Advertisement

Cost Effectiveness of Community-Based Physical Activity Interventions

      Background

      Physical inactivity is associated with the increased risk of many chronic diseases. Such risks decrease with increases in physical activity. This study assessed the cost-effectiveness of population-wide strategies to promote physical activity in adults and followed disease incidence over a lifetime.

      Methods

      A lifetime cost-effectiveness analysis from a societal perspective was conducted to estimate the costs, health gains, and cost-effectiveness (dollars per quality-adjusted life year [QALY] gained, relative to no intervention) of seven public health interventions to promote physical activity in a simulated cohort of healthy U.S. adults stratified by age, gender, and physical activity level. Interventions exemplifying each of four strategies strongly recommended by the Task Force on Community Preventive Services were evaluated: community-wide campaigns, individually adapted health behavior change, community social-support interventions, and the creation of or enhanced access to physical activity information and opportunities. Each intervention was compared to a no-intervention alternative. A systematic review of disease burden by physical activity status was used to assess the relative risk of five diseases (coronary heart disease, ischemic stroke, type 2 diabetes, breast cancer, and colorectal cancer) across a spectrum of physical activity levels. Other data were obtained from clinical trials, population-based surveys, and other published literature.

      Results

      Cost-effectiveness ratios ranged between $14,000 and $69,000 per QALY gained, relative to no intervention. Results were sensitive to intervention-related costs and effect size.

      Conclusions

      All of the evaluated physical activity interventions appeared to reduce disease incidence, to be cost-effective, and—compared with other well-accepted preventive strategies—to offer good value for money. The results support using any of the seven evaluated interventions as part of public health efforts to promote physical activity.

      Introduction

      Physical inactivity is a global public health problem, and has been identified as a national public health priority.
      WHO
      A global strategy for diet, physical activity, and health.
      WHO
      Preventing chronic diseases: a vital investment.
      USDHHS
      Physical activity and health: a report of the Surgeon General.
      There is clear evidence to link physical inactivity with an increased risk of many chronic diseases, including coronary heart disease (CHD), ischemic stroke, type 2 diabetes, breast cancer, and colorectal cancer.
      WHO
      A global strategy for diet, physical activity, and health.
      WHO
      Preventing chronic diseases: a vital investment.
      USDHHS
      Physical activity and health: a report of the Surgeon General.
      The negative health effects of physical inactivity are paralleled by staggering economic consequences: the annual cost directly attributable to inactivity in the U.S. is an estimated $24 billion–$76 billion, or 2.4%–5.0% of national healthcare expenditures.
      • Colditz G.A.
      Economic costs of obesity and inactivity.
      • Pratt M.
      • Macera C.A.
      • Wang G.
      Higher direct medical costs associated with physical inactivity.
      • Finkelstein E.A.
      • Wang G.
      • Lee I.M.
      National and state-specific inactivity-attributable medical expenditures for six diseases Final report prepared for the Centers for Disease Control and Prevention by the Research Triangle Institute.
      Fortunately, modest increases in physical activity have the potential to produce substantial health benefits.
      • Haskell W.L.
      • Lee I.
      • Pate R.R.
      • et al.
      Physical activity and public health: updated recommendation for adults from the American College of Sports Medicine and the American Heart Association.
      Further, systematic reviews of population-based interventions to promote health and prevent disease have provided strong evidence that public health efforts can successfully increase physical activity.
      • Kahn E.B.
      • Ramsey L.T.
      • Brownson R.C.
      • et al.
      The effectiveness of interventions to increase physical activity: a systematic review.
      • Truman B.I.
      • Smith-Akin C.K.
      • Hinman A.R.
      • et al.
      Developing the Guide to Community Preventive Services—overview and rationale The Task Force on Community Preventive Services.
      It appears that sustained physical activity initiatives could make substantial contributions to the control of chronic diseases. However, the potential benefits of physical activity promotion have not yet been fully realized, and the majority of adults in the U.S. (54.1%) do not engage in sufficient physical activity to meet public health recommendations.
      • Sapkota S.
      • Bowles H.R.
      • Ham S.A.
      • Kohl III, H.W.
      Adult participation in recommended levels of physical activity—U.S., 2001 and 2003.
      This situation presents an important opportunity to evaluate the merits of competing physical activity promotion strategies and to develop effective public health policy. The need for decision making in an environment of uncertainty, scarcity, and competing priorities makes the use of cost-effectiveness analysis (CEA) attractive for public health planning.
      These analyses attempt to answer the question How much health improvement can be gained when an intervention is compared, dollar for dollar, with an alternative? Rather than promoting cost savings, the goal of CEA is to determine how money can be spent with maximum public health benefit. Decision-analytic models have been used routinely to guide important public health policy decisions, from airbag regulation in motor vehicles and the widespread use of pneumococcal vaccine in older adults
      • Graham J.D.
      • Thompson K.M.
      • Goldie S.J.
      • Segui-Gomez M.
      • Weinstein M.C.
      The cost-effectiveness of air bags by seating position.
      • Hirschmann J.V.
      Pneumococcal vaccination in older adults.
      to the improvement of cervical cancer screening practices globally.
      • Goldie S.J.
      • Gaffikin L.
      • Goldhaber-Fiebert J.D.
      • et al.
      Cost-effectiveness of cervical-cancer screening in five developing countries.
      Models have also been used to influence clinical practice guidelines for cardiovascular disease prevention
      • Gaspoz J.M.
      • Coxson P.G.
      • Goldman P.A.
      • et al.
      Cost effectiveness of aspirin, clopidogrel, or both for secondary prevention of coronary heart disease.
      • Kuntz K.M.
      • Fleischmann K.E.
      • Hunink M.G.
      • Douglas P.S.
      Cost-effectiveness of diagnostic strategies for patients with chest pain.
      and to promote national anti-tobacco education efforts.
      • Tengs T.O.
      • Osgood N.D.
      • Chen L.L.
      The cost-effectiveness of intensive national school-based anti-tobacco education: results from the tobacco policy model.
      Markov models, which consider probabilistic events over time, are particularly well-suited to evaluate population-based health promotion efforts and to examine outcomes over an extended period of time.
      The purpose of this study was to evaluate the cost-effectiveness (dollars per quality-adjusted life year [QALY] gained) of seven exemplar interventions to increase physical activity, relative to no intervention, as well as to follow disease incidence over a lifetime. This study is the first to integrate the best available epidemiologic and intervention data on physical activity into a CEA of the most promising public health interventions for physical activity promotion.

      Methods

      The CDC Measurement of the Value of Exercise (MOVE) Model

      A comprehensive, flexible, state-transition Markov model was developed from a societal perspective to estimate the lifetime costs, health gains, and cost-effectiveness of population interventions that promote physical activity among U.S. adults. Cost-effectiveness is defined as the ratio of incremental costs (dollars) to incremental QALYs. The incremental cost is the difference between the total expected cost of the intervention and the total expected cost of no intervention. Incremental QALYs are the difference between the total expected QALYs associated with the intervention and the total expected QALYs with no intervention. The performance of interventions evaluated by this model was measured using the cost-effectiveness ratio (dollars per QALY gained). The Guide to Community Preventive Services (Community Guide), an evidence-based review of physical activity promotion, strongly recommends four strategies for adults: community-wide campaigns, individually adapted health behavior change, social-support interventions in community settings, and the creation of or enhanced access to places for physical activity combined with informational outreach activities.
      • Kahn E.B.
      • Ramsey L.T.
      • Brownson R.C.
      • et al.
      The effectiveness of interventions to increase physical activity: a systematic review.
      Exemplar interventions representing each of these strategies were selected for the model (discussed below in Interventions).

      Study Design

      The simulation was started with a closed cohort the size of the U.S. adult population aged 25–64 years in 2004. The cohort was stratified by age, gender, and level of physical activity. At the beginning of the model, it was assumed that all cohort members were well, which was defined as the absence of five diseases: CHD, ischemic stroke, type 2 diabetes, breast cancer, and colorectal cancer. These are the diseases for which the strongest evidence exists associating regular physical activity with lower disease risk.
      USDHHS
      Physical activity and health: a report of the Surgeon General.
      • Kesaniemi Y.K.
      • Danforth Jr., E.
      • Jensen M.D.
      • Kopelman P.G.
      • Lefebvre P.
      • Reeder B.A.
      Dose-response issues concerning physical activity and health: an evidence-based symposium.
      Because of the complex relationship between physical activity and obesity, the status of obesity as an intermediate variable along the pathways to disease, and the potential to double-count costs, obesity was not included as a disease outcome in this model. However, the model cohort is based on the 2003 Behavioral Risk Factor Surveillance Survey (BRFSS) and is representative of the population prevalence of overweight (37.9%) and obesity (25.2%).
      CDC
      Behavioral Risk Factor Surveillance System: 2003 survey data.
      The model then simulates the progression of the cohort as they change activity levels or develop illness over a lifetime. The likelihood of changing physical activity levels, developing disease, or dying is specified with probabilities.
      Immediately following the 1-year physical activity promotion intervention, cohort members faced intervention-specific probabilities of improving their physical activity levels. Then, depending on the sustainability of the intervention effect, in subsequent years cohort members had an annual probability of either remaining in the new physical activity level or moving to a lower activity level. In the absence of the intervention, the change in physical activity levels from year to year was based on a natural history model, developed from age- and gender-specific physical activity data from the BRFSS.
      CDC
      Behavioral Risk Factor Surveillance System: 2003 survey data.
      Among the well, the risk of developing one of the five diseases depends on age, gender, and activity level. The risk of death depends on age, gender, and disease status. Thus, the ten states represented in the model constitute the four activity levels among the well, the five physical activity–related diseases, and death. Because there are insufficient published data describing differential mortality risk by both physical activity level among diseased adults and race/ethnicity, these factors were not included in this model. A schematic of the model is shown in Figure 1.
      Figure thumbnail gr1
      Figure 1Conceptual overview of the CDC MOVE model. Illustration of the 10-state Markov process represented as a state-transition diagram. In this process, circles represent possible health states, and arrows represent allowed transitions among these discrete health states. In each cycle of the Markov model, transition probabilities denote the likelihood with which people within a particular health state will stay in that state (represented by the tight curvilinear arrows to and from a single circle); transition to a new health state; or die. Death is an absorbing state from which no future transitions are possible. The output from the Markov process is depicted by the box; a running tally of the total costs and quality-of-life benefits generated during each cycle as a result of being in a series of health states over time. MOVE, measurement of the value of exercise

      Data Sources

      Population demographics

      Data on age and gender distribution of the initial population
      U.S. Census Bureau
      Projections of the resident population by age, sex, race, and Hispanic origin.
      were obtained from the U.S. Census Bureau (Table 1
      CDC
      Behavioral Risk Factor Surveillance System: 2003 survey data.
      U.S. Census Bureau
      Projections of the resident population by age, sex, race, and Hispanic origin.
      • King A.C.
      • Kiernan M.
      • Oman R.F.
      • Kraemer H.C.
      • Hull M.
      • Ahn D.
      Can we identify who will adhere to long-term physical activity? Signal detection methodology as a potential aid to clinical decision making.
      • Marcus B.H.
      • Dubbert P.M.
      • Forsyth L.H.
      • et al.
      Physical activity behavior change: issues in adoption and maintenance.
      • Wilson P.W.
      • D'Agostino R.B.
      • Levy D.
      • Belanger A.M.
      • Silbershatz H.
      • Kannel W.B.
      Prediction of coronary heart disease using risk factor categories.
      • Wolfe C.D.
      • Rudd A.G.
      • Howard R.
      • et al.
      Incidence and case fatality rates of stroke subtypes in a multiethnic population: the South London Stroke Register.
      • Brown R.D.
      • Whisnant J.P.
      • Sicks J.D.
      • O'Fallon W.M.
      • Wiebers D.O.
      Stroke incidence, prevalence, and survival: secular trends in Rochester, Minnesota, through 1989.
      CDC
      National diabetes fact sheet: U.S., 2003.
      CDC
      National diabetes surveillance system: incidence of diabetes.
      National Cancer Institute
      Surveillance, epidemiology, and end results.
      • Katzmarzyk P.T.
      • Janssen I.
      The economic costs associated with physical inactivity and obesity in Canada: an update.
      • Hu F.B.
      • Stampfer M.J.
      • Colditz G.A.
      • et al.
      Physical activity and risk of stroke in women.
      National Cancer Institute
      Surveillance, epidemiology, and end results.
      National Cancer Institute
      Surveillance, epidemiology, and end results.
      CDC
      Self-reported heart disease and stroke among adults with and without diabetes—U.S., 1999–2001.
      CDC
      Diabetes surveillance system: prevalence of diabetes.
      • Brown R.D.
      • Whisnat J.P.
      • Sicks J.D.
      • O'Fallon E.M.
      • Wiebers D.O.
      Stroke incidence, prevalence, and survival: secular trends in Rochester, Minnesota, through 1989.
      ). Based on the public health physical activity recommendations of CDC and the American College of Sports Medicine,
      • Haskell W.L.
      • Lee I.
      • Pate R.R.
      • et al.
      Physical activity and public health: updated recommendation for adults from the American College of Sports Medicine and the American Heart Association.
      the baseline population for this study's model was further divided into four levels of physical activity (inactive, irregularly active, sufficiently active to meet public health recommendations, and highly active) by age and gender, using data from the 2003 BRFSS
      CDC
      Behavioral Risk Factor Surveillance System: 2003 survey data.
      (Table 2
      • Ainsworth B.E.
      • Haskell W.L.
      • Whitt M.C.
      • et al.
      Compendium of physical activities: an update of activity codes and MET intensities.
      American College of Sports Medicine
      ACSM's guidelines for exercise testing and prescription.
      ). The BRFSS does not include questions that allow further stratification of the starting population by the five diseases included in this model.
      Table 1Sources of data used in the CDC MOVE physical activity policy model
      ParameterSource
      U.S. population by age and gender
      Hypothetical cohort of 100,000 U.S. adults defined by gender and age distribution from U.S. Census Bureau and physical activity level distribution of 2003 BRFSS data, using predefined cut points
      U.S. Census Bureau
      Projections of the resident population by age, sex, race, and Hispanic origin.
      U.S. population by physical activity level
      Hypothetical cohort of 100,000 U.S. adults defined by gender and age distribution from U.S. Census Bureau and physical activity level distribution of 2003 BRFSS data, using predefined cut points
      CDC
      Behavioral Risk Factor Surveillance System: 2003 survey data.
      Physical activity
       Probabilities for moving among levels by age group and gender
      These probabilities are conditional, and represent the probability of transitioning to another age, gender, and physical activity group given that an individual is already in a particular age, gender, and physical activity group.
      CDC
      Behavioral Risk Factor Surveillance System: 2003 survey data.
       Maintenance of a particular physical activity level
      • King A.C.
      • Kiernan M.
      • Oman R.F.
      • Kraemer H.C.
      • Hull M.
      • Ahn D.
      Can we identify who will adhere to long-term physical activity? Signal detection methodology as a potential aid to clinical decision making.
      ,
      • Marcus B.H.
      • Dubbert P.M.
      • Forsyth L.H.
      • et al.
      Physical activity behavior change: issues in adoption and maintenance.
      Disease incidence
       Coronary heart disease
      • Wilson P.W.
      • D'Agostino R.B.
      • Levy D.
      • Belanger A.M.
      • Silbershatz H.
      • Kannel W.B.
      Prediction of coronary heart disease using risk factor categories.
       Ischemic stroke
      • Wolfe C.D.
      • Rudd A.G.
      • Howard R.
      • et al.
      Incidence and case fatality rates of stroke subtypes in a multiethnic population: the South London Stroke Register.
      ,
      • Brown R.D.
      • Whisnant J.P.
      • Sicks J.D.
      • O'Fallon W.M.
      • Wiebers D.O.
      Stroke incidence, prevalence, and survival: secular trends in Rochester, Minnesota, through 1989.
       Type 2 diabetes
      CDC
      National diabetes fact sheet: U.S., 2003.
      ,
      CDC
      National diabetes surveillance system: incidence of diabetes.
       Breast cancer
      National Cancer Institute
      Surveillance, epidemiology, and end results.
       Colorectal cancer
      National Cancer Institute
      Surveillance, epidemiology, and end results.
      RR of disease by physical activity level
      Relative risk values are non–gender-specific, do not control for BMI, and are a linear interpolation across physical activity levels, with the most active group set as the reference group with a RR of 1.0.
       Coronary heart disease (RR for inactive=1.5)
      • Finkelstein E.A.
      • Wang G.
      • Lee I.M.
      National and state-specific inactivity-attributable medical expenditures for six diseases Final report prepared for the Centers for Disease Control and Prevention by the Research Triangle Institute.
      ,
      • Katzmarzyk P.T.
      • Janssen I.
      The economic costs associated with physical inactivity and obesity in Canada: an update.
       Ischemic stroke (RR for inactive=1.3)
      • Hu F.B.
      • Stampfer M.J.
      • Colditz G.A.
      • et al.
      Physical activity and risk of stroke in women.
       Type 2 diabetes (RR for inactive=1.7)
      • Finkelstein E.A.
      • Wang G.
      • Lee I.M.
      National and state-specific inactivity-attributable medical expenditures for six diseases Final report prepared for the Centers for Disease Control and Prevention by the Research Triangle Institute.
      ,
      • Katzmarzyk P.T.
      • Janssen I.
      The economic costs associated with physical inactivity and obesity in Canada: an update.
       Breast cancer (RR for inactive=1.3)
      • Finkelstein E.A.
      • Wang G.
      • Lee I.M.
      National and state-specific inactivity-attributable medical expenditures for six diseases Final report prepared for the Centers for Disease Control and Prevention by the Research Triangle Institute.
      ,
      • Katzmarzyk P.T.
      • Janssen I.
      The economic costs associated with physical inactivity and obesity in Canada: an update.
       Colorectal cancer (RR for inactive=1.6)
      • Finkelstein E.A.
      • Wang G.
      • Lee I.M.
      National and state-specific inactivity-attributable medical expenditures for six diseases Final report prepared for the Centers for Disease Control and Prevention by the Research Triangle Institute.
      ,
      • Katzmarzyk P.T.
      • Janssen I.
      The economic costs associated with physical inactivity and obesity in Canada: an update.
      Disease-specific mortality by physical activity level, age group, and gender
      CDC
      National diabetes fact sheet: U.S., 2003.
      ,
      National Cancer Institute
      Surveillance, epidemiology, and end results.
      , ,
      National Cancer Institute
      Surveillance, epidemiology, and end results.
      ,
      National Cancer Institute
      Surveillance, epidemiology, and end results.
      ,
      CDC
      Self-reported heart disease and stroke among adults with and without diabetes—U.S., 1999–2001.
      ,
      CDC
      Diabetes surveillance system: prevalence of diabetes.
      ,
      • Brown R.D.
      • Whisnat J.P.
      • Sicks J.D.
      • O'Fallon E.M.
      • Wiebers D.O.
      Stroke incidence, prevalence, and survival: secular trends in Rochester, Minnesota, through 1989.
      BRFSS, Behavioral Risk Factor Surveillance Survey; MOVE, measurement of the value of exercise; RR, relative risk
      a Hypothetical cohort of 100,000 U.S. adults defined by gender and age distribution from U.S. Census Bureau and physical activity level distribution of 2003 BRFSS data, using predefined cut points
      b These probabilities are conditional, and represent the probability of transitioning to another age, gender, and physical activity group given that an individual is already in a particular age, gender, and physical activity group.
      c Relative risk values are non–gender-specific, do not control for BMI, and are a linear interpolation across physical activity levels, with the most active group set as the reference group with a RR of 1.0.
      Table 2Distribution of original U.S. population cohort aged 25–64 years across physical activity levels
      Physical activity levelMET-min/wk range
      1000 MET-minutes per week of physical activity is roughly equal to 40 minutes per day of brisk walking at an intensity of 3.5 METs.36 One MET represents the metabolic rate equivalent to consuming 3.5 milliliters of oxygen per kilogram of body weight per minute and is equivalent to a resting metabolic rate.37
      Cohort frequency distribution (%)
      MenWomen
      Inactive
      <10 minutes of physical activity per week
      0–15018.321.7
      Irregularly active
      >10 minutes per week of physical activity, but ≤ recommended
      151–85529.138.1
      Meets CDC recommendations
      ≥30 minutes (but <60 minutes) of moderate physical activity on 5 or more days a week, or ≥20 minutes vigorous physical activity on 3 or more days a week, or both
      856–228028.424.7
      Highly active
      ≥60 minutes of moderate physical activity on 7 days a week, or ≥20 minutes of vigorous physical activity on 5 or more days a week, or both
      >228024.315.5
      a 1000 MET-minutes per week of physical activity is roughly equal to 40 minutes per day of brisk walking at an intensity of 3.5 METs.
      • Ainsworth B.E.
      • Haskell W.L.
      • Whitt M.C.
      • et al.
      Compendium of physical activities: an update of activity codes and MET intensities.
      One MET represents the metabolic rate equivalent to consuming 3.5 milliliters of oxygen per kilogram of body weight per minute and is equivalent to a resting metabolic rate.
      American College of Sports Medicine
      ACSM's guidelines for exercise testing and prescription.
      b <10 minutes of physical activity per week
      c >10 minutes per week of physical activity, but ≤ recommended
      d ≥30 minutes (but <60 minutes) of moderate physical activity on 5 or more days a week, or ≥20 minutes vigorous physical activity on 3 or more days a week, or both
      e ≥60 minutes of moderate physical activity on 7 days a week, or ≥20 minutes of vigorous physical activity on 5 or more days a week, or both

      Interventions

      Seven interventions studies were chosen; they are described in the Appendix.
      • Jeffery R.W.
      • Wing R.R.
      • Thorson C.
      • Burton L.R.
      Use of personal trainers and financial incentives to increase exercise in a behavioral weight-loss program.
      • Kriska A.M.
      • Bayles C.
      • Cauley J.A.
      • LaPorte R.E.
      • Sandler R.B.
      • Pambianco G.
      A randomized exercise trial in older women: increased activity over two years and the factors associated with compliance.
      • Linenger J.M.
      • Chesson 2nd, C.V.
      • Nice D.S.
      Physical fitness gains following simple environmental change.
      • Lombard D.N.
      • Lombard T.N.
      • Winett R.A.
      Walking to meet health guidelines: the effect of prompting frequency and prompt structure.
      • Knowler W.C.
      • Barrett-Connor E.
      • Fowler S.E.
      • et al.
      Reduction in the incidence of type 2 diabetes with lifestyle intervention or metformin.
      • Reger B.
      • Cooper L.
      • Booth-Butterfield S.
      • et al.
      Wheeling Walks: a community campaign using paid media to encourage walking among sedentary older adults.
      • Young D.R.
      • Haskell W.L.
      • Taylor C.B.
      • Fortmann S.P.
      Effect of community health education on physical activity knowledge, attitudes, and behavior The Stanford Five-City Project.
      Six strongly recommended exemplar interventions for the public health promotion of physical activity among adults were selected from those reviewed and categorized by the Task Force on Community Preventive Services (the Task Force). They dealt with the four strategies recommended for adults: community-wide campaigns, individually adapted health behavior change, social-support interventions in community settings, and the creation of or enhanced access to places for physical activity combined with informational outreach activities. A seventh, newer intervention study was added to better represent community-wide campaigns.
      • Jeffery R.W.
      • Wing R.R.
      • Thorson C.
      • Burton L.R.
      Use of personal trainers and financial incentives to increase exercise in a behavioral weight-loss program.
      • Kriska A.M.
      • Bayles C.
      • Cauley J.A.
      • LaPorte R.E.
      • Sandler R.B.
      • Pambianco G.
      A randomized exercise trial in older women: increased activity over two years and the factors associated with compliance.
      • Linenger J.M.
      • Chesson 2nd, C.V.
      • Nice D.S.
      Physical fitness gains following simple environmental change.
      • Lombard D.N.
      • Lombard T.N.
      • Winett R.A.
      Walking to meet health guidelines: the effect of prompting frequency and prompt structure.
      • Knowler W.C.
      • Barrett-Connor E.
      • Fowler S.E.
      • et al.
      Reduction in the incidence of type 2 diabetes with lifestyle intervention or metformin.
      • Reger B.
      • Cooper L.
      • Booth-Butterfield S.
      • et al.
      Wheeling Walks: a community campaign using paid media to encourage walking among sedentary older adults.
      • Young D.R.
      • Haskell W.L.
      • Taylor C.B.
      • Fortmann S.P.
      Effect of community health education on physical activity knowledge, attitudes, and behavior The Stanford Five-City Project.
      Criteria for inclusion were (1) a controlled study of adult subjects without established disease, (2) a detailed study protocol available for costing, (3) the availability of a physical activity outcome measure, and (4) a study duration of 3 months or longer. Excluded were interventions that did not provide baseline and post-intervention effect-size data or a reference group, did not report a measurable effect-size metric, or reported one that was not amenable to conversion to the metric used in this model.
      AppendixStrategy and intervention descriptions
      Community-wide campaign
      • Involves many community sectors in highly visible, broad-based, multiple-intervention approaches to increasing physical activity
      • Communication techniques are a common element in all of the campaigns and are directed at large populations.
      • Addresses sedentary behavior as well as cardiovascular disease risk factors (diet and smoking)
      • Messages are in various media formats (e.g., radio, newspaper, mailings, billboards, paid advertisements, press releases), and can be a combination of two or more of these approaches.
      • Interventions typically include some combination of social support, risk-factor screening, counseling, and education in a variety of settings (e.g., worksites, schools).
      AuthorDescription
      Young (1996)
      • Young D.R.
      • Haskell W.L.
      • Taylor C.B.
      • Fortmann S.P.
      Effect of community health education on physical activity knowledge, attitudes, and behavior The Stanford Five-City Project.
      Stanford Five-City Project was a 6-year, integrated, community-wide multifactorial health education intervention for improving physical activity (print materials, radio, TV, seminars, community walking events, worksite- and school-based programs).
      Reger (2002)
      • Reger B.
      • Cooper L.
      • Booth-Butterfield S.
      • et al.
      Wheeling Walks: a community campaign using paid media to encourage walking among sedentary older adults.
      Wheeling Walks was an 8-week intensive community-wide intervention that promoted walking among sedentary adults aged 50–65 years using paid media (TV, radio, newspapers, websites, billboards); public relations; and public health activities at worksites, churches, and local organizations.
      Social support
      • Interventions focus on changing physical activity behavior through building, strengthening, and maintaining social networks that provide supportive relationships for behavior change.
      • Can be achieved either by creating new social networks or working within existing networks in a social setting outside the family
      • Interventions typically involve setting up a buddy system, making a contract with others to achieve specified levels of physical activity, or setting up walking or other groups to provide companionship and support while being physically active.
      AuthorDescription
      Kriska (1986)
      • Kriska A.M.
      • Bayles C.
      • Cauley J.A.
      • LaPorte R.E.
      • Sandler R.B.
      • Pambianco G.
      A randomized exercise trial in older women: increased activity over two years and the factors associated with compliance.
      Use of organized walking groups, social gatherings, phone calls, cards, home visits, and a newsletter to enhance exercise compliance and promote physical activity
      Lombard (1995)
      • Lombard D.N.
      • Lombard T.N.
      • Winett R.A.
      Walking to meet health guidelines: the effect of prompting frequency and prompt structure.
      After an initial training session involving walking maps and handouts on strategies and support (walking partner or walking group) for starting and maintaining a walking program, the frequency and duration of phone calls were varied to prompt participants' walking.
      Individually adapted health behavior change programs
      • Programs are tailored to the individual's readiness for change, specific interests, and preferences.
      • Teaches participants specific behavioral skills that enable them to incorporate moderate-intensity physical activity into daily routines
      • Behaviors may be planned (scheduled walk) or unplanned (taking stairs when the opportunity arises).
      • Interventions use constructs from one or more established health behavior change models such as the social cognitive theory, the health belief model, or the transtheoretical model of change.
      • All programs incorporate the following behavioral approaches: (1) setting goals for physical activity and self-monitoring progress toward goals, (2) building social support for new behavioral patterns, (3) behavioral reinforcement through self-reward and positive self-talk, (4) structured problem solving geared to maintenance of the behavior change, and (5) the prevention of relapse into sedentary behaviors. All of the interventions evaluated were delivered to people either in group settings or by mail, telephone, or directed media.
      AuthorDescription
      Jeffery (1998)
      • Jeffery R.W.
      • Wing R.R.
      • Thorson C.
      • Burton L.R.
      Use of personal trainers and financial incentives to increase exercise in a behavioral weight-loss program.
      Use of personal trainers, standard behavior-therapy sessions, financial incentives, and phone calls to participants to increase physical activity
      Knowler (DPP) (2002)
      • Knowler W.C.
      • Barrett-Connor E.
      • Fowler S.E.
      • et al.
      Reduction in the incidence of type 2 diabetes with lifestyle intervention or metformin.
      Intensive lifestyle-modification program for adults at high risk of developing type 2 diabetes, involving exercise testing, written information, and individual counseling sessions; a 16-lesson curriculum covering diet, exercise, and behavior modification; individual and group exercise sessions; and in-person visits and phone calls to participants
      Enhanced access to places for physical activity combined with informational outreach activities
      • Interventions involved the efforts of worksites, coalitions, agencies, and communities to create or provide access to places and facilities where people can be physically active (providing access to weight and aerobic fitness equipment in fitness centers or community centers, creating walking trains, and providing access to nearby fitness centers).
      • Also incorporated components such as training on equipment, health behavior education and techniques, seminars, counseling, risk screening, health forums and workshops, referrals to physicians or additional services, health and fitness programs, and support or buddy systems
      AuthorDescription
      Linenger (1991)
      • Linenger J.M.
      • Chesson 2nd, C.V.
      • Nice D.S.
      Physical fitness gains following simple environmental change.
      Exposure to an environment that emphasizes and supports a more active lifestyle (bike paths, extended fitness facility hours, opening of a new fitness center, cycling clubs, marked running courses, organized athletic events)
      DPP, Diabetes Prevention Program

      Disease

      To estimate the annual probability of developing each disease, population-based, disease-specific incidence data
      • Wilson P.W.
      • D'Agostino R.B.
      • Levy D.
      • Belanger A.M.
      • Silbershatz H.
      • Kannel W.B.
      Prediction of coronary heart disease using risk factor categories.
      • Wolfe C.D.
      • Rudd A.G.
      • Howard R.
      • et al.
      Incidence and case fatality rates of stroke subtypes in a multiethnic population: the South London Stroke Register.
      • Brown R.D.
      • Whisnant J.P.
      • Sicks J.D.
      • O'Fallon W.M.
      • Wiebers D.O.
      Stroke incidence, prevalence, and survival: secular trends in Rochester, Minnesota, through 1989.
      CDC
      National diabetes fact sheet: U.S., 2003.
      CDC
      National diabetes surveillance system: incidence of diabetes.
      National Cancer Institute
      Surveillance, epidemiology, and end results.
      were combined with relative risks derived from epidemiologic studies, specific for physical activity level and disease (Table 1).
      • Finkelstein E.A.
      • Wang G.
      • Lee I.M.
      National and state-specific inactivity-attributable medical expenditures for six diseases Final report prepared for the Centers for Disease Control and Prevention by the Research Triangle Institute.
      • Katzmarzyk P.T.
      • Janssen I.
      The economic costs associated with physical inactivity and obesity in Canada: an update.
      • Hu F.B.
      • Stampfer M.J.
      • Colditz G.A.
      • et al.
      Physical activity and risk of stroke in women.
      The median relative risk of each disease among people in the lowest activity level (inactive) compared with the highest level (highly active is approximately equivalent to ≥2000 kilocalories [kcal] per week), not controlling for medical conditions that are consequences of physical inactivity, including obesity, was obtained from a review of 120 studies.
      • Finkelstein E.A.
      • Wang G.
      • Lee I.M.
      National and state-specific inactivity-attributable medical expenditures for six diseases Final report prepared for the Centers for Disease Control and Prevention by the Research Triangle Institute.
      The relative risk for ischemic stroke versus all stroke was adjusted based on data from a large cohort study.
      • Hu F.B.
      • Stampfer M.J.
      • Colditz G.A.
      • et al.
      Physical activity and risk of stroke in women.
      To estimate the relative risk for the two intermediate physical activity levels, linear interpolation, assuming a dose–response relationship, was used.
      • Kesaniemi Y.K.
      • Danforth Jr., E.
      • Jensen M.D.
      • Kopelman P.G.
      • Lefebvre P.
      • Reeder B.A.
      Dose-response issues concerning physical activity and health: an evidence-based symposium.
      • Lee I.M.
      • Skerrett P.J.
      Physical activity and all-cause mortality: what is the dose–response relation?.
      • Oguma Y.
      • Sesso H.D.
      • Paffenbarger Jr, R.S.
      • Lee I.M.
      Physical activity and all cause mortality in women: a review of the evidence.

      Mortality

      A common method was used to estimate the annual probability of death, conditioned on having a particular disease, for each 5-year age group and gender. In addition to disease-specific prevalence data,
      CDC
      National diabetes fact sheet: U.S., 2003.
      National Cancer Institute
      Surveillance, epidemiology, and end results.
      CDC
      Self-reported heart disease and stroke among adults with and without diabetes—U.S., 1999–2001.
      CDC
      Diabetes surveillance system: prevalence of diabetes.
      • Brown R.D.
      • Whisnat J.P.
      • Sicks J.D.
      • O'Fallon E.M.
      • Wiebers D.O.
      Stroke incidence, prevalence, and survival: secular trends in Rochester, Minnesota, through 1989.
      data from the 2002 National Vital Statistics Reports were used to estimate the annual probability of death in people with CHD, ischemic stroke, or type 2 diabetes, while the Surveillance Epidemiology and End Results database was used to estimate the annual probability of death from breast or colorectal cancer (Table 1).
      National Cancer Institute
      Surveillance, epidemiology, and end results.
      National Cancer Institute
      Surveillance, epidemiology, and end results.
      • Lee I.M.
      • Skerrett P.J.
      Physical activity and all-cause mortality: what is the dose–response relation?.
      To estimate the annual probability of death for well adults, available mortality data, excluding disease-specific death rates for the five modeled diseases, were adjusted for age group and gender.

      Quality of life

      Quality-of-life (QOL) data were obtained for all disease and activity states from new analyses of the 2001 National Health Interview Survey, using previously validated scales for quality of well-being (QWB) widely accepted for assessing health-related QOL.
      • Tengs T.O.
      • Lin T.H.
      A meta-analysis of quality of life estimates for stroke.
      • Kaplan R.M.
      • Anderson J.P.
      • Ake C.F.
      Gender differences in quality-adjusted life expectancy: results from the National Health Interview Survey.
      • Kaplan R.M.
      • Anderson J.P.
      The general health policy model: an integrated approach.
      • Kaplan R.M.
      • Anderson J.P.
      • Patterson T.L.
      • et al.
      Validity of the quality of well-being scale for persons with human immunodeficiency virus infection HNRC Group, HIV Neurobehavioral Research Center.
      Multiple regressions to estimate QOL as a function of age, gender, disease, and physical activity level were performed (Table 3
      • Diamond L.H.
      • Sherbine K.
      AMA national summit on obesity: costs of obesity and communications implications for the U.S. healthcare system MarketScan database.
      ).
      Table 3Select model parameters
      Value
      INTERVENTION-SPECIFIC PARAMETERSEffect size at 1 year (MET-min/wk)1-year per-person costs
      Considered direct medical, direct nonmedical, and time costs as a function of age group for each exemplar intervention. Direct medical costs included staff labor wages, provider fees, and laboratory or monitoring test fees. Direct nonmedical costs captured costs associated with educational materials, physical activity monitoring devices, and participant out-of-pocket expenses. Time costs, where applicable, represented the monetary value of time spent participating in the intervention. The values shown are for the group aged 40–49 years.
      (2003 $U.S.)
      InterventionStrategyMenWomenMenWomenSource
      Stanford Five-City Project (see Appendix for description)Community-wide campaign503213951239
      • Young D.R.
      • Haskell W.L.
      • Taylor C.B.
      • Fortmann S.P.
      Effect of community health education on physical activity knowledge, attitudes, and behavior The Stanford Five-City Project.
      Wheeling Walks (see Appendix for description)Community-wide campaign31631618371618
      • Reger B.
      • Cooper L.
      • Booth-Butterfield S.
      • et al.
      Wheeling Walks: a community campaign using paid media to encourage walking among sedentary older adults.
      Use of organized walking groups, social gatherings, phone calls, cards, home visits, and a newsletter to enhance physical activity compliance and promote physical activitySocial support228
      Estimated for men and adjusted for average body weight
      27525542308
      • Kriska A.M.
      • Bayles C.
      • Cauley J.A.
      • LaPorte R.E.
      • Sandler R.B.
      • Pambianco G.
      A randomized exercise trial in older women: increased activity over two years and the factors associated with compliance.
      Initial training session involving walking maps and handouts on strategies and support (walking partner/walking group) for starting and maintaining a walking program; frequency and duration of phone calls varied to prompt participants to walkSocial support21021015831504
      • Lombard D.N.
      • Lombard T.N.
      • Winett R.A.
      Walking to meet health guidelines: the effect of prompting frequency and prompt structure.
      Use of personal trainers, standard behavior therapy sessions, financial incentives, and phone calls to prompt participants to increase physical activityIndividually adapted behavior change45655140703895
      • Jeffery R.W.
      • Wing R.R.
      • Thorson C.
      • Burton L.R.
      Use of personal trainers and financial incentives to increase exercise in a behavioral weight-loss program.
      Intensive lifestyle modification program for adults at high risk of developing type 2 diabetes (see Appendix for description)Individually adapted behavior change43543546734673
      • Knowler W.C.
      • Barrett-Connor E.
      • Fowler S.E.
      • et al.
      Reduction in the incidence of type 2 diabetes with lifestyle intervention or metformin.
      Exposure to an environment that emphasized and supported a more active lifestyle (see Appendix for description)Enhanced access954.31151.753084646
      • Linenger J.M.
      • Chesson 2nd, C.V.
      • Nice D.S.
      Physical fitness gains following simple environmental change.
      ValueSource
      DISEASE-SPECIFIC PARAMETERSAnnual per-person medical costs
      Longitudinal database from a major claims processor in the central U.S. The database contains 7 years of data and a stable population of approximately 350,000 members. Effective annual cost includes the entire stream of costs for each illness, including the lead-in costs in the years prior to diagnosis. To improve the generalizability of cost estimates, they were adjusted by nationally representative Medical Expenditure Panel Survey data.
      (2003 $US)
      DiseaseMenWomen
      Coronary heart disease10,80012,500
      • Diamond L.H.
      • Sherbine K.
      AMA national summit on obesity: costs of obesity and communications implications for the U.S. healthcare system MarketScan database.
      Ischemic stroke12,90017,700
      Type 2 diabetes8,90010,900
      Breast cancerN/A11,000
      Colorectal cancer15,30016,600
      Well (absence of the five diseases)2,5003,100
      Quality of well-being score
      Quality-of-life (QOL) data by disease state, physical activity level, age, and gender were obtained from mean 2001 National Health Interview Survey–imputed quality of well-being (QWB) scores. Multiple regression equation for individuals with disease: QOL = B0+B1 (Gender)+B2 (Age)+B3  (Age2)+∑i=15 B (i+3) (Disease i). Multiple regression equation for well people: QOL = B0 + B1 (Gender) + B2 (Age) + B3 (Age2) + B4 (MET-minutes).
      (age-specific)
      Men (aged 40)Women (aged 40)Source
      Coronary heart disease0.6470.623
      • Tengs T.O.
      • Lin T.H.
      A meta-analysis of quality of life estimates for stroke.
      ,
      • Kaplan R.M.
      • Anderson J.P.
      • Ake C.F.
      Gender differences in quality-adjusted life expectancy: results from the National Health Interview Survey.
      Ischemic stroke0.5990.599
      Type 2 diabetes0.6630.650
      Breast cancer0.7340.686
      Colorectal cancer0.6720.691
      Well (absence of the five diseases): inactive0.8060.791
      Well (absence of the five diseases): irregularly active0.8100.796
      Well (absence of the five diseases): meeting public health physical activity guidelines0.8200.807
      Well (absence of the five diseases): highly active0.8290.818
      a Considered direct medical, direct nonmedical, and time costs as a function of age group for each exemplar intervention. Direct medical costs included staff labor wages, provider fees, and laboratory or monitoring test fees. Direct nonmedical costs captured costs associated with educational materials, physical activity monitoring devices, and participant out-of-pocket expenses. Time costs, where applicable, represented the monetary value of time spent participating in the intervention. The values shown are for the group aged 40–49 years.
      b Estimated for men and adjusted for average body weight
      c Longitudinal database from a major claims processor in the central U.S. The database contains 7 years of data and a stable population of approximately 350,000 members. Effective annual cost includes the entire stream of costs for each illness, including the lead-in costs in the years prior to diagnosis. To improve the generalizability of cost estimates, they were adjusted by nationally representative Medical Expenditure Panel Survey data.
      d Quality-of-life (QOL) data by disease state, physical activity level, age, and gender were obtained from mean 2001 National Health Interview Survey–imputed quality of well-being (QWB) scores. Multiple regression equation for individuals with disease: QOL=B0+B1(Gender)+B2(Age)+B3(Age2)+i=15B(i+3)(Diseasei).Multiple regression equation for well people: QOL = B0 + B1 (Gender) + B2 (Age) + B3 (Age
      WHO
      Preventing chronic diseases: a vital investment.
      ) + B4 (MET-minutes).

      Costs

      Through direct communication with the authors of original investigations in combination with a review of manuscript protocols, each original intervention was itemized to determine all associated costs. In addition to the cost of materials and intervention delivery, out-of-pocket expenses paid by participants—such as required exercise clothing and equipment—were estimated. Participants' time was also valued as a cost, using age- and gender-specific wages.
      • Haddix A.
      • Teutsch S.M.
      • Corso P.S.
      Prevention effectiveness: a guide to decision analysis and economic evaluation.
      For the enhanced-access intervention,
      • Linenger J.M.
      • Chesson 2nd, C.V.
      • Nice D.S.
      Physical fitness gains following simple environmental change.
      the costs associated with developing and maintaining the infrastructural components were included (e.g., physical activity facilities, trails; Table 3).
      To derive medical cost estimates, a longitudinal medical claims database
      • Diamond L.H.
      • Sherbine K.
      AMA national summit on obesity: costs of obesity and communications implications for the U.S. healthcare system MarketScan database.
      was used to analyze claims for the five disease states by ICD-9 codes. From these costs, the effective annual cost was calculated for each of the diseases over a lifetime. An annual medical inflation factor of 8%
      U.S. Division of Labor, Centers for Medicare & Medicaid Services, Office of the Actuary
      National medical care price indicators.
      was applied, and the costs were discounted back to the present at 3% per year.
      • Gold M.
      • Siegel J.
      • Russell L.
      • Weinstein M.
      Cost-effectiveness in health and medicine.
      To improve national representativeness, medical-claims data were adjusted using Medical Expenditure Panel Survey data (Table 3).
      Agency for Healthcare Research and Quality
      Medical expenditure panel survey, 2001.

      Modeling the Effect of Interventions

      The four physical activity levels were characterized, using the MET-minutes per week, which captures the intensity, duration, and frequency of physical activity (Table 2). One MET represents the metabolic rate equivalent to consuming 3.5 milliliters of oxygen per kilogram of body weight per minute, and is equivalent to a resting metabolic rate.
      American College of Sports Medicine
      ACSM's guidelines for exercise testing and prescription.
      One thousand MET-minutes per week of physical activity is roughly equal to 40 minutes per day of brisk walking at an intensity of 3.5 METs.
      • Ainsworth B.E.
      • Haskell W.L.
      • Whitt M.C.
      • et al.
      Compendium of physical activities: an update of activity codes and MET intensities.
      To estimate the probability of moving to a higher physical activity level after intervention, intervention-specific MET-minutes were added to current levels, and the percentage of the cohort that moved from one level of physical activity to another as a result of an intervention was noted. This percentage was then used as a transition probability. For example, if adding a certain number of MET-minutes per week to all members of the inactive group caused 25% of them to move up to the irregularly active group, then the probability of moving from inactive to irregularly active was estimated as 0.25 in the first year following the intervention. Intervention effect-size estimates are shown in Table 3.
      The impact of an intervention was assumed to decline after the intervention had ended, and the decline in maintenance of physical activity levels over time was modeled based on the limited data from the research literature on the long-term maintenance of increased physical activity resulting from interventions.
      • King A.C.
      • Kiernan M.
      • Oman R.F.
      • Kraemer H.C.
      • Hull M.
      • Ahn D.
      Can we identify who will adhere to long-term physical activity? Signal detection methodology as a potential aid to clinical decision making.
      • Marcus B.H.
      • Dubbert P.M.
      • Forsyth L.H.
      • et al.
      Physical activity behavior change: issues in adoption and maintenance.
      For all of the interventions, with the exception of the enhanced-access intervention,
      • Linenger J.M.
      • Chesson 2nd, C.V.
      • Nice D.S.
      Physical fitness gains following simple environmental change.
      a 50% decline in physical activity in Year 2 was modeled. For the enhanced-access study, a 33% decline was modeled, because the environmental enhancements persisted long after the intervention had ended.
      Following the substantial decline (33%–50%) in physical activity modeled post-intervention, in Year 2, cohort members were transitioned into a natural-history model, which models the general decline in physical activity that occurs with age.
      USDHHS
      Physical activity and health: a report of the Surgeon General.
      • Sapkota S.
      • Bowles H.R.
      • Ham S.A.
      • Kohl III, H.W.
      Adult participation in recommended levels of physical activity—U.S., 2001 and 2003.
      • Pratt M.
      • Macera C.A.
      • Blanton C.
      Levels of physical activity and inactivity in children and adults in the U.S.: current evidence and research issues.
      Thus each year participants faced gender- and age-specific probabilities of moving to a lower level or remaining at the same physical activity level.

      Estimating Cost Effectiveness

      The cost effectiveness of each intervention was estimated, using methods consistent with the guidelines established by the Panel on Cost-Effectiveness in Health and Medicine.
      • Gold M.
      • Siegel J.
      • Russell L.
      • Weinstein M.
      Cost-effectiveness in health and medicine.
      The model was used to project lifetime costs as well as gains in both life-years (survival) and QALYs associated with the seven evaluated physical activity promotion interventions and with the no-intervention scenario. Consistent with the panel's recommendations, the societal perspective was adopted, and future costs and benefits were discounted to the present at an annual rate of 3%.
      • Gold M.
      • Siegel J.
      • Russell L.
      • Weinstein M.
      Cost-effectiveness in health and medicine.
      The average relative performance of each intervention was assessed compared to no intervention, using a ratio of the additional expected cost of each program divided by the additional expected QALYs gained relative to the no-intervention alternative. The number of cases of disease averted was also estimated. To determine the robustness of the final results, sensitivity analyses were conducted. These simulations included one-way, two-way, and probabilistic sensitivity analyses, with particular emphasis on intervention effect size and cost estimates. Last, as a form of model validation, a parallel model that substituted all-cause mortality for disease-specific mortality was developed.

      Results

      Average Cost Effectiveness

      Estimates of the population health and economic outcomes associated with each intervention are shown in Table 4. Summarized are the average costs, effectiveness, and cost-effectiveness ratios associated with a one-time application of each physical activity promotion intervention relative to no intervention. Results are cumulative over a 40-year time-horizon for the whole U.S. population, aged 25–64 years, but average per-person values are reported here. Absent any new intervention to improve physical activity levels (by natural history), the average discounted quality-adjusted life expectancy was calculated to be 14.77 years, and lifetime costs were approximately $195,000. Intervention participation improved average QALYs by 0.7 to 5.3 weeks (i.e., equivalent to 0.014–0.102 QALYs in Table 4), with the Linenger
      • Linenger J.M.
      • Chesson 2nd, C.V.
      • Nice D.S.
      Physical fitness gains following simple environmental change.
      intervention yielding the greatest gain in QALYs compared to no intervention.
      Table 4Cost effectiveness of each intervention compared to no intervention
      StudyIntervention strategyTotal cost (2003 $US)Total LYsTotal QALYInc costInc LYsInc QALYCost/LY ($/LY)Cost/QALY ($/QALY)
      No intervention195,01319.37014.767
      Reger (2002)
      • Reger B.
      • Cooper L.
      • Booth-Butterfield S.
      • et al.
      Wheeling Walks: a community campaign using paid media to encourage walking among sedentary older adults.
      Community-wide campaign195,71319.40114.816$7000.0310.04922,65414,286
      Lombard (1995)
      • Lombard D.N.
      • Lombard T.N.
      • Winett R.A.
      Walking to meet health guidelines: the effect of prompting frequency and prompt structure.
      Social support195,72519.38714.793$7120.0160.02643,66327,373
      Linenger (1991)
      • Linenger J.M.
      • Chesson 2nd, C.V.
      • Nice D.S.
      Physical fitness gains following simple environmental change.
      Enhanced access197,92519.43314.869$2,9120.0630.10246,44228,548
      Jeffery (1998)
      • Jeffery R.W.
      • Wing R.R.
      • Thorson C.
      • Burton L.R.
      Use of personal trainers and financial incentives to increase exercise in a behavioral weight-loss program.
      Individually adapted health behavior196,91819.41014.831$1,9050.0400.06448,09629,759
      Kriska (1986)
      • Kriska A.M.
      • Bayles C.
      • Cauley J.A.
      • LaPorte R.E.
      • Sandler R.B.
      • Pambianco G.
      A randomized exercise trial in older women: increased activity over two years and the factors associated with compliance.
      Social support196,24419.38914.798$1,2300.0190.03165,44739,690
      Knowler (2002)
      • Knowler W.C.
      • Barrett-Connor E.
      • Fowler S.E.
      • et al.
      Reduction in the incidence of type 2 diabetes with lifestyle intervention or metformin.
      (DPP)
      Individually adapted health behavior197,73419.40614.825$2,7210.0360.05875,58346,914
      Young (1996)
      • Young D.R.
      • Haskell W.L.
      • Taylor C.B.
      • Fortmann S.P.
      Effect of community health education on physical activity knowledge, attitudes, and behavior The Stanford Five-City Project.
      Community-wide campaign195,97319.37914.781$9600.0090.014110,32268,557
      Note: Values are rounded to no more than three decimal places.
      DPP, Diabetes Prevention Program; Inc, incremental; LY, life-year; QALY, quality-of-life year
      Cost-effectiveness ratios ranged between $14,000 and $69,000 per QALY gained. In one example, the Lombard intervention
      • Lombard D.N.
      • Lombard T.N.
      • Winett R.A.
      Walking to meet health guidelines: the effect of prompting frequency and prompt structure.
      (one of the social-support strategies) increased quality-adjusted life expectancy by 14.79 years and cost $27,370 per QALY gained relative to no intervention. In another case,
      • Knowler W.C.
      • Barrett-Connor E.
      • Fowler S.E.
      • et al.
      Reduction in the incidence of type 2 diabetes with lifestyle intervention or metformin.
      also in comparison with no intervention, the physical activity component of the diabetes prevention trial (an individually adapted health behavior–change strategy) had a cost-effectiveness ratio of $46,910 per QALY gained. All interventions appeared to reduce disease incidence: reductions ranged from 5–15 cases per 100,000 for colorectal cancer, to 15–58 cases per 100,000 for breast cancer, to 59–207 cases per 100,000 for type 2 diabetes, and to as many as 140–476 cases per 100,000 for CHD. The impact of interventions on the prevention of ischemic stroke (which often occurs later in life) did not follow the trend of disease-incidence reduction, probably because the model permitted the development of only one illness and because interventions increased longevity.

      Sensitivity Analyses

      Several assumptions were varied in a one-way sensitivity analysis. Repeating the intervention once after 20 years had a small effect on cost-effectiveness (results not shown). Similarly, varying the dissipation of the effect sizes of the interventions had a marginal impact.
      Shortening the analytic time-horizon from 40 to 30, 20, or 10 years influenced cost-effectiveness substantially. For example, while the Lombard intervention
      • Lombard D.N.
      • Lombard T.N.
      • Winett R.A.
      Walking to meet health guidelines: the effect of prompting frequency and prompt structure.
      had a cost per QALY of $27,000 over the original 40-year time-horizon, it had a cost per QALY of $147,000 over a 10-year horizon.
      To assess the impact of uncertain intervention cost and effect-size parameter estimates on uncertainty in cost-effectiveness, a probabilistic sensitivity analysis (Monte Carlo analysis) was performed.
      • Gold M.
      • Siegel J.
      • Russell L.
      • Weinstein M.
      Cost-effectiveness in health and medicine.
      When running this analysis, a distribution of 600 ratios drawn from the underlying distributions of costs and effect sizes was obtained.
      Using the distributions from this analysis, the probability was assessed that the cost-effectiveness of each intervention was below various thresholds that are commonly used to determine whether interventions provide good value for money. Results are shown by the acceptability curves in Figure 2. For example, there is a 55% chance that the cost per QALY of the Young intervention
      • Young D.R.
      • Haskell W.L.
      • Taylor C.B.
      • Fortmann S.P.
      Effect of community health education on physical activity knowledge, attitudes, and behavior The Stanford Five-City Project.
      is less than $50,000 per QALY, a commonly used benchmark of cost-effectiveness.
      • Hirth R.A.
      • Chernew M.E.
      • Miller E.
      • Fendrick A.M.
      • Weissert W.G.
      Willingness to pay for a quality-adjusted life year: in search of a standard.
      Likewise, using $200,000 per QALY as a more contemporary threshold, given that many widely accepted public health initiatives have cost-effectiveness ratios exceeding this value,
      • Hirth R.A.
      • Chernew M.E.
      • Miller E.
      • Fendrick A.M.
      • Weissert W.G.
      Willingness to pay for a quality-adjusted life year: in search of a standard.
      there is 100% chance (i.e., all 600 randomly selected pairs of costs and effect sizes produced cost-per-QALY estimates below $200,000) that the associated cost per QALY falls below it, suggesting that this intervention is cost effective, or acceptable. Thus, despite model parameter and resultant cost-per-QALY ratio uncertainties, it is likely that this intervention is an acceptable use of societal resources.
      Figure thumbnail gr2
      Figure 2Results from probabilistic sensitivity analyses: acceptability curves representing variation of intervention costs and effect sizes

      Conclusion

      All of the evaluated physical activity interventions were found to be cost-effective and offered good value for money, with gains in both survival and health-related QOL, and with reasonable cost-per-QALY and cost-per-life-year ratios. These results support using any of the seven evaluated interventions as part of public health efforts to promote physical activity. Factors specific to implementing interventions within a given strategy and setting may introduce greater variability in cost-effectiveness than exists among strategies. The results do not suggest that any one of these recommended interventions is clearly more cost-effective than the others. Interestingly, even the most complex and expensive intervention (the physical activity component of the diabetes prevention program) proved to be cost-effective, suggesting that targeting high-risk populations with intensive interventions can be a good use of public health funds.
      Although comparisons of public health strategies evaluated by separate economic analyses should be undertaken with caution, cost-effectiveness estimates of other well-known strategies provide context for these results. Risk-reduction counseling for CHD and targeted screening for type 2 diabetes cost $74,000 and $34,000 per QALY gained, respectively.
      • Salkeld G.
      • Phongsavan P.
      • Oldenburg B.
      • et al.
      The cost-effectiveness of a cardiovascular risk reduction program in general practice.
      • Hoerger T.J.
      • Harris R.
      • Hicks K.A.
      • Donahue K.
      • Sorensen S.
      • Engelgau M.
      Screening for type 2 diabetes mellitus: a cost-effectiveness analysis.
      Further, both policymakers and the general public have reported being willing to pay more than $200,000 per QALY for gains in health.
      • Hirth R.A.
      • Chernew M.E.
      • Miller E.
      • Fendrick A.M.
      • Weissert W.G.
      Willingness to pay for a quality-adjusted life year: in search of a standard.
      The physical activity promotion strategies assessed in this analysis compare favorably with other well-accepted public health strategies that are deemed to be cost-effective.
      • Salkeld G.
      • Phongsavan P.
      • Oldenburg B.
      • et al.
      The cost-effectiveness of a cardiovascular risk reduction program in general practice.
      • Hoerger T.J.
      • Harris R.
      • Hicks K.A.
      • Donahue K.
      • Sorensen S.
      • Engelgau M.
      Screening for type 2 diabetes mellitus: a cost-effectiveness analysis.
      • Neumann P.J.
      • Rosen A.B.
      • Greenberg D.
      • et al.
      Can we better prioritize resources for cost-utility research?.
      • Stone P.W.
      • Teutsch S.
      • Chapman R.H.
      • Bell C.
      • Goldie S.J.
      • Neumann P.J.
      Cost-utility analyses of clinical preventive services: published ratios, 1976–1997.
      Although rigorous methodology was used to identify and incorporate data into the model, the sensitivity analyses were also revealing. Estimated cost-effectiveness was most influenced by intervention cost and effect size. However, probabilistic sensitivity analyses revealed that varying parameter estimates across a wide range of uncertainty still resulted in cost-effectiveness ratios below widely accepted thresholds for public health value.
      Cost-effectiveness is also influenced by the time-horizon over which the analysis is evaluated. Shorter periods of evaluation are associated with less-favorable cost-effectiveness, suggesting that perspectives motivated by short-term gains may not see as much benefit from increasing physical activity, which prevents longer-term morbidity and premature mortality. Specifically, in the short term, upfront intervention costs that are less diluted by discounting outweigh the medical cost savings due to diseases that occur late in life.
      Cost-effectiveness analyses are often susceptible to the design weaknesses and generalizability issues of their component studies, and analysts usually must make assumptions in model development. Incorporating additional detail into a model can help illustrate the complexity of reality, but a carefully validated parsimonious model is less susceptible to the inconsistencies, ambiguities, and redundancies of a more complex approach.
      • Forster M.R.
      Key concepts in model selection: performance and generalizability.
      The strength of the current analysis lies in the systematic way in which these issues have been addressed. The study design has emphasized the use of the highest-quality data available from the fewest sources necessary to create a valid, streamlined model.
      Given that it is conceivable that more than one of the evaluated interventions could be simultaneously implemented and that the intervention effect sizes of combining two or more of these interventions remain unknown in the literature, average cost-effectiveness ratios were reported. For completeness, incremental analyses (not reported) were performed. All of these issues are believed to be important foci for future analyses.
      As with all models, several assumptions were necessary to carry out this analysis. In each instance, the most conservative and parsimonious approach was chosen. Key assumptions are noted here.
      Because this model focuses on prevention and a long-term societal perspective and is based on the availability of data, the starting population was defined to be a cohort of well people aged 25–64 years. A separate analysis, not reported here, was conducted specifically for older adults. Due to limited data on race/ethnicity-specific disease outcomes and physical activity and intervention effects, it was not possible to extend the model to assess the cost-effectiveness of interventions in subpopulations by race or ethnicity.
      People not falling into one of the five disease categories of interest were considered to be members of the well population, as done in other models of behavior change and primary prevention.
      • Tengs T.O.
      • Osgood N.D.
      • Chen L.L.
      The cost-effectiveness of intensive national school-based anti-tobacco education: results from the tobacco policy model.
      This assumption, while contributing to the model's parsimony, probably underestimates the potential impact of physical inactivity on other important diseases. The results of an alternate all-cause mortality model, which accounted for the direct effects of physical activity on mortality and QOL, yielded slightly more cost-effective results than the disease-specific model.
      It was assumed that the lowest and highest physical activity categories from the BRFSS matched the lowest and highest physical activity categories of the epidemiologic studies from which the associations between physical activity and the relative risk of disease are derived. In general, the highest physical activity category in epidemiologic studies involving men is approximately 2000 kcals/week, slightly lower than this study's highest category of 2280 MET-minutes per week. Most epidemiologic studies involving women set lower thresholds for the highest physical activity category, so that the use of the 2280 MET-minutes per week level is conservative for men and extremely conservative for women.
      • Finkelstein E.A.
      • Wang G.
      • Lee I.M.
      National and state-specific inactivity-attributable medical expenditures for six diseases Final report prepared for the Centers for Disease Control and Prevention by the Research Triangle Institute.
      The BRFSS includes people with and without disease, while the entry cohort of this model is defined as well. It is possible that this biases the current sample toward a slightly larger percentage of the population's being sedentary than would be expected from an entirely healthy one. However, additional analyses of physical activity levels among self-reported diabetics and well people using 2003 BRFSS-weighted prevalences indicate that the increase in sedentary people is only 1.2%.
      In the main analysis, it was assumed that each intervention was applied to the entire population, even though some of the interventions modeled (e.g., social-support or individually adapted health behavior–change strategies) are generally applied only to more-sedentary participants. Applying them to the whole population likely results in less-favorable cost-per-QALY estimates, given that people starting in the highest physical activity level accrue no further benefit in the model from increasing physical activity.
      From the QWB analysis conducted, well people with higher levels of physical activity were found to report higher QOL levels than did well people who were less physically active. Given that these data are cross-sectional, causality cannot be inferred. Unfortunately, there are no longitudinal, population-based estimates for changes in QOL resulting from physical activity interventions.
      Cost-effectiveness analyses have been criticized for using utility values for the entire population instead of those specific to the subpopulation likely to choose the treatment.
      • Meltzer D.
      • Huang E.
      • Jin L.
      • Shook M.
      • Chin M.
      Effects of self-selection on medical cost-effectiveness analysis: impact in intensive therapy for type 2 diabetes mellitus among the elderly.
      This paper shares that limitation. One opportunity for future research is to investigate how the inputs into the CEA model (conditional disease and mortality probabilities, and QWB) differ for likely intervention participants relative to the general population, and how use of the latter values affects the estimates of cost-effectiveness.
      The results of these analyses have several important implications for research and public health practice:
      • This study demonstrates that it is possible to carry out complex prevention modeling of community-based interventions using a decision-analytic approach previously focused on clinical analyses. The complexity and rigor of this method must be carefully balanced with parsimony to ensure the clarity and practicality of the results.
      • The modeling approach that was employed is a useful adjunct to the rigorous evidence-based review carried out by the Task Force to identify recommended interventions for the Community Guide.
      • Careful interpretation, translation, and communication of both the methods and results will be required if the potential of prevention modeling for guiding public health decision making is to be fully realized.
      • Systematic reviews and CEA are important decision-making tools for public health practitioners and advocates.
        • Brownson R.C.
        • Baker E.A.
        • Leet T.L.
        • Gillespie K.N.
        Evidence-based public health.
        • Friedlaender E.
        • Winston F.
        Evidence based advocacy.
        Yet for a majority of public health interventions, reliable CEA data are lacking.
        • Zaza S.
        • Briss P.A.
        • Harris K.W.
        The guide to community preventive services: what works to promote health?.
        However, these results demonstrate that community-intervention strategies for physical activity are cost-effective, suggesting that the widespread implementation of these four intervention strategies is justified. CEA results such as these should be applied in context with information on program reach, effectiveness, feasibility, and community priorities and resources.
        • Glasgow R.E.
        • Vogt T.M.
        • Boles S.M.
        Evaluating the public health impact of health promotion interventions: the RE-AIM framework.
      This study applied high-quality data in a cost-effectiveness framework to extend the findings of the Community Guide to decision making regarding the allocation of societal resources. In taking a flexible, comprehensive, and rigorous approach with input from numerous health and economic sources, a new method has emerged to bring fundamental population-level research to life in setting health policy. The physical activity promotion interventions evaluated in this analysis provide good value for money. With few examples of prevention modeling in the literature to date, the results of this study provide compelling evidence to encourage similar evaluations for other public health prevention strategies and settings, and clearly demonstrate that these physical activity interventions should be implemented at the population level.
      This project integrated the expertise and dedication of both a large multidisciplinary team of accomplished investigators and policy leaders from academic centers across the U.S. and colleagues in the Physical Activity and Health Branch at the CDC. The authors acknowledge the work of: The Robert Wood Johnson Foundation: Terry L. Bazzarre, Pamela G. Russo, Lori Melichar, Kathryn A. Thomas; CDC Foundation: C. Adam Brush, Connie L. Carmack, John R. Moore; UCLA School of Public Health, Los Angeles County, Department of Health Services: Brian Cole, Jonathan E. Fielding, Robert M. Kaplan; Research Triangle Institute/University of North Carolina Center of Excellence in Health Promotion Economics: Eric Finkelstein; Stanford Prevention Research Center at the Stanford University School of Medicine: William L. Haskell; University of Chicago: David Meltzer; The Physical Activity and Health Branch in the Division of Nutrition, Physical Activity and Obesity at the CDC: Chantelle Avery, Laura Biazzo, Mario Bracco, Casey J. Hannan, Carrie Heitzler, Diana Parra, and Guijing Wang; and Milliman Inc.
      This scale of collaboration was made possible by the commitment and support of the Robert Wood Johnson Foundation and the CDC Foundation and their project officers. The principal investigators of this study, Larissa Roux and Michael Pratt, had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. The views expressed in this paper are those of the authors, not the funder, and no conflicts of interest, financial or other, have been identified in the scope of this work.
      No financial disclosures were reported by the authors of this paper.

      References

        • WHO
        A global strategy for diet, physical activity, and health.
        WHO, Geneva2004
        • WHO
        Preventing chronic diseases: a vital investment.
        WHO, Geneva2005
        • USDHHS
        Physical activity and health: a report of the Surgeon General.
        CDC, Atlanta GA1996
        • Colditz G.A.
        Economic costs of obesity and inactivity.
        Med Sci Sports Exerc. 1999; 31: S663-S667
        • Pratt M.
        • Macera C.A.
        • Wang G.
        Higher direct medical costs associated with physical inactivity.
        Phys Sportsmed. 2000; 28: 63-70
        • Finkelstein E.A.
        • Wang G.
        • Lee I.M.
        National and state-specific inactivity-attributable medical expenditures for six diseases.
        CDC and Research Triangle Institute, Research Triangle Park NC2004
        • Haskell W.L.
        • Lee I.
        • Pate R.R.
        • et al.
        Physical activity and public health: updated recommendation for adults from the American College of Sports Medicine and the American Heart Association.
        Med Sci Sports Exerc. 2007; 39: 1423-1434
        • Kahn E.B.
        • Ramsey L.T.
        • Brownson R.C.
        • et al.
        The effectiveness of interventions to increase physical activity: a systematic review.
        Am J Prev Med. 2002; 22: 73-107
        • Truman B.I.
        • Smith-Akin C.K.
        • Hinman A.R.
        • et al.
        Developing the Guide to Community Preventive Services—overview and rationale.
        Am J Prev Med. 2000; 18: 18-26
        • Sapkota S.
        • Bowles H.R.
        • Ham S.A.
        • Kohl III, H.W.
        Adult participation in recommended levels of physical activity—U.S., 2001 and 2003.
        MMWR Morb Mortal Wkly Rep. 2005; 54: 1208-1212
        • Graham J.D.
        • Thompson K.M.
        • Goldie S.J.
        • Segui-Gomez M.
        • Weinstein M.C.
        The cost-effectiveness of air bags by seating position.
        JAMA. 1997; 278: 1418-1425
        • Hirschmann J.V.
        Pneumococcal vaccination in older adults.
        N Engl J Med. 2003; 349: 712-714
        • Goldie S.J.
        • Gaffikin L.
        • Goldhaber-Fiebert J.D.
        • et al.
        Cost-effectiveness of cervical-cancer screening in five developing countries.
        N Engl J Med. 2005; 353: 2158-2168
        • Gaspoz J.M.
        • Coxson P.G.
        • Goldman P.A.
        • et al.
        Cost effectiveness of aspirin, clopidogrel, or both for secondary prevention of coronary heart disease.
        N Engl J Med. 2002; 346: 1800-1806
        • Kuntz K.M.
        • Fleischmann K.E.
        • Hunink M.G.
        • Douglas P.S.
        Cost-effectiveness of diagnostic strategies for patients with chest pain.
        Ann Intern Med. 1999; 130: 709-718
        • Tengs T.O.
        • Osgood N.D.
        • Chen L.L.
        The cost-effectiveness of intensive national school-based anti-tobacco education: results from the tobacco policy model.
        Prev Med. 2001; 33: 558-570
        • Kesaniemi Y.K.
        • Danforth Jr., E.
        • Jensen M.D.
        • Kopelman P.G.
        • Lefebvre P.
        • Reeder B.A.
        Dose-response issues concerning physical activity and health: an evidence-based symposium.
        Med Sci Sports Exerc. 2001; 33: S351-S358
        • CDC
        Behavioral Risk Factor Surveillance System: 2003 survey data.
        • U.S. Census Bureau
        Projections of the resident population by age, sex, race, and Hispanic origin.
        • King A.C.
        • Kiernan M.
        • Oman R.F.
        • Kraemer H.C.
        • Hull M.
        • Ahn D.
        Can we identify who will adhere to long-term physical activity?.
        Health Psychol. 1997; 16: 380-389
        • Marcus B.H.
        • Dubbert P.M.
        • Forsyth L.H.
        • et al.
        Physical activity behavior change: issues in adoption and maintenance.
        Health Psychol. 2000; 19: 32-41
        • Wilson P.W.
        • D'Agostino R.B.
        • Levy D.
        • Belanger A.M.
        • Silbershatz H.
        • Kannel W.B.
        Prediction of coronary heart disease using risk factor categories.
        Circulation. 1998; 97: 1837-1847
        • Wolfe C.D.
        • Rudd A.G.
        • Howard R.
        • et al.
        Incidence and case fatality rates of stroke subtypes in a multiethnic population: the South London Stroke Register.
        J Neurol Neurosurg Psychiatry. 2002; 72: 211-216
        • Brown R.D.
        • Whisnant J.P.
        • Sicks J.D.
        • O'Fallon W.M.
        • Wiebers D.O.
        Stroke incidence, prevalence, and survival: secular trends in Rochester, Minnesota, through 1989.
        Stroke. 1996; 27: 373-380
        • CDC
        National diabetes fact sheet: U.S., 2003.
        • CDC
        National diabetes surveillance system: incidence of diabetes.
        • National Cancer Institute
        Surveillance, epidemiology, and end results.
        • Katzmarzyk P.T.
        • Janssen I.
        The economic costs associated with physical inactivity and obesity in Canada: an update.
        Can J Appl Physiol. 2004; 29: 90-115
        • Hu F.B.
        • Stampfer M.J.
        • Colditz G.A.
        • et al.
        Physical activity and risk of stroke in women.
        JAMA. 2000; 283: 2961-2967
      1. National Vital Statistics Reports.
        (Dec 19 2002;51(3))
        • National Cancer Institute
        Surveillance, epidemiology, and end results.
        • National Cancer Institute
        Surveillance, epidemiology, and end results.
        • CDC
        Self-reported heart disease and stroke among adults with and without diabetes—U.S., 1999–2001.
        MMWR Morb Mortal Wkly Rep. 2003; 52: 1065-1070
        • CDC
        Diabetes surveillance system: prevalence of diabetes.
        • Brown R.D.
        • Whisnat J.P.
        • Sicks J.D.
        • O'Fallon E.M.
        • Wiebers D.O.
        Stroke incidence, prevalence, and survival: secular trends in Rochester, Minnesota, through 1989.
        Stroke. 1996; 27: 373-380
        • Ainsworth B.E.
        • Haskell W.L.
        • Whitt M.C.
        • et al.
        Compendium of physical activities: an update of activity codes and MET intensities.
        Med Sci Sports Exerc. 2000; 32: S498-S504
        • American College of Sports Medicine
        ACSM's guidelines for exercise testing and prescription.
        6th ed. Lippincott Williams & Wilkins, Baltimore MD2000
        • Jeffery R.W.
        • Wing R.R.
        • Thorson C.
        • Burton L.R.
        Use of personal trainers and financial incentives to increase exercise in a behavioral weight-loss program.
        J Consult Clin Psychol. 1998; 66: 777-783
        • Kriska A.M.
        • Bayles C.
        • Cauley J.A.
        • LaPorte R.E.
        • Sandler R.B.
        • Pambianco G.
        A randomized exercise trial in older women: increased activity over two years and the factors associated with compliance.
        Med Sci Sports Exerc. 1986; 18: 557-562
        • Linenger J.M.
        • Chesson 2nd, C.V.
        • Nice D.S.
        Physical fitness gains following simple environmental change.
        Am J Prev Med. 1991; 7: 298-310
        • Lombard D.N.
        • Lombard T.N.
        • Winett R.A.
        Walking to meet health guidelines: the effect of prompting frequency and prompt structure.
        Health Psychol. 1995; 14: 164-170
        • Knowler W.C.
        • Barrett-Connor E.
        • Fowler S.E.
        • et al.
        Reduction in the incidence of type 2 diabetes with lifestyle intervention or metformin.
        N Engl J Med. 2002; 346: 393-403
        • Reger B.
        • Cooper L.
        • Booth-Butterfield S.
        • et al.
        Wheeling Walks: a community campaign using paid media to encourage walking among sedentary older adults.
        Prev Med. 2002; 35: 285-292
        • Young D.R.
        • Haskell W.L.
        • Taylor C.B.
        • Fortmann S.P.
        Effect of community health education on physical activity knowledge, attitudes, and behavior.
        Am J Epidemiol. 1996; 144: 264-274
        • Lee I.M.
        • Skerrett P.J.
        Physical activity and all-cause mortality: what is the dose–response relation?.
        Med Sci Sports Exerc. 2001; 33 (discussion S493–4): S459-S471
        • Oguma Y.
        • Sesso H.D.
        • Paffenbarger Jr, R.S.
        • Lee I.M.
        Physical activity and all cause mortality in women: a review of the evidence.
        Br J Sports Med. 2002; 36: 162-172
      2. National Vital Statistics Reports.
        (Sept 18 2003;52(3))
        • Tengs T.O.
        • Lin T.H.
        A meta-analysis of quality of life estimates for stroke.
        Pharmacoeconomics. 2003; 21: 191-200
        • Kaplan R.M.
        • Anderson J.P.
        • Ake C.F.
        Gender differences in quality-adjusted life expectancy: results from the National Health Interview Survey.
        Clin J Wom Health. 2001; 1: 191-197
        • Kaplan R.M.
        • Anderson J.P.
        The general health policy model: an integrated approach.
        in: Spilker B. Quality of life and pharmacoeconomics in clinical trials. Raven, New York1996
        • Kaplan R.M.
        • Anderson J.P.
        • Patterson T.L.
        • et al.
        Validity of the quality of well-being scale for persons with human immunodeficiency virus infection.
        Psychosom Med. 1995; 57: 138-147
        • Diamond L.H.
        • Sherbine K.
        AMA national summit on obesity: costs of obesity and communications implications for the U.S. healthcare system.
        Thomson Medstat, Ann Arbor MI2004
        • Haddix A.
        • Teutsch S.M.
        • Corso P.S.
        Prevention effectiveness: a guide to decision analysis and economic evaluation.
        2nd ed. Oxford University Press, Oxford2003
        • U.S. Division of Labor, Centers for Medicare & Medicaid Services, Office of the Actuary
        National medical care price indicators.
        • Gold M.
        • Siegel J.
        • Russell L.
        • Weinstein M.
        Cost-effectiveness in health and medicine.
        Oxford University Press, New York1996
        • Agency for Healthcare Research and Quality
        Medical expenditure panel survey, 2001.
        • Pratt M.
        • Macera C.A.
        • Blanton C.
        Levels of physical activity and inactivity in children and adults in the U.S.: current evidence and research issues.
        Med Sci Sports Exerc. 1999; 31: S526-S533
        • Hirth R.A.
        • Chernew M.E.
        • Miller E.
        • Fendrick A.M.
        • Weissert W.G.
        Willingness to pay for a quality-adjusted life year: in search of a standard.
        Med Decis Making. 2000; 20: 332-342
        • Salkeld G.
        • Phongsavan P.
        • Oldenburg B.
        • et al.
        The cost-effectiveness of a cardiovascular risk reduction program in general practice.
        Health Policy. 1997; 41: 105-119
        • Hoerger T.J.
        • Harris R.
        • Hicks K.A.
        • Donahue K.
        • Sorensen S.
        • Engelgau M.
        Screening for type 2 diabetes mellitus: a cost-effectiveness analysis.
        Ann Intern Med. 2004; 140: 689-699
        • Neumann P.J.
        • Rosen A.B.
        • Greenberg D.
        • et al.
        Can we better prioritize resources for cost-utility research?.
        Med Decis Making. 2005; 25: 429-436
        • Stone P.W.
        • Teutsch S.
        • Chapman R.H.
        • Bell C.
        • Goldie S.J.
        • Neumann P.J.
        Cost-utility analyses of clinical preventive services: published ratios, 1976–1997.
        Am J Prev Med. 2000; 19: 15-23
        • Forster M.R.
        Key concepts in model selection: performance and generalizability.
        J Math Psychol. 2000; 44: 205-231
        • Meltzer D.
        • Huang E.
        • Jin L.
        • Shook M.
        • Chin M.
        Effects of self-selection on medical cost-effectiveness analysis: impact in intensive therapy for type 2 diabetes mellitus among the elderly.
        Med Decis Making. 2002; 22: 535
        • Brownson R.C.
        • Baker E.A.
        • Leet T.L.
        • Gillespie K.N.
        Evidence-based public health.
        Oxford University Press, New York2003
        • Friedlaender E.
        • Winston F.
        Evidence based advocacy.
        Inj Prev. 2004; 10: 324-326
        • Zaza S.
        • Briss P.A.
        • Harris K.W.
        The guide to community preventive services: what works to promote health?.
        Oxford University Press, New York2005
        • Glasgow R.E.
        • Vogt T.M.
        • Boles S.M.
        Evaluating the public health impact of health promotion interventions: the RE-AIM framework.
        Am J Public Health. 1999; 89: 1322-1327