Long-term Health and Economic Impact of Preventing and Reducing Overweight and Obesity in Adolescence

Journal of Adolescent Health 46 (2010) 467–473

Original article

Long-term Health and Economic Impact of Preventing and Reducing Overweight and Obesity in Adolescence Li Y. Wang, M.B.A., M.A.a,*, Maxine Denniston, M.S.a, Sarah Lee, Ph.D.a, Deborah Galuska, Ph.D., M.P.H.b, and Richard Lowry, M.D., M.S.a a

Division of Adolescent and School Health, National Center for Chronic Disease Prevention and Health Promotion, Centers for Disease Control and Prevention, Atlanta, Georgia b Division of Nutrition, Physical Activity and Obesity Prevention, National Center for Chronic Disease Prevention and Health Promotion, Centers for Disease Control and Prevention, Atlanta, Georgia Manuscript received July 16, 2009; manuscript accepted November 13, 2009

Abstract

Purpose: Using data from the 2000 National Medical Expenditure Panel Survey and estimates from published studies, this study projected the long-term health and economic impacts of preventing and reducing overweight and obesity in today’s adolescents. Methods: We developed a body mass index progression model to project the impact of a 1% point reduction in both overweight and obese adolescents aged 16–17 years at present on the number of nonoverweight, overweight, and obese adults at age 40 years. We then estimated its impact on the lifetime medical costs and quality-adjusted life years (QALYs) after age 40. Medical costs (in 2007 dollars) and QALYs were discounted to age 17 years. Results: A 1% point reduction in both overweight and obese adolescents ages 16–17 years at present could reduce the number of obese adults by 52,821 in the future. As a result, lifetime medical care costs after age 40 years would decrease by $586 million and lifetime QALYs would increase by 47,138. In the worst case scenario, the 1% point reduction would lower medical costs by $463 million and increase QALYs by 34,394; in the best case scenario, it would reduce medical costs by $691 million and increase QALYs by 57,149. Conclusions: Obesity prevention in adolescents goes beyond its immediate benefits; it can also reduce medical costs and increase QALYs substantially in later life. Therefore, it is important to include long-term health and economic benefits when quantifying the impact of obesity prevention in adolescents. Ó 2010 Society for Adolescent Medicine. All rights reserved.

Keywords:

Obesity prevention; Adolescents; Long-term impact; Medical costs; Quality-adjusted life years

Over the past three decades, the prevalence of obesity (defined as a body mass index [BMI]  95th percentile [1]) among children aged 6–11 years has increased from 6.5% to 17.0%, and among adolescents aged 12–19 years, it has increased even more, from 5.0% to 17.6% [2,3]. The primary concern about these trends is the potential impact on physical

The findings and conclusions in this report are those of the authors and do not necessarily represent the official positions of the Centers for Disease Control and Prevention. *Address correspondence to: Li Yan Wang, M.B.A., M.A., Division of Adolescent and School Health, NCCDPHP, CDC, 4770 Buford Hwy, MS K33, Chamblee, GA 30341. E-mail address: [email protected]

health, not only in childhood but also in adulthood. Obesity can affect a child’s health immediately through physical conditions such as hypercholesterolemia, impaired fasting glucose, hepatitis, sleep apnea, and hypertension [4,5]. In addition, the metabolic and physiologic changes associated with obesity in childhood and adolescence, along with the obesity itself, tend to track into adult life and eventually increase the individual’s risk of disease, disability, and death [6]. If current trends continue, one can anticipate in adults an even greater increase in obesity-related health problems and in obesity-related economic costs such as lost productivity, disability, morbidity, and premature death. Because the probability of transitioning successfully from obesity to normal body weight in adulthood through voluntary weight loss is

1054-139X/10/$ – see front matter Ó 2010 Society for Adolescent Medicine. All rights reserved. doi:10.1016/j.jadohealth.2009.11.204

468

L.Y. Wang et al. / Journal of Adolescent Health 46 (2010) 467–473

low [7], prevention of obesity in children and adolescents has become a public health strategy to reduce overall prevalence of obesity and its health and social consequences. Since the early 1990s, numerous dietary and physical activity intervention programs aimed at preventing and reducing obesity in children and adolescents have been developed, implemented, and evaluated across the nation. However, most current evaluation studies assess only the short-term impact of intervention programs [8–12], such as increases in physical activity, reduction in fat intake, decreases in BMI, or reduction in percent body fat. Few studies have examined the long-term impact of obesity prevention programs for youths [13,14], because study of remote outcomes typically requires very long-term, costprohibitive follow-up to obtain information on morbidity and mortality. Although some programs are shown to be effective in the short-term, such as by lowering the prevalence of obesity in children or adolescents, their long-term impacts remain unknown. It seems helpful to take a step back from the short-term efficacy of specific programs to consider more fundamentally the potential long-term health and economic impacts of preventing and reducing obesity in children and adolescents. For example, if obesity prevention interventions for children and adolescents reduce the current prevalence of overweight and obesity by 1% point, what potential difference will this make in adulthood? In this study, using data from various sources, we projected the potential long-term impact of a hypothetical 1% point reduction in the prevalence of overweight and obesity among adolescents aged 16–17 years. Our goal is to offer insight into the relationship between a program’s short-term success (i.e., lowering overweight and obesity prevalence in adolescence) and its long-term health and economic benefits (i.e., increase in quality-adjusted life years and decrease in medical care cost in adulthood). Methods The analysis of this study was conducted through two major steps. First, because most obesity-related adverse health outcomes occur after age 40 [15], we developed a BMI progression model to project the number of nonoverweight, overweight, and obese adults at age 40 on the basis of BMI status in adolescence. Using this model, we estimated the impact of a 1% point reduction in the prevalence of overweight and obesity among youth on the number of adults in each weight category. Second, we estimated the impact of the 1% point reduction on lifetime medical costs and qualityadjusted life years (QALYs) after age 40. We conducted analyses for males and females separately. Because early studies suggest that the tracking of BMI is strongest from adolescence to adulthood [16,17], we selected adolescents aged 16–17 years as our study cohort. The size of the cohort was based on the 2000 U.S. Census data; there were approximately 4.2 million males and 3.9 million females who were 16 or 17 years old in 2000. Using

the 2003–2004 prevalence estimates for U.S. adolescents aged 12–19 years [18], we first divided the cohort into three groups: nonoverweight (BMI < 85th percentile), overweight (85th  BMI < 95th percentile), and obese (BMI  95th percentile). For each group, we then projected the number of adults at age 40 being nonoverweight (BMI < 25), overweight (25  BMI < 30), and obese (BMI  30) using the most recent published probability estimates on the tracking of BMI from adolescence to adulthood [19]. This study used data from the National Longitudinal Study of Youth 1979 to examine the association between BMI in adolescence and obesity in adulthood. Among overweight adolescents, 62% of the males and 73% of the females became obese adults; among obese adolescents, 80% of the males and 92% of the females became obese adults [19]. Finally, we combined the number of adults in the same weight category across all three baseline BMI groups as the projected number of nonoverweight, overweight, and obese adults on the basis of current prevalence of overweight and obesity in adolescence. We used the same model to project the total number of nonoverweight, overweight, and obese adults in a hypothetical scenario where the current prevalence of overweight and obesity among adolescents each decreased 1% point. We assumed that a 1% point reduction in obesity would result in a 1% point increase in overweight, while a 1% point reduction in overweight would result in a 1% point increase in nonoverweight (see Table 1). The tracking probabilities remain the same in both scenarios. The impact of a 1% point reduction in overweight and obesity among 16- and 17-year-old adolescents was estimated as the change in the number of nonoverweight, overweight, and obese adults between the current prevalence scenario and the 1% point reduction scenario. Table 1 presents the prevalence estimates and the BMI tracking probability estimates that we used for each scenario. To estimate the impact of a 1% point prevalence reduction on lifetime medical costs and QALYs after age 40, we needed to know the lifetime medical costs and QALYs associated with being nonoverweight, overweight, and obese after age 40. Although many studies have investigated the relationships between obesity and medical costs [20–24], and between obesity and health-related quality of life (HRQL) [25,26], no study in the existing literature provided the medical costs and QALY estimates we needed for this study. Using data from the 2000 National Medical Expenditure Panel Survey (MEPS), we derived estimates of mean annual medical costs and HRQL after age 40 by age group, sex, and weight status. Using life table approach, we generated life expectancy at age 40 by weight status based on the survival probability estimates derived by Finkelstein et al. [27] and the percent distribution of weight status, smoking status, and race/ethnicity of the 2000 MEPS sample. We then combined those cost and HRQL estimates with those life expectancy estimates to calculate lifetime medical costs and QALYs associated with being nonoverweight, overweight, and obese after age 40.

L.Y. Wang et al. / Journal of Adolescent Health 46 (2010) 467–473

469

Table 1 Baseline prevalences and BMI tracking probabilities Baseline BMI status

Male Nonoverweight (BMI < 85th percentile) Overweight (85th  BMI < 95th percentile) Obese (BMI  95th percentile) Total Female Nonoverweight (BMI < 85th percentile) Overweight (85th  BMI < 95th percentile) Obese (BMI  95th percentile) Total a

Baseline prevalence

BMI tracking probablitiesa

Current

1% reduction

BMI < 25

25  BMI < 30

BMI  30

63.20% 18.50% 18.30% 100%

64.20% 18.50% 17.30% 100%

24% 4% 4% na

56% 34% 16% na

20% 62% 80% na

68.30% 15.30% 16.40% 100%

69.30% 15.30% 15.40% 100%

46% 2% 0% na

34% 25% 8% na

20% 73% 92% na

BMI tracking probability, the probability of transitioning from a particular weight category in adolescence into a particular weight category in adulthood.

To generate medical cost estimates, we selected 14,143 observations from the 2000 MEPS for our analysis. The sample included persons who were 18 years or older and had data on both height and weight. Two-part models are often used to model cost data that include many zero observations [28]. Thus, this approach is well suited for predicting mean expenditures using the MEPS data as 19.8% of 2000 MEPS respondents reported no medical expenditures. Because cost data often contain a small number of extremely large values (i.e., are skewed to the right), the second part of a two-part model is often an ordinary least-squares model with a log transformed outcome variable. However, when the outcome variable is log transformed it must be retransformed in order to draw conclusions about the original variable. This retransformation can result in biased estimates in a particular range of estimated costs if the error terms (residuals) are not normally distributed or if they vary with cost. Generalized linear models (GLM) can be used to directly model both the mean and variance functions on the original scale, thus avoiding the need to retransform results. We followed the procedures in Manning and Mullahy [29] and Buntin and Zaslavsky [28] to test the distribution of the cost data and found a two-part GLM with a gamma distribution and log link to be the most appropriate model for the cost analysis. That is, our analyses used a GLM in which expenditures were modeled on the log scale and the variance was assumed to be proportional to the mean squared expenditures (i.e., variance has the form of a gamma-like distribution). We first used a logistic regression model to predict the probability of having a positive expenditure for each sample subject. Next, we used a GLM with log link and gamma distribution to derive the predicted mean annual

expenditures by age group (40–49, 50–59, 60–69, and 70), sex, and BMI category (18.5  BMI < 25, 25  BMI < 30, and BMI  30). The expected expenditures were calculated as the product of the probability of having a positive expenditure and the predicted expenditure level conditional on the presence of expenditures. We started both models with 10 independent variables: age group, sex, race and ethnicity, BMI category, marital status, poverty status, education attainment, health insurance coverage, smoking status, and physical activity. We tested for collinearity among all variables and checked interaction among BMI category, age group, and sex. Several insignificant variables were excluded from the final models (marital status, poverty status, and physical activity out of the first model; smoking status out of the second model). Delete-1 Jackknife method was used to estimate standard error and 95% confidence intervals (CI) for the expected expenditures. In 2000, MEPS began to collect HRQL data from adult participants using the EuroQol (EQ-5D) instrument. The EQ-5D contains five domains, mobility, self-care, usual activity, pain/discomfort, and anxiety/depression. Each domain has three levels, no problem, some problem, and major problem. The instrument generates 243 health states, including ‘‘unconscious’’ and ‘‘dead,’’ making 245 in all. The weight for each health state can be determined from population-based preference weights and can be incorporated into QALYs, which combine gains or losses in quantity and quality of life. The strength of EQ-5D is that it can provide population-based preference scores; the weakness is that it still requires primary data collection. The instrument has been widely used in the United States and abroad and extensively validated in clinical settings and population-based

Table 2 Projected number of nonoverweight, overweight, and obese adults at age 40 based on baseline prevalence Number of male adults (n ¼ 4,157,619)

Current prevalence 1% Prevalence reduction Change with 1% point prevalence reduction

Number of female adults (n ¼ 3,899,236)

BMI < 25

25  BMI < 30

BMI  30

BMI < 25

25  BMI < 30

BMI  30

679,026 687,259 8,232

1,851,783 1,868,413 16,630

1,626,810 1,601,947 (24,863)

1,229,324 1,247,104 17,781

1,096,750 1,106,927 10,177

1,573,162 1,545,205 (27,958)

4,034–4,097 4,227–4,286 5,195–5,214 5,238–5,250 4,066 4,256 5,205 5,244 2,788–2,827 3,453–3,520 4,005–4,019 5,108–5,118 2,808 3,486 4,012 5,113 2,231–2,275 3,738–3,783 3,698–3,739 5,776–5,795 2,253 3,761 3,719 5,786 3,695–3,812 4,938–5,029 6,087–6,245 6,715–6,730 HRQL ¼ Health-related quality of life.

1,831–1,894 2,983–3,057 5,085–5,172 6,150–6,166 1,863 3,020 5,128 6,158

3,753 4,983 6,166 6,722

.79–.83 .77–.81 .76–.82 .73–.80 .81 .79 .79 .77 .84–.87 .81–.85 .81–.86 .80–.84 .85 .83 .83 .82 .87–.90 .84–.87 .84–.88 .83–.87 .89 .85 .86 .85 .83–.86 .79–.85 .80–.85 .77–.86 .87–.89 .84–.87 .82–.87 .81–.85 .88 .86 .85 .83

.85 .82 .82 .82

na 39.75 na 42.09 na 42.64 na 35.85 na 38.96

Life expectancy at age 40 37.45 na HRQL within each age interval 40–49 .89 .87–.90 50–59 .86 .84–.88 60–69 .85 .82–.88 >70 .82 .79–.85 Annual medical costs within each age interval ($) 40–49 2,471 2,414–2,529 50–59 3,809 3,740–3,879 60–69 5,777 5,731–5,823 >70 5,322 5,292–5,352

Mean 95% CI Mean Mean 95% CI 95% CI

BMI  30

95% CI

Mean Mean Mean

95% CI

BMI  30 18.5  BMI < 25 25  BMI < 30 18.5  BMI < 25

25  BMI < 30 Female Male

studies. In this study, we used the EQ-5D preference weights that were generated from samples of the U.S. population [30]. Although 15,438 adults provided EQ-5D data for the MEPS 2000 survey, 11,412 adults were included in this study after excluding those with proxy reports and those with missing height or weight information. Multilinear regression was used to model the relationship between BMI categories and HRQL scores and to generate age- and sex-specific HRQL scores for each BMI category (least square means), while controlling for other covariates (race and ethnicity, marital status, poverty status, education attainment, health insurance coverage, smoking status, and physical activity). Analyses of medical costs and HRQL were conducted using SUDAAN version 9.13 and STATA version 10.1 (for the second model of cost analyses) to account for the complex sample design. Two studies have previously examined the impact of overweight and obesity on life expectancy [31,32]; however, one study [31] was based on a cohort that was not nationally representative, the other study [32] reported life expectancy by BMI units not BMI categories. A recent study conducted by Finkelstein et al. [27] reported age-specific survival probabilities by gender, BMI and race categories for ages 18–85 based on analyses conducted using linked data from the 1986–2002 National Health Interview Survey and the 1986–2002 National death Index. Based on Finkelstein’s survival probabilities and the percent distribution of smoking status, BMI category, and race/ethnicity of the 2000 MEPS sample, we generated the weighted average life expectancy at age 40 in each BMI category using life table approach. We combined these life expectancy estimates with the mean annual medical expenditure and mean HRQL estimates to calculate lifetime cumulative medical costs and QALYs after age 40 years for each BMI category. Assuming that annual medical costs and HRQL would remain constant within each age interval we first calculated cumulative medical costs and QALYs over each age interval and discounted them to age 17 years using a 3% discount rate. Then we added up the discounted costs and QALYs across all age intervals as the discounted lifetime cumulative medical costs and QALYs after age 40 years. For the last age interval of 70 years and older, we calculated the cumulative costs and QALYs over the life expectancy after age 70 years (e.g., 9.75, 12.09, and 12.64 years for obese, overweight, and nonoverweight women, respectively). All costs were adjusted to 2007 dollars. To test whether the results of our base-case analysis were dependent on the accuracy of the parameter estimates that were derived in this or a previous study, we conducted both univariate and multivariate sensitivity analyses on three variables: BMI tracking probabilities; age-, sex-, and BMI-specific annual medical costs after age 40 years; and age-, sex-, and BMI-specific quality of life after age 40 years. We employed the 95% CI for each of the three parameters as a plausible range for variation. In multivariate analysis, Monte Carlo simulation of 10,000 trials was performed using @RISK (Palisade Corp.,

95% CI

L.Y. Wang et al. / Journal of Adolescent Health 46 (2010) 467–473

Table 3 Life expectancy, HRQL, and medical costs at age 40 and older

470

L.Y. Wang et al. / Journal of Adolescent Health 46 (2010) 467–473

471

Table 4 Lifetime medical costs saved and QALYs gained after age 40 as a result of 1% point prevalence reduction in adolescence (base case results)

Male Per capita lifetime medical costs after age 40, discounted to age 17 ($) Per capita lifetime QALYs after age 40, discounted to age 17 Increase in number of adults Increase in total medical costs ($) Increase in total QALYs Female Per capita lifetime medical costs after age 40, discounted to age 17 ($) Per capita lifetime QALYs after age 40, discounted to age 17 Increase in number of adults Increase in total medical costs (million $) Increase in total QALYs

18.5  BMI < 25

25  BMI < 30

BMI  30

All

43,896 9.71 8,232 361,355,957 79,937

39,846 9.90 16,630 662,657,876 164,599

54,203 9.12 (24,863) (1,347,617,781) (226,841)

na na na (323,603,948) 17,695

43,120 10.43 17,781 766,696,251 185,509

43,474 10.04 10,177 442,433,159 102,177

52,646 9.24 (27,958) (1,471,864,324) (258,244)

na na na (262,734,914) 29,443

QALYs ¼ quality-adjusted life years.

Newfield, NY). Parameter values for each simulation trial were selected randomly from each 95% CI, assuming a triangular distribution of values for each parameter. Results Table 2 shows the projected number of nonoverweight, overweight, and obese adults in each of the two baseline prevalence scenarios as well as the increase or decrease in the number of adults in each BMI category due to the 1% point reduction in adolescence. Among the 8,056,855 persons in the study cohort, the number of expected obese adults decreased by 52,821 and the number of expected overweight adults and nonoverweight adults increased by 26,807 and 26,013, respectively. Table 3 shows the life expectancy estimates, the adjusted mean annual medical expenditures, and the adjusted mean HRQL scores. The annual medical costs generally increased with increasing age. Obese adults had the highest medical costs relative to normal weight and overweight adults with one exception of female aged 70 years and older, but overweight adults did not necessarily have higher medical costs relative to normal weight adults. The HRQL scores generally decreased with increasing age. In females, the scores decreased with increasing BMI, but in males the scores were not very different between overweight adults and normal weight adults.

Table 4 shows the average lifetime medical costs and QALYs after age 40 as well as the lifetime medical costs saved and QALYs gained as a result of a 1% point reduction in adolescence. In females, lifetime costs increased and lifetime QALYs decreased with increasing BMI. In males, obese adults had the highest costs and lowest QALYs, but overweight adults had lower costs and higher QALYs than normal weight adults. As a result of the 1% point reduction in overweight and obesity in adolescence, total lifetime medical costs would decrease by $586.3 million ($73 per capita) and total QALYs would increase by 47,138 (0.0059 per capita). Table 5 shows the results of the sensitivity analyses. Comparing the univariate analysis results with the multivariate analysis result, we found that our results were mainly sensitive to progression probability estimates, but not sensitive to the cost estimates or HRQL estimates. In the worst case scenario, the 1% point reduction would lower medical costs by $463 million and increase QALYs by 34,394; in the best case scenario, such a reduction would reduce medical costs by $691 million and increase QALYs by 57,149. Discussion Using data from the 2000 MEPS and estimates from published studies, this study projected the long-term health and economic impacts of preventing and reducing overweight and obesity in today’s adolescents. Our

Table 5 Lifetime medical costs saved and QALYs gained after age 40 as a result of 1% point prevalence reduction in adolescence (sensitivity analysis results) Male

Univariate analyses Progression probabilities HRQL Annual medical costs Multivariate analysis

Female

All

Medical costs saved (millions $)

QALYs gained

Medical costs saved (millions $)

QALYs gained

Medical costs saved (millions $)

QALYs gained

247–401 324 316–331 247–401

13,546–21,844 14,447–20,988 17,695 12,657–23,335

216–289 263 258–267 216–290

22,668–31,744 25,859–33,057 29,443 21,737–33,814

463–690 587 574–598 463–691

36,214–53,588 40,306–54,045 47,138 34,394–57,149

QALYs ¼ quality-adjusted life years.

472

L.Y. Wang et al. / Journal of Adolescent Health 46 (2010) 467–473

projections show that a 1% point reduction in both overweight and obese adolescents aged 16–17 years today could reduce the number of obese adults by 52,821 in the future. As a result, lifetime medical costs would decrease by $586.3 million ($73 per capita) and total QALYs would increase by 47,138 (0.0059 per capita). These results suggest that obesity prevention in adolescents goes beyond its immediate benefits; it can also reduce medical costs and increase QALYs substantially in later life. Therefore, it is important to include long-term health and economic benefits when quantifying the impact of obesity prevention in adolescents. Several studies have previously investigated the relationships between obesity and medical costs [20–24], and between obesity and HRQL [25,26]. Unfortunately, due to the differences in study population (i.e., different age groups, different racial groups, different BMI groups), we cannot directly compare our estimates with those of other studies. However, the patterns of our estimates are generally consistent with other studies. First, annual medical costs increase with increasing age, and the costs attributable to overweight are small or negative but the costs attributable to obesity are positive with the exception of females 70 years and older. One possible explanation for obese females 70 years and older not having higher costs might be a survivor effect. Second, HRQL decrease with increasing age and decrease with increasing category of BMI with the exception of overweight males whose HRQL are slightly higher than normal weight males. One possible cause for overweight males having lower costs and slightly higher HRQL than normal weight males might be that many diseases that are associated with low BMI, such as cancer and AIDS, result in higher medical care costs and lower HRQL. The model developed in this study can be used to project long-term benefits regardless of the type of prevalence change (i.e., in overweight, obesity, or both) or the magnitude of change. The magnitude of the benefits depends on the combination of the type of prevalence reduction and the magnitude of the prevalence reduction. For example, the projected benefits from our base-case analysis would double if both overweight and obesity prevalence decreased by 2% points and would decrease by half if both overweight and obesity prevalence decreased by 0.5% point. Using projected long-term benefits, researchers can assess the cost-effectiveness of interventions. If data on intervention costs can be obtained, one can directly calculate the costeffectiveness ratio as net intervention costs (intervention costs minus costs averted by the intervention) divided by the effectiveness of the intervention (QALYs gained). An intervention is generally considered cost-effective if the cost-effectiveness ratio is less than or equal to $30,000 per QALY saved [33–35]. Using the projected benefits, researchers can also investigate the maximum investment that an intervention can spend and still meet acceptable standards of cost-effectiveness. Some limitations of this study should be noted. First, height and weight data from the 2000 MEPS were self-reported.

Self-reporting tends to result in underestimates for weight and overestimates for height [36,37]; therefore, the findings in this study likely underestimate the difference in medical costs and HRQL between different BMI groups. In other words, our projected long-term impact could be an underestimate of true impact. Second, because no published prevalence data are available for adolescents aged 16–17 years, we used the published prevalence for adolescents 12–19 years. Third, for simplicity, we assumed that weight status remains constant after age 40 years. The impact of this assumption on the cost savings and QALYs gained is not clear. On one hand, some overweight or obese persons might lose weight after age 40; on the other hand, some normal-weight persons might become overweight or obese after age 40 [38,39]. Fourth, in this study, we used cross-sectional data to generate cost and HRQL estimates and applied them to our study cohort. Although this approach could take into account age effect, it misses out on cohort effect. A recent study found that increase in BMI with aging is underestimated in all age groups when studying cross-sectional data only [40]. Finally, we assumed that the whole cohort would move to age 40 years without any change in race distribution. As the nation’s minority population continues to increase, our assumption should result in an underestimation of the number of overweight and obese adults. This study fills a void in the current literature by projecting the long-term health and economic impacts of preventing and reducing overweight and obesity in today’s adolescents. It showed that successful obesity prevention in adolescents today could achieve substantial health and economic benefits in later life. The findings of this study warrant careful considerations by policy makers when they must make resource allocation decisions among competing programs. We hope our findings will provide additional argument for schools, communities, and health care professionals to justify funds for intensified prevention programs and policies. The models developed in this study can be used to project an intervention’s long-term benefits based on its short-term success. This study also provides important data projections that will enable researchers to conduct cost-effectiveness analysis of obesity prevention interventions. References [1] Barlow SE. the Expert Committee. Expert committee recommendations regarding the prevention, assessment, and treatment of child and adolescent overweight and obesity: Summary report. Pediatrics 2007; 120:S164–92. [2] Hedley AA, Ogden CL, Johnson CL, et al. Prevalence of overweight and obesity among US children, adolescents, and adults, 1999–2002. JAMA 2004;291:2847–50. [3] Ogden CL, Carroll MD, Flegal KM. High body mass index for age among US children and adolescents, 2003–2006. JAMA 2008;299: 2401–5. [4] Dietz WH. Health consequences of obesity in youth: Childhood predictors of adult disease. Pediatrics 1998;101:518–25. [5] Must A, Strauss RS. Risks and consequences of childhood and adolescent obesity. Int J Obes Relat Metab Disord 1999;23(Suppl 2):S2–11.

L.Y. Wang et al. / Journal of Adolescent Health 46 (2010) 467–473 [6] Koplan JP, Liverman CT, Kraak VA, eds. Preventing Childhood Obesity: Health in the Balance. Washington, DC: The National Academies Press, 2005. [7] Wadden TA, Phelan S. Behavioral assessment of the obese patient. In: Wadden TA, Stunkard AJ, eds. Handbook of Obesity Treatment. New York, NY: Guilford Press, 2002:186–226. [8] Luepker RV, Perry CL, McKinlay SM, et al. Outcomes of a field trial to improve children’s dietary patterns and physical activity. The Child and Adolescent Trial for Cardiovascular Health. CATCH collaborative group. JAMA 1996;275:768–76. [9] Gortmaker SL, Peterson K, Wiecha J, et al. Reducing obesity via a school-based interdisciplinary intervention among youth: Planet health. Arch Pediatr Adolesc Med 1999;153:409–18. [10] Robinson TN. Reducing children’s television viewing to prevent obesity: A randomized controlled trial. JAMA 1999;282:1561–7. [11] Yin Z, Gutin B, Johnson MH, et al. An environmental approach to obesity prevention in children: Medical College of Georgia FitKid Project year 1 results. Obes Res 2005;13:2153–61. [12] Coleman KJ, Tiller CL, Sanchez J, et al. Prevention of the epidemic increase in child risk of overweight in low-income schools: The El Paso coordinated approach to child health. Arch Pediatr Adolesc Med 2005;159:217–24. [13] Wang LY, Yang Q, Lowry R, et al. Economic analysis of a schoolbased obesity prevention program. Obes Res 2003;11:1313–24. [14] Brown HS, Perez A, Li Y, et al. The cost-effectiveness of a schoolbased overweight program. Int J Behav Nutr Phys Act 2007;4:47. [15] Centers for Disease Control and Prevention. Summary health statistics for U.S. adults: National Health Interview Survey. Vital Health Stat 2006;2007(235):1–153. [16] Freedman DS, Khan LK, Serdula MK, et al. The relation of childhood BMI to adult adiposity: The Bogalusa Heart Study. Pediatrics 2005; 115:22–7. [17] Whitaker RC, Wright JA, Pepe MS, et al. Predicting obesity in young adulthood from childhood and parental obesity. N Engl J Med 1997; 337:869–73. [18] Ogden CL, Carroll MD, Curtin LR, et al. Prevalence of overweight and obesity in the United States, 1999–2004. JAMA 2006;295: 1549–55. [19] Wang LY, Chyen D, Lee S, et al. The association between body mass index in adolescence and overweight or obesity in adulthood. J Adolesc Health 2008;42:512–8. [20] Wee CC, Phillips RS, Legedza AT, et al. Health care expenditures associated with overweight and obesity among US adults: Importance of age and race. Am J Public Health 2005;95:159–65. [21] Daviglus ML, Liu K, Yan LL, et al. Relation of body mass index in young adulthood and middle age to Medicare expenditures in older age. JAMA 2004;292:2743–9. [22] Thompson D, Brown JB, Nichols GA, et al. Body mass index and future healthcare costs: A retrospective cohort study. Obes Res 2001; 9:210–8.

473

[23] Andreyeva T, Sturm R, Ringel JS. Moderate and severe obesity have large differences in health care costs. Obes Res 2004;12:1936–43. [24] Finkelstein EA, Trogdon JG, Brown DS, et al. The lifetime medical cost burden of overweight and obesity: Implications for obesity prevention. Obesity 2008;16:1843–8. [25] Muennig P, Lubetkin E, Jia H, et al. Gender and the burden of disease attributable to obesity. Am J Public Health 2006;96:1662–8. [26] Jia H, Lubetkin EL. The impact of obesity on health-related qualityof-life in the general adult US population. J Public Health 2005;27: 156–64. [27] Finkelstein E, Brown D, Wrage L, et al. Individual and aggregate years of life lost associated with overweight and obesity. Obesity (Sliver Spring), in press. [28] Buntin M, Zaslavsky A. Too much ado about two-part models and transformation? Comparing methods of modeling Medicare expenditures. J Health Econ 2004;23:525–42. [29] Manning W, Mullahy J. Estimating log models: To transform or not to transform? J Health Econ 2001;20:461–94. [30] Shaw JW, Johnson JA, Coons SJ. US valuation of the EQ-5D health states: Development and testing of the D1 valuation model. Med Care 2005;43:203–20. [31] Peeters A, Barendregt JJ, Willekens F, et al. Obesity in adulthood and its consequences for life expectancy: A life-table analysis. Ann Intern Med 2003;138:24–32. [32] Fontaine KR, Redden DT, Wang C, et al. Years of life lost due to obesity. JAMA 2003;289:187–93. [33] Laupacis A, Feeny D, Detsky A, et al. Tentative guidelines for using clinical and economic evaluations revisited. CMAJ 1993; 148:927–9. [34] Owens DK, Nease RF, Harris R. Use of cost-effectiveness and value of information analyses to customize guidelines for specific clinical practice settings [abstract]. Med Decis Making 1993;13:395. [35] Tolley GL, Fabian R. Valuing Health for Policy: An Economic Approach. Chicago, IL: University of Chicago Press, 1994. [36] Kuczmarski MF, Kuczmarski RJ, Najjar M. Effects of age on validity of self-reported height, weight, and body mass index: Findings from the Third National Health and Nutrition Examination Survey, 1988–1994. J Am Diet Assoc 2001;101:28–34. [37] Ezzati M, Martin H, Skjold S, et al. Trends in national and state-level obesity in the USA after correction for self-report bias: Analysis of health surveys. J R Soc Med 2006;99:250–7. [38] He XZ, Baker DW. Changes in weight among a nationally representative cohort of adults aged 51 to 61, 1992 to 2000. Am J Prev Med 2004; 27:8–15. [39] Sheehan TJ, duBrava S, Dechello LM, et al. Rates of weight change for black and white Americans over a twenty year period. Int J Obes Relat Metab Disord 2003;27:498–504. [40] Nooyens AC, Visscher TL, Verschuren WM, et al. Age, period and cohort effects on body mass index in adults: The Doetinchem Cohort Study. Public Health Nutr 2009;12:862–70.