- Email: [email protected]
Physical inactivity
Physical Inactivity Direct Cost to a Health Plan Nancy A. Garrett, PhD, Michelle Brasure, PhD, MSPH, Kathryn H. Schmitz, PhD, MPH, MSEd, Monica M. Schultz, MS, Michael R. Huber, MHSA Purpose:
The purpose of this study was to estimate the total medical expenditures attributable to physical inactivity patterns among members of a large health plan, Blue Cross Blue Shield of Minnesota.
Methods:
The study used a cost-of-illness approach to attribute medical and pharmacy costs for specific diseases to physical inactivity in 2000. Relative risks come from the scientific literature, demonstrating that heart disease, stroke, hypertension, type 2 diabetes, colon cancer, breast cancer, osteoporosis, depression, and anxiety are directly related to individual physical activity patterns in adults. Data sources were the 2000 Behavioral Risk Factor Surveillance System and medical claims incurred in 2000 among 1.5 million health plan members aged ⱖ18 years. Primary analysis was completed in 2002.
Results:
Nearly 12% of depression and anxiety and 31% of colon cancer, heart disease, osteoporosis, and stroke cases were attributable to physical inactivity. Heart disease was the most expensive outcome of physical inactivity within the health plan population, costing $35.3 million in 2000. Total health plan expenditures attributable to physical inactivity were $83.6 million, or $56 per member.
Conclusions: This study confirms the growing body of research quantifying physical inactivity as a serious and expensive public health problem. The costs associated with physical inactivity are borne by taxpayers, employers, and individuals in the form of higher taxes to subsidize public insurance programs and increased health insurance premiums. (Am J Prev Med 2004;27(4):304 –309) © 2004 American Journal of Preventive Medicine
Introduction
I
n spite of the growing evidence for the importance of physical activity, most Americans have a sedentary lifestyle. In 2000, 74% of adults failed to meet recommended guidelines for physical activity of 30 minutes of moderate-intensity activity on most days of the week.1 This paper quantifies the cost of physical inactivity in a health plan population, providing information for managers and policymakers on the economic implications of this health behavior. In addition, the analytic method is described in detail, providing a blueprint for estimating the direct cost of physical activity inexpensively using administrative data and publicly available survey data.
From HealthPartners (Garrett), Division of Epidemiology, University of Minnesota (Schmitz), and Regulatory and Clinical Research Institute, Inc. (Schultz), Minneapolis, Minnesota; Minnesota Department of Health (Brasure), St. Paul, Minnesota; and Blue Cross and Blue Shield of Minnesota (Huber), Eagan, Minnesota Note: This study was conducted while NAG and MMS were employees of Blue Cross and Blue Shield of Minnesota. Address correspondence to: Nancy Garrett, PhD, Health Services Analysis and Reporting, HealthPartners, 8100 34th Ave. S, Mail code 21108Q, Minneapolis MN 55440-1309. E-mail: Nancy.A.Garrett@ HealthPartners.com
304
Diet and physical activity patterns together rank second only to tobacco use as the leading risk factor for all causes of death.2,3 Physical inactivity increases the risk of heart disease, stroke, hypertension, type 2 diabetes, colon cancer, breast cancer, osteoporosis, depression, anxiety, and injuries from falls among the elderly.4,5 In addition to preventing disease, greater levels of physical activity can lessen complications among people with chronic diseases.6 Many studies have quantified the impact of physical activity on the incidence and severity of disease as an independent risk factor. Meta-analyses of these studies have provided relative risks for inactive versus active people for conditions attributable to physical inactivity. Colditz7 developed relative risks for ischemic heart disease (2.0), hypertension (1.5), type 2 diabetes (1.5), osteoporosis and related fractures (2.0), and colon cancer (2.0). Based on a review of studies by Wannamethee et al.8 in 1999, Bricker et al.5 calculated that inactive people were twice as likely to experience a stroke compared to active people. In a meta-analysis, Stephenson et al.9 found that sedentary people were 1.3 times as likely to experience depression compared to active people, and Friedenreich et al.10 found that average relative risks for inactive versus active people
Am J Prev Med 2004;27(4) © 2004 American Journal of Preventive Medicine • Published by Elsevier Inc.
0749-3797/04/$–see front matter doi:10.1016/j.amepre.2004.07.014
Table 1. Condition categories, ICD-9 codes, and relative risks Condition category
Code
Ischemic heart disease Hypertension Stroke Depression and anxiety
Breast cancer
ICD-9: 410–414.xx ICD-9: 401–404.xx ICD-9: 430–438.xx ICD-9: 296.2, 296.3, 300.0, 300.00, 300.01, 300.02, 300.09, 300.4, 311, 313, 313.0, 313.1 ICD-9: 250.x0, 250.x2, 357.2, 362.01, 362.02, 366.41 ICD-9: 174-174.x
Osteoporosis and related fractures Colon cancer
ICD-9: 733-733.xx ICD-9: 153, 153.x, 233.4
Diabetes, type 2
Source paper on risk for physically active relative to sedentary
Relative risk
Colditz (1999)7 Colditz (1999)7 Bricker (2001)5 based on Wannamethee (1999)8 Stephenson (2000)9
2.0 1.5 2.0 1.3
Colditz (1999)7
1.5
Friedenreich (2001)10 (average RR 1.43–1.67; this study uses the midpoint of 1.5) Colditz (1999)7 Colditz (1999)7
1.5 2.0 2.0
ICD-9, International Classification of Diseases, 9th revision.
for breast cancer ranged from 1.43 to 1.67. Injuries from falls among the elderly have also been linked to physical inactivity.11 Communicating the importance of interventions to increase physical activity can be difficult in spite of the epidemiologic evidence, because many diseases caused by inactivity develop over years rather than months. To help make the case for health promotion activities more tangible, researchers have translated the causal evidence linking physical activity patterns to disease into costs using epidemiologic methods. Some studies have calculated the direct medical costs of inactivity,9,12–15 while others have added indirect costs from absenteeism, lost productivity, and years of life lost.7 Because of increasing knowledge of the role of lifestyle and behaviors in causing and exacerbating chronic illness in the United States, health plans, employers, and policymakers are looking for ways to support and reward lifestyle changes among populations as an important component of primary, secondary, and tertiary prevention. The first step in constructing evidence justifying interventions at the plan or similar population level is documenting the extent and magnitude of the problem. This study used epidemiologic methods to estimate the direct cost of physical inactivity within the membership of a health plan, Blue Cross Blue Shield of Minnesota. The analysis had several advantages over previous studies: the estimates were based on paid claim amounts rather than charges, and pharmacy data were included, an increasingly important proportion of total healthcare costs. In addition, the problem was approached from the point of view of a health plan, a key stakeholder in striving to improve population health.
Methods Study Population This study was designed to estimate the total medical expenditures attributable to physical activity patterns for Blue Cross
Blue Shield of Minnesota. Blue Cross provides insurance for about one third of Minnesota residents. All 1.5 million adult members (aged ⬎18) were included in this study, including fee-for-service and managed care products. The study population also included commercial insurance and several government programs, as well as people living in the Twin Cities metropolitan area, in Greater Minnesota, and outside Minnesota.
Data Two sources of data provided information on medical care utilization of Blue Cross enrollees and their physical activity patterns. The first data source was inpatient and outpatient medical claims records from Blue Cross with dates of service in 2000. Facility, professional, and x-ray/lab claims were included for conditions with a convincing level of scientific evidence linking them to physical inactivity. These conditions included heart disease, stroke, hypertension, type 2 diabetes, colon cancer, breast cancer, osteoporosis, depression, and anxiety.5,9,10,15 Injuries from falls among the elderly were not included because of limitations in identifying the source of injury through the claims data system. Breast cancer and other hormonally related cancers were also excluded, as there is still debate in the literature about their association with physical activity. The specific condition categories and their corresponding ICD-9 (International Classification of Diseases, Ninth Revision) codes appear in Table 1. The second source was the 2000 Behavioral Risk Factor Surveillance System (BRFSS), a national system of statewide surveys conducted by the Centers for Disease Control and Prevention (CDC). These statewide telephone surveys provide information on adult activity patterns, including whether respondents participated in any leisure-time physical activities and, if so, the frequency and intensity of those activities. The BRFSS results for Minnesota provide an approximation of the physical activity patterns of Blue Cross members. The age distribution in Minnesota of adult Blue Cross members is slightly younger than that for the state as a whole: 12% of Blue Cross adult members are aged ⬎65 years compared to 16% in Minnesota. In addition, Blue Cross members are less likely to live in the Twin Cities metropolitan
Am J Prev Med 2004;27(4)
305
area (36%) than the general population (54%).16 To the extent that physical activity patterns vary by age and metropolitan versus nonmetropolitan counties, the BRFSS results may incorrectly state levels of physical activity in the Blue Cross population. Because patterns of physical activity in Minnesota are not very different from national patterns, the Minnesota BRFSS is likely to be a reasonable estimate of activity levels for members who live outside Minnesota.
Measures Physical activity is defined in a manner consistent with that of the consensus recommendations from the Surgeon General, the American Academy of Sports Medicine, and the American Heart Association.4 This definition includes three categories of physical activity: (1) inactive, (2) irregularly active, and (3) regularly active. Survey respondents were classified as regularly active if they engaged in vigorous leisure-time physical activity for at least 20 minutes for a minimum of 3 days per week, or if they engaged in moderate physical activity totaling at least 150 minutes per week and on at least 5 days per week. Respondents were classified as inactive if they reported performing no leisure-time physical activity at all. Irregularly active respondents were those who report some amount of physical activity, but not enough to classify them as regularly active. All medical claims from Blue Cross members aged ⱖ18 years (approximately 1.5 million) were aggregated. The primary analysis was completed in 2002. For pharmacy claims, which in most cases do not contain an ICD-9 code, a method described by Bernhardt et al.18 was applied to assign drug classes to diagnosis clusters. This method assigns drug classes on an aggregate basis to diagnosis groups—a necessary step because diagnosis codes do not appear on most prescriptions. (Contact the corresponding author for details on this method.) Claims were then aggregated by condition category, using the paid claim amount as the measure of cost. This amount included both the payment that the health plan makes to providers and the payment the member makes, and so represents an approximation of the cost to society of medical services.
Analysis The study used a cost-of-illness methodology to attribute costs to the risk factor of insufficient physical activity. The approach was specifically modeled after reports from Georgia5 and Colditz.7 Treatment of the chronic conditions addressed in this study was expensive; however, only a portion of the cases of each condition was attributable to physical activity patterns. For example, 35% of coronary heart disease in the United States is attributable to physical activity patterns.19 This 35% is called the “population attributable risk proportion” (PARP), and it is an estimate of the proportion of disease that can be attributed to the risk factor of interest—in this case, inactivity. The PARP is calculated using both (1) the relative risk of disease for inactive versus active people and for irregularly active versus active people, and (2) the prevalence of the risk factor in the population of interest, using the following formula:
306
PARP ⫽
⌺Pexp共i兲 ⴱ 共RRi ⫺ 1兲
1 ⫹ ⌺关Pexp共i兲 ⴱ 共RRi ⫺ 1兲兴
⫻ 100.
The RR is relative risk, which is calculated as the cumulative incidence of disease or death among the exposed (the physically inactive) divided by the cumulative incidence among the unexposed. Pexp is the population prevalence of exposure to the risk factor, and i is the level of exposure. This analysis included three levels of exposure: inactive, irregularly active, and regularly active. Diseases included in the analysis have been directly linked to physical inactivity in the research literature. The RR represents a meta-analysis of the findings of published studies comparing the risk of disease incidence and/or adverse events resulting from disease for people who are inactive versus regularly active. The sources for these relative risks are listed in Table 1. Multiplying the total estimated expenditures for each condition category by the relevant PARP (specific condition, inactive vs irregularly active) provides the estimate of expenditures attributable to physical inactivity. The RR for the irregularly active category is the geometric mean of the risk for regularly active versus inactive people (as suggested by Bricker et al.5).
Results One quarter of Minnesota adults aged ⱖ18 were inactive in 2000, 49% were irregularly active, and 27% were regularly active, according to weighted BRFSS data received from the CDC (S. Ham, CDC, personal communication, March 27, 2002). Table 2 applies the prevalence of physical activity to the RR of disease for each condition category to calculate the PARP for each condition. Almost one third (31%) of costs related to heart disease, stroke, colon cancer, and osteoporosis in this population were attributable to physical inactivity. Applying the PARP for each condition and activity category to total costs yielded a total of $83.6 million in expenditures for medical treatment attributable to physical inactivity in 2000 (Table 3). This worked out to $56 per member in 2000, which could have been avoided if the entire population was active. Heart disease accounted for 42% of total expenditures in 2000, or $35.3 million. Stroke was the next most costly condition, followed by depression and diabetes.
Discussion This study estimated that physical inactivity cost $83.6 million in 2000 for inpatient, outpatient, and pharmacy claims in a health plan population of 1.5 million members, or $56 per member. The results are not directly comparable to other studies, however, because the analysis included different conditions, outpatient and pharmacy costs, and actual paid amounts rather than charges. A health plan population may also be healthier on average than the population of an entire state, which includes uninsured individuals as well as
American Journal of Preventive Medicine, Volume 27, Number 4
Table 2. Population-attributable risk proportion (PARP) of insufficient physical activity Activity category Condition category Ischemic heart disease Hypertension Stroke Depression and anxiety Diabetes, type 2 Breast cancer Osteoporosis Colon cancer
Pexp RR PARP RR PARP RR PARP RR PARP RR PARP RR PARP RR PARP RR PARP
Inactive
Irregularly active
Regularly activea
0.25 2.0 0.17 1.5 0.10 2.0 0.17 1.3 0.07 1.5 0.10 1.5 0.02 2.0 0.17 2.0 0.17
0.49 1.4 0.14 1.2 0.08 1.4 0.14 1.1 0.06 1.2 0.08 1.2 0.02 1.4 0.14 1.4 0.14
0.26 1.0
Total PARPb
0.31 1.0 0.18 1.0 0.31 1.0 0.12 1.0 0.18 1.0 0.19 1.0 0.31 1.0 0.31
Pexp, population prevalence of exposure to risk factor; PARP, population-attributable risk proportions; RR, relative risk. a Regular physical activity: ⱖ5 days a week, ⱖ150 minutes total, or ⱖ3 days a week of vigorous activity for ⱖ20 minutes each session b PARP ⫽ (Pexp*(RR ⫺ 1))/(1 ⫹ sum((Pexp (RR ⫺ 1)))).
those covered exclusively by Medicare. In addition, Minnesota is one of the lowest cost areas of the country for Medicare services.20 Yet the results of this study are on the same order of magnitude as other studies. Colditz7 attributed $128 per capita to physical inactivity in the United States in 1995. State estimates of the charges related to physical inactivity were $477 million in Georgia ($79 per capita)5 and $157 million in South Carolina ($78 per capita).15 A total of $1 billion in direct costs were related to physical inactivity in New York ($70 per capita), and $81.6 million inpatient hospital charges was attributed to physical inactivity in Washington ($19 per capita).21 Per capita costs for each study were estimated using the 2000 adult population from the Census.17 Other studies have addressed the short-term costs of physical inactivity by comparing the average healthcare expenditures of physically active individuals versus physically inactive individuals.13,14 The advantage of this type of analysis is that all healthcare expenditures are included, not just those related to the specific Table 3. Blue Cross Blue Shield of Minnesota costs attributable to physical inactivity ($ millions, 2000 dollars) Disease
Cost
Ischemic heart disease Hypertension Stroke Depression and anxiety Diabetes, type 2 Breast cancer Osteoporosis Colon cancer Total
$35.3 10.8 9.2 9.1 7.2 4.6 4.5 2.9 $83.6
diseases with a proven association to physical inactivity. Pratt et al.22 compared medical expenditures between regularly active (defined as spending ⱖ30 minutes in moderate or strenuous physical activity at least three times per week) and inactive participants of the 1987 National Medical Expenditures Survey, with the net annual benefit of physical activity at $330 per person in 1987 dollars. This is substantially higher than the $56 per member attributable to physical inactivity in the current study (in 2000 dollars). The difference is likely a result of the difference in methods: the current study included only medical costs associated with a subset of conditions, while the Pratt et al.22 report included all medical costs related to inactivity. In summary, this study’s estimates of the costs of physical inactivity are of a similar scale to those of the growing literature on this subject. Because the analysis included the entire healthcare dollar, not just inpatient costs, and used actual paid amounts rather than charges, it may provide a more accurate assessment of costs than previous studies. This study focused on physical activity as a risk factor for disease. The epidemiologic studies used to determine relative risks for active versus inactive individuals examined physical activity as an independent risk factor. However, because activity and obesity are strongly correlated, they may be confounded in this analysis. Since physical inactivity is a risk factor for obesity, and obesity is itself a risk factor for the diseases included in this study, the analysis may have considerably underestimated the true cost of physical inactivity. Since physical activity and healthcare cost data were not linked at the individual level, it is possible that the reported associations were due to unmeasured conAm J Prev Med 2004;27(4)
307
founding at the individual level and not due to physical inactivity. In addition, the study did not control for pre-existing conditions that might limit an individual’s ability to be physically active. However, individual-level analyses have found that inactive individuals incur more healthcare costs than active individuals.13,22 This cost-of-illness approach is an efficient, practical way to model the impact of a behavioral risk factor in a large population. Another problem inherent with this methodology is that current inactivity may not correlate well with prior levels of physical inactivity in individuals. Many of the costs associated with diseases related to physical inactivity are a result of past, rather than current, inactivity. However, BRFSS data show that the proportion of the population that is inactive in Minnesota has changed very little since 1986.23 Therefore, the estimates should be robust at the aggregate level. There were also several limitations of the practical implications of this study. First, even the most effective interventions to date have achieved only a modest reduction in physical inactivity.24 Therefore, even with the application of recommended strategies to increase physical activity,24 some proportion of the population will likely be inactive and irregularly active. Second, when a population becomes more active, some potential cost savings may take years to realize, especially those related to the prevention of chronic disease. However, recent evidence indicates that the effect of exercise on blood pressure and blood insulin and glucose25 are realized immediately after an exercise session, and effects on depression and anxiety are prompt as well.26 The results of this study estimating the costs of physical inactivity point to the need for cost– benefit analyses of physical activity interventions.27,28 These studies put the costs of physical inactivity in the context of the impact and costs of interventions to increase physical activity. Cost-benefit analyses will help stakeholders to make decisions about the most feasible approaches to change.
Conclusions This analysis puts a dollar figure on the direct cost of physical inactivity in a health plan population in 2000. While $83.6 million is most likely an underestimate of the medical costs of inactivity since it includes a narrow group of conditions and does not reflect the influence of physical inactivity on obesity, it puts the abstract concept of physical activity as a public health issue into a concrete form that is meaningful to a wide variety of stakeholders. The data used in this study were easily accessible to individual health plans, and all of the information needed for other plans to conduct similar analyses is presented here. Communities and coalitions of groups within those communities (such as health 308
What This Study Adds . . . This study adds to the literature on epidemiologic estimates of the costs of physical inactivity by using more precise data on healthcare costs. Specifically, the estimates were based on paid claim amounts rather than charges and pharmacy data were included. In addition, the problem was approached from the point of view of a health plan, providing a blueprint for estimating the direct cost of physical activity inexpensively by using administrative data and publicly available survey data.
plans, employers who purchase health insurance, providers, and local governments) may be more willing to invest time and money in physical activity promotion efforts when they learn that not making this investment is a costly decision. We are grateful to Sanne Magnan, Patricia Ball, and Dan Anderson of Blue Cross Blue Shield of Minnesota for their helpful contributions, and Karen Fitzner and Nadine Caputo of the Blue Cross Blue Shield Association for their support of this project.
References 1. Centers for Disease Control and Prevention. Prevalence of Physical Activity, Including Lifestyle Activities Among Adults—United States, 2000 –2001. MMWR Morb Mortal Wkly Rep 2003;52:764 – 69. 2. McGinnis JM, Foege WH. Actual causes of death in the United States. JAMA 1993;270:2207–12. 3. Mokdad AH, Marks JS, Stroup DF, Gerberding JL. Actual causes of death in the United States, 2000. JAMA 2004;291:1238 – 45. 4. Centers for Disease Control and Prevention. Physical activity and health: a report of the Surgeon General. Atlanta GA: U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, 1996. 5. Bricker SK, Powell KE, Parasha U, et al. Physical activity report. Atlanta GA: Georgia Department of Human Resources, Division of Public Health, American Heart Association, Southeast Affiliate, September 2001. 6. Hu FB, Stampfer MJ, Solomon C, et al. Physical activity and risk for cardiovascular events in diabetic women. Ann Intern Med 2001;134:96 –105. 7. Colditz GA. Economic costs of obesity and inactivity. Med Sci Sports Exerc 1999(suppl 11):S663–7. 8. Wannamethee SG, Shaper AG. Physical activity and the prevention of stroke. J Cardiovasc Risk 1999;6:213– 6. 9. Stephenson J, Bauman A, Armstrong T, Smith B, Bellew B. The costs of illness attributable to physical activity in Australia: a preliminary study. Commonwealth of Australia, Canberra: Population Health Division Publications, 2000. 10. Friedenreich CM. Physical activity and cancer prevention: from observational to intervention research. Cancer Epidemiol Biomarkers Prev 2001;10:287–301. 11. Gillespie L, Gillespie W, Robertson M, Lamb S, Cumming R. Interventions for preventing falls in elderly people. Cochrane Database Syst Rev 2001;3: CD000340. 12. Colditz G. Economic costs of obesity and inactivity. Med Sci Sports Exerc 2000(suppl 10):S663–7. 13. Pronk N, Goodman M, O’Connor P, Martinson B. Relationship between modifiable health risks and short-term health care charges. JAMA 1999;282:2235–9. 14. Goetzel RZ, Anderson DR, Whitmer RW, Ozminkowski RJ, Dunn RL, Wasserman J. The relationship between modifiable health risks and health
American Journal of Preventive Medicine, Volume 27, Number 4
15.
16.
17. 18.
19.
20. 21.
care expenditures. An analysis of the multi-employer HERO health risk and cost database. The Health Enhancement Research Organization (HERO) Research Committee. J Occup Environ Med 1998;40:843–54. Powell KE, Greany M, Huang Y, Whitt M. Physical activity in South Carolina. Columbia: University of South Carolina Prevention Research Center, School of Public Health, 1999. Garrett N, Alesci N, Schultz M, Foldes S, Magnan S, Manley M. The relationship of stage of change for smoking cessation to stage of change for fruit/vegetable consumption and physical activity in a health plan population. J Health Promotion 2004. In press. U.S. Census Bureau. 2000 Adult census. Available at: http://quickfacts.census. gov/gfd. Accessed October 15, 2003. Bernhardt TS, Foldes S, Heaton A, et al. Using diagnosis clusters: a case study of the specification of the cost of care for a health plan population and selected sub-populations. Paper presented at: Minnesota Health Services Research Conference, Minneapolis MN, February 23, 1999. Brownson RC, Remington PL, Davis JR. Chronic disease epidemiology and control, 2nd ed. Washington DC: American Public Health Association, 1998. Wennberg J, Cooper M. The Dartmouth atlas of health care 1999. Chicago IL: American Hospital Association Press, 1999. Claybrooke C, LeMier M, Bensley L,et al. Physical activity in Washington State. Olympia WA: Washington State Department of Health, 2000.
22. Pratt M, Macera CA, Blanton C. Levels of physical activity and inactivity in children and adults in the United States: current evidence and research issues. Med Sci Sports Exerc 1999(suppl 11):S526 –33. 23. Centers for Disease Control and Prevention. Physical activity statistics 1986 – 2000. Available at: www.cdc.gov/nccdphp/dnpa/physical/stats/stats.htm. Accessed June 14, 2004. 24. Kahn EB, Ramsey LT, Brownson RC, et al. The effectiveness of interventions to increase physical activity: a systematic review. Am J Prev Med 2002;22(suppl 4):73–107. 25. Kelley DE, Goodpaster BH. Effects of exercise on glucose homeostasis in type 2 diabetes mellitus. Med Sci Sports Exerc 2001;33(suppl 6):S495–501. 26. Dunn AL, Trivedi MH, O’Neal HA. Physical activity dose–response effects on outcomes of depression and anxiety. Med Sci Sports Exerc 2001; 33(suppl 6):S587–597. 27. Roux L, Hu D, Yore M, Yanagawa T, Pratt M. The health and economic impact of promoting physical activity: a decision analytic approach. Paper presented at annual meeting of the American College of Sports Medicine, Indianapolis IN, June 4, 2004. 28. Proper K, Bruyne M, Hildebrandt V, van der Beek A, Jan Merrding W, van Mechelen W. Cost-effectiveness of a worksite physical activity counseling program: results of a randomized controlled trial. Paper presented at annual meeting of the American College of Sports Medicine, Indianapolis IN, June 4, 2004.
Am J Prev Med 2004;27(4)
309
The purpose of this study was to estimate the total medical expenditures attributable to physical inactivity patterns among members of a large health plan, Blue Cross Blue Shield of Minnesota.
Methods:
The study used a cost-of-illness approach to attribute medical and pharmacy costs for specific diseases to physical inactivity in 2000. Relative risks come from the scientific literature, demonstrating that heart disease, stroke, hypertension, type 2 diabetes, colon cancer, breast cancer, osteoporosis, depression, and anxiety are directly related to individual physical activity patterns in adults. Data sources were the 2000 Behavioral Risk Factor Surveillance System and medical claims incurred in 2000 among 1.5 million health plan members aged ⱖ18 years. Primary analysis was completed in 2002.
Results:
Nearly 12% of depression and anxiety and 31% of colon cancer, heart disease, osteoporosis, and stroke cases were attributable to physical inactivity. Heart disease was the most expensive outcome of physical inactivity within the health plan population, costing $35.3 million in 2000. Total health plan expenditures attributable to physical inactivity were $83.6 million, or $56 per member.
Conclusions: This study confirms the growing body of research quantifying physical inactivity as a serious and expensive public health problem. The costs associated with physical inactivity are borne by taxpayers, employers, and individuals in the form of higher taxes to subsidize public insurance programs and increased health insurance premiums. (Am J Prev Med 2004;27(4):304 –309) © 2004 American Journal of Preventive Medicine
Introduction
I
n spite of the growing evidence for the importance of physical activity, most Americans have a sedentary lifestyle. In 2000, 74% of adults failed to meet recommended guidelines for physical activity of 30 minutes of moderate-intensity activity on most days of the week.1 This paper quantifies the cost of physical inactivity in a health plan population, providing information for managers and policymakers on the economic implications of this health behavior. In addition, the analytic method is described in detail, providing a blueprint for estimating the direct cost of physical activity inexpensively using administrative data and publicly available survey data.
From HealthPartners (Garrett), Division of Epidemiology, University of Minnesota (Schmitz), and Regulatory and Clinical Research Institute, Inc. (Schultz), Minneapolis, Minnesota; Minnesota Department of Health (Brasure), St. Paul, Minnesota; and Blue Cross and Blue Shield of Minnesota (Huber), Eagan, Minnesota Note: This study was conducted while NAG and MMS were employees of Blue Cross and Blue Shield of Minnesota. Address correspondence to: Nancy Garrett, PhD, Health Services Analysis and Reporting, HealthPartners, 8100 34th Ave. S, Mail code 21108Q, Minneapolis MN 55440-1309. E-mail: Nancy.A.Garrett@ HealthPartners.com
304
Diet and physical activity patterns together rank second only to tobacco use as the leading risk factor for all causes of death.2,3 Physical inactivity increases the risk of heart disease, stroke, hypertension, type 2 diabetes, colon cancer, breast cancer, osteoporosis, depression, anxiety, and injuries from falls among the elderly.4,5 In addition to preventing disease, greater levels of physical activity can lessen complications among people with chronic diseases.6 Many studies have quantified the impact of physical activity on the incidence and severity of disease as an independent risk factor. Meta-analyses of these studies have provided relative risks for inactive versus active people for conditions attributable to physical inactivity. Colditz7 developed relative risks for ischemic heart disease (2.0), hypertension (1.5), type 2 diabetes (1.5), osteoporosis and related fractures (2.0), and colon cancer (2.0). Based on a review of studies by Wannamethee et al.8 in 1999, Bricker et al.5 calculated that inactive people were twice as likely to experience a stroke compared to active people. In a meta-analysis, Stephenson et al.9 found that sedentary people were 1.3 times as likely to experience depression compared to active people, and Friedenreich et al.10 found that average relative risks for inactive versus active people
Am J Prev Med 2004;27(4) © 2004 American Journal of Preventive Medicine • Published by Elsevier Inc.
0749-3797/04/$–see front matter doi:10.1016/j.amepre.2004.07.014
Table 1. Condition categories, ICD-9 codes, and relative risks Condition category
Code
Ischemic heart disease Hypertension Stroke Depression and anxiety
Breast cancer
ICD-9: 410–414.xx ICD-9: 401–404.xx ICD-9: 430–438.xx ICD-9: 296.2, 296.3, 300.0, 300.00, 300.01, 300.02, 300.09, 300.4, 311, 313, 313.0, 313.1 ICD-9: 250.x0, 250.x2, 357.2, 362.01, 362.02, 366.41 ICD-9: 174-174.x
Osteoporosis and related fractures Colon cancer
ICD-9: 733-733.xx ICD-9: 153, 153.x, 233.4
Diabetes, type 2
Source paper on risk for physically active relative to sedentary
Relative risk
Colditz (1999)7 Colditz (1999)7 Bricker (2001)5 based on Wannamethee (1999)8 Stephenson (2000)9
2.0 1.5 2.0 1.3
Colditz (1999)7
1.5
Friedenreich (2001)10 (average RR 1.43–1.67; this study uses the midpoint of 1.5) Colditz (1999)7 Colditz (1999)7
1.5 2.0 2.0
ICD-9, International Classification of Diseases, 9th revision.
for breast cancer ranged from 1.43 to 1.67. Injuries from falls among the elderly have also been linked to physical inactivity.11 Communicating the importance of interventions to increase physical activity can be difficult in spite of the epidemiologic evidence, because many diseases caused by inactivity develop over years rather than months. To help make the case for health promotion activities more tangible, researchers have translated the causal evidence linking physical activity patterns to disease into costs using epidemiologic methods. Some studies have calculated the direct medical costs of inactivity,9,12–15 while others have added indirect costs from absenteeism, lost productivity, and years of life lost.7 Because of increasing knowledge of the role of lifestyle and behaviors in causing and exacerbating chronic illness in the United States, health plans, employers, and policymakers are looking for ways to support and reward lifestyle changes among populations as an important component of primary, secondary, and tertiary prevention. The first step in constructing evidence justifying interventions at the plan or similar population level is documenting the extent and magnitude of the problem. This study used epidemiologic methods to estimate the direct cost of physical inactivity within the membership of a health plan, Blue Cross Blue Shield of Minnesota. The analysis had several advantages over previous studies: the estimates were based on paid claim amounts rather than charges, and pharmacy data were included, an increasingly important proportion of total healthcare costs. In addition, the problem was approached from the point of view of a health plan, a key stakeholder in striving to improve population health.
Methods Study Population This study was designed to estimate the total medical expenditures attributable to physical activity patterns for Blue Cross
Blue Shield of Minnesota. Blue Cross provides insurance for about one third of Minnesota residents. All 1.5 million adult members (aged ⬎18) were included in this study, including fee-for-service and managed care products. The study population also included commercial insurance and several government programs, as well as people living in the Twin Cities metropolitan area, in Greater Minnesota, and outside Minnesota.
Data Two sources of data provided information on medical care utilization of Blue Cross enrollees and their physical activity patterns. The first data source was inpatient and outpatient medical claims records from Blue Cross with dates of service in 2000. Facility, professional, and x-ray/lab claims were included for conditions with a convincing level of scientific evidence linking them to physical inactivity. These conditions included heart disease, stroke, hypertension, type 2 diabetes, colon cancer, breast cancer, osteoporosis, depression, and anxiety.5,9,10,15 Injuries from falls among the elderly were not included because of limitations in identifying the source of injury through the claims data system. Breast cancer and other hormonally related cancers were also excluded, as there is still debate in the literature about their association with physical activity. The specific condition categories and their corresponding ICD-9 (International Classification of Diseases, Ninth Revision) codes appear in Table 1. The second source was the 2000 Behavioral Risk Factor Surveillance System (BRFSS), a national system of statewide surveys conducted by the Centers for Disease Control and Prevention (CDC). These statewide telephone surveys provide information on adult activity patterns, including whether respondents participated in any leisure-time physical activities and, if so, the frequency and intensity of those activities. The BRFSS results for Minnesota provide an approximation of the physical activity patterns of Blue Cross members. The age distribution in Minnesota of adult Blue Cross members is slightly younger than that for the state as a whole: 12% of Blue Cross adult members are aged ⬎65 years compared to 16% in Minnesota. In addition, Blue Cross members are less likely to live in the Twin Cities metropolitan
Am J Prev Med 2004;27(4)
305
area (36%) than the general population (54%).16 To the extent that physical activity patterns vary by age and metropolitan versus nonmetropolitan counties, the BRFSS results may incorrectly state levels of physical activity in the Blue Cross population. Because patterns of physical activity in Minnesota are not very different from national patterns, the Minnesota BRFSS is likely to be a reasonable estimate of activity levels for members who live outside Minnesota.
Measures Physical activity is defined in a manner consistent with that of the consensus recommendations from the Surgeon General, the American Academy of Sports Medicine, and the American Heart Association.4 This definition includes three categories of physical activity: (1) inactive, (2) irregularly active, and (3) regularly active. Survey respondents were classified as regularly active if they engaged in vigorous leisure-time physical activity for at least 20 minutes for a minimum of 3 days per week, or if they engaged in moderate physical activity totaling at least 150 minutes per week and on at least 5 days per week. Respondents were classified as inactive if they reported performing no leisure-time physical activity at all. Irregularly active respondents were those who report some amount of physical activity, but not enough to classify them as regularly active. All medical claims from Blue Cross members aged ⱖ18 years (approximately 1.5 million) were aggregated. The primary analysis was completed in 2002. For pharmacy claims, which in most cases do not contain an ICD-9 code, a method described by Bernhardt et al.18 was applied to assign drug classes to diagnosis clusters. This method assigns drug classes on an aggregate basis to diagnosis groups—a necessary step because diagnosis codes do not appear on most prescriptions. (Contact the corresponding author for details on this method.) Claims were then aggregated by condition category, using the paid claim amount as the measure of cost. This amount included both the payment that the health plan makes to providers and the payment the member makes, and so represents an approximation of the cost to society of medical services.
Analysis The study used a cost-of-illness methodology to attribute costs to the risk factor of insufficient physical activity. The approach was specifically modeled after reports from Georgia5 and Colditz.7 Treatment of the chronic conditions addressed in this study was expensive; however, only a portion of the cases of each condition was attributable to physical activity patterns. For example, 35% of coronary heart disease in the United States is attributable to physical activity patterns.19 This 35% is called the “population attributable risk proportion” (PARP), and it is an estimate of the proportion of disease that can be attributed to the risk factor of interest—in this case, inactivity. The PARP is calculated using both (1) the relative risk of disease for inactive versus active people and for irregularly active versus active people, and (2) the prevalence of the risk factor in the population of interest, using the following formula:
306
PARP ⫽
⌺Pexp共i兲 ⴱ 共RRi ⫺ 1兲
1 ⫹ ⌺关Pexp共i兲 ⴱ 共RRi ⫺ 1兲兴
⫻ 100.
The RR is relative risk, which is calculated as the cumulative incidence of disease or death among the exposed (the physically inactive) divided by the cumulative incidence among the unexposed. Pexp is the population prevalence of exposure to the risk factor, and i is the level of exposure. This analysis included three levels of exposure: inactive, irregularly active, and regularly active. Diseases included in the analysis have been directly linked to physical inactivity in the research literature. The RR represents a meta-analysis of the findings of published studies comparing the risk of disease incidence and/or adverse events resulting from disease for people who are inactive versus regularly active. The sources for these relative risks are listed in Table 1. Multiplying the total estimated expenditures for each condition category by the relevant PARP (specific condition, inactive vs irregularly active) provides the estimate of expenditures attributable to physical inactivity. The RR for the irregularly active category is the geometric mean of the risk for regularly active versus inactive people (as suggested by Bricker et al.5).
Results One quarter of Minnesota adults aged ⱖ18 were inactive in 2000, 49% were irregularly active, and 27% were regularly active, according to weighted BRFSS data received from the CDC (S. Ham, CDC, personal communication, March 27, 2002). Table 2 applies the prevalence of physical activity to the RR of disease for each condition category to calculate the PARP for each condition. Almost one third (31%) of costs related to heart disease, stroke, colon cancer, and osteoporosis in this population were attributable to physical inactivity. Applying the PARP for each condition and activity category to total costs yielded a total of $83.6 million in expenditures for medical treatment attributable to physical inactivity in 2000 (Table 3). This worked out to $56 per member in 2000, which could have been avoided if the entire population was active. Heart disease accounted for 42% of total expenditures in 2000, or $35.3 million. Stroke was the next most costly condition, followed by depression and diabetes.
Discussion This study estimated that physical inactivity cost $83.6 million in 2000 for inpatient, outpatient, and pharmacy claims in a health plan population of 1.5 million members, or $56 per member. The results are not directly comparable to other studies, however, because the analysis included different conditions, outpatient and pharmacy costs, and actual paid amounts rather than charges. A health plan population may also be healthier on average than the population of an entire state, which includes uninsured individuals as well as
American Journal of Preventive Medicine, Volume 27, Number 4
Table 2. Population-attributable risk proportion (PARP) of insufficient physical activity Activity category Condition category Ischemic heart disease Hypertension Stroke Depression and anxiety Diabetes, type 2 Breast cancer Osteoporosis Colon cancer
Pexp RR PARP RR PARP RR PARP RR PARP RR PARP RR PARP RR PARP RR PARP
Inactive
Irregularly active
Regularly activea
0.25 2.0 0.17 1.5 0.10 2.0 0.17 1.3 0.07 1.5 0.10 1.5 0.02 2.0 0.17 2.0 0.17
0.49 1.4 0.14 1.2 0.08 1.4 0.14 1.1 0.06 1.2 0.08 1.2 0.02 1.4 0.14 1.4 0.14
0.26 1.0
Total PARPb
0.31 1.0 0.18 1.0 0.31 1.0 0.12 1.0 0.18 1.0 0.19 1.0 0.31 1.0 0.31
Pexp, population prevalence of exposure to risk factor; PARP, population-attributable risk proportions; RR, relative risk. a Regular physical activity: ⱖ5 days a week, ⱖ150 minutes total, or ⱖ3 days a week of vigorous activity for ⱖ20 minutes each session b PARP ⫽ (Pexp*(RR ⫺ 1))/(1 ⫹ sum((Pexp (RR ⫺ 1)))).
those covered exclusively by Medicare. In addition, Minnesota is one of the lowest cost areas of the country for Medicare services.20 Yet the results of this study are on the same order of magnitude as other studies. Colditz7 attributed $128 per capita to physical inactivity in the United States in 1995. State estimates of the charges related to physical inactivity were $477 million in Georgia ($79 per capita)5 and $157 million in South Carolina ($78 per capita).15 A total of $1 billion in direct costs were related to physical inactivity in New York ($70 per capita), and $81.6 million inpatient hospital charges was attributed to physical inactivity in Washington ($19 per capita).21 Per capita costs for each study were estimated using the 2000 adult population from the Census.17 Other studies have addressed the short-term costs of physical inactivity by comparing the average healthcare expenditures of physically active individuals versus physically inactive individuals.13,14 The advantage of this type of analysis is that all healthcare expenditures are included, not just those related to the specific Table 3. Blue Cross Blue Shield of Minnesota costs attributable to physical inactivity ($ millions, 2000 dollars) Disease
Cost
Ischemic heart disease Hypertension Stroke Depression and anxiety Diabetes, type 2 Breast cancer Osteoporosis Colon cancer Total
$35.3 10.8 9.2 9.1 7.2 4.6 4.5 2.9 $83.6
diseases with a proven association to physical inactivity. Pratt et al.22 compared medical expenditures between regularly active (defined as spending ⱖ30 minutes in moderate or strenuous physical activity at least three times per week) and inactive participants of the 1987 National Medical Expenditures Survey, with the net annual benefit of physical activity at $330 per person in 1987 dollars. This is substantially higher than the $56 per member attributable to physical inactivity in the current study (in 2000 dollars). The difference is likely a result of the difference in methods: the current study included only medical costs associated with a subset of conditions, while the Pratt et al.22 report included all medical costs related to inactivity. In summary, this study’s estimates of the costs of physical inactivity are of a similar scale to those of the growing literature on this subject. Because the analysis included the entire healthcare dollar, not just inpatient costs, and used actual paid amounts rather than charges, it may provide a more accurate assessment of costs than previous studies. This study focused on physical activity as a risk factor for disease. The epidemiologic studies used to determine relative risks for active versus inactive individuals examined physical activity as an independent risk factor. However, because activity and obesity are strongly correlated, they may be confounded in this analysis. Since physical inactivity is a risk factor for obesity, and obesity is itself a risk factor for the diseases included in this study, the analysis may have considerably underestimated the true cost of physical inactivity. Since physical activity and healthcare cost data were not linked at the individual level, it is possible that the reported associations were due to unmeasured conAm J Prev Med 2004;27(4)
307
founding at the individual level and not due to physical inactivity. In addition, the study did not control for pre-existing conditions that might limit an individual’s ability to be physically active. However, individual-level analyses have found that inactive individuals incur more healthcare costs than active individuals.13,22 This cost-of-illness approach is an efficient, practical way to model the impact of a behavioral risk factor in a large population. Another problem inherent with this methodology is that current inactivity may not correlate well with prior levels of physical inactivity in individuals. Many of the costs associated with diseases related to physical inactivity are a result of past, rather than current, inactivity. However, BRFSS data show that the proportion of the population that is inactive in Minnesota has changed very little since 1986.23 Therefore, the estimates should be robust at the aggregate level. There were also several limitations of the practical implications of this study. First, even the most effective interventions to date have achieved only a modest reduction in physical inactivity.24 Therefore, even with the application of recommended strategies to increase physical activity,24 some proportion of the population will likely be inactive and irregularly active. Second, when a population becomes more active, some potential cost savings may take years to realize, especially those related to the prevention of chronic disease. However, recent evidence indicates that the effect of exercise on blood pressure and blood insulin and glucose25 are realized immediately after an exercise session, and effects on depression and anxiety are prompt as well.26 The results of this study estimating the costs of physical inactivity point to the need for cost– benefit analyses of physical activity interventions.27,28 These studies put the costs of physical inactivity in the context of the impact and costs of interventions to increase physical activity. Cost-benefit analyses will help stakeholders to make decisions about the most feasible approaches to change.
Conclusions This analysis puts a dollar figure on the direct cost of physical inactivity in a health plan population in 2000. While $83.6 million is most likely an underestimate of the medical costs of inactivity since it includes a narrow group of conditions and does not reflect the influence of physical inactivity on obesity, it puts the abstract concept of physical activity as a public health issue into a concrete form that is meaningful to a wide variety of stakeholders. The data used in this study were easily accessible to individual health plans, and all of the information needed for other plans to conduct similar analyses is presented here. Communities and coalitions of groups within those communities (such as health 308
What This Study Adds . . . This study adds to the literature on epidemiologic estimates of the costs of physical inactivity by using more precise data on healthcare costs. Specifically, the estimates were based on paid claim amounts rather than charges and pharmacy data were included. In addition, the problem was approached from the point of view of a health plan, providing a blueprint for estimating the direct cost of physical activity inexpensively by using administrative data and publicly available survey data.
plans, employers who purchase health insurance, providers, and local governments) may be more willing to invest time and money in physical activity promotion efforts when they learn that not making this investment is a costly decision. We are grateful to Sanne Magnan, Patricia Ball, and Dan Anderson of Blue Cross Blue Shield of Minnesota for their helpful contributions, and Karen Fitzner and Nadine Caputo of the Blue Cross Blue Shield Association for their support of this project.
References 1. Centers for Disease Control and Prevention. Prevalence of Physical Activity, Including Lifestyle Activities Among Adults—United States, 2000 –2001. MMWR Morb Mortal Wkly Rep 2003;52:764 – 69. 2. McGinnis JM, Foege WH. Actual causes of death in the United States. JAMA 1993;270:2207–12. 3. Mokdad AH, Marks JS, Stroup DF, Gerberding JL. Actual causes of death in the United States, 2000. JAMA 2004;291:1238 – 45. 4. Centers for Disease Control and Prevention. Physical activity and health: a report of the Surgeon General. Atlanta GA: U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, 1996. 5. Bricker SK, Powell KE, Parasha U, et al. Physical activity report. Atlanta GA: Georgia Department of Human Resources, Division of Public Health, American Heart Association, Southeast Affiliate, September 2001. 6. Hu FB, Stampfer MJ, Solomon C, et al. Physical activity and risk for cardiovascular events in diabetic women. Ann Intern Med 2001;134:96 –105. 7. Colditz GA. Economic costs of obesity and inactivity. Med Sci Sports Exerc 1999(suppl 11):S663–7. 8. Wannamethee SG, Shaper AG. Physical activity and the prevention of stroke. J Cardiovasc Risk 1999;6:213– 6. 9. Stephenson J, Bauman A, Armstrong T, Smith B, Bellew B. The costs of illness attributable to physical activity in Australia: a preliminary study. Commonwealth of Australia, Canberra: Population Health Division Publications, 2000. 10. Friedenreich CM. Physical activity and cancer prevention: from observational to intervention research. Cancer Epidemiol Biomarkers Prev 2001;10:287–301. 11. Gillespie L, Gillespie W, Robertson M, Lamb S, Cumming R. Interventions for preventing falls in elderly people. Cochrane Database Syst Rev 2001;3: CD000340. 12. Colditz G. Economic costs of obesity and inactivity. Med Sci Sports Exerc 2000(suppl 10):S663–7. 13. Pronk N, Goodman M, O’Connor P, Martinson B. Relationship between modifiable health risks and short-term health care charges. JAMA 1999;282:2235–9. 14. Goetzel RZ, Anderson DR, Whitmer RW, Ozminkowski RJ, Dunn RL, Wasserman J. The relationship between modifiable health risks and health
American Journal of Preventive Medicine, Volume 27, Number 4
15.
16.
17. 18.
19.
20. 21.
care expenditures. An analysis of the multi-employer HERO health risk and cost database. The Health Enhancement Research Organization (HERO) Research Committee. J Occup Environ Med 1998;40:843–54. Powell KE, Greany M, Huang Y, Whitt M. Physical activity in South Carolina. Columbia: University of South Carolina Prevention Research Center, School of Public Health, 1999. Garrett N, Alesci N, Schultz M, Foldes S, Magnan S, Manley M. The relationship of stage of change for smoking cessation to stage of change for fruit/vegetable consumption and physical activity in a health plan population. J Health Promotion 2004. In press. U.S. Census Bureau. 2000 Adult census. Available at: http://quickfacts.census. gov/gfd. Accessed October 15, 2003. Bernhardt TS, Foldes S, Heaton A, et al. Using diagnosis clusters: a case study of the specification of the cost of care for a health plan population and selected sub-populations. Paper presented at: Minnesota Health Services Research Conference, Minneapolis MN, February 23, 1999. Brownson RC, Remington PL, Davis JR. Chronic disease epidemiology and control, 2nd ed. Washington DC: American Public Health Association, 1998. Wennberg J, Cooper M. The Dartmouth atlas of health care 1999. Chicago IL: American Hospital Association Press, 1999. Claybrooke C, LeMier M, Bensley L,et al. Physical activity in Washington State. Olympia WA: Washington State Department of Health, 2000.
22. Pratt M, Macera CA, Blanton C. Levels of physical activity and inactivity in children and adults in the United States: current evidence and research issues. Med Sci Sports Exerc 1999(suppl 11):S526 –33. 23. Centers for Disease Control and Prevention. Physical activity statistics 1986 – 2000. Available at: www.cdc.gov/nccdphp/dnpa/physical/stats/stats.htm. Accessed June 14, 2004. 24. Kahn EB, Ramsey LT, Brownson RC, et al. The effectiveness of interventions to increase physical activity: a systematic review. Am J Prev Med 2002;22(suppl 4):73–107. 25. Kelley DE, Goodpaster BH. Effects of exercise on glucose homeostasis in type 2 diabetes mellitus. Med Sci Sports Exerc 2001;33(suppl 6):S495–501. 26. Dunn AL, Trivedi MH, O’Neal HA. Physical activity dose–response effects on outcomes of depression and anxiety. Med Sci Sports Exerc 2001; 33(suppl 6):S587–597. 27. Roux L, Hu D, Yore M, Yanagawa T, Pratt M. The health and economic impact of promoting physical activity: a decision analytic approach. Paper presented at annual meeting of the American College of Sports Medicine, Indianapolis IN, June 4, 2004. 28. Proper K, Bruyne M, Hildebrandt V, van der Beek A, Jan Merrding W, van Mechelen W. Cost-effectiveness of a worksite physical activity counseling program: results of a randomized controlled trial. Paper presented at annual meeting of the American College of Sports Medicine, Indianapolis IN, June 4, 2004.
Am J Prev Med 2004;27(4)
309