menz H. A Physiological Profile Approach for Falls Prevention. Falls in Older People. Risk Factors and Strategies for Prevention.  Cambridge: Cambridge University Press, 2001. 15.  Lord SR, Menz HB, Tiedemann A. A physiological profile approach to falls risk asses sment and prevention. Phys Ther 2003;83(3):237-52. 16.  Liu-Ambrose T, Ashe MC, Marra C, Conditions Research Team PA. Among Older Adults with Multiple Chronic Conditions, Physical Activity is Independently and Inversely Associated with Health Care Utilization. Br J Sports Med 2008. 17.  Maetzel A, Li LC, Pencharz J, Tomlinson G, Bombardier C. The economic burden associated with osteoarthritis, rheumatoid arthritis, and hypertension: a comparative study. Ann Rheum Dis 2004;63(4):395-401. 18.  Fitzpatrick R, Davey C, Buxton MJ, Jones DR. Evaluating patient-based outcome measures for use in clinical trials. Health Technol Assess 1998;2(14):i-iv, 1-74. 19.  Dolan P. Modeling valuations for EuroQol health states. Med Care 1997;35(11):1095-108.   192 20.  Kroenke CH, Kubzansky LD, Adler N, Kawachi I. Prospective change in health-related quality of life and subsequent mortality among middle-aged and older women. Am J Public Health 2008;98(11):2085-91. 21.  Barton GR, Sach TH, Avery AJ, Jenkinson C, Doherty M, Whynes DK, e t al. A comparison of the performance of the EQ-5D and SF-6D for individuals aged >or= 45 years. Health Econ 2008;17(7):815-32. 22.  Bharmal M, Thomas J, 3rd. Comparing the EQ-5D and the SF-6D descriptive systems to assess their ceiling effects in the US general population. Value Health 2006;9(4):262-71. 23.  Pare P, Gray J, Lam S, Balshaw R, Khorasheh S, Barbeau M, et al. Health-related quality of life, work productivity, and health care resource utilization of subjects with irritable bowel syndrome: baseline results from LOGIC (Longitudinal Outcomes Study of Gastrointestinal Symptoms in Canada), a naturalistic study. Clin Ther 2006;28(10):1726-35; discussion 1710-1. 24.  Claxton K. The irrelevance of inference: a decision-making approach to the stochastic evaluation of health care technologies. J Health Econ 1999;18(3):341-64. 25.  Mooney G, Wiseman V. Burden of disease and priority setting. Health Econ 2000;9(5):369-72. 26.  Wiseman V, Mooney G. Burden of illness estimates for priority setting: a debate revisited. Health Policy 1998;43(3):243-51. 27.  Drummond MF, Sculpher MJ, Torrance GW, O'Brien B, Stoddart GL. Methods for the economic evaluation for health care programmes. Third edition. New York. United States of America: Oxford University Press, 2005. 28.  Walters SJ, Brazier JE. Comparison of the minimally important difference for two health state utili ty measures: EQ-5D and SF-6D. Qual Life Res 2005;14(6):1523-32.    fA version of this chapter was submitted for publication. DAVIS JC, Marra CA, Najafzadeh M, Liu-Ambrose T. The independent contribution of executive functions to health related quality of life in older women. (Revised and re-submitted Jan 29, 2010).  193 7 The independent contribution of executive functions to health related quality of life in older womenf 7.1 Introduction  Health related quality of life is an important construct that describes an individual‘s overall health status. It is commonly used in economic evaluations1 as a measure of health benefit following treatment, and may be more responsive among populations with conditions associated with high morbidity. 2 HRQL is defined by several domains,3 with general agreement that emotion, physical and social are core domains. These reflect the World Health Organization‘s definition of health as ‗a state of complete physical, mental, and social well-being, and not merely the absence of disease or infirmity‘.4 However, given that HRQL is ―the subjective assessment of the impact of disease and treatment across the physical, psychological, social, and somatic domains of functioning and well-being‖,5 the specific contribution of HRQL to overall quality of life remains unknown.6  The use of a generic, preference-based instrument is one method commonly used to  assess HRQL.7 The EQ-5D developed by the EuroQol Group, is one such instrument and is the most widely used generic instrument that uses a utility-based scoring approach.8 The EQ-5D captures 243 unique health states within the following domains using a short five-item questionnaire: 1) mobility, 2) self-care, 3) usual activities, 4) pain and 5) anxiety/depression.8 Individuals‘ preferences for the scoring of the EQ-5D were estimated using the time trade off technique in a random sample of adults living in the York (UK) region (N=3000). 9 The EQ- 5D yields a single summary score on a common scale to facilitate comparison across different health conditions and patient populations.10,11 The single summary score, known as a health state utili ty value (HSUV) is anchored at zero – a health state equivalent to death – and 1.0 indicates a state of ―full health.‖ HSUV scores of less than zero indicate health states considered worse than death.   194  HSUVs (measured using instruments such as the EQ-5D) are then used to calculate Quality Adjusted Life Years (QALYs), a measure of HRQL.  QALYs indicate the quality of life of a patient in a given health state and the time spent in that health state. Briefly, the QALY is a useful measure of health benefit because it simultaneously captures both quantity and quality of life gains or losses. 12 A key benefit of the QALY is that it enables direct comparison of patient outcomes across diseases and diverse health interventions. 12 Also, it accounts for changes in both morbidity and mortality under a common metric. QALYs are defined as  a measure of health benefit in terms of time spent in a series of quality -weighted health states, in which the quality weights reflect the desirability of living in the state, typically anchored at ―perfect‖ health (weighted 1.0) and dead (weighted 0.0).13 The quality weights spent in each state are multiplied by the time spent in each state. The sum of all these products is the total number of QALYs.  A number of factors may work together or independently contribute to HRQL. Some clinical measures that are associated with HRQL include cognition, physical disability and chronic conditions such as rheumatoid arthritis, sex, social functioning and physical activity.14 Specifically, two studies demonstrated that adults with a physical disability, asthma or who are female had significantly increased odd ratios for poor HRQL.15,16 Among individuals with Alzheimer‘s disease, global cognition as measured by the MMSE, was associated with HSUVs estimated using the EQ-5D.17  Cognition is a multidimensional construct and to my knowledge, no previous studies have examined the independent contribution of specific domains of cognition to HRQL. I hypothesize that executive functions may be of particular importance to HRQL. Executive functions are higher-order cognitive processes that control planning, initiation, sequencing and monitoring of complex goal directed behavior. 18,19 These cognitive processes are essential to the person‘s ability to carry out health-promoting behaviours,20 such as medication management, dietary and lifestyle changes, self-monitoring of responses, and follow-up with health care professionals. Hence, in this study, I examined whether executive functions in community -   195 dwelling older women are independently associated with HRQL in community dwelling older women, calculated using the EQ-5D HSUVs at three time points, after accounting for global cognitive function and known confounders. 7.2 Methods  7.2.1 Study design and participants  The total sample for this analysis consisted of 135 women who consented and completed a one year randomized controlled trial of exercise (NCT00426881; Brain Power study) that aimed to examine the effect of once weekly and twice weekly resistance training on cognitive performance of executive functions. The design and the primary results of the Brain Power study have been reported elsewhere.21 Briefly, women enrolled in Brain Power were: aged 65 to 75 years, community dwelling and had a MMSE score ≥ 24. 7.2.2 Functional comorbidity index  Functional Comorbidity Index was calculated to estimate the degree of comorbidity associated with physical functioning.22 This scale‘s score is the total number of comorbidities. 7.2.3 Global cognition measures – Mini Mental State Examination  The MMSE is a widely used and well-known questionnaire used to screen for cognitive impairment (i.e., MMSE <24).23 It is scored on a 30-point scale, with a score <24 indicating cognitive impairment and a median score of 28 for more normal octogenarians with more than 12 years of education.23 The MMSE may underestimate cognitive impairment for frontal system disorders24 because it has no items specifically addressing frontal lobe function.23      196 7.2.4 Central executive functions—set shifting, updating and response inhibition  The assessment of executive functions at baseline were composed of three tests that measure different aspects of executive functions:18 1) Trail Making Test Part B, 2) Verbal Digits Backward Test and 3) Stroop Colour-Word Test. 7.2.4.1 Trail making part B  I used the Trail Making Part B test to assess set shifting. Set shifting refers to an individuals ability to go back and forth between multiple tasks or mental sets.25 The test consists of one page with circled letters (A- L) and numbers (1-13). Participants were instructed to draw a single line  as quickly and accurately as possible from 1 to A, A to 2, 2 to B and so forth until the task was completed. The number of errors and the length of time the task took were recorded. To index set shifting, we calculated the difference between Part B and Part A completion time. Smaller difference scores indicate more cognitive flexibility. Reliability scores for the Trail Making Part B varied from moderate to excellent.26 7.2.4.2 Verbal digits backward  I used the Verbal Digits Backward test to assess working memory. 27 Working memory (updating) refers to an individuals ability filter incoming information for relevance to the  current task and subsequently update informational content replacing old non-relevant information with new relevant incoming information.25 Seven pairs of random number sequences were read aloud by an assessor at one number per second. The first sequence consists of three numbers and the sequence was increased by one number up to a length of nine digits. Participants repeated each sequence in exactly the reverse order until they failed two attempts of the same sequence length. It was scored on a 14-point scale with higher scores indicating a better performance. For the verbal digits forward test, the participant‘s task is to repeat each sequence exactly as it is given. The difference between the verbal digits forward test score and the verbal digits backward test score was used as an index of the central executive component of working memory. Smaller difference scores indicate better working memory.   197 7.2.4.3 Stroop test  The Stroop Test assessed response inhibition28 including deliberate inhibition of automatic, dominant or routine responses.26 For the primary test condition, participants were presented colour-words printed in incongruent coloured inks (e.g., the word ―BLUE‖ printed in red ink) and were required name the ink colour that the words were printed while ignoring the word itself. The time participants took to read each condition was recorded. The ability to selectively attend and control response output was calculated as the time difference between the test condition and the priming condition (e.g., coloured X‘s). Smaller time differences indicate better selective attention and conflict resolution. 7.2.5 Preference based measures – HSUV instrument  The HSUV instrument I used was the EQ-5D. Major differences between the EQ-5D and other preference based measures were outlined previously.8 The EQ-5D does not directly measure cognition; another generic preference based instrument, the Health Utili ties Index Mark 3 (HUI3) does. 29 To my knowledge, no previous studies have examined the association between executive functions and HRQL using the HUI3. Therefore, I chose the EQ-5D given that is it the most widely used generic preference based utility instrument that has been used among individuals with cognitive decline. The EQ-5D short structure was considered a strength in terms of high response rates, participant burden and feasibility30 and a weakness in terms of its responsiveness and sensitivity.31 I used the EQ-5D to calculate QALYs as an assessment of an individual‘s HRQL according to the following five EQ -5D domains: mobility, self-care, usual activities, pain, and anxiety/depression. Each domain has three levels that either indicate no problems, some problems or severe problems. The EQ-5D HSUVs at each time point are bounded from -0.54 to 1.00 where a score of less than zero is indicative of a health state worse than death. I used three HSUVs for each individual from the EQ-5D at baseline, 6 months and 12 months to calculate QALYs for each individual. Specific to this study only, QALYs are a measure of HRQL because no participants died and all participants were followed for the same time period, thus any changes in QALYs   198 are due to quality of life, rather than quantity of time spent in a given health state. 7.2.6 Timed up and go  I used the TUG to assess general mobility.32 Participants were instructed to rise from a chair with their arms crossed (seat height 45 cm), walk a distance of three meters, turn around, walk back to the chair, and sit down with their arms crossed around their chest. Each trial was timed and I and took the mean of two trials for my statistical analysis. 7.2.7 Data analysis  I analysed all data using STATA version 10.0. My base case analysis included 135 women based on recommendations for multiple imputation of missing cost and HSUV data. 33 For all three time points, I used a combination of multiple imputation and bootstrapping to estimate uncertainty caused by missing values and I report both the imputed data set analysis and a complete case analysis. My complete case analysis consisted of 89 participants for the EQ-5D who had all three HSUVs (EQ-5D results) at baseline, 6-months and 12-months.  I report descriptive data for all variables of interest. For data that are normally distributed I report mean and standard deviation and frequencies depending on the measure. For data that were significantly skewed, I report median and interquartile range. I used the Pearson product moment correlation coefficient to determine the level of association between QALYs gained during the one year trial and age, group, education, average waist girth, functional comorbidity index, general mobility, global cognition and executive functions. The baseline data for age, group (i.e., twice weekly balance and tone, once weekly resistance training and twice weekly resistance training), educatio n, average waist girth, functional comorbidity index, general mobility, global cognition and executive functions were used for all statistical analyses.    199 In the multiple linear regression model, we controlled for age, intervention group, education, average  waist girth, functional comorbidity index, general mobility and global cognition were statistically controlled by forcing these six variables into the regression model. These independent variables were determined based on the results of the Pearson product moment coefficient analyses (i.e., alpha level < 0.05) but those with assumed biological relevance, such as MMSE and waist girth, were entered into the model regardless of the results of the correlation analyses. The executive functions (i.e., Trail Making Part B, Digits Backwards) was then entered into the model. Those that significantly added to the model (i.e., significant change in R2) were kept in the model. Age, intervention group, education, average waist girth, functional comorbidity index, general mobility and global cognition were entered last into the model. I assessed the assumptions of normality of the residuals and heteroscedasticity. 7.3 Results  I report the results of both the imputed case analysis (base case) and the complete case analysis. 7.3.1 Sample  Table 7-1 reports descriptive statistics for descriptive variables (age, baseline EQ-5D HSUV, group, education, average waist girth, functional comorbidity index, Trail Making Part A, Trail Making Part B, Digits Forward, MMSE and TUG) and our outcome of interest (QALYs). Participants included in our imputed and case analysis were similar on demographic characteristics. Overall, this cohort of community-dwelling senior women were high functioning individuals as indicated by their baseline EQ-5D scores (HSUVs) of 0.82 (SD: 0.19) and 0.85 (SD: 0.18) for the imputed and complete case sets, respectively. Further, the mean MMSE score was greater than 28 (maximum 30 points).    200 Table 7-1. Characteristics of the Brain Power cohort at baseline Variable at Baseline Imputed Data Set (N=135) Complete Case Set (N=89)  Mean Standard Deviation Mean Standard Deviation QALY (EQ-5D) 0.83 0.17 0.83 0.17 Age (years) 69.6 3.0 69.7 3.0 Baseline EQ-5D HSUV 0.82 0.19 0.85 0.18 Average waist girth (cm) 86.3 13.0 87.3 12.4 Function Comorbidity Index 2.1 1.7 2.0 1.6 Trail A (sec) 55.3 18.3 54.4 17.7 Trail B (sec) 101.2 41.7 97.0 36.7 Trail B - Trail A (sec) 42.5 29.9 46.2 34.8 Digits Forward (maximum 14 points) 7.9 2.3 7.9 2.3 Digits Backward (maximum 14 points) 4.5 2.4 4.3 2.4 Digits Forward – Digits Backward 3.7 2.3 3.4 2.3 MMSE score (maximum 30) 28.6 1.3 28.7 1.4 Timed Up and Go Test (sec) 6.6 1.4 6.7 1.5  7.3.2 Correlation coefficients  Table 7-2 reports the correlation coefficients between variables of interest and QALYs. Age, group education, baseline EQ-5D HSUV, average waist girth, functional comorbidity index, TUG, set shifting (assessed by the difference score for Trail Making Part B and A) and working memory (assessed by the difference score for Digits Forward and Backward) were significantly associated with QALYs calculated from the EQ-5D (p < 0.05). Group allocation and response inhibition (assessed by using the Stroop Colour- Word test) were not significantly associated with QALYs calculated from the EQ-5D (p > 0.05).    201 Table 7-2. Correlation coefficient matrix ‡ Variable at Baseline Imputed Data Set QALYs (EQ-5D) Age (years) -0.298** Group (0=twice weekly balance and tone, 1=once weekly resistance training, 2=twice weekly resistance training) 0.0913 Education level (1=did not finish high school, 2= finished high school, 3 = post-secondary) 0.311** Average waist girth (cms) -0.232* Function Comorbidity Index score -0.488** Trail B – Trail A (sec) -0.108* Digits Forward – Digits Backward (sec) -0.171** MMSE score (max 30) 0.051 Timed Up and Go (sec) -0.598** Stroop test (sec) -0.113 ‡ Results from both imputed and complete case analysis were identical to x decimal places. * p < 0.05 ** p < 0.01 7.3.3 Multivariate linear regression results for QALYs calculated from the EQ-5D  The Trail Making Part B was a significant and independent predictor for HRQL as assessed by the EQ-5D (p < 0.01). Digits Backward was also a significant predictor for HRQL (p < 0.01) based on the EQ-5D results of the imputed data set (for complete case set, p = 0.09). Adding the Trail Making Part B and the Digits Backward variables to this model resulted in an R2 change of 4% (p < 0.01). The total variance accounted for by our final model was 50% (Table 7-3). The R2 and R2 change from both imputed and complete case analysis were identical to two decimal places. The Stroop Colour-Word task results did not significantly improve the model after accounting for age, group, education waist girth, functional comorbidity index, general mobility and global cognition.   202 Table 7-3. Multiple linear regression summary for predicting QALYs in older women as calculated from EQ- 5D HSUVs ‡  Imputed Data Set (N=135) Complete Case Set (N=89) Independent Variables Unstandardized ß (Standard Error) P-value Unstandardized ß (Standard Error) P-value Model  R2  0.536   R2  0.536   Trail B – Trail A (sec) 0.0012 (0.0002) 0.00** 0.0012 (0.0005) 0.03* Digits Forward – Digit Backward (sec) -0.011 (0.003) 0.00** -0.011 (0.007) 0.08 Age (years) 0.0008 (0.0022) 0.717 0.0008 (0.0052) 0.88 Group (0=twice weekly balance and tone, 1=once weekly resistance training, 2=twice weekly resistance training) -0.008 (0.008) 0.296 -0.008 (0.018) 0.66 Education level (1=did not finish high school, 2= finished high school, 3 = post-secondary) 0.025 (0.005) 0.00** 0.02 (0.01) 0.02* Average waist girth (cm) -0.0006 (0.0005) 0.240 -0.0006 (0.0012) 0.62 Functional Comorbidity Index score -0.036 (0.004) 0.00** -0.036 (0.009) 0.00** Timed Up and Go test (sec) -0.062 (0.005) 0.00** -0.06 (0.01) 0.00** MMSE score (max 30) -0.017 (0.005) 0.001 -0.02 (0.01) 0.15 ‡ R2 and R2 change were the same for both imputed and complete case analyses * p < 0.05 ** p < 0.01 7.4 Discussion 7.4.1 Relationship between executive functions and QALYs – HRQL  Persons who experience cognitive decline have a reduced quality of life. 34 To my knowledge, this study is the first to demonstrate the independent association between key executive processes as measured by standard neuropsychological tests, and QALYs measured prospectively over one year among high functioning community dwelling senior women. Of particular importance, this independent association persisted after accounting for age, waist girth, functional comorbidity index, general mobility and global cognition. Also, my final model explained 50% of the variation in QALYs; regression models in clinical research often do not often account for such a large amount of variance.35    203 I specifically found that, both set shifting and working memory, were independently associated with HRQL, measured by QALYs calculated from the EQ-5D scores (HSUVs). These results extend previous findings that set shifting, as measured by the Trail Making B Test, was associated with factors that may influence QALYs: 1) mobility;36 37 2) medication adherence;38 3) driving performance;39 4) anxiety and emotional regulation.40 I highlight that mobility and anxiety/depression are domains in the EQ-5D. Additionally, my discovery that set shifting also contributes to HRQL extends previous findings that working memory is associated with pain severity. Pain is one of the five domains of the EQ-5D; therefore, I would expect an association between pain and QALYs.41 7.4.2 Relating working memory and health related quality of life  My finding of both set shifting and working memory contributing to health related quality of life extend previous studies examining the association of cognitive function and instrumental activities of daily living. Instrumental activities of daily living include the ability to prepare a balanced meal, remember appointments, keep financial records and take medications as prescribed. 42 Health related quality of life is related to one‘s ability to perform instrumental activities of daily living 43 and one‘s overall mobility.44 Previous studies have demonstrated that executive functions are associated with instrumental activities of daily living and functional s tatus among older adults.45,46 Specifically, the Trail Making B Test is an independent predictor of the instrumental activities of daily living.45,46 7.4.3 Response inhibition and health related quality of life – comparison with another study  My findings for the association between the Stroop test and health related quality of life differ from those of previous research.47 Specifically, one study among 72 older adults with stable cardiovascular disease found a significant association between response inhibition and instrumental activities of daily living. 47 Differences in the population studied may explain the conflict. Participants of the Brain Power cohort were high functioning individuals. Hence, there may have been a ceiling effect for both instrumental activities of daily   204 living and health related quality of life as assessed by the EQ-5D. Further research is needed to better understand the contribution of response inhibition to health related quality of life. 7.4.4 Contrasting the imputed and complete case analyses  The lack of a significant association between global cognition and HRQL for my complete case analysis was contrary to the results of my imputed data set analysis. This was likely due to the smaller sample size of the complete case analysis. A previous study found a linear relationship between HRQL, assessed using the Assessment of Quality of Life instrument, and MMSE in individuals with Alzheimer's disease  -- a finding similar to that of my imputed data set analysis.48 Further, the ―absence of evidence is not evidence of absence.‖49 Bland and Altman highlighted that if s tudy findings are not statistically significant this is not an indication that these findings are indeed nonsignificant or not of clinical important. Ra ther, when studies lack the necessary power to detect real, and clinically worthwhile, differences in treatment, we should not interpret or conclude that this is necessarily evidence of no effect. Therefore, because our findings are consistent with one previous study,48 we interpret the discrepant results as a lack of statistical power to detect a difference given the smaller sample size in our complete case analysis. 7.4.5 Timed up and go was a key explanatory variable in our model  I found that the TUG32 was most strongly associated with HRQL in my bivariate analyses, accounting for 27% of the variation in QALYs. One previous study found that functional ability/pain explained most of the variation in global utility score; however, this assessment was not based on a specific measure of mobility such at the TUG.50 Three previous studies investigated the association between the TUG and the Physical Function domain of the SF-3651-53 and two indicated the TUG explained approximately 20% of the variation.52,53 Therefore, my findings, the first using the EQ-5D in this context, demonstrated that the clinically useful TUG test is strongly associated with EQ-5D HSUVs.    205 7.4.6 Conclusions  As my small study sample consisted only of older community dwelling women who were cognitively intact, I cannot generalize to older women with mild cognitive impairment or dementia, older men, other age groups, or to adults in care facilities. Thus, my study highlights the need for future prospective studies to ascertain whether this finding applies to other clinical populations and whether changes in executive functions are causally linked to changes in HRQL assessed using generic preference based HSUV instruments, such as the EQ-5D. These findings indicate that EQ-5D HSUVs over time can be largely explained by baseline measures of age, waist girth, functional comorbidity index, general mobility, global cognition and executive functions. Given that executive functions at a certain time point explain a statistically significant amount of variability in QALYs in the following year, clinicians may need to consider assessing executive functions to indicate the person‘s likely quality of life.       206 7.5 References  1.  Sadatsafavi M, Marra CA, Ayas NT, Stradling J, Fleetham J. Cost-effectiveness of oral appliances in the treatment of obstructive sleep apnoea-hypopnoea. Sleep Breath 2009;13(3):241-52. 2.  Sach TH, Foss AJ, Gregson RM, Zaman A, Osborn F, Masud T, et al. Falls and health status in elderly women following first eye cataract surgery: an economic evaluation conducted alongside a randomised controlled trial. Br J Ophthalmol 2007;91(12):1675-9. 3.  Matza LS, Swensen AR, Flood EM, Secnik K, Leidy NK. Assessment of health-related quality of life in children: a review of conceptual, methodological, and regulatory issues. Value Health 2004;7(1):79-92. 4.  World Health Organization: Constitution of the World Health Organization Geneva: WHO; 1948. 5.  Revicki DA, Osoba D, Fairclough D, Barofsky I, Berzon R, Leidy NK, et al. Recommendations on health-related quality of life research to support labeling and promotional claims in the United States. Qual Life Res 2000;9(8):887-900. 6.  Coons SJ. Health-related quality of life: let's measure and report it appropriately. Clin Ther 2007;29(12):2746-7. 7.  Guyatt GH, Feeny DH, Patrick DL. Measuring health-related quality of life. Ann Intern Med 1993;118(8):622-9. 8.  Marra CA, Woolcott JC, Kopec JA, Shojania K, Offer R, Brazier JE, et al. A comparison of generic, indirect utility measures (the HUI2, HUI3, SF-6D, and the EQ-5D) and disease-specific instruments (the RAQoL and the HAQ) in rheumatoid arthritis. Soc Sci Med 2005;60(7):1571-82. 9.  Dolan P. Modeling valuations for EuroQol health states. Med Care 1997;35(11):1095-108. 10.  Torrance GW. Measurement of health state utilities for economic appraisal. J Health Econ 1986;5(1):1-30.   207 11.  Torrance GW. Utility approach to measuring health-related quality of life. J Chronic Dis 1987;40(6):593-603. 12.  Drummond MF, Sculpher MJ, Torrance GW, O'Brien B, Stoddart GL. Methods for the economic evaluation for health care programmes. Third edition. New York. United States of America: Oxford University Press, 2005. 13.  Neumann PJ, Goldie SJ, Weinstein MC. Preference-based measures in economic evaluation in health care. Annu Rev Public Health 2000;21:587-611. 14.  Morey MC, Snyder DC, Sloane R, Cohen HJ, Peterson B, Hartman TJ, et al. Effects of home - based diet and exercise on functional outcomes among older, overweight long -term cancer survivors: RENEW: a randomized controlled trial. JAMA 2009;301(18):1883-91. 15.  Jiang Y, Hesser JE. Patterns of health-related quality of life and patterns associated with health risks among Rhode Island adults. Health Qual Life Outcomes 2008;6:49. 16.  Rabins PV, Black BS. Measuring quality of life in dementia: purposes, goals, challenges and progress. Int Psychogeriatr 2007;19(3):401-7. 17.  Jonsson L, Andreasen N, Kilander L, Soininen H, Waldemar G, Nygaard H, et al. Patient- and proxy-reported utility in Alzheimer disease using the EuroQoL. Alzheimer Dis Assoc Disord 2006;20(1):49-55. 18.  Royall DR, Lauterbach EC, Cummings JL, Reeve A, Rummans TA, Kaufer DI, et al. Executive control function: a review of its promise and challenges for clinical research. A report from the Committee on Research of the American Neuropsychiatric Association. J Neuropsychiatry Clin Neurosci 2002;14(4):377-405. 19.  Stuss DT, Alexander MP. Executive functions and the frontal lobes: a conceptual view. Psychol Res 2000;63(3-4):289-98. 20.  Kuo HK, Lipsitz LA. Cerebral white matter changes and geriatric syndromes: is there a link? J Gerontol A Biol Sci Med Sci 2004;59(8):818-26.   208 21.  Liu-Ambrose T, Nagamatsu LH, Graf P, Beattie L, Ashe MC, Handy T. The effect of resistance training on brain function among community-dwelling senior women: a randomized controlled trial. Archives of Internal Medicine 2010;170(2):170-8. 22.  Groll DL, To T, Bombardier C, Wright JG. The development of a comorbidity index with physical function as the outcome. J Clin Epidemiol 2005;58(6):595-602. 23.  Folstein MF, Folstein SE, McHugh PR. "Mini-mental state". A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 1975;12(3):189-98. 24.  Royall DR, Polk M. Dementias that present with and without posterior cortical features: an important clinical distinction. J Am Geriatr Soc 1998;46(1):98-105. 25.  Miyake A, Emerson MJ, Friedman NP. Assessment of executive functions in clinical settings: problems and recommendations. Semin Speech Lang 2000;21(2):169-83. 26.  Spreen O, Strauss E. A compendium of Neurological Tests, 2nd edition. New York: Oxford University Press, 1998. 27.  Wechsler D. Wechsler Adult Intelligence Scale--Revised San Antonio: Psychological Corporation, 1981. 28.  Trenerry M, Crosson B, De Boe J. Stroop Neuropsychological Screening Test. Odessa, Florida: Psychological Assessment Resources, Incoorporated, 1989. 29.  Feeny D, Furlong W, Torrance GW, Goldsmith CH, Zhu Z, DePauw S, et al. Multiattribute and single-attribute utility functions for the health utili ties index mark 3 system. Med Care 2002;40(2):113-28. 30.  Brooks R, Robin R, de Charro F. The measurement and valuation of health status using EQ-5D: a European perspective (evidence from the EuroQoL BIOMED research programme), . Netherlands: Kluwer Academic Publishers, 2003. 31.  Conner-Spady B, Suarez-Almazor ME. Variation in the estimation of quality-adjusted life-years by different preference-based instruments. Med Care 2003;41(7):791-801.   209 32.  Podsiadlo D, Richardson S. The timed "Up & Go": a test of basic functional mobility for frail elderly persons. J Am Geriatr Soc 1991;39(2):142-8. 33.  Oostenbrink JB, Al MJ. The analysis of incomplete cost data due to dropout. Health Econ 2005;14(8):763-76. 34.  Ostbye T, Crosse E. Net economic costs of dementia in Canada. Cmaj 1994;151(10):1457-64. 35.  Kesse-Guyot E, Castetbon K, Estaquio C, Czernichow S, Galan P, Hercberg S. Association between the French nutritional guideline-based score and 6-year anthropometric changes in a French middle-aged adult cohort. Am J Epidemiol 2009;170(6):757-65. 36.  Iersel MBv, Kessels RPC, Bloem BR, Verbeek ALM, Olde Rikkert MGM. Executive Functions Are Associated With Gait and Balance in Community-Living Elderly People. J Gerontol A Biol Sci Med Sci 2008;63(12):1344-1349. 37.  Liu-Ambrose T, Katarynych LA, Ashe MC, Nagamatsu LS, Hsu CL. Dual -task gait performance among community-dwelling senior women: the role of balance confidence and executive functions. J Gerontol A Biol Sci Med Sci 2009;64(9):975-82. 38.  Stoehr GP, Lu SY, Lavery L, Bilt JV, Saxton JA, Chang CC, et al. Factors associated with adherence to medication regimens in older primary care patients: the Steel Valley Seniors Survey. Am J Geriatr Pharmacother 2008;6(5):255-63. 39.  Kantor B, Mauger L, Richardson VE, Unroe KT. An analysis of an older driver evaluation program. J Am Geriatr Soc 2004;52(8):1326-30. 40.  Johnson DR. Emotional attention set-shifting and its relationship to anxiety and emotion regulation. Emotion 2009;9(5):681-90. 41.  Weiner DK, Rudy TE, Morrow L, Slaboda J, Lieber S. The relationship between pain, neuropsychological performance, and physical function in community -dwelling older adults with chronic low back pain. Pain Med 2006;7(1):60-70.   210 42.  Lawton MP, Brody EM. Assessment of older people: self-maintaining and instrumental activities of daily living. Gerontologist 1969;9(3):179-86. 43.  Chan SW, Chiu HF, Chien WT, Goggins W, Thompson D, Hong B. Predictors of change in health- related quality of life among older people with depression: a longitudinal study. Int Psychogeriatr 2009;21(6):1171-9. 44.  Devlin N, Parkin D, Brown J. Using the EQ-5D as a performance measurement tool in the NHS. 2009. 45.  Cahn-Weiner DA, Boyle PA, Malloy PF. Tests of executive function predict instrumental activities of daily living in community-dwelling older individuals. Appl Neuropsychol 2002;9(3):187-91. 46.  Carlson MC, Fried LP, Xue QL, Bandeen-Roche K, Zeger SL, Brandt J. Association between executive attention and physical functional performance in community -dwelling older women. J Gerontol B Psychol Sci Soc Sci 1999;54(5):S262-70. 47.  Jefferson AL, Paul RH, Ozonoff A, Cohen RA. Evaluating elements of executive functioning as predictors of instrumental activities of daily living (IADLs). Arch Clin Neuropsychol 2006;21(4):311- 20. 48.  Wlodarczyk JH, Brodaty H, Hawthorne G. The relationship between quality of life, Mini -Mental State Examination, and the Instrumental Activities of Daily Living in patients with Alzheimer's disease. Arch Gerontol Geriatr 2004;39(1):25-33. 49.  Altman DG, Bland JM. Absence of evidence is not evidence of absence. BMJ 1995;311(7003):485. 50.  Marra CA, Esdaile JM, Guh D, Kopec JA, Brazier JE, Koehler BE, et al. A comparison of four indirect methods of assessing utility values in rheumatoid arthritis. Med Care 2004;42(11):1125-31. 51.  Peri K, Kerse N, Robinson E, Parsons M, Parsons J, Latham N. Does functionally  based activity make a difference to health status and mobility? A randomised controlled trial in residential care facilities (The Promoting Independent Living Study; PILS). Age Ageing 2008;37(1):57-63.   211 52.  Syddall HE, Martin HJ, Harwood RH, Cooper C, Aihie Sayer A. The SF-36: a simple, effective measure of mobility-disability for epidemiological studies. J Nutr Health Aging 2009;13(1):57-62. 53.  Teixeira LE, Silva KN, Imoto AM, Teixeira TJ, Kayo AH, Montenegro -Rodrigues R, et al. Progressive load training for the quadriceps muscle associated with proprioception exercises for the prevention of falls in postmenopausal women with osteoporosis: a randomized controlled trial. Osteoporos Int 2009.   212 8 Integrated Discussion The purpose of this chapter is to examine my research from a wider perspective – to review Chapters 2 through 7 (Studies 1 to 6) with respect to the question, ‗What does this body of research mean and what does it add to the current state of literature?‘ I posit that my research adds to aging re search in three ways. First, the research contributes health economics information in a field that had been largely devoid of any form of economic enquiry (i.e., ‗new questions‘). Second, I applied some contemporary economic methods to provide more robust answers than would have been possible in the 1990s (i.e., ‗new methods in this field‘). Third, certain findings have immediate relevance to health policy decisions (i.e., ‗new information‘). Tempering these claims of innovation, I also highlight challenges I encountered. The chapter concludes with my vision for how this research area can be advanced in the near term (3-5 years).  This thesis includes the first studies to: quantify the cost of falls in community dwelling seniors, discern the best value for money among falls prevention strategies and hone in on health economic benefits of resistance training in older women. The rationale for the innovations was outlined in the 'Overview of thesis' and 'Literature review' (Chapter 1). In the sections below I outline where these studies extend methods that had previously been applied in this field of research (8.1) or have immediate applicability for decision- makers (8.2). 8.1 Contemporary economic methods and methodological advances In the following five subsections I discuss five economic methods (Sections 8.1.1-8.1.5) developed between 1994 and 2008 that I used in my six studies and that had not previously been applied in this field. Their usefulness emphasizes the importance of having expert health economists (i .e., my economic mentors) engage in research into significant health problems (e.g., falls in seniors) so that these newly recognized problems benefit from the advances that have already been made in fields where health economics has a   213 longer track record.1,2 The first area where I applied established economics methods to falls was in developing cost items for my CEA and CUAs. 8.1.1 Development of cost items for cost-effectiveness and cost-utility analyses Prior to my papers in Chapters 4 and 5, most economic evaluations of falls prevention strategies had not quantified both total and fall related health care resource use when estimating incremental cost effectiveness ratios. To address this shortcoming, I included both these categories of costs in my papers and also developed and published a list of essential cost items that should be ascertained for future falls prevention economic evaluations (Chapter 3, Table 3-3). Broadly, these cost categories included: 1) cost of implementing the intervention, 2) comparator group intervention costs, 3) fall related health care resource use, 4) total health care resource use and 5) personal out-of-pocket expenses. There is some debate among academics regarding the importance of including implementation costs of the intervention, but it is essential to allow decisions makers to consider maintaining status quo, investment and disinvestment. If this list of essential cost items gains traction, it will mean future economic evaluations will have better comparability and consistency. 8.1.2 Valuation of cost items and assessment of study quality For cost of illness studies, Rice3 first formalized the broad list of cost categories that need to be ascertained. These are: 1) direct costs, 2) indirect costs and 3) intangible costs. Despite these broad categories, there has been little research detailing specific components included in each of these three categories in the context of falls in older adults. I found that the economic costs of falls are likely greater than Canadian policy makers currently appreciate (Chapter 2) because estimates internationally are larger than previously reported in Canada. The estimates produced in Canada for the cost of fall related injuries are currently in the order of magnitude of hundreds of thousands of Canadian dollars, 4 whereas pro rata these estimates should be in the millions of dollars pe r year (UK estimate 981 million pounds sterling in 20005 and US estimates in the order of billions of US dollars in 2000). 6 Some reasons for differences in   214 international cost of falls estimates are likely due to different: 1) population sizes, 2) methodologies for cost of illness studies including the different ways falls were classified and 3) quality of included studies. Items two and three are methodology factors that can be  improved in future studies; therefore, I developed a list of categories (Chapter 2, Tables 2-1 to 2-3) that should be reported in cost of falls studies, and I applied a modified version of Drummond‘s checklist for economic evaluations 7 that should be considered when interpreting cost of falls studies (Table 2-4). The following should be included in all cost of falls studies: population denominator, time horizon, perspective, currency, country, year, costing approach (incidence versus prevalence) and method of collecting costs (self report versus database). Cost of falls studies should also include where relevant the following breakdown of cost items: 1) inpatient, 2) healthcare professional, 3) home healthcare, 4) emergency department, 5) ambulance,  6) medications, 7) out of pocket (personal), 8) funder and 9) non-injurious falls. To my knowledge, this was the first cost of falls systematic review that designed a method for detailing cost items and that categorized these estimates by geography and design of each study.8  Of note, a recent systematic review that detailed cost of falls by study setting did not define specific cost items that are a necessary inclusion in cost of falls studies.8 This study compared mean costs per fall, per fall related injury or per person based on study setting. However, given that each country has a different health care system and different population demographics, a more appropriate method of providing an international estimate for the cost of falls is by geographic region and study design. Thus, my cost of falls study can build upon recent literature by detailing key economic components for conducting, reporting and estimating the cost of falls in the future. 8.1.3 Multiple imputation of missing data for economic evaluations alongside clinical trials As missing data are almost inevitable in clinical trials, there is a need to minimize data loss. Oostenbrink and Briggs9-11 pioneered multiple imputation of missing data for economic evaluations alongside clinical   215 trials;9-11 these techniques are well-described in some other areas of health care.12,13  I applied them to this field Chapters 4 and 5 (Study 3 and Study 4) and these are the first examples of an economic evaluation in a falls prevention RCT that included multiple imputation of missing data.9-11 The benefit of using multiple imputation for missing economic data is two-fold: it minimizes bias in estimating the ICER and there is no loss of power that is caused by missing data. Of note, there are no established rules for the extent of missing data under which it is no longer appropriate to carry out the analysis, but previous economic evaluations generally have imputed up to 30% of missing data.12,13 8.1.4 Probabilistic sensitivity analyses to ascertain the uncertainty around the point estimates of costs and health benefits A relatively recent innovation in health economics has been probabilistic sensitivity analyses to ascertain the degree of uncertainty around the point estimates of costs and health outcomes. This was relevant to Chapters 4 and 5 (Study 3 and Study 4) where I applied the methods of bootstrapping, selective one way and probabilistic sensitivity analyses to this field.14 In Study 3 (Chapter 4), I concluded that over 80% of ICER point estimates were in the southeast quadrant of the cost-effectiveness plane when comparing resistance training with balance and tone classes (Chapter 1, Figure 1-6). Although these techniques are becoming widely accepted in other areas of health care, they are not the norm for research among older adults in any field.  Previous economic evaluations reported the incremental costs per fall prevented solely as a point estimate with or without a test of significance or p-value for the effectiveness outcome measure. Only one falls prevention study reported estimates of uncertainty using probabilistic sensitivity analyses. 15 My thesis extends this particular s tudy by providing estimates for uncertainty that are not constrained by a p -value.16 To date, many researchers adhered to the premise that there needs to be a statistically significant benefit in the primary or secondary outcomes for a new intervention in order to warrant the need for an economic   216 evaluation.16 My paper will draw attention to this issue, and I hope the results and discussion in that paper (Sections 4.3 and 4.4) will provide useful edification. Further, my research may enable more appropriate consideration of the estimates of uncertainty around the mean, a significant improvement in the field of falls prevention research because of the variability in estimating both falls and fall related HRU.  To conclude, there has been a widely held dogma in relation to economic evaluations that if there is no significant difference between groups in the primary or secondary clinical o utcome upon which the economic evaluation is based, this mitigates the motive for conducting an economic evaluation. This notion is false. Karl Claxton16 states: ―Decisions should be based only on the posterior mean irrespective of the level of significance or whether [the confidence intervals for the ICER] falls outside a Bayesian range of equivalence.‖ In practical terms, this makes sense at a policy level because policy makers are faced with making decisions between competing alternatives (i.e., there must be a choice and  for every choice there is an opportunity cost between mutually exclusive alternatives).16 8.1.5 In depth comparison of the EQ-5D and the SF-6D among older women As medicine strives to be continually more objective and based on evidence, it means that when multiple instruments exist, they need to be compared against each other and tested in specific settings and amongst specific populations. Two widely used instruments for measuring HRQL are the EQ-5D and the SF-6D9-13 (see sections 1.2.4.1.3.1 and 1.2.4.1.3.2) but they have rarely been compared among a group of older adults.17 In a cohort of older women, I carefully compared the ICERs in my CUA in Study 3 (Chapter 4) calculated using the EQ-5D and the SF-6D, and for differences in key predictors that explain variation in QALYs estimated from the EQ-5D and SF-6D (see Study 5 (Chapter 6)). This led to the important finding that the incremental cost effectiveness ratio point estimate is dependent in part on the instrument used to measure HRQL, but this does not make a difference in terms of the final conclusion as represented on the cost-effectiveness plane (Chapter 1, Figure 1-6).   217  I hope that the detailed comparison of the EQ-5D and the SF-6D in this thesis may mark a turning point in a movement to encourage the use of generic preference based utility instruments among aging adults. The reason that CUAs may not currently be prominent in the field of falls prevention in older adults may be the: 1) challenge of identifying and measuring appropriate outcomes and 2) lack of sensitivity of the outcome measurement scale.18 The SF-6D and the EQ-5D may not be responsive among fallers, thus these instruments may fail to capture benefits that may accrue from the intervention. 18 In clinical trials, the SF-6D and EQ-5D are used to estimate QALYs, capturing both length and quality of life. Harwood and colleagues18 argue that these are not the only important goals. ―Success‖ could also be measured by: 1) prevention or curing disease or related complications, 2) relief of distressing symptoms or 3) improving physical and social function(s).18 These types of valuable changes may not easily generate a meaningful change in a HSUV and thus not generate any significant change in QALYs. The solution to this problem is not simple, but I see essential first steps as being: 1) compare these generic preference based utility instruments (as in Study 5, Chapter 6) in individuals at low, moderate and high risk of falls with those at no risk, and 2) assess the responsiveness of each of these instruments in specific population of fallers (please see section 8.4). 8.2 New information and knowledge transfer for health policy Ultimately, the goal of clinical research is to inform policy decisions, whether they be through maintaining status quo, investment or disinvestment.19 To adequately address falls and the burden of fall related injury, it is essential to engage policy makers by providing information about specific interventions and their value for money when delivered to a specific population.19 Although there is increasing use of evidence-informed policy making based on efficacy and effectiveness data, this trend is not currently seen in the context of falls prevention research.20,21 Based on my findings, I recommend that future researchers studying fall prevention strategies engage policy makers when designing the trial and outcome measures.22-24 This will   218 enable policy makers to inform researchers of the outcomes they use to make decisions and will give researchers the opportunity to use their research for appropriate knowledge transfer.22-25  Because of my commitment to knowledge translation and what one might call ‗impact research‘ (research that has an impact in the near-term) my thesis included two studies that had potential to influence policy makers immediately. Chapter 2 of this thesis provides the ‘benchmark’ health economic cost of illness for falls. The paper is descriptive in nature (i.e., no meta-analysis) but this provides an essential platform for more analytical studies in the future. The paper informs policy makers of the potential efficiency of falls prevention related resource use. Further, and highly relevant to fall prevention research is the point that very few health care interventions are understood in terms of their effectiveness in specific populations and proven to be cost-effective.26,27 Therefore, a reasonable starting point is the assumption that the higher the cost of illness estimate, the greater the scope to increase productive efficiency.26,27 This in turn may result in a greater gain in benefits from health care resources from effective health care interventions. 26,27  The second study that should influence policy makers has already been discussed in the New Zealand parliament (Wellington, October 2009). I determined which falls prevention strategies provided the best value for money (Chapter 3).28 Specifically, this included the following high risk groups:  1) a multifactorial program targeted at eight fall risk factors, 2) the home-based OEP delivered to people aged ≥ 80 years, and 3) a home safety program for those recently discharged from hospital, if delivered to the subgroup of participants with a previous fall. These findings highlight the importance of determining specific populations for which falls prevention interventions provide the best health and economic benefits. As many governments contribute in at least some way to falls prevention efforts, and the OEP is used in at least five countries I know of, this study has immediate relevance in not only New Zealand and Canada (where the OEP is used in BC) but also Australia, the UK, the US and throughout the world. The paper gives guidance   219 to all countries on best value for money for falls prevention, it is not solely limited to where the OEP is used currently.  An outcome of increasing importance to health care policy makers is patient reported HRQL. 29 In exploring the independent contributors to HRQL in older women, I discovered the independent association between key executive processes as measured by standard neuropsychological tests, and QALYs measured prospectively over one year among high functioning community dwelling senior women (Study 6, Chap ter 7). Noteworthy is that this independent association can now provide evidence to health care professionals that cognitive performance is an important contributor to HRQL. In turn, this study can provide a basis for future interventions and health technologies aimed at improving HRQL in older adults. 8.3 Limitations With respect to my interpretation of these thesis findings, I acknowledge the important limitations that I detailed within each specific chapter (i.e., Chapters 2-7). In this section, I provide an overview of the most prominent limitations in this thesis. The most important limitation of this thesis is that the idea to perform the economic evaluation of the Brain Power study postdated the design of the clinical trial. The economic evaluation preceded trial initiation, but there was little flexibility in design given the competing interests of the Brain Power clinical study. As a result, HRU data were only collected for nine of the 12 months for the first year economic evaluation. In general, economic evaluation endpoints should be at least equal length to the clinical trial.  A key limitation is that the Brain Power study was not powered to see a reduction in falls. It was powered for a change in cognitive performance via neuropsychological testing. A reduction in falls risk profile (not falls per se) was a secondary outcome of the Brain Power study. Further, the EQ-5D was administered via mail out for the entire study period. This mode of delivery is appropriate, but it did contribute to the issue of missing data, with approximately 30% of data missing after the 12 -month intervention period.   220  Although comparisons between the SF-6D and the EQ-5D were designed a priori in the economic evaluation for the Brain Power study, they were not initially includ ed in the clinical trial protocol. Therefore, these comparisons were also not included in the original power calculations. Thus, the comparisons between the SF-6D and the EQ-5D are classified as exploratory analyses. This study included women only. Men were excluded from the Brain Power study because men and women‘s cognitive responses to exercise may differ. This limits the generalizability of some of the results from this thesis. 8.4 Future directions for economic research among older adults In order to determine future directions for efficient delivery of falls prevention strategies, it is important to acknowledge where the specific field is positioned within a conceptual ‗research cycle‘ (Chapter 1, Figure 1-1). The first step is in ascertaining the burden o f disease (i.e., cost of illness).30 Falls among older people living independently in the community are costly (Study 1, Chapter 2), but preventable (Study 2, Chapter 3).31 Given that we have substantive evidence for effective falls prevention, future research should be focused on efficiency. Efficiency is rooted in generating: 1) cost of illness estimates and 2) evaluations of value for money. With regard to efficiency, I provide a brief overview of why cost of illness studies should be considered an essential component of our future literature base. In considering the future  directions for economic research among older adults, it may also be important to stratify each analysis on the basis of sex. 8.4.1 The current limitations of fall-related economic evaluations – new potential for methodological innovations for future studies Considering costs and benefits together, there are a number of methodological issues of current economic evaluations of falls prevention strategies, which I will detail in this section. 1) Of the tested methods for prospective collection of falls data, fall d iaries that are completed daily and returned each month by participants is considered the current gold standard.32 For economic evaluations, the measure of   221 effectiveness needs to be objective, when possible.7 A fall is a tangible unit (an event). It is not free from issues related to self report (i.e., recall and response bias), but using monthly falls diaries is a tested reliable and valid method of tracking falls.33 2) Another limitation is that the health interventions designed to reduce falls may also have other health benefits that are not covered by measuring the number of falls prevented.33 3) Further, a crude estimate of the number of falls prevented does not account for differences in group sizes, drop out rates or followup times (i.e., person-time at risk). These three factors limit the conclusions that can be made for health policy. I propose that future cost effectiveness analyses of falls prevention strategies determine the incremental cost per average number of fal ls prevented. This methodology will be consistent with CEAs in other disciplines.12 8.4.1.1 Relative use for cost effectiveness or cost utility analyses Ideally, an opportune point to conduct a cost effectiveness analysis is when you have a given budget allocated for a limited range of options within a specific field (i.e., falls prevention). 7,34 One example of relevant units in the context of falls prevention would be to compare a home safety intervention with a home based strength and balance training intervention in terms of falls prevented. One key feature of cost effectiveness analyses is that the measure of effectiveness should be as objective as possible. 7,34  Further, interventions are often designed to result in multiple health benefits. It is possible that the number of falls prevented may not be the best possible outcome for all falls prevention economic evaluations because other important benefits such as quality of life or cognition may also be obtained. If this is suspected, there are two possible approaches that can be taken: 1) for each of the al ternative interventions, present an array of differences for each of the effectiveness dimensions measured or 2) conduct a cost utility analysis.7,34   222 8.4.1.2 Recommendation for the use of cost-utility analyses Despite the potential benefit of CUAs in falls prevention research, the lack of published CUAs is a current limitation.7 Cost-utili ty analyses should be used in the following situations when: 1) HRQL is the, or an, important outcome, 2) the intervention affects both quality and quantity of life, 3) comparator programs have a wide and large range of outcomes, 4) comparing with other programs that were evaluated using CUAs, 5) the budget is limited and/or 6) the goal is to allocate limited resources most effectively by comparing all possible options.7 In the context of falls prevention, the addition of CUAs to the current body of research would be useful so that in the future, falls prevention strategies are valued in the ‗total body of literature‘ that is being considered when decision makers are allocating health resources. 8.4.1.3 Consistency in future economic evaluations – uniform assessment of health benefits From the comparison of the EQ-5D and SF-6D, my study in Chapter 6 demonstrates that depending on the generic preference based utility instrument used, the conclusions of CUAs may differ. 35,36 This suggests that a uniform approach be adopted for decision makers to allocate health resources consistently. The UK National Institute of Health and Clinical Excellence (NICE) aims to follow this principle. UK research funders routinely request inclusion of an economic component in grant proposals; the EQ-5D must be administered in a number of clinical settings.29 These data are influential and essential in prioritizing decisions made by NICE. Therefore, before these economic evaluation data are used to inform policy, it will be valuable to provide a set list of criteria for the assessment of health benefits of new health care technologies. 8.4.1.3.1 Consensus on the utility instrument used that is most responsive fo r population of fallers: what is the construct validity of the utility instruments for falls? Construct validity in the context of utility instruments is defined as the ability of the instrument, in this case the EQ-5D and SF-6D, to detect expected differences in severity and health care utilizations.35 The logic of testing construct validity is to determine whether the data are consistent with theoretical constructs. As suggested by consultation with Dr Marra and colleagues,35 the ideal and novel study design would be one   223 where each of the utility instruments are administered monthly for 12 months among a population of fallers. These longitudinal data could then be used to assess the ‗responsiveness‘ of each of the preference -based utility instruments by comparing falls with changes in HRQL as assessed by each of the utili ty instruments.35 Specific domains for each instrument could also  be compared.35 Together, these analyses constitute the next step for fall prevention studies that use HRQL as a primary outcome. 8.4.1.3.2 Reporting of ICER as cost/mean falls instead of cost/number of falls Future economic evaluations of falls prevention strategies should detail the ICER as the average incremental cost per average number of falls prevented instead of the crude number of falls prevented. However it is also argued that the crude number of falls prevented is an appropriate denominator for a ICER because falls are count data, and are often not normally distributed; therefore, presenting a mean for falls data is incorrect. This would be consistent with other areas of applied health economics. 12 Currently, economic evaluations report a crude estimate of the number of falls prevented. 37-40 This is reasonable when there is no loss to followup and group sizes are equivalent but this is rarely the case in clinical t