| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
RESEARCH ARTICLE |
1 Department of Exercise Science, University of South Carolina, Columbia.
2 Department of Family & Preventive Medicine, University of South Carolina School of Medicine, Columbia.
Address correspondence to Sara Wilcox, PhD, Department of Exercise Science, Arnold School of Public Health, University of South Carolina, 1300 Wheat Street, Columbia, SC 29208. E-mail: swilcox{at}sc.edu
 |
Abstract |
|---|
 TOP Abstract MethodsResultsDiscussionReferences |
|---|
THE physical and mental health benefits of physical activity (PA) are well established and include a reduction in the risk of coronary heart disease, diabetes mellitus, certain types of cancer, obesity, hypertension, and all-cause mortality, as well as improved mood, decreased symptoms of depression and anxiety, and improved health-related quality of life (U.S. Department of Health and Human Services, 1996
). PA has an important role in helping older adults preserve independence, control weight, and maintain muscle, joint, and bone health (American College of Sports Medicine, 1998).
Older women are the least active segment of the population, with 51% of women aged 65–74 years and 66% of women aged 75 years and over reporting no leisure-time PA that lasts 10 min or longer (Schoenborn & Barnes, 2002). Women living in rural regions have lower rates of PA participation than women living in urban areas, and African American women in both of these regions are the least active (Schoenborn & Barnes, 2002; Wilcox, Castro, King, Housemann, & Brownson, 2000).
Studying the factors that influence PA participation can be useful for identifying segments of the population with low rates of PA participation that could benefit from interventions; it can also be useful for identifying targets for these interventions (Sallis & Owen, 1999, chap. 7). Social cognitive theory (Bandura, 1986), which emphasizes the interactions among individual or personal factors, behavior, and the environment (social and physical), has frequently been applied to PA determinants and intervention research (Baranowski, Perry, & Parcel, 1997). Although the theory's breadth and consideration of multiple levels of influence are useful for understanding PA, most research has focused on individual influences within this model (e.g., self-efficacy or outcome expectations), and less research has examined environmental influences.
Relative to other population segments, little is known about the factors that facilitate or hinder PA participation in older African American and White women who live in rural settings. Rural women often experience greater barriers to health behaviors than urban women, including higher rates of poverty and chronic disease and illness, lower levels of education, and greater distance to travel for health care and other services (Mulder et al., 2000). Furthermore, African American women who live in rural settings have the highest rates of poverty (Mulder et al., 2000). Rurality, African American race, female gender, and older age likely interact in their relationship to PA, underscoring the importance of studying PA correlates in older and racially diverse samples of rural women.
Being of an older age and having a greater number of family members have been negatively associated with PA, and having higher education and higher income, having had a physician discuss PA, having greater exercise knowledge, and attempting to lose weight have been positively associated with PA in rural African American women (Duelberg, 1992; Frank, Stephens, & Lee, 1998; Macera, Croft, Brown, Ferguson, & Lane, 1995; Mobily, Lemke, Drube, Wallace, & Leslie, 1987; Washburn, Kline, Lackland, & Wheeler, 1992). In a study of diverse and rural older women, African American and American Indian race, older age, and greater personal barriers were negatively associated with PA, and higher education, greater enjoyable scenery, frequency of seeing others exercise, and greater social support were positively associated with PA (Wilcox et al., 2000). Greater self-efficacy (i.e., confidence in one's ability to engage in PA, even when faced with obstacles) and social support, positive attitudes toward PA, higher education, higher income, and nonsmoking status have been reported as positive correlates of PA in White rural women (Duelberg, 1992; Eaton, Nafziger, Strogatz, & Pearson, 1994; Horne, 1994; Mobily et al., 1987). In summary, most studies that have examined factors that influence PA in rural African American and White women have focused on sociodemographic variables. Relatively few studies have examined psychological, social, and environmental influences that are theoretically believed to be important and that have been shown to be important in studies with other population segments (Trost, Owen, Bauman, Sallis, & Brown, 2002).
The present study was designed to improve our understanding of the factors that influence PA in older African American and White women in rural settings. These factors were conceptualized within a social cognitive theory framework and included individual (sociodemographic and psychological) and perceived environmental (social and physical) influences.
Our choice of measures was guided by empirical evidence and social cognitive theory. For example, age, education, and race are well-established correlates of PA (King, 2001; Schoenborn & Barnes, 2002). There is also some evidence that being married is associated with more favorable health practices (Umberson, 1987, 1992). Greater social support, higher self-efficacy, and greater perceived benefits or pros of PA and fewer perceived barriers or cons of PA are perhaps the most commonly reported psychological correlates of PA among diverse groups (Eyler et al., 2002; King, 2001; Wilcox, Tudor-Locke, & Ainsworth, 2002). Social support from a health care provider may also be important for older adults (Cousins, 1995). Furthermore, there is consistent evidence that greater symptoms of depression, and possibility anxiety and stress, are associated with lower levels of PA (Dunn, Trivedi, & O'Neal, 2001). For example, in the Alameda County study, declines in PA over time were independently associated with higher baseline depression (Kaplan, Lazarus, Cohen, & Leu, 1991). Finally, although relatively few studies have examined physical environmental correlates of PA (e.g., perceived safety or presence of sidewalks), there has been great interest and encouragement to examine these variables, as they are consistent with social cognitive theory and social ecological models of behavior (Humpel, Owen, & Leslie, 2002; Sallis, Bauman, & Pratt, 1998).
We hypothesized that individual (sociodemographic and psychological), social, and perceived physical environment variables would be uniquely associated with PA in our population of rural older women, consistent with social cognitive theory. For sociodemographic variables, we hypothesized that PA levels would be negatively associated with age and African American race and positively associated with education levels. For psychological variables, we hypothesized that PA levels would be negatively associated with depressive symptoms and perceived stress and positively associated with PA self-efficacy, PA social support, and greater perceived benefits than barriers to PA (i.e., pros outweighing cons). For social variables, we hypothesized that having greater levels of social support from family and friends and having had a health care provider discuss PA would be associated with higher levels of PA. Finally, we hypothesized that environmental supports for PA would consist of greater perceived safety, the presence of sidewalks, light automobile traffic, streetlights, parks, and the perceived absence of stray dogs in one's neighborhood.
|
METHODS |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
2(1, N = 101) = 4.15, p =.04. Fairfield County is nonmetropolitan (U.S. Department of Agriculture, 1993), with 23,454 residents (U.S. Census Bureau, 2001). One third (33%) of residents have less than a 12th grade education, 35.7% are high school graduates (including equivalency), and 31.3% have some college coursework or completed college. Close to one quarter (24.1%) of adults aged 65 years and older are living below the poverty level. The largest farm crop is timber; thus most residents are not engaged in active farm labor.
|
Prior to completing self-administered questionnaires, participants read and signed an informed consent form approved by the Institutional Review Board at the University of South Carolina. The option of having an interviewer read the questionnaires was offered to participants so that those with low literacy, vision impairments, or both could still participate (n = 10). All participants were given a $10 gift card.
Measures
Sociodemographics
Age, race, highest grade of school completed, marital status, and occupation were obtained by a self-report questionnaire.
Physical activity
PA was measured with the Physical Activity Scale for the Elderly (PASE; Washburn, Smith, Jette, & Janney, 1993). Participants reported the frequency and duration of their participation in six categories of leisure-time PA in the past 7 days. For frequency and duration, participants selected a category that represented a range. They also reported whether they participated in six household activities over the past 7 days and whether they worked for pay or as a volunteer in the past 7 days, and, if so, the number of hours per week and the amount of PA required for their work. A summary score was derived that represented total PA from leisure activities, household tasks, and pay or volunteer work (Washburn et al., 1993). Because participants report frequency and duration from a range rather than a precise number, one cannot calculate the percentage of participants who meet national PA recommendations.
The PASE is a reliable and valid measure of PA in older adults. Excellent test–retest reliability over a 3- to 7-week interval has been reported (r =.75; Washburn et al., 1993). The PASE has been validated in older adults against two PA gold standards: doubly labeled water (r =.58; Schuit, Schouten, Westerterp, & Saris, 1997) and a portable accelerometer (r =.64 for those over the age of 70 years; Washburn & Ficker, 1999). It also correlated highly with the 6-min walk in older women (r =.63 in older women; Harada, Chiu, King, & Stewart, 2001) and was related to scores on grip strength (r =.40), static balance (r =.33), leg strength (r =.30), and the sickness impact profile (r = -.37) in older women (Washburn et al., 1993).
Self-efficacy for physical activity
Self-efficacy for PA was measured with a 12-item questionnaire that assesses one's confidence to engage in a PA or exercise program when one is faced with common barriers (Sallis, Pinski, Grossman, Patterson, & Nader, 1988). Participants were asked to rate their confidence on a scale ranging from 1 (very sure I could not do it) to 5 (very sure I could do it). The participant's self-efficacy score was calculated by summing the scores on the individual items; total scores could range from 12 to 60 (higher = greater self-efficacy). Good test–retest reliability (r =.68), internal consistency (
=.85), and validity (exercisers had significantly higher self-efficacy than nonexercisers; p <.001) have been reported for this measure (Sallis et al., 1988). In the present study, internal consistency was high for White ( =.96) and African American ( =.95) women.
Social support for physical activity
Social support for PA from friends (15 items) and family (5 items) was assessed with a measure in which participants rated how often family and friends engaged in acts that were supportive of PA in the past 3 months, from 1 (none) to 5 (very often; Sallis, Grossman, Pinski, Patterson, & Nader, 1987). Participants could also select "does not apply." As recommended (Sallis et al., 1987), family was defined as "members of the household" and friends were defined as "friends, acquaintances, or coworkers." Scores were averaged across items separately for family and friends, with a possible range of 1 to 5 (higher = greater social support). Good test–retest reliability (r =.57 to r =.86) and internal consistency ( =.83 to =.87) have been reported for this measure (Sallis et al., 1987). Criterion-related validity has also been reported in that social support for PA has been significantly associated with actual PA (r =.23 to r =.46; Sallis et al., 1987). In the present study, internal consistency was high for friend and family support in White ( =.89 and =.90, respectively) and in African American ( =.86 and =.92, respectively) women. Because social support from family and friends was correlated (r =.36; p <.01), we used an average of the two summary scores in all analyses.
Participants were also asked whether they had seen their health care provider in the past year for a nonemergency visit, and, if so, whether their health care provider talked to them about exercise. Participants who had seen a health care provider and reported that their health care provider talked to them about exercise scored 1 on health care provider support. All other participants scored 0.
Physical activity decisional balance (pros and cons of PA)
Participants completed a 16-item measure of the perceived benefits or pros of PA (10 items; e.g., increased energy and better ability to perform routine tasks) and the perceived barriers or cons to PA (6 items; e.g., too tired and not enough time; see Marcus, Rakowski, & Rossi, 1992). Participants rated how important each of the 16 items was in their decision whether or not to be regularly physically active, from 1 (not at all important) to 5 (extremely important). The pros items were summed to produce raw scores that could range from 10 to 50. The cons items were summed to produce raw scores that could range from 6 to 30. As recommended (Marcus et al., 1992), raw summary scores for the pros and the cons scales were each converted to t scores so that their scaling was comparable. Decisional balance represents the difference between t scores for the pros and cons of PA. A positive score indicates more pros than cons, and a negative score indicates more cons than pros. Good internal consistency ( =.95 for pros and =.79 for cons) and validity (pros increased and cons decreased across advanced stages of change; p <.0001) have been reported for this measure (Marcus et al., 1992). In the present study, internal consistency was high for pros and adequate for cons in White ( =.93 and =.76, respectively) and African American ( =.93 and =.75, respectively) women. Although this scale was developed from a transtheoretical model framework, the construct of decisional balance is conceptually very similar to outcome expectations.
Depression and stress
Participants completed a five-item version of the Geriatric Depression Scale (Hoyl et al., 1999). Summed scores could range from 0 to 5 (higher = greater depressive symptoms). As compared to a clinical evaluation as the gold standard for depression, this five-item measure has been shown to have good sensitivity (.97), specificity (.85), positive predictive value (.85), negative predictive value (.97), accuracy (.90) for predicting depression, and internal consistency ( =.80; Hoyl et al., 1999). The agreement between a clinical diagnosis and this five-item index was 
= 0.81. In the present study, however, internal consistency was somewhat low in White ( =.62) and African American ( =.41) women. For both groups, deleting the item "Do you prefer to stay at home rather than going out and doing new things?" improved internal consistency ( =.68 for White and =.49 for African American women). Thus, this item was removed from the summary score.
Participants also reported how much stress they experienced in the past 2 weeks by answering the following question: "During the past 2 weeks, would you say that you experienced a lot of stress, a moderate amount of stress, relatively little stress, or almost no stress at all?" Scores ranged from 0 (almost none at all), to 1 (relatively little), to 2 (moderate amount), to 3 (a lot). This item is from the National Health Interview Survey (Thornberry, Wilson, & Golden, 1986); reliability and validity data are not published.
Perceived physical environment
To our knowledge, no single established and validated self-report measure of the physical environment currently exists. A variety of measures and questions have been used (Humpel et al., 2002). Our questions were based on the research of Ainsworth and colleagues (SIP 4-99 Research Group, 2002) and Brownson and colleagues (Brownson et al., 1999).
Participants rated the safety of their neighborhood during the day, at night, and overall (1 = very unsafe, 2 = unsafe, 3 = neutral, 4 = safe, and 5 = very safe). Because these items were highly correlated (r =.53 to r =.61; p <.001), we used the item average to represent perceived neighborhood safety. Participants also described motorized traffic (1 = heavy, 2 = moderate, and 3 = light), street lighting (1 = very poor or no lighting, 2 = poor, 3 = fair, 4 = good, and 5 = very good), and unattended dogs (1 = not a problem at all, 2 = not very much of a problem, 3 = somewhat of a problem, and 4 = a big problem) in their neighborhood. Finally, participants reported whether they lived within walking distance of a park (within 
to 1 mile; 0 = no and 1 = yes) and whether they had sidewalks in their immediate neighborhood (0 = no and 1 = yes).
Although these questions have face validity, data on other types of validity are not yet available. In a recent survey of racially and ethnically diverse women aged 20 to 50 years across the United States, these questions (proximity to parks was not assessed) had acceptable test–retest reliability in both African American and White women, with intraclass correlation coefficients ranging from.39 to.92, with most values above.60 (Evenson, Eyler, Wilcox, Thompson, & Burke, in press). Intraclass correlation coefficients of.4 to.6 indicate moderate agreement (Landis & Koch, 1977).
Open-ended questionnaire
Participants were asked to provide a written response to printed open-ended questions regarding PA barriers ("What gets in your way of exercising or exercising more?"), motivators ("What would motivate you to exercise or exercise more?"), and perceived risks ("What are the risks of exercising for older women, if any?"). These questions were designed to elicit factors most salient to participants that might be missed when only closed-ended, forced-choice surveys were used. We used the term "exercise" rather than "physical activity" because previous research has shown that the term "physical activity" can be confusing and broader than intended (Eyler et al., 1998). The lead author read all of the responses and developed codes for a content analysis. Two additional authors (L. Oberrecht and M. Bopp) then independently coded all written responses for content. There were 96 barrier, 70 motivator, and 70 perceived risk responses coded. Agreement between the two independent coders was high (92.7% for barriers, 94.1% for motivators, and 89.8% for perceived risks). When there was a disagreement between raters (19 of 236 codes), the lead author resolved the discrepancy. The frequency with which each code was used was summarized.
Statistical Analysis
Race differences in continuous variables were examined with t tests, and race differences in categorical variables were examined with 2 statistics. Bivariate associations between each independent variable and PA were examined. Multicollinearity in independent variables was examined with regression diagnostic procedures (Neter, Kutner, Nachtsheim, & Wasserman, 1996) using SPSS. A hierarchical regression analysis examined the independent associations of sociodemographic, psychological, social, and physical environmental characteristics with PA. Four sets of variables were entered in the order listed herein, and the increment in explained variance (i.e., IR2) was tested (Cohen & Cohen, 1983). We viewed this approach as most conservative in that shared variance was attributed to established variables (e.g., sociodemographics), and more exploratory independent variables were entered last (e.g., environment). Race interactions for each of the independent variables were also tested to see if they contributed explained a significant percentage of variance beyond the main effects (i.e., IR2).
|
RESULTS |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
2(1, N = 102) = 4.75, p =.03; having lower scores on the stress item, t(98) = 2.23, p =.03, and the Geriatric Depression Scale, t(99) = 2.02, p =.03; having higher scores on family social support for PA, t(97) = -2.01, p =.05; and having a park within walking distance, 2(1, N = 102) = 11.61, p =.001. As shown in Table 2, most of the sociodemographic, psychological, social, and physical environmental variables were associated with PA. Younger age; higher education; fewer depressive symptoms; greater perceived stress, PA self-efficacy, pros than cons to PA, PA social support, and perceived neighborhood safety; less perceived neighborhood traffic; and the absence of sidewalks were all associated with higher PA levels.
|
The complete hierarchical model explained 47.4% of the variance in PA, F(13,73) = 5.06, adjusted R2 =.38, p =.00, with the sociodemographic set (age, race, education, and marital status) explaining 22.8% of the variance, F(4,82) = 6.06, adjusted R2 =.19, p =.00; the psychological set (depressive symptoms, perceiving greater pros than cons, self-efficacy, and perceived stress) explaining an additional 8.8% of the variance, F change (4,78) = 2.51, p =.048; the social set (PA social support; health care provider discussion of PA) explaining an additional 6.3% of the variance, F change (2,76) = 3.89, p =.025; and the physical environmental set (sidewalks, safety, and traffic) explaining a final 9.4% of the variance, F change (3,73) = 4.35, p =.01.
Finally, a second simultaneous regression analysis was conducted to trim to model. This trimmed model was statistically significant, as it explained 46.3% of the variance in PA: F(9,77) = 7.38, adjusted R2 =.40, p =.000. Younger age (ß = -0.28, p =.004), a more positive decisional balance (pros outweighing cons; ß = 0.22, p =.02), fewer depressive symptoms (ß = -0.21, p =.03), not having a health care provider discuss PA (ß = -0.27, p =.003), the reported absence of sidewalks (ß = -0.21, p =.02), and greater perceived safety (ß = 0.20, p =.03) were significantly associated with higher levels of PA. There was a trend for greater social support from family and friends (ß = 0.16, p =.09), greater perceived stress (ß = 0.20, p =.06), and lighter perceived traffic (ß = 0.14, p =.11) to be associated with higher levels of PA.
Open-Ended Barriers, Motivators, and Perceived Risks
The results of the content analysis are summarized in Table 3. There was no limit to the number of responses participants could give, but the maximum given was three for barriers, two for motivators, and four for risks.
|
Of the 63 women who listed a PA motivator, social support and health-related factors were the most common, each representing approximately 25% of the responses. For social support, an interest in exercising with a partner (e.g., "Someone walking with me would be a great motivation"), exercising in a group, or receiving support for PA from family or friends (e.g., "A friend or someone to encourage you") were most common. Health-related responses generally took two forms: first, participants would be more motivated to exercise if their health was better, and second, achieving better health was seen as a motivation to be more active. The former, even though it can be conceptualized as a barrier (i.e., poor health is a barrier to being active), was labeled as a motivator because participants specifically listed this response in response to what would motivate them to exercise. Other common motivators were self-motivation, often described as "determination," "self-discipline," or "motivating myself" (8%), weight loss or improved appearance (8%), physical environmental factors such as facilities, paved roads, and good walking conditions (6%), and mental health benefits ("to feel better"; 6%). Six percent of participants said that nothing would motivate them.
Finally, of the 56 women who identified risks to PA for older women, close to 30% reported that there were no risks or few risks associated with PA if one is careful. Falls (14%), injuries such as pulled muscles, "getting hurt," or "physical injuries" (13%), heart attacks (11%), broken bones (7%), other health problems such as a stroke or "severe arthritis" (10%), and "overdoing it" (4%) were the most commonly cited perceived risks of exercise.
|
DISCUSSION |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
) and African American women in particular (Eyler et al., 2002; Wilcox, 2002) have received increased attention in recent years, yet relatively few studies have examined factors that influence PA in older women who live in rural settings. Our study focused on sociodemographic, psychological, and perceived environmental influences in older and rural African American and White women.
Support of Social Cognitive Theory
Consistent with social cognitive theory (Bandura, 1986), sociodemographic, psychological, social, and perceived physical environment characteristics were all uniquely associated with PA in older rural women. Our findings underscore the importance of simultaneously studying and intervening on multilevel influences on PA. Traditionally, PA interventions have focused on the individual, but there is a growing interest in more ecological approaches to PA promotion in general (Booth et al., 2001; Sallis et al., 1998) and in older adults (Partnership for Prevention, 2001; The Robert Wood Johnson Foundation, 2001).
Correlates of Physical Activity
We hypothesized that older age, African American race, and lower education would be associated with lower levels of PA. As hypothesized, older age was consistently and independently associated with lower PA. Although age is a consistently reported determinant of PA in large epidemiological studies (Schoenborn & Barnes, 2002), it is important to note that smaller studies have not always found this relationship (Ebrahim & Rowland, 1996; Frank et al., 1998; Macera et al., 1995; Ransdell & Wells, 1998), particularly when the age range was restricted as in our study. Thus, our findings suggest that the association between age and PA may be particularly strong among rural women. Sociodemographic correlates cannot be changed; however, age appears to be a marker for modifiable characteristics. In particular, age was negatively related to self-efficacy (consistent with Cousins, 1997; Scharff, Homan, Kreuter, & Brennan, 1999; Wilcox & Storandt, 1996) and social support (consistent with Cousins, 1995).
Contrary to hypotheses and large epidemiological studies (Schoenborn & Barnes, 2002), we found no race differences in PA or PA correlates. In most studies, however, race and socioeconomic status are confounded, but in our sample age and education levels were similar for African American and White women. Our findings suggest that among older women, education level and age may be more influential than race or ethnicity.
We hypothesized that fewer depressive symptoms, lower perceived stress, greater self-efficacy, and the perception of greater pros than cons would be psychological variables associated with higher levels of PA. In regression analyses, only lower levels of depressive symptoms and greater pros than cons remained independent predictors of higher PA. The open-ended responses provided additional depth regarding pros and cons to PA that are most salient to older rural women. Health problems were common barriers and health-related factors were common motivators in our sample. In addition, as in other studies of adult women (Eyler et al., 2002; Marcus & Forsyth, 1998), social role constraints appear important for older rural women.
We hypothesized that the social and physical environment would play an important role in influencing PA. Social support did not emerge as an independent predictor in our regression analysis, although it approached significance. However, in response to an open-ended question in our study, social support was a common motivator to be more physically active. Contrary to expectations, we found that having had a health care provider discuss exercise with the participant in the past year was associated with lower levels of PA. It is possible that these women had greater health risk factors and poorer health status, including a sedentary lifestyle, which prompted provider discussion of exercise. It is difficult, however, to disentangle cause and effect in correlational studies.
Finally, we hypothesized that greater perceived safety, sidewalks, light traffic, street lights, and parks, and the absence of stray dogs, would be physical environment variables associated with higher levels of PA. Perceived safety remained an independent and positive predictor of PA in the regression analysis, whereas lighter traffic only approached significance. Aspects related to the physical environment were also cited by women in response to open-ended questions as both barriers and motivators, but to a lesser extent than personal and social influences. Contrary to expectations, women who reported sidewalks in their immediate neighborhood had lower levels of PA than those who reported no sidewalks. Few participants in this study of rural women, however, reported sidewalks in their neighborhood (29%). Having sidewalks may be an indicator of some other influence that we did not assess. The study of environmental influences on PA is still in its infancy (Humpel et al., 2002), and we are only beginning to explore these associations. King and colleagues (King, Castro, et al., 2000) also reported unexpected associations between PA and physical environment in a large sample of diverse older women. Studies that rely on both self-reports and objective assessments are needed to better understand relationships between the environment and PA.
The lack of independent associations of both self-efficacy and social support with PA might be explained by their association with age: age was the variable most strongly associated with PA, and it was also significantly and negatively associated with self-efficacy and social support. These variables may still be important from an intervention perspective, particularly because age is not a modifiable variable.
Implications for Intervention
Our findings underscore the importance of intervening on multiple levels to have an impact on PA in older rural women. Interventions with older rural women should focus on emphasizing the benefits for older women, addressing the perceived barriers, and working with families, communities, and health care providers to support PA among older women.
Intervening on the physical environment is likely more complicated, especially in rural settings. Facilitating walking groups, neighborhood watches, and traffic containment (e.g., reduced speed limits) may address perceived safety and traffic. Activities that can be done in the comfort and safety of one's own home may also be particularly appealing to older rural women, who have more limited access to facilities (King, Pruitt, et al., 2000).
Study Limitations
There are several limitations to this study. First, the nonrepresentative sampling prohibits us from generalizing our findings to other older rural women, particularly those outside of the Southern United States. Second, we did not assess health status and participant living arrangement, two variables that might have explained additional variance in PA. Empirical studies have shown that poor health is associated with lower levels of PA (King, 2001; Wilcox, 2002; Wilcox et al., 2002). Although being married is thought to have a positive effect on health behaviors (Umberson, 1987, 1992), living alone is not synonymous with being unmarried. Older women who live alone might actually be more active than those who live with others, because they may not have anyone to share household responsibilities that require energy expenditure. Third, the PA measure we chose provides a unitless PA score that does not allow one to quantify leisure-time PA from various sources (e.g., leisure time, occupational, or household) or to classify whether participants are meeting national recommendations for PA. Finally, we relied on self-report physical environment rather than an objective measure.
Conclusions
These limitations notwithstanding, our study provides useful information regarding PA barriers and enablers in older African American and White women who live in rural settings, a group that is typically underactive and at increased risk for chronic disease. Our findings demonstrate the importance of considering multilevel influences on behavior. Consistent with our hypotheses, younger age, having more pros than cons to PA, fewer depressive symptoms, and greater perceived safety of one's neighborhood appear to be important and unique factors influencing PA. It is notable that the variables chosen in our model explained close to half of the variance in PA behavior, with each level of influence explaining a significant and meaningful amount of unique variance.
|
Acknowledgments |
|---|
We gratefully thank the women of Fairfield County, South Carolina, who participated in this study and the staff at the Martin Primary Health Care Center for all of their assistance in completing this study.
Received for publication July 19, 2002. Accepted for publication June 30, 2003.
|
References |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  L. Fredman, J. A. Cauley, S. Satterfield, E. Simonsick, S. M. Spencer, H. N. Ayonayon, T. B. Harris, and for the Health ABC Study Group Caregiving, Mortality, and Mobility Decline: The Health, Aging, and Body Composition (Health ABC) Study Arch Intern Med, October 27, 2008; 168(19): 2154 - 2162. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Bopp, D. Lattimore, S. Wilcox, M. Laken, L. McClorin, R. Swinton, O. Gethers, and D. Bryant Understanding physical activity participation in members of an African American church: a qualitative study Health Educ. Res., December 1, 2007; 22(6): 815 - 826. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. S. Mujahid, A. V. Diez Roux, J. D. Morenoff, and T. Raghunathan Assessing the Measurement Properties of Neighborhood Scales: From Psychometrics to Ecometrics Am. J. Epidemiol., April 15, 2007; 165(8): 858 - 867. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. Resnick, D. Orwig, L. Wehren, S. Zimmerman, M. Simpson, and a. J. Magaziner The Exercise Plus Program for Older Women Post Hip Fracture: Participant Perspectives Gerontologist, August 1, 2005; 45(4): 539 - 544. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
