Correlates of physical activity: Who's active and who's not?



A large body of research has established that regular physical activity reduces the risk of premature death and disability from a variety of disease and conditions, including heart disease, diabetes, colon cancer, osteoarthritis (OA), and osteoporosis (1). Despite the proven health benefits of physical activity, more than 70% of US adults either are sedentary or only irregularly active (2). Chronic conditions, such as arthritis, exacerbate sedentary behavior. Persons reporting symptoms of arthritis have higher rates of sedentary behavior than those not reporting symptoms (43% versus 38%) and lower rates of both moderate and vigorous physical activity (3).

The most frequent cause for disability is arthritis. Arthritis is one of the most prevalent diseases in the US, with an estimated 7 million persons reporting it as a major cause of activity limitation (4). Arthritis affects 50% of people aged 65 years and older, however, most people with arthritis are younger than 65 years and are of working age (5, 6).

In the past, increased physical activity was avoided as a mode of treatment for arthritis because it was thought that use of affected joints would worsen the problem (7). However, recent studies demonstrate the safety and efficacy of physical activity for people with arthritis and chronic musculoskeletal complaints (7–9).

In addition, physical activity may play an important role, although indirect, in the prevention of certain types of arthritis. For example, obesity has been convincingly demonstrated to have a causal role in OA of the knee. In cross-sectional studies, obesity is a strong risk factor for both unilateral and bilateral knee OA in both sexes (10, 11); and in longitudinal studies, obesity predicted the development of knee OA in both sexes (12). In another study, weight loss decreased the risk of developing symptomatic knee OA among women (13). Because increasing physical activity can aid in weight loss, it has the potential to reduce the risk for OA.

To encourage people to become more active, public health recommendations added a “lifestyle” approach to the traditional planned “exercise” recommendation for increasing physical activity (14). The lifestyle approach encourages common activities, such as brisk walking, climbing stairs, housework, yard work, and recreational physical activity. To achieve recommended levels of physical activity, a person must accumulate 30 minutes in at least 10-minute bouts of moderate-intensity physical activity at least 5 times per week or vigorous-intensity physical activity at least 3 times per week for ≥20 minutes at a time (2, 14). This level of activity is believed to be realistic and achievable for the vast majority of adults, including many adults who have moderate limitations due to disability.

The purpose of this article is to summarize the current understanding about correlates of physical activity in the adult population and to promote more research in this area with people with arthritis. Examining the characteristics of those less likely to be physically active has important implications for developing and understanding the initiation and adoption of this behavior. Even though people with arthritis may have added barriers to becoming physically active, understanding other factors that correlate to physical activity behavior in the general population can provide a good basis for more descriptive correlate research among populations with arthritis.


To understand the ways of promoting physical activity, researchers have studied correlates of physical activity that typically are documented via observational studies. There are factors or attributes that show repeated correlation with behavior. Correlates can be organized into 2 broad categories: individual and environmental. Individual characteristics include sociodemographics, biologic or health factors, and psychological factors. Environmental factors include social characteristics, environmental and policy factors, and physical environmental factors. There have been many published reviews on the correlates of physical activity for the general population over the past 2 decades (15). These reviews mainly identify correlates of leisure-time physical activity (LTPA) in general rather than specific components of fitness (e.g., strength training). Although there are many published studies on physical activity and arthritis, only 1 study was found that specifically looked at correlates of physical activity among persons with arthritis (16).


Individual factors.


The prevalence of physical activity for adults in the US varies by sociodemographic characteristics. Women consistently have lower rates of both moderate and vigorous LTPA than men (17, 18). The 1994 Behavioral Risk Factor Surveillance System survey indicates that 32.6% of women reported no LTPA in the previous month compared with 28% of men (19). Results from the 1990 National Health Interview Survey (NHIS) are similar and show a greater number of women who are sedentary (18). Participation in LTPA also declines with age (15, 20). For example, in the 1990 NHIS, 19.3% of adults aged 18–24 years were sedentary versus 33.9% of those aged 65 and older (18). Additionally, this decline in activity with age seems to be greater for women than for men (21).

The majority of evidence from past reviews identifies that level of education is positively associated with physical activity (15). This is true across sex and racial/ethnic groups. However, when types of activity are more specifically identified in research, results vary. Sternfeld et al (22) found that for women, education was positively associated with sports/exercise and LTPA, but was negatively associated with household or caregiving physical activities.

When race/ethnicity are considered, nonwhites are more likely to be sedentary (15). African Americans and other minority populations are less active than white Americans (14), and this disparity is more pronounced for women (23). The difference in activity level by race may be confounded by other factors. Some believe that the differences in education and socioeconomic status account for most, if not all, of the differences in LTPA (24). Others still show a significant difference in activity level by race/ethnicity after adjusting for education and income (20, 25).

The relationships of income and occupation to physical activity are not clear. Some studies show a modest positive relationship between income and LTPA (20, 26), but income was unrelated to physical activity in several other studies (27–29). The association between occupation and physical activity also remains unclear (26, 30).

Marital status may also be linked to physical activity level. One review summarized that unmarried people are the most active, and married women are the least physically active (1), however, other studies identify marital status as having no association to physical activity level (29). The relationship between marital status and physical activity level differs when various types of physical activity are reported. Some studies suggest that being married is negatively related to vigorous physical activity but is positively related to household or caregiving physical activity (22, 25).

Biologic and health factors.

Perceived health is positively associated with physical activity level. People who perceive themselves to be healthy are more active than people who report poor health. This association remains true across sex and race/ethnicity (15, 26, 31). Actual health status or illness is also positively related to physical activity level. It has been consistently demonstrated that healthy people are more active than persons with medical problems (1, 26, 32). Musculoskeletal problems, symptoms of arthritis (e.g., joint pain) in particular, are also related to less physical activity (3).

In addition to medical conditions, weight status is often predictive of physical activity level. Research shows that overweight individuals are less active than those who are not overweight (25, 26, 33). The question remains as to whether the greater weight is the cause of inactivity or if the inactivity causes the increased weight.

Health behaviors.

Despite a seemingly obvious link between smoking and inactivity, the negative relationship between cigarette smoking status and physical activity level is relatively modest (26). However, research does show that smokers have an increased risk of dropping out of supervised physical activity programs (15).

Attempting weight loss has been positively correlated with physical activity level (33). One of the most common reasons given for exercising is weight control, and dieting to control weight is positively associated with frequency of participation in both high-intensity and moderate-intensity physical activity (34).

Other health behaviors, such as alcohol use and preventive health screenings, have not been studied enough to produce dependable associations with physical activity and warrant further research (15, 31).

Psychological factors.

Self efficacy is defined as one's confidence of being able to successfully perform a specific activity or behavior (35). Among the psychological correlates of physical activity that have been examined, self efficacy is the strongest and most consistent predictor of physical activity behavior (33). Self efficacy ratings are related not only to the behavior itself, but also to maintenance of physical activity and adherence to exercise prescription in cardiac patients (26, 36). Because self efficacy level can be enhanced, interventions aimed at increasing a person's confidence to be physically active may lead to increased participation (37).

As stated in the Health Belief Model, beliefs about physical activity, perceived benefits from being physically active, and barriers to physical activity play an important role in the behavior (38). People with greater belief in the health benefits of physical activity are more likely to adopt or currently participate in a physically active lifestyle (15, 37). Perceived benefits of being physically active differ by sex. Women are more likely to report social factors and release of tension as the beneficial factors, whereas men tend to describe benefits of activity in terms of fitness and health (33). Influencing the perceived benefits shows potential as an effective aspect of intervention. However, just giving people information about the benefits of physical activity may not be enough. There has been repeated documentation of lack of association with knowledge about physical activity and physical activity level itself (15). Influencing knowledge about physical activity may not be desirable as the sole choice for intervention, but may be effective as part of a comprehensive program.

Self motivation is another psychological factor shown in many studies to be positively associated with physical activity level (15, 26, 36). Even though self motivation and other external motivating factors have been correlated with physical activity, relatively little information is available on how motivation affects initiation, maintenance, or discontinuation of physical activity behavior (33).

Barriers to physical activity often represent attitudes and beliefs about physical activity. The most commonly reported barrier is lack of time (15, 31, 39). For many people, this may represent a lack of interest in physical activity, because population surveys indicate that regularly active people are as likely as the sedentary to view time as a barrier (26, 39). Teaching people to change their perception of this time barrier seems to be an important aspect of future interventions. In fact, time management was a suggested intervention in a recent study of physical activity barriers and women (40).

Another psychological attribute found to be correlated with level of physical activity is enjoyment. Perceived exercise enjoyment is shown to predict higher levels of physical activity (41, 42). Additionally, the belief that physical activity is a physical stressor that could result in injury may have negative implications (15, 33).

High levels of stress may also be associated with lower levels of physical activity (33). In one study, level of perceived stress and number of stressful events were both associated with less physical activity (43). Similarly, cross-sectional research indicates that those who engage in higher levels of physical activity report lower levels of perceived stress (44).

Past experience with physical activity warrants mention. Research shows that physical activity history during youth or college years is inconsistently predictive of adult exercise (15). However, recent past participation in physical activity seems to be a stronger predictor of current physical activity (37).

Environmental correlates.

The environment plays an important role in behavior, as stated in several health behavior theories. For example, Bandura's social learning theory emphasizes the environment, social interaction, and the dynamic interaction of person, behavior and environment (45). McLeroy and colleagues propose an ecological model that encompasses several levels of influence on behavior. According to their theory, in addition to personal factors, the social environment, institutional and community factors, and public policy all play integral roles in behavior (46).

Social environment.

Social support is one of the most commonly studied correlates of physical activity (31). Individuals who are regular exercisers report more support for activity from people in their home and work environments (33). Sources of support that have been linked to greater participation in physical activity include support from one's spouse and other family members, friends, coworkers, program staff, and other program participants (15, 37). Support from one's health care provider also shows potential as a source of motivation to exercise (47). Support from a health care provider may be especially important to those under a physician's care for arthritis (47). Although the importance of social support has been well documented, the amount and type necessary to begin or maintain physical activity behavior is difficult to identify. It should also be mentioned that the relationship between physical activity and social support is a dynamic process in which the sources of support or need for support may change over time and through the phases of adoption and maintenance of this health behavior (48).

Environment and policy.

Environmental and policy strategies are aimed at changing the physical and sociopolitical environments (49). Environmental and policy approaches may be more effective than individual behavior and lifestyle modification strategies at reaching target populations because they can benefit all people exposed to the environment rather than focusing on changing the behavior of one person at a time (49–51).

In 1 report of attitudes toward policy measures to support physical activity, the majority of respondents in the study favored zoning regulations for walking trails and bike paths, and they agreed that local government funds should be spent to build and maintain places where people can exercise (52). Lack of access to places to exercise has been reported as a barrier to physical activity (31, 53). Policies to build or improve access to existing facilities may influence physical activity at the community level. The extent to which physical environments are conducive to exercise (i.e., walking/biking paths, safe streets) likely has a strong impact on the physical activity level of residents (33, 53). Variations in environmental and policy correlates may help to explain part of the well-documented variations in physical activity behavior across socioeconomic strata (53).

Physical environment.

The physical environment plays an important role in health behaviors, particularly physical activity. However, associations between physical environmental factors and physical activity level have received relatively little empirical study. Climate or bad weather, lessened daylight hours, lack of personal safety or neighborhood crime, and lack of transportation are environmental factors commonly viewed as barriers to physical activity (31, 54). Among physical environmental factors, one predictor of physical activity is the distance individuals are required to travel to reach exercise facilities (55, 56).


For the general population and those with arthritis, the decision to be active or sedentary ultimately resides with the individual (39). However, this decision is influenced by a variety of factors, individual and environmental. For example, a person with arthritis may live in a community conducive to physical activity, but may lack the self confidence to initiate a program. Also, regular participation in physical activity or exercise must be viewed as a dynamic process in which correlates and barriers may change over time. Maturation, life events, social interactions, and environmental climate may interact and alter adoption or maintenance of this health behavior.

Sociodemographic research reports that women, minority populations, the less educated, and the elderly stand out as the most consistently reported inactive groups in terms of overall physical activity (36). Knowledge of who is less likely to be active can help tailor interventions to these specific groups.

Important modifiable correlates for adults include self efficacy, the perceived benefits of physical activity, enjoyment, and social support (15). The physical environment and policies, also modifiable, seem to play an important role in physical activity. Improved access, safety, and incorporating cues hold potential for increasing physical activity levels.

The challenge to practitioners is to first alter any perceived barriers of physical activity due to arthritis, then focus on the modifiable correlates found to be associated with physical activity for the general population. Personal factors such as self efficacy, social support, and perceived health benefits are all mutable and show potential for increasing the chance of individual behavior change in people with arthritis. In addition to focusing on individual correlates, environmental and policy strategies, such as increasing access, improving safety, and promoting exercise-friendly communities, may motivate people with arthritis to begin and maintain physical activity behavior.

Even though there have been decades of physical activity correlates research, there is still much to be revealed about the nature and extent of physical activity behavior. First, there needs to be more published research on correlates of physical activity among groups with varied health statuses and health conditions such as arthritis. Because becoming more physically active can improve the outcome of many diseases and conditions, it is important to identify determining factors within these populations and use the information to promote appropriate physical activity. Furthermore, future research should include refined assessments of physical activity and exercise. These assessments should be more focused on lifestyle versus traditional exercise and be tailored to and tested on such lesser-studied groups as women, older adults, minorities, and people with specific diseases and conditions. For example, preliminary research demonstrates that for women, correlates for sports and exercise vary from those for housework and caregiving physical activities (22). The correlates also vary somewhat by race/ethnicity (31). These situations need to be investigated further and the information used in intervention planning for these specific groups.

Additionally, more research is needed to better quantify certain correlates. For example, how much social support is needed to initiate physical activity behavior or at what level of self efficacy can one maintain a physical activity program? Similarly, it would be helpful to identify the synergistic effects of physical activity correlates. Perhaps a supportive social network coupled with a positive physical environment are 2 factors, that when combined, strongly correlate with physical activity level. Knowledge of this can help make the content of interventions more efficient.

Also, environmental and policy strategies hold great potential to affect the physical activity behavior of large groups of people, but there are sparse data on the patterns and effects of these approaches on a population-wide basis (1, 50, 53). Future agendas should include this important aspect of research.

Because of its dynamic nature, increasing physical activity level in all populations is challenging. Despite national goals, modifications in recommendations, and increased number of interventions, there has been little change in the prevalence of physical activity in the past decade (1). Research on correlates of physical activity can play an important part in increasing the number of people who are physically active, especially those with arthritis.