1. Obesity, Medication Use and Expenditures among Nonelderly Adults with Asthma Eric M. Sarpong AHRQ Conference September 10, 2012

2. Introduction Prevalence of asthma in adults – chronic and complex health condition increased over the past decade (Zahran et al., 2011) Prevalence of asthma in adults – chronic and complex health condition increased over the past decade (Zahran et al., 2011) Prevalence of obesity, an important risk factor for asthma remains high (Ogden et al., 2012) Prevalence of obesity, an important risk factor for asthma remains high (Ogden et al., 2012) Asthma more difficult to control in obese asthma patients (Lavoie et al., 2006 Saint-Pierre et al., 2006 Dixon et al., 2006) Asthma more difficult to control in obese asthma patients (Lavoie et al., 2006 Saint-Pierre et al., 2006 Dixon et al., 2006) Both conditions result in increased resource use and costs Both conditions result in increased resource use and costs – Estimated healthcare costs of asthma in the U.S. - $18 billion (Sullivan et al., 2011) – 2008 estimated healthcare costs of obesity in the U.S. - $147 billion (Finkelstein et al., 2009) The presence of obesity in asthma patients may exacerbate medication use and expenditures The presence of obesity in asthma patients may exacerbate medication use and expenditures

3. Research Objective Provide insights on the role of obesity in generating health resource use and costs for asthma treatment Provide insights on the role of obesity in generating health resource use and costs for asthma treatment Study used nationally representative data on nonelderly adults with treatment for asthma to examine the relationship between obesity and; Study used nationally representative data on nonelderly adults with treatment for asthma to examine the relationship between obesity and; – Medication use Asthma medication and all prescribed medications Asthma medication and all prescribed medications – Expenditures Asthma medication, all prescribed medications and total health care Asthma medication, all prescribed medications and total health care

4. Previous Literature Relationship between asthma and obesity well documented (Ford, 2005; Shore and Johnston, 2006; Shore, 2006, 2007, 2008; Dixon et al., 2010) Relationship between asthma and obesity well documented (Ford, 2005; Shore and Johnston, 2006; Shore, 2006, 2007, 2008; Dixon et al., 2010) Few studies in the U.S., however, have examined the contribution of obesity to increased medication use and expenditures in adult asthma patients Few studies in the U.S., however, have examined the contribution of obesity to increased medication use and expenditures in adult asthma patients – Taylor et al (2008): obese asthma patients had increased medication use compared to non-overweight asthma patients – Mosen et al (2008): obese individuals more likely to report use of oral corticosteroids – Suh et al (2011) estimated medical costs attributable to obesity in asthma patients - $1,087

5. Contributions Previous literature limited Previous literature limited – Uses administrative claims data, key variables unavailable or uses regional samples – Analyses do not examine the use of all prescribed medications in addition to asthma medications – Previous studies differ from this study on a number of dimensions u Time periods and population (e.g., ≥ 18 years, ≥ 35 years, patients with diagnosed asthma) u Degree to which confounding variables are controlled for across bodyweight categories This study uses regression-based modeling approaches to help inform policymakers about how obesity exacerbates medication use and expenditures This study uses regression-based modeling approaches to help inform policymakers about how obesity exacerbates medication use and expenditures

6. Data 2005-2009 Medical Expenditure Panel Survey (MEPS) 2005-2009 Medical Expenditure Panel Survey (MEPS) – Nationally representative data on U.S. civilian non- institutionalized population – Detailed information on drug therapeutic classifications, quantities purchased, and sources of payment (OOP payments and private and public insurance payments) Detailed information on health conditions, economic and socio-demographic variables Detailed information on health conditions, economic and socio-demographic variables Analytical sample of adults (ages 18-64) with reported treatment for asthma Analytical sample of adults (ages 18-64) with reported treatment for asthma – Reported treatment implies health service use associated with asthma – Sample of 3,580 (964 = normal weight: 18.5 kg/m 2 > BMI ≤ 25 kg/m 2 ; 985 overweight: 25 kg/m 2 BMI ≤ 25 kg/m 2 ; 985 overweight: 25 kg/m 2 < BMI < 30 kg/m 2, and 1,631 obese: 30 kg/m 2 ≥ BMI ≤ 100 kg/m 2

7. Analytic Approach Describe differences in treated prevalence of asthma, medication use and expenditures by BMI categories Describe differences in treated prevalence of asthma, medication use and expenditures by BMI categories Use generalized linear models (GLM) to estimate the effects of BMI categories on: Use generalized linear models (GLM) to estimate the effects of BMI categories on: – Number of asthma and all prescribed medications used (Poisson family and log link function) – Asthma and all prescribed medications expenditures (gamma family and power link function) – Total health care expenditures (gamma family and log link function) All GLM estimates control for age, sex, race-ethnicity, health insurance, family income, employment status, marital status, family size, health status, comorbidities, medication beliefs All GLM estimates control for age, sex, race-ethnicity, health insurance, family income, employment status, marital status, family size, health status, comorbidities, medication beliefs – Effects of BMI presented as differences in observed and predicted change – Effects of Characteristics presented as marginal effects

8. Nonelderly Adults with Treatment for Asthma by BMI Categories Source: MEPS, 2005–2009. BMI = Body mass index (normal weight: 18.5 kg/m 2 > BMI ≤ 25 kg/m 2 ; overweight: 25 kg/m 2 < BMI < 30 kg/m 2, and obese: 30 kg/m 2 ≥ BMI ≤ 100 kg/m 2 ). Estimate is significantly different from normal weight category at: ** p<0.05, *p<0.10. Estimate is significantly different from overweight category at: †† p<0.05, †p<0.10.

9. Medication Use Among Adults With Treatment for Asthma by BMI Categories Source: MEPS, 2005–2009. BMI = Body mass index (normal weight: 18.5 kg/m 2 > BMI ≤ 25 kg/m 2 ; overweight: 25 kg/m 2 < BMI < 30 kg/m 2, and obese:30 kg/m 2 ≥ BMI ≤ 100 kg/m 2 ). Estimate is significantly different from normal weight category at: ** p<0.05, *p<0.10.

10. Summary: Differences in Treated Prevalence and Medication Use by BMI Categories Among nonelderly adults with reported treatment for asthma Among nonelderly adults with reported treatment for asthma – 42.5% percent were obese, 27.7% were overweight, and 29.9% were normal weight – Obese patients were prescribed 1.8 asthma medicines on average compared with 1.7 for normal weight patients – Obese patients filled 40.4 prescribed medications on average compared with 26.1 for overweight patients and 23.7 for normal weight patients

11. Expenditures for Adults with Treatment for Asthma by BMI Categories Source: MEPS, 2005–2009. BMI = Body mass index (normal weight: 18.5 kg/m 2 > BMI ≤ 25 kg/m 2 ; overweight: 25 kg/m 2 < BMI < 30 kg/m 2, and obese:30 kg/m 2 ≥ BMI ≤ 100 kg/m 2 ). All expenditures for all years are CPI-U adjusted to 2009 U.S. dollars. Estimate is significantly different from normal weight category at: ** p<0.05, *p<0.10. Estimate is significantly different from overweight category at: †† p<0.05, †p<0.10.

12. Summary: Differences in Medications and Total Health Care Expenditures by BMI Categories Among nonelderly adults with reported treatment for asthma Among nonelderly adults with reported treatment for asthma – Asthma medications expenditures were 19.3 percent higher for obese patients ($867) compared with those for normal weight patients ($726) – All prescribed medication expenditures were more than 40 percent higher for obese patients ($3,251) than those for overweight patients ($2,243) and normal weight patients ($2,019) – Total health care expenditures were about 30 percent higher for obese patients ($9,750) than those for overweight patients ($7,468) and normal weight patients ($7,486)

13. Selected Characteristics of Nonelderly Adults with Reported Treatment for Asthma by BMI Categories Source: MEPS, 2005–2009. BMI = Body mass index (normal weight: 18.5 kg/m 2 > BMI ≤ 25 kg/m 2 ; overweight: 25 kg/m 2 < BMI < 30 kg/m 2, and obese: 30 kg/m 2 ≥ BMI ≤ 100 kg/m 2 ). ). NH = non-Hispanic; FPL = Federal poverty line. Estimate is significantly different from normal weight category at: ** p<0.05, *p<0.10. VariablesCategoriesNormal weight Overweight Obese Age in years:18 to 44 62.78 46.77** 42.62** 45 to 64 37.22 53.23** 57.38** Race-ethnicity:NH White 81.15 78.09 72.03** NH Black 9.62 12 15.23** Hispanic 9.24 9.9 12.74** Health insurance status:Any private 74.7 75.59 64.17** Public only 17.7 15.07 27.00** Uninsured 7.6 9.33 8.82 Family income (% of FPL):Middle/high income 72.65 70.8 62.58** Low income 9.81 12.05 13.30** Poor/near poor 17.54 17.15 24.12** Marital status:Not married 57.9 45.63** 51.10** Married 42.1 54.37** 48.90** Perceived health status: Excellent/very good/good 69.41 64.39 43.98** Fair/poor 30.59 35.61 56.02** Comorbidity:No comorbid condition 57.17 49.67** 34.23** Comorbid condition 42.83 50.33** 65.77**

14. Summary: Selected Characteristics of Nonelderly Adults with Reported Treatment for Asthma Among nonelderly adults with reported treatment for asthma Among nonelderly adults with reported treatment for asthma – Obese adults were more likely than normal weight adults: To be older (ages 45-64), NH Black and Hispanic, covered by public insurance, poor and low income, married, in fair or poor health To be older (ages 45-64), NH Black and Hispanic, covered by public insurance, poor and low income, married, in fair or poor health To have comorbidities To have comorbidities

15. Effects of BMI on Medication Use and Expenditures. Source: MEPS, 2005–2009. ‡ Differences in observed and predicted changes in BMI categories on outcomes. (a) GLM with Poisson family and log link; (b) GLM with gamma family and power link; (c) GLM with gamma family and log link. BMI = Body mass index (normal weight: 18.5 kg/m 2 > BMI ≤ 25 kg/m 2 ; overweight: 25 kg/m 2 < BMI < 30 kg/m 2, and obese: 30 kg/m 2 ≥ BMI ≤ 100 kg/m 2 ). NH = non-Hispanic; FPL = Federal poverty line. All expenditures for all years are CPI-U adjusted to 2009 U.S. dollars. GLM Models are estimated with controls for age, sex, race-ethnicity, health insurance, family income, employment status, marital status, family size, health status, comorbidities, medication beliefs and year dummies. Significance level: *** p<0.01, ** p<0.05, *p<0.10. Medication Use Expenditures Counterfactual BMI categories ‡ Asthma Medications a. All Medications a. Asthma Medications b. All Medications b. Health Care c Overweight as normal weight -0.03 -1.68 -137 -157 -372 Obese as normal weight -0.07** -12.09** -93* -992** -2433** Obese as overweight -0.03 -10.20** -89** -829** -2073**

16. Interpretation of Effects of BMI on Outcome Variables The study finds that, controlling for other socio-demographic, economic and health characteristics The study finds that, controlling for other socio-demographic, economic and health characteristics If obese nonelderly adults were counterfactually switched to normal weight or overweight If obese nonelderly adults were counterfactually switched to normal weight or overweight – Mean number of all prescribed medication would decrease by 12.1 and 10.2 fills – Expenditures on Asthma medications would decrease by $93 and $89 Asthma medications would decrease by $93 and $89 All prescribed medications would decrease by $992 and $829 All prescribed medications would decrease by $992 and $829 Total health care would decrease by $2,433 and $2,073 Total health care would decrease by $2,433 and $2,073

17. Effects of Selected Characteristic on Medication Use and Expenditures. Source: MEPS, 2005–2009. ‡Marginal effects of characteristics on outcomes: (a) GLM with Poisson family and log link; (b) GLM with gamma family and power link; (c) GLM with gamma family and log link. BMI = Body mass index (normal weight: 18.5 kg/m 2 > BMI ≤ 25 kg/m 2 ; overweight: 25 kg/m 2 < BMI < 30 kg/m 2, and obese: 30 kg/m 2 ≥ BMI ≤ 100 kg/m 2 ). NH = non-Hispanic; FPL = Federal poverty line. All expenditures for all years are CPI-U adjusted to 2009 U.S. dollars. GLM Models are estimated with controls for age, sex, race-ethnicity, health insurance, family income, employment status, marital status, family size, health status, comorbidities, medication beliefs and year dummies. Significance level: *** p<0.01, ** p<0.05, *p<0.10. Medication Use Expenditures VariablesAsthma Medications a. All Medications a. Asthma Medications b. All Medications b. Health Care c Age in years (18 to 44) -- 45 to 64 0.19*** 8.27*** 355.83*** 758.79*** 2427.08*** Race-ethnicity (NH White) -- NH Black -0.18** -3.56 -362.61*** -284.71 -1560.63** Hispanic -0.24*** -8.08*** -537.94*** -729.56*** -1971.89*** Health insurance status (Any private) -- Public only -0.0911.84***215.19711.58***-1335.72 Uninsured -0.28*** -7.23*** -332.37*** -1031.39*** -6363.33*** Family income (% of FPL) (Middle/high income) -- Poor/near poor -0.07 -4.81*** -151.12* -468.25** -2192.58** Perceived health status (Excellent/very good/ good) -- Fair/poor 0.20*** 13.08*** 323.27*** 1145.48*** 4957.14*** Comorbidity (No comorbid condition) -- Comorbid condition -0.01 19.95*** 126.03 1785.90*** 3450.94***

18. Interpretation of Marginal Effects of Characteristics on Outcome Variables Several characteristics were significantly related to the outcome variables. Several characteristics were significantly related to the outcome variables. Age 45-64 and fair or poor health status increase Age 45-64 and fair or poor health status increase – Expected number of asthma medications and all prescribed medication fills – Expenditures for asthma medication, all prescribed medications and total health care Both NH Black and Hispanic race-ethnicity decrease Both NH Black and Hispanic race-ethnicity decrease – Expected number of all prescribed medication fills – Expenditures for asthma medication, all prescribed medications and total health care Both public insurance and low income increase Both public insurance and low income increase – Expected number of all prescribed medication fills and corresponding expenditures Comorbid conditions increase Comorbid conditions increase – expected number of all prescribed medication fills, and expenditures for all prescribed medications and total health care

19. Limitations BMI calculated using self-reported measures of height and weight BMI calculated using self-reported measures of height and weight – May result in underestimate of the true effects of BMI on medication use and expenditures Omitted variables and residual confounding effects cannot be excluded Omitted variables and residual confounding effects cannot be excluded – E.g., asthma severity may play a critical role in the effects of BMI on medication use and expenditures – Results may change if severity differs across BMI groups Inclusion of comorbidities – an intermediate pathway through which BMI affects health services use and expenditures may affect results Inclusion of comorbidities – an intermediate pathway through which BMI affects health services use and expenditures may affect results Non-causal regression models Non-causal regression models

20. Conclusions Study demonstrates that obesity is associated with increased medication use and expenditures in nonelderly adults with asthma Study demonstrates that obesity is associated with increased medication use and expenditures in nonelderly adults with asthma Multivariate analysis showed that counterfactually switching obese nonelderly adults to normal weight would decrease medication use and expenditures Multivariate analysis showed that counterfactually switching obese nonelderly adults to normal weight would decrease medication use and expenditures Implications: Implications: – There appears to be an association between obesity and high costs of care for the treatment of asthma – The study suggest maintaining a normal weight could reduce both asthma related and overall health care costs for nonelderly adults with asthma