Inactivity-associated medical costs among US adults with arthritis

Authors

  • Guijing Wang,

    Corresponding author
    1. Division of Nutrition and Physical Activity, National Center for Chronic Disease Prevention and Health Promotion, Centers for Disease Control and Prevention, Atlanta, Georgia;
    • Centers for Disease Control and Prevention, 4770 Buford Highway, Atlanta, Georgia 30341-3717

    Search for more papers by this author
  • Charles G. Helmick,

    1. Division of Adult and Community Health, National Center for Chronic Disease Prevention and Health Promotion, Centers for Disease Control and Prevention, Atlanta, Georgia
    Search for more papers by this author
  • Carol Macera,

    1. Division of Nutrition and Physical Activity, National Center for Chronic Disease Prevention and Health Promotion, Centers for Disease Control and Prevention, Atlanta, Georgia;
    Search for more papers by this author
  • Ping Zhang,

    1. Division of Adult and Community Health, National Center for Chronic Disease Prevention and Health Promotion, Centers for Disease Control and Prevention, Atlanta, Georgia
    Search for more papers by this author
  • Mike Pratt

    1. Division of Nutrition and Physical Activity, National Center for Chronic Disease Prevention and Health Promotion, Centers for Disease Control and Prevention, Atlanta, Georgia;
    Search for more papers by this author

Abstract

Objective

To analyze direct medical costs among US adults with arthritis and estimate the proportion associated with inactivity.

Methods

In the 1987 National Medical Expenditure Survey, arthritis was defined using questions on self-reported, doctor-diagnosed arthritis or rheumatism. Physical activity was defined using a self-report question on level of activity. Inactivity-associated medical costs were derived by subtracting costs for active adults from costs for inactive adults after controlling for functional limitation.

Results

Among 5,486 adults with arthritis, inactive persons had higher medical costs than did active persons in all demographic groups examined. In multivariate models adjusting for key covariates, the proportion of costs associated with inactivity averaged 12.4% ($1,250 in 2000 dollars) and ranged from 7.8% to 14.3% among various demographic groups.

Conclusion

Inactivity-associated medical costs among persons with arthritis are considerable. Physical activity interventions may be a cost-effective strategy for reducing the burden of arthritis.

INTRODUCTION

Arthritis and other rheumatic conditions are a major health problem in the United States, affecting 15% of the population (43 million persons) in 1990 and a projected 18.2% (60 million persons) by 2020(1). Arthritis is a leading cause of chronic pain; is associated with a variety of health problems such as obesity, physical inactivity, and gastrointestinal bleeding related to use of nonsteroidal antiinflammatory drugs; is associated with poor health-related quality of life (2); and is the leading cause of disability in the US (1, 3). More than 60% of persons with arthritis have limited physical activity to some extent (4). Disability related to arthritis among both working-age (18–64 years) and older (≥65 years) adults is significant, and trend data show disability rates for persons with arthritis are increasing, even when adjusted for age (5).

The economic burden of arthritis is also enormous. Direct and indirect costs have been increasing in the US, from $21 billion (about 1% of the gross national product) in 1980 to $54.6 billion in 1988 and $65 billion in 1992 (6–8). About 23% of the costs were attributable to the direct costs of medical care; the remainder reflected indirect costs due to lost productivity (8).

Because of the large economic and health burdens of arthritis, interest in arthritis as a public health problem is growing. Recently, the Arthritis Foundation, the Association of State and Territorial Health Officials, the Centers for Disease Control and Prevention (CDC), and 90 other organizations developed the National Arthritis Action Plan: A Public Health Strategy, which calls for developing intervention strategies to prevent the occurrence and progression of arthritis (9). One recommended strategy is to decrease the high rate of physical inactivity among adults with arthritis, which is far higher than the rate among adults without arthritis (34.8% and 27.7%, respectively) (10). The physical inactivity associated with arthritis can have serious health consequences, such as loss of function and independence, increased risk for cardiovascular disease, diminished quality of life, and unnecessary disability (11, 12). The general health benefits of physical activity are well known; many recent studies have demonstrated the positive effects of physical activity on persons with arthritis as well (13–19). Experts now recognize that physical activity is an important element in the prevention and management of disability due to arthritis, and they recommend that physical activity be included in the overall treatment of patients with osteoarthritis, rheumatoid arthritis, and ankylosing spondylitis (13, 15, 20–24).

Although the health benefits of reducing physical inactivity among persons with arthritis seem clear, the associated economic benefits in this population have not been explored. To intelligently consider the benefits of various interventions (including physical activity) to reduce the impact of arthritis, policymakers and others who set funding priorities need information on the costs associated with physical inactivity among persons with arthritis. In this paper, we analyze direct medical costs among US adults with arthritis and estimate the proportion associated with physical inactivity.

MATERIALS AND METHODS

We used data from the 1987 National Medical Expenditure Survey (NMES) conducted by the Agency for Healthcare Research and Quality (formerly the Agency for Health Care Policy and Research), which is the most recent national survey on health care costs with physical activity information available. The survey was a nationally representative sample of the US civilian, noninstitutionalized population. A detailed description of the sample design is provided elsewhere (25). Information on medical costs was collected during an in-person interview, confirmed by medical providers and insurance claims, and summed for services occurring from January 1 through December 31, 1987. We took a societal perspective by including all payments from out of pocket, insurance policies, government programs, and other sources.

Arthritis status, functional limitation, and physical activity status were determined from a self-administered questionnaire in a supplement of the NMES. For this survey component, 30,008 persons completed the questionnaire. Participants were classified as having arthritis if they answered “yes” to either of these questions: 1) “Did a doctor ever tell you that you had arthritis?” or 2) “Did a doctor ever tell you that you had rheumatism?” We included rheumatism, because a large proportion (30–53%) of persons with arthritis and other rheumatic conditions do not know their type of arthritis and probably report it as rheumatism (26).

Severity of arthritis, a potential confounder related to both physical activity level and direct medical costs, could not be measured directly with the data available. Functional limitation, an indirect surrogate measure for severity of arthritis, was determined from activity limitation data. Participants were classified as having functional limitation if they answered “yes” to either of these questions on musculoskeletal activities: 1) “Does your health limit the kinds or amount of moderate activities you can do such as moving a table, carrying groceries, or bowling?” and 2) “Do you have trouble bending, lifting, or stooping because of your health?”

In the questionnaire, one question (“Which comes closest to describing your physical activity?”) pertained to work and leisure-time physical activity. Two answers were possible. Participants who chose “I often spend at least half an hour in moderate or strenuous physical activity (3 or more times a week)” were classified as physically active. Participants who chose “I am not very active when I'm at work or working around the house, and (except for ordinary activity of daily living) I don't spend too much time in physical activities” were classified as physically inactive.

We included only adults (persons aged 18 years or older) in our analysis, because the NMES data did not contain physical activity information for children. We derived inactivity-associated medical costs by subtracting medical costs for physically active adults with arthritis from medical costs for physically inactive adults with arthritis. Because 3% of the adults with arthritis had no medical costs, and because medical cost data were skewed, we added trivial amount ($1) dollar to each person's medical costs and took the natural log of the value to correct the skewness problem (27).

We first determined unadjusted inactivity-associated medical costs for the entire cohort, then stratified the unadjusted costs by functional limitation, age, sex, and race subgroups. Next, we used regression techniques to model the proportion of medical costs associated with physical inactivity, adjusting for other important factors that can affect direct medical costs. These other factors included functional limitation, age, sex, race, smoking status (never smoker, former smoker, current smoker), body mass index (underweight, <18.5 kg/m2; healthy weight, ≥18.5 to <25 kg/m2; overweight, ≥25 to <30 kg/m2; obese, ≥30 kg/m2), socioeconomic status (low, middle, high), and region of residence (Northeast, Midwest, South, West.) Body mass index was calculated from self-reported weight and height; the index categories were set according to the newly released clinical guidelines on identification of overweight and obesity (28). Socioeconomic status was used to assess a person's ability to seek health care services and was measured by federal poverty lines that used household income adjusted for household size. The 1987 poverty line ranged from $5,778 for a one-person family to $23,105 for a family of 9. Region of residence was included to adjust for recognized geographic cost differences.

In calculating the proportion of medical costs associated with physical inactivity (% MCinactive), we estimated medical costs from one regression model using our sample (MC1). We also estimated medical costs from a second regression model assuming that all the persons were physically active (MC2). With formula % MCinactive = (MC1 − MC2)/MC1 × 100% to derive the percentage of medical costs associated with physical activity. We used SUDAAN (Triangle Park, NC), a statistical software designed for incorporating complex sample design effect into standard error estimation, for the statistical analyses (29). Costs in 1987 dollars were adjusted to 2000 dollars using measured (30) and projected (31) trends in personal health care expenditures.

RESULTS

Our study cohort consisted of 5,486 adults, representing an estimated 33.4 million adults with arthritis in the civilian, noninstitutionalized US population in 1987. With weighted estimates, 78% met the case definition of arthritis by reporting doctor-diagnosed arthritis only, 3% by reporting doctor-diagnosed rheumatism only, and 19% by reporting both (Table 1). Most persons with arthritis were female (63%), white (86%), aged 45 years or older (82%), overweight or obese (57%), physically inactive (59%), former or current smokers (56%), and residents of the Midwest or South (63%). The cohort was evenly split by functional limitation (yes, no) and by socioeconomic status (low, middle, high).

Table 1. Selected characteristics of US adults with arthritis
CharacteristicSample sizeWeighted* proportion
n(%)
  • *

    Sample of 5,486 weighted to the 1987 US civilian, noninstitutionalized population.

  • Active, ≥30 minutes of moderate or strenuous physical activity on ≥3 days per week; inactive, less than this amount.

  • Never smoker = smoked ≤100 cigarettes in lifetime; former smoker = smoked >100 cigarettes in lifetime and did not smoke at time of questionnaire; current smoker = smoked >100 cigarettes in lifetime and smoked at time of questionnaire.

  • §

    Yes = limited by health in the kinds or amount of moderate activities (n = 2,245; 72.6% of all severe arthritis) or have trouble bending, lifting, or stooping (n = 2,661; 84.9% of all severe arthritis); No = no such activity limitation.

  • Socioeconomic status was defined using household income as related to the federal poverty line adjusted for household size. High, >400% of federal poverty line; middle, >200%–400% of federal poverty line; low, ≤200% of federal poverty line.

Type of arthritis
 Doctor-diagnosed arthritis only4,16078.1
 Doctor-diagnosed rheumatism only1803.3
 Both1,14618.7
Sex
 Male1,93237.4
 Female3,55462.7
Race
 White4,36986.4
 African American94010.2
 Other1773.4
Age
 18–44 years78918.3
 45–64 years1,88541.0
 ≥ 65 years2,81240.7
Body mass index
 Underweight (<18.5 kg/m2)1793.1
 Healthy weight (≥18.5 to <25 kg/m2)2,14140.1
 Overweight (≥25 to <30 kg/m2)2,01136.7
 Obese (≥30 kg/m2)1,15520.1
Physical activity status
 Active2,07441.2
 Inactive3,41258.9
Smoking status
 Never smoker2,55643.7
 Former smoker1,68531.7
 Current smoker1,24524.6
Functional limitation§
 Yes3,02451.1
 No2,46248.9
Region of residence
 Northeast1,01219.7
 Midwest1,38525.9
 South2,14536.9
 West94417.5
Socioeconomic status
 High1,74633.1
 Middle1,67732.7
 Low2,06334.1

Annual 1987 medical costs averaged $3,366 among adults with arthritis. Inactivity-associated medical costs were $1,714; $2,018 occurred among persons with functional limitation, and $186 occurred among persons with no functional limitation (Table 2). Among all age, sex, race, and sex-and-age groups examined, inactive persons with any arthritis incurred higher medical costs than active persons; these differences were statistically significant (P < 0.05) except for African Americans and persons aged 18 to 44 years. Inactivity-associated medical costs were higher among women than men, were highest among women aged 65 years and older and lowest among women between the ages of 18 and 44 years, and increased with age among both sexes.

Table 2. Inactivity-associated medical costs (in 1987 dollars) among US adults aged 18 and older with arthritis, by physical activity status*
 Physically active ($)Physically inactive ($)Inactivity-associated medical costs ($, inactive − active)P value
  • *

    Weighted estimates for the 1987 US civilian, noninstitutionalized population based on a sample of 5,486.

Total2,3574,0711,7140.00
Functional limitation
 Yes3,3645,3822,0180.00
 No1,7851,9711860.31
Sex
 Male2,6604,3251,6450.00
 Female2,1103,9511,8410.00
Race
 White2,3094,0621,7520.00
 African American3,2554,3261,0710.26
 Other1,4933,4661,9730.07
Age
 18–44 years1,7172,2965800.13
 45–64 years2,2363,3001,0630.01
 ≥65 years2,9915,2632,2720.00
Men
 18–44 years1,4161,9975810.42
 45–64 years2,8133,7849720.26
 ≥65 years3,3695,5092,1400.00
Women
 18–44 years1,9712,4314600.26
 45–64 years1,7713,0611,2900.00
 ≥65 years2,6875,1502,4620.00

Among adults with arthritis and functional limitation, except for men aged 45 to 64 years, inactive persons had higher medical costs than active persons (Table 3). Many of these differences were statistically significant (P < 0.05). The difference for men between the ages of 45 and 64 years was negative, but this difference was not statistically significant (P = 0.89). Inactivity-associated medical costs were higher among women than men, were highest among persons of “other” race and lowest among African Americans, and were higher among persons aged 65 years and older than among younger persons.

Table 3. Inactivity-associated medical costs (in 1987 dollars) among US adults aged 18 and older with arthritis and functional limitation, by physical activity status*
 Physically active ($)Physically inactive ($)Inactivity-associated medical costs ($, inactive − active)P value
  • *

    Weighted estimates for the 1987 US civilian, noninstitutionalized population based on a sample of 3,024.

Total3,3645,3822,0180.00
Sex
 Male4,0355,7531,7180.06
 Female2,8675,2032,3360.00
Race
 White3,3425,4062,0640.00
 African American4,2145,2139990.51
 Other8505,4984,6480.01
Age
 18–44 years2,2973,5911,2940.04
 45–64 years3,4884,4961,0080.21
 ≥65 years3,7506,2992,5490.00
Men
 18–44 years2,0764,2242,1480.20
 45–64 years5,1754,933−2420.89
 ≥65 years3,8736,6912,8180.01
Women
 18–44 years2,5183,3248050.34
 45–64 years2,3314,2431,9120.00
 ≥65 years3,6636,1292,4660.00

In the multivariate model created to adjust for other factors associated with medical costs, persons with physical inactivity, persons with functional limitation, older adults, women, whites, former smokers, those with higher socioeconomic status, and residents of the South independently incurred significantly higher medical costs than their counterparts (data not presented). On average, 12.4% of medical costs of inactive persons was associated with physical inactivity, corresponding to $504 per person in 1987 dollars ($1,250 in 2000 dollars) (Table 4). Among all age, sex, and race groups, inactive persons incurred higher medical costs; the percentage of medical costs associated with physical inactivity ranged from 7.8% ($156 in 1987 dollars, $387 in 2000 dollars) among men aged 18 to 44 years to 14.3% ($617 in 1987 dollars, $1,530 in 2000 dollars) among African Americans. Absolute dollar values in 1987 ranged from $156 ($387 in 2000 dollars) among men aged 18 to 44 years to $725 ($1,798 in 2000 dollars) among women aged 65 years and older.

Table 4. Proportion and estimate of medical costs associated with physical inactivity among inactive adults with arthritis*
 Proportion of adjusted medical costs associated with inactivity (%)Mean adjusted medical costs associated with physical inactivity
1987 dollars2000 dollars
  • *

    Weighted estimates for the 1987 US civilian, noninstitutionalized population (based on a sample of 5,486) used in a regression model adjusting for functional limitation, age, sex, race, smoking status, body mass index, socioeconomic status, and region of residence.

Total12.45041,250
Sex
 Male10.94721,171
 Female13.05141,275
Race
 White12.24941,225
 African American14.36171,530
 Other13.44631,148
Age
 18–44 years9.9228565
 45–64 years11.2369915
 ≥65 years13.57091,758
Men
 18–44 years7.8156387
 45–64 years10.0377935
 ≥65 years12.16671,647
Women
 18–44 years10.9266660
 45–64 years11.7359890
 ≥65 years14.17251,798

DISCUSSION

Physical inactivity was associated with considerable medical costs among adults with arthritis; this finding held true for most age, sex, and race groups. Not surprisingly, inactivity-associated costs unadjusted for other factors occurred primarily among persons with functional limitation. However, physical inactivity was associated with greater medical costs even independent of functional limitation and other important factors in the adjusted model.

Our findings imply that along with the recognized health benefits, physical activity among adults with arthritis will provide economic benefits as well. Although not all inactivity-associated medical costs may be amenable to physical activity interventions, the large size of inactivity-associated medical costs found in this study, even after adjustment for other important cost-related factors, suggests that many of the costs may be reduced, and that physical activity intervention may be a cost-effective strategy for reducing the burden of arthritis in the population. Women and individuals aged 65 years and older, whose inactivity-associated costs were particularly large, may represent the most cost-effective targets of physical activity interventions.

Several limitations of our study should be noted. First, only direct medical costs were included in the analysis. Arthritis is also a leading cause of work-related disability (1, 32, 33), and it directly affects a person's quality of life (2). Indirect costs of arthritis (including productivity loss and loss of general well-being) have been estimated to be 3 to 4 times higher than the direct costs. If we had included indirect costs, the inactivity-associated costs would likely be even greater. Second, our data were cross-sectional and not longitudinal, so we could not attribute the excess costs directly to physical inactivity. Although these excess costs might be due to other factors, such as treatment for inactivity-related severe pain and disability, we attempted to adjust for some of these factors in the regression model. Further, the large size of these excess costs suggests that at least a portion are attributable to inactivity itself. Third, we used definitions that were cruder than we would prefer, because ours was a secondary analysis of preexisting data. Specifically, our definition of arthritis relied on a doctor's diagnosis of arthritis or rheumatism. Although this is a reasonable compromise for a retrospective analysis, it misses the 16% of persons who do not see a doctor for their arthritis (34) and may miss persons with some of the numerous (≥100) arthritic and other rheumatic conditions who did not consider their condition to be arthritis or rheumatism. Fourth, the type, duration, and severity of arthritis are potential confounders, because they can affect both physical activity and costs; they could not be fully addressed, however, because there was no specific arthritis diagnosis and no direct measure (such as pain or clinical assessment) available in the data set. As a surrogate measure for severity we used functional limitation, as defined by available questions on functional measures. These questions did not directly attribute the limitation to arthritis, although they focused on nominal musculoskeletal activities. A specific arthritis diagnosis and a more complete measure of arthritis severity would help better define the relationship of physical inactivity and costs. Last, persons who were classified as physically inactive by our definition included those with some minimal level of activity; this misclassification would tend to minimize the differences observed.

Several strengths of our study should also be noted. First, we used a large national representative sample to derive the economic impact of physical activity on medical costs among persons with arthritis. This large sample allowed us to analyze selected demographic groups and to adjust for potential confounders in multivariate modeling. Second, medical costs were confirmed through providers and insurance claims information and were not based solely on self-report. Third, although numerous studies have pointed out the enormous economic costs of arthritis, and others have demonstrated the health benefits of physical activity among adults with arthritis, no study other than ours has linked the 2 quantitatively.

Promoting physical activity among persons with arthritis presents a public health challenge (23). Traditionally, persons with arthritis have been cautioned by doctors to rest and avoid vigorous activity (21, 22). Because doctors and other health care providers play a critical role in supplying information, resources, and continuing reinforcement for adoption and maintenance of appropriate physical activity behaviors, they need to be aware of the benefits of physical activity among persons with arthritis and the interventions available to promote physical activity among this population group (23, 35). Various physical activity interventions exist that accommodate the special needs of persons with arthritis, such as walking, PACE (People with Arthritis Can Exercise), and YMCA Aquatics (20, 36), but few of these have been evaluated for their cost-effectiveness. Cost-effectiveness evaluations of these and future physical activity interventions can quantify their benefit and allow consideration of their reimbursement by insurers and others. Existing public health efforts that address physical activity per se, such as the 1996 Surgeon General's Report on Physical Activity and Health, Healthy People 2010 objectives, and CDC's It's Everywhere You Go physical activity program, may not address the special needs of persons with arthritis. Because persons with arthritis represent more than 24% of the physically inactive adult population (10), achieving these national goals may require efforts to target this large group.

Future research on physical inactivity-associated medical costs among persons with arthritis should focus on elaborating the role of arthritis severity as a confounder and measuring the proportion of these costs that are amenable to physical activity interventions. This research would provide a more accurate assessment of the economic benefits of such intervention. Research should also focus on such costs among children. Establishing good exercise habits among children with arthritis may provide long-term health benefits and, hence, even more substantial cost savings than those achieved with physical activity interventions among adults.

Ancillary