The benefits of physical activity (PA) on physical and psychological health are well known (12). Regular PA is associated with reduced risks of cardiovascular disease, thromboembolic stroke, hypertension, type 2 diabetes, osteoporosis, obesity, colon cancer, breast cancer, anxiety, and depression (8). The American College of Sports Medicine (ACSM) and the Centers for Disease Control and Prevention (CDC) recommend at least 30 min·d−1 of moderate activity on most days of the week or at least 20 min·d−1 of vigorous activity on 3 or more d·wk−1 to maintain and improve health (6,12).
One in 10 adults in the United States is a military veteran (10). To prepare for military missions, military personnel must achieve and maintain high levels of physical fitness. Physical training, typically involving vigorous activities such as running, climbing, and lifting and carrying loads, is a requirement of active service; individuals must pass twice-a-year physical fitness tests for retention and promotion. There are few data available on PA levels in individuals who are no longer in active duty. To our knowledge, only one previous study has evaluated the prevalence of PA in a national sample of veterans (16). Generalizability of findings from this study is limited because the study included obese individuals only.
Measuring PA levels in veterans is useful for several reasons. First, veterans represent a unique and large population of individuals who at one time were extremely physically active. We might expect that exposure to such high levels of activity would lead to higher levels of activity, or at least reduce the prevalence of inactivity, later in life. Quantifying the prevalence of adequate PA in veterans is also of interest because the US Department of Veterans Affairs (VA) provides health care to more than 5 million veterans at public expense. Effective disease prevention and health promotion could not only improve veterans' health but could also reduce costs borne by taxpayers. Moreover, the VA, through its integrated health care system, which includes more than 150 medical centers and 731 community-based outpatient clinics (10), is in a unique position to implement health promotion campaigns and policies related to PA. Information on PA prevalence may be helpful in focusing VA-based PA promotion programs toward the most sedentary veterans.
The first aim of this study was to describe and compare the prevalence of inactivity, insufficient activity, and meeting PA recommendations in two paired comparisons: 1) veterans and nonveterans and 2) veterans who use VA care (VA users) and veterans who do not (VA nonusers), overall and stratified by various sociodemographic and health-related characteristics. The second aim was to examine the degree to which any differences in the prevalence of inactivity, insufficient activity, and meeting PA recommendations could be explained by sociodemographic and health-related characteristics in these same paired comparisons. We hypothesized that after controlling for differences in demographic characteristics, veterans would be more likely to meet PA recommendations than nonveterans because of 1) meeting stringent physical health and capabilities criteria for entering the military and 2) the high level of PA required during their service. We also hypothesized that VA users would be less likely to meet PA recommendations than VA nonusers because of the eligibility criteria for VA care, which is largely on the basis of presence of service-connected conditions and financial need. Those with service-connected conditions rated ≥50% disabling have the highest priority, followed by those with service-connected disabilities rated as less than 50% and/or with limited financial means.
The Behavioral Risk Factor Surveillance System (BRFSS) is a collaboration between the CDC and US states and territories. The BRFSS was initiated in 1984 to collect uniform data on preventive health practices and risk behaviors that are linked to chronic diseases, injuries, and preventable infectious diseases in the adult population. The current study uses data from the 2003 BRFSS national survey, the most recent BRFSS survey in which data were available both on veteran status and PA. Information was collected using computer-assisted telephone interviews on sociodemographic characteristics (e.g., age, education, race/ethnicity), health conditions and status (self-rated health, diabetes, hypertension, and hypercholesterolemia), and health habits (e.g., smoking, physical activity), among other characteristics. Details on the questionnaire can be found elsewhere (4-6). Adults aged 18 yr and older, living in households and having a landline telephone, were sampled. Although approximately 95% of US households have telephones, coverage varies by state (range = 87%-98%) and subgroup. For example, people living in the South, minorities, and those in lower socioeconomic groups typically have lower telephone coverage (2). Telephone interviews were conducted during each calendar month, and calls were made 7 d·wk−1, during both daytime and evening hours. Standard procedures were followed for rotation of calls during days of the week and time of day (5).
We describe below the questions used to form the primary comparison groups and outcome measures. Individuals were classified as veterans if they reported that they had ever served on active duty in the US Armed Forces, either in the regular military or in a National Guard or military reserve unit, and were currently retired, medically discharged, or discharged from the military. We further classified veterans as VA users if they reported receiving some or all of their health care from VA facilities in the last 12 months. VA nonusers were veterans who reported receiving none of their health care from the VA in the past 12 months. Respondents were classified as meeting PA recommendations if they reported 30 min or more of moderate activity (described as "brisk walking, bicycling, vacuuming, gardening, or anything else that causes some increase in breathing or heart rate") on 5 or more days per week or 20 min or more of vigorous activity (described as "running, aerobics, heavy yard work, or anything else that causes large increases in breathing or heart rate") on 3 or more days per week, during a usual week. Those who performed some PA, but did not meet the recommended amount or frequency, were classified as "insufficiently active." Those who reported no moderate or vigorous PA of at least 10 min in duration were classified as inactive. Exact wording of questions can be found elsewhere (6).
The interview also elicited information on demographic characteristics, smoking, body mass index, and health conditions, including hypertension, hypercholesterolemia, diabetes, joint pain, and disability. We classified individuals on the basis of their self-reported height and weight as normal weight (<25 kg·m−2), overweight (25-29.9 kg·m−2), or obese (≥30 kg·m−2). Individuals were classified as having joint pain if they reported either having symptoms of pain, aching, or stiffness in or around a joint in the past 30 d or having a doctor tell them that they had some form of arthritis, rheumatoid arthritis, gout, lupus, or fibromyalgia. Individuals were classified as having a disability if they reported being limited in activities because of physical, mental, or emotional problems or if they reported having a health problem that required them to use special equipment such as a cane, wheelchair, a special bed, or a special telephone. To ensure sufficient numbers in subgroups of interest, we collapsed categories of general health status (excellent/very good, good/fair/poor), education (less than high school graduate, high school graduate, some college, and college graduate), and employment status (employed, unemployed, unable to work, and retired).
Each respondent in the data set was assigned a final sampling weight on the basis of (a) his or her overall probability of selection and (b) a poststratification factor that ensured that the age and gender distribution of the weighted sample would agree with population estimates from the US Census Bureau, thus adjusting for demographic differences in probability of selection and nonresponse. Poststratification weights may also partially correct for any bias caused by telephone noncoverage. The value of the weight for each person in the sample was the estimated number of people that he or she represented in the target population. The size and distribution of the source population could thus be reconstructed from the sample by summing these weights across all sampled persons with a certain characteristic.
We used model-based direct rate adjustment (14) to the distribution of all veterans to estimate the prevalence of inactivity, insufficient activity, and meeting PA recommendations, overall, and stratified by various demographic, health behavior, and occupational characteristics. The prevalence estimates were adjusted for age group and gender, where noted. To evaluate whether differences in the prevalence of PA categories between 1) veterans and nonveterans and 2) VA users and VA nonusers were statistically significant, we used Pearson's chi-square tests corrected for the survey design (13) and converted into an F statistic using the svy (survey) commands in Stata 9.2 (Stata Corp., College Station, TX). To evaluate the extent to which observed differences in PA categories were due to confounding, we conducted analyses unadjusted, and then successively adjusted for demographic characteristics (age, gender, race/ethnicity, education), behavioral risk factors (smoking), and finally for various health conditions that may limit ability to engage in exercise (diabetes, hypertension, hypercholesterolemia, disability, and joint pain). We selected these characteristics and conditions because they were available in the 2003 BRFSS survey, might differ between groups (i.e., veterans and nonveterans and/or VA users and VA nonusers) and be related to PA prevalence. However, because these health conditions might be a consequence of either low PA (e.g., diabetes, hypertension, and hypercholesterolemia) or high PA (e.g., joint pain) rather than a predictor, these results must be interpreted cautiously.
BRFSS data are in the public domain. This study was therefore exempt from Institutional Review Board review.
In 2003, a total of 264,684 adults completed the BRFSS survey. We excluded 7783 (2.9%) individuals who either refused to answer the question on veteran status or answered "do not know/not sure" and 11,337 individuals (4.4%) who refused or were unable to answer the questions on PA, leaving 245,564 for analyses. Of the 34,863 self-identified veterans, 2712 (7.4%) refused or answered "do not know/not sure" to the questions on receipt of care at VA facilities and thus were excluded from analyses in veterans only.
As we would expect, there were large sociodemographic differences between veterans and nonveterans (Table 1). A greater proportion of veterans were older; were male; were non-Hispanic white; were high school graduates or higher; reported good, fair, or poor health; were former smokers; were overweight or obese; and were retired. Veterans were also more likely to report having a disability, diabetes, hypercholesterolemia, and joint pain compared with nonveterans. Approximately 17% of veterans reported receiving some of their care at VA facilities. VA users were older than nonusers. Other differences between VA users and nonusers reflect, in part, the selection criteria for coverage of VA care discussed briefly above. Specifically, compared with VA nonusers, a greater proportion of VA users were non-Hispanic black; were less educated; reported good, fair, or poor health; were unable to work; or were retired. VA users were also more likely to have a disability, diabetes, hypertension, joint pain, and hypercholesterolemia than VA nonusers.
Overall and across all subgroups, veterans were less likely to be inactive and more likely to meet PA recommendations than nonveterans (Table 2); absolute differences in meeting PA recommendations between veterans and nonveterans were greatest in those who were older; were African American or other nonwhite race; had less than a high school education; had poorer reported health; were obese; or were retired (all P < 0.01).
There was little difference in the prevalence of meeting PA recommendations between VA users and nonusers. However, VA users were more likely to be inactive and less likely to be insufficiently active overall (P < 0.0001) and in most subgroups (Table 3). Specifically, such differences between VA users and nonusers in the prevalence of inactivity and insufficient activity were large and statistically significant in the following subgroups: ≥50 yr of age, men, white, in good/fair/poor health, overweight, obese, retired, and in those whose occupational activity involved mainly walking. In those who were college graduates and employed, compared with VA nonusers, VA users were more likely to meet PA recommendations and less likely to be insufficiently active; this pattern was also observed in VA users <35 yr and with "other" race, although small numbers reduced power to detect associations.
To evaluate the extent to which confounding might explain some of the differences in PA, we conducted analyses unadjusted; adjusted for age and gender only (Model 1); adjusted for Model 1 factors plus race/ethnicity, education, and smoking (Model 2), and adjusted for Model 2 factors plus diabetes, hypertension, hypercholesterolemia, disability, and joint pain (termed "chronic conditions"). There was little difference in PA between veterans and nonveterans (approximately 16% of veterans and nonveterans were inactive and 46% met PA recommendations) on the basis of unadjusted estimates. When the prevalence of PA in nonveterans was standardized to the age and gender distribution of veterans, veterans were statistically significantly less likely to be inactive and more likely to meet recommendations (presented in Table 2). Additional adjustment for Model 2 factors (presented in Fig. 1) and Model 3 factors (data not presented) did not substantially change estimates. In unadjusted analyses comparing VA users to nonusers, VA users were more likely to be inactive (22.6% vs 14.9%) and less likely to meet PA recommendations (42.6% vs 46.7%; overall P < 0.0001). Adjustment for Model 1 factors (presented in Table 3) and Model 2 factors (presented in Fig. 2) resulted in some attenuation of the differences between VA users and nonusers in inactivity and meeting recommendations, with little change in the difference in insufficient activity. After further adjustment for Model 3 factors, there was no longer much difference in inactivity (16.8% vs 15.2%), but there continued to be an approximately 4.5 point difference in insufficient activity. Moreover, the differences in meeting PA recommendations reversed, with a higher prevalence in VA users (49.5%) than in nonusers (46.5%; overall P = 0.009).
Only 46% of veterans met PA recommendations for moderate or vigorous activity, short of the Healthy People 2010 objective of 50% for adults (objective 22-2). Veterans, however, were more likely to meet PA recommendations than nonveterans. In addition, although there was little evidence of confounding due to education, race/ethnicity, smoking, and chronic conditions after accounting for differences in age and gender, the prevalence of meeting PA recommendations varied by subgroup. Specifically, differences between veterans and nonveterans were largest in the groups in which meeting PA recommendations was lowest (e.g., age ≥70 yr, nonwhite race/ethnicity, less educated, obese, and retired).
In analyses comparing VA users and nonusers, in unadjusted analyses and analyses adjusted for demographic characteristics, VA users were more likely to be inactive and less likely to meet PA recommendations. After further adjustment for race/ethnicity, education, smoking, and chronic conditions, VA users were more likely to meet recommendations than nonusers, suggesting that the higher prevalence of chronic conditions among VA users may at least partially explain their lower PA participation, or conversely that lower PA participation among VA users may have led to a greater prevalence of chronic conditions. Because of the cross-sectional nature of the data, we are unable to distinguish between these possibilities. Nevertheless, we did find that in veterans who were younger, highly educated, and employed, VA users were more likely to meet PA recommendations than VA nonusers without adjustment for chronic conditions. These characteristics may be proxies for a healthier population, suggesting that in veteran groups that are healthier, VA users have higher PA levels than nonusers. Reasons for use of VA health care may be different for these groups than for other groups. Future research could try to untangle the temporal sequence and causal associations of chronic conditions and PA in veterans.
Differences between veterans and nonveterans in meeting PA recommendations could be a consequence of selection criteria for entry into the military ("healthy soldier" effect) and/or a direct consequence of being more physically ac tive during their military service. Only those who were physically fit for military service could have become veterans. Nonveterans would consequently represent a heterogeneous population including both those who were sufficiently healthy to meet criteria for entrance into the military, but chose not to, and those with chronic conditions or disabilities that rendered them unfit for military service. Alternatively or additionally, PA during military service may have increased the likelihood of PA after separation from the military, possibly because of improved physical fitness, increased self-efficacy for exercise, or generally being more motivated to exercise because of enjoyment and perceived benefits in health and well-being (1,15).
We know of only one other study using recent national data that estimated the prevalence of meeting PA recommendations in a population of veterans. Using data from the 2000 BRFSS, Wang et al. observed no difference in meeting PA recommendations between obese VA users and a population of obese nonveterans and veterans who did not use VA facilities (17.8% vs 16.2%). Their results for this subgroup were similar to ours, although these investigators did not report prevalence of inactivity, where we observed large differences. Furthermore, the prevalence of meeting PA recommendations was likely lower in the prior study because the PA questions in the BRFSS changed in 2001 to cover more activity domains (e.g., household, transportation, and leisure time), to profile activities in a usual week rather than reporting the top two activities during the preceding month (such that less frequent activities that might not have been mentioned in the 2000 question format could be included in the 2003 overall activity profile), and to inquire about moderate- and vigorous-intensity activities separately (increasing the potential to recall less intense lifestyle activities) (3).
There are several limitations to this study. Only individuals with a landline telephone were eligible to participate in the BRFSS. Thus, poorer individuals may be underrepresented in this survey. Second, small sample sizes for some subgroups limited precision and statistical power, particularly for analyses comparing VA users to nonusers. Third, PA was based on self-report and thus susceptible to errors in recall and reporting. In addition, the survey questions were not designed to assess whether a combination of moderate and vigorous PA satisfied the requirement for meeting recommendations for regular PA when the two activity types measured separately did not. Consequently, the prevalence of meeting recommendations may be underestimated. However, we do not have a reason to suspect that these limitations would affect the within-sample comparisons evaluated in the current study; comparisons of prevalences between groups defined by veteran status or health care use should be valid. We were also unable to adjust for ability to be physically active. The BRFSS includes two broad questions on mental and physical disabilities, but does not differentiate between the two. Both veterans and VA users were more likely to report having a disability than nonveterans and VA nonusers, respectively. Adjustment for disability status did not substantially change estimates in comparisons of veterans and nonveterans; in comparisons of VA users and VA nonusers, there was only a slight attenuation of the difference in inactivity and meeting PA recommendations. Finally, we do not have information on time since separation from the military. As time since military service separation may vary considerably both within and between subgroups, we are unable to ascertain whether differences in PA prevalence by age are due to aging per se, to cohort effects (e.g., individuals born in the 1940s may have been exposed to a different environment than those born earlier or later), or to differences in time since separation from the military.
These results suggest several avenues for future research and interventions. Future studies could investigate transitions from military to civilian life and individual characteristics as well as aspects of that transition that are associated with PA maintenance. Because of the network of medical centers and outpatient facilities, the VA is in a unique position to reach and intervene on millions of individuals to increase PA and improve health status. Moving from inactivity to small amounts of PA can result in significant health benefits (7,9,11). Thus, for individuals who are sedentary, emphasizing that even some amounts of PA (e.g., 10-min bouts) throughout the day and week can be beneficial, and setting initial PA goals that are below PA recommendations may be important. To this end, the VA implemented a national weight loss program beginning in 2007 called The Managing Overweight and Obesity in Veterans Everywhere (MOVE!), which emphasizes dietary modification and PA promotion. If effective, this program could have wide-ranging impact on the ∼60% of veterans who are inactive or insufficiently active.
This material is based on work supported in part by the Office of Research and Development Cooperative Studies Program, Department of Veterans Affairs. The results of the present study do not constitute endorsement by ACSM.
The authors have no conflicts of interest to report.
1. Bandura A. Self-efficacy: toward a unifying theory of behavioral change. Psychol Rev
2. Bureau of the Census. Phoneless in America
. Washington (DC): US Department of Commerce, Bureau of the Census; 1994.
3. Centers for Disease Control and Prevention (CDC). Prevalence of physical activity, including lifestyle activities among adults-United States, 2000-2001. Morb Mortal Wkly Rep
4. Centers for Disease Control and Prevention (CDC). 2003 BRFSS Overview
. Atlanta (GA): Centers for Disease Control and Prevention; 2003. p. 1-6.
5. Centers for Disease Control and Prevention (CDC). Behavioral Risk Factor Surveillance System: Operational and User's Guide
. Atlanta (GA): Centers for Disease Control and Prevention; 2006. p. 1-112.
7. Haskell WL, Lee IM, Pate RR, et al. Physical activity and public health: updated recommendation for adults from the American College of Sports Medicine and the American Heart Association. Med Sci Sports Exerc
8. Kesaniemi YK, Danforth E Jr, Jensen MD, Kopelman PG, Lefebvre P, Reeder BA. Dose-response issues concerning physical activity and health: an evidence-based symposium. Med Sci Sports Exerc
. 2001;33(6 suppl):S351-8.
9. Kohl HW 3rd. Physical activity and cardiovascular disease: evidence for a dose response. Med Sci Sports Exerc
. 2001;33(6 suppl):S472-83; discussion S493-4.
11. Oguma Y, Shinoda-Tagawa T. Physical activity decreases cardiovascular disease risk in women: review and meta-analysis. Am J Prev Med
12. Pate RR, Pratt M, Blair SN, et al. Physical activity and public health. A recommendation from the Centers for Disease Control and Prevention and the American College of Sports Medicine. JAMA
13. Rao JNK, Scott AJ. On chi-squared tests for multiway contingency tables with cell proportions estimated from survey data. Ann Stat
14. Rosenbaum PR. Model-based direct adjustment. J Am Stat Assoc
15. Ryan RM, Frederick C, Lepes D, Rubio N, Sheldon K. Intrinsic motivation and exercise adherence. Int J Psychol
16. Wang A, Kinsinger LS, Kahwati LC, et al. Obesity and weight control practices in 2000 among veterans using VA facilities. Obes Res
Keywords:©2009The American College of Sports Medicine
EXERCISE; MODERATE; VIGOROUS; BEHAVIORAL RISK FACTORS SURVEILLANCE SYSTEM; EXERCISE MAINTENANCE