Approximately 200,000 to 300,000 premature deaths occur each year in the United States because of physical inactivity (1-4). Despite the benefits of regular activity, only 31% of adults in the United States report engaging in recommended amounts of physical activity (i.e., 30 minutes of moderate-intensity activity 5 or more days per week or 20 minutes of vigorous-intensity activity 3 or more days per week), and 38% report no leisure-time regular physical activity (5). Activity levels are even lower among people with disability; for example, Healthy People 2010 reports that 56% of people with disability reported no leisure-time physical activity, compared with 36% of people without disability, and rates of participation in regular moderate and vigorous physical activity are also lower for people with disability (6). This is particularly important because physical activity is similarly beneficial for people with or without disability and has been shown to improve quality of life and reduce functional impairment among people with disability (7-14). Accordingly, the goal of increasing physical activity is one of 10 leading indicator areas within Healthy People 2010, and people with disability are specifically included within the target population (15).

Correlates of physical activity among adults without disability are well-known and consistent across many studies; they include sex, age, income, race, education, obesity, and general health status. There has been less research on the correlates of physical activity for adults with disability. Disability has often been included as one item on a list of barriers to physical activity, often addressed within an item asking about injury or disability or as a barrier or correlate, such as arthritis, obesity, and asthma (16-19). There have also been studies on activity levels of people with medical conditions that can be disabling, such as arthritis (20).

Recently, a few researchers have begun to study the correlates of activity level for people with disability. For instance, Simonsick et al examined walking activity in a group of elderly women with moderate to severe disability and found that even when degree of disability was considered, race, psychosocial factors, and specific impairments remained significant predictors of activity level (21). Kinne et al found that barriers, motivation, and self-efficacy were predictors of exercise maintenance in a group of people with mobility impairments, but demographic factors were not (22). Shifflett et al found that perceived benefits, facilities barriers, and health barriers were important predictors of activity level in people with disability (23). Rimmer et al identified several barriers to physical activity for people with disability, including cost, lack of transportation, and inaccessibility (24-26). Two instruments for measuring activity level for people with disability have also been reported (27,28), indicating interest in evaluating physical activity in this population. However, no studies have examined large samples of people with broadly defined classes of disability for the purpose of finding correlates of physical activity that may be used to construct large-scale interventions for people with disability.

Table 1 presents results of the four BRFSS disability classification questions for the entire study population (n = 46,372) as well as results of the questions on major impairment or health problem and length of disability for people who answered yes to either question 1 or question 2 (n = 8348). Responses are provided by sex and two age categories (younger than 65 or 65 and older). Results are also presented for combinations of BRFSS disability classification questions. Chi-square tests were used to test differences in responses between men and women and people younger and older than age 65 on the disability questions. Adults aged 65 or older were significantly more likely to answer yes to three of the four disability classification questions (not to "need help with personal care") and to multiple questions (indicating a higher level of disability). Women were significantly more likely than men to answer affirmatively to the disability questions. Major impairment or health problem differed by age: adults 65 and older were most likely to report arthritis (26.5%) followed by a walking problem (11.5%), whereas adults under 65 were most likely to report back or neck problems (18.5%), followed by arthritis/rheumatism (13.5%). Women were most likely to report arthritis/rheumatism as their chief problem, followed by back or neck problems, whereas men were most likely to report back or neck problems (17.2%), followed by fractures or bone or joint injury (12.4%). Men were significantly more likely than women to have had disability for 10 years or longer, as were people 65 years and older compared with people younger than 65 years.

Table 2 presents descriptive characteristics of the study population with disability and without disability; chi-square tests were used to test the differences between the samples with and without disability. All group differences except race were significant at the .05 level. About half as many adults with disability met the moderate activity standard (25.4%) as adults without disability (43.3%). Adults with disability had lower incomes and less education and were older, more likely to be female, less likely to be married, more likely to live alone, and less likely to be employed than adults without disability. They were also in worse health; more likely to have diabetes, arthritis, or asthma; and more likely to be obese.

Unadjusted and adjusted ORs for each correlate are presented in Table 3. The unadjusted ORs show relationships similar to those found in studies of the general population. For adults with disability, male sex, increasing age, higher educational level and income, and good health were all associated with increased likelihood of meeting the moderate activity standard. African American race was negatively associated with activity level, as was obesity and having diabetes or arthritis. Alcohol consumption was positively related to activity level, but smoking did not have a significant relationship with activity. Being unable to work had a strong negative relationship with activity level, as did not currently working.

After adjustment for other correlates, only a few variables remain significant predictors of activity level. Increasing age is negatively related to activity level. African American race is negatively associated with sufficient activity level. Only the two highest income categories ($50,000–$74,999 and ≥$75,000) are significant predictors of activity compared to the lowest category (<$10,000). The only employment category that was a significant predictor was being unable to work compared with working. The three lowest categories of general health (good, fair, and poor) are significant predictors of insufficient activity level compared with the highest category (excellent). Of the three chronic medical conditions included, only asthma is a significant predictor, and paradoxically it is positively associated with sufficient activity.

Adults with disability in the United States are not achieving activity levels recommended by CDC and ACSM, and in fact only about half as many meet the recommendations for moderate activity compared with adults without disability. This suggests that a special effort should be made to encourage physical activity among people with disability. Any effort to develop interventions for people with disability must be based on knowledge of correlates of physical activity for that population. This study is the first to investigate correlates of physical activity in a large sample of people from numerous geographical regions using a broad definition of disability.

Some of the correlates of physical activity for the population with disability found in this study are similar to correlates found in studies of the general population. Among these are African American race (negatively associated with physical activity), higher income (positively associated with activity), older age (negatively associated with activity), and poor general health (negatively associated with activity). Other correlates that we expected to be important (because they have been found to be related to activity levels in studies of the general population) were not significant predictors after adjustment for other covariates. These correlates include years of activity limitation, body mass index, education level, having diabetes or arthritis, and smoking behavior. Surprisingly, years of activity limitation was not related to physical activity level after controlling for other covariates.

The definition of disability used in this study is based on functional status rather than diagnosis of disease or medical condition. There are many ways to define and measure disability, and no definition is perfect; however, we believe that the definition we selected is appropriate for large-scale survey instruments that are administered to the general population and that must obtain disability information by using only a few questions. The combination of questions used in this study yields a broad classification, and people classified as having disability by these questions will certainly be heterogeneous on medical condition and personal limitations. However, national efforts to increase physical activity among people with disability cannot be designed to target separately each and every type and degree of disability but must use broad categories and common correlates in planning interventions. Analyses of large-sample surveys such as the BRFSS are an important part of identifying these correlates.

This study has several limitations. One is that some people with disability are excluded by design because the BRFSS only surveys the noninstitutionalized population and requires that individuals have a telephone and be willing and able to answer the survey questions. A second limitation is the broad classification of disability and the subjective questions used to make the classification: two people could have similar levels of impairment or disability by medical or legal definition and yet answer the classification questions differently. However, the current definition of disability is applied as a social and demographic descriptor and not a medical or legal definition, so these self-definitions are appropriate for this purpose. In addition, the questions are the product of extensive national discussions and constituent feedback, and the first two questions (the first on limitations and the second on the requirement of special equipment) are also used in the National Health and Nutrition Examination Survey and the National Health Interview Survey (39). A third limitation is that the 2001 BRFSS did not include questions about some topics that have been shown to be strongly related to exercise in people with disability; these include barriers such as cost (24) and inaccessible built environments (25,26) as well as social support and psychological factors such as perceived competence and perceived benefits of exercise and exercise self-efficacy (22,23). A fourth limitation is that the data were drawn from only the District of Columbia and the 12 states that administered the Quality of Life and Caregiving Module of the BRFSS in 2001. However, we have no reason to suspect that these relationships would vary if all state BRFSS respondents were represented.