Cardiovascular disease is currently the leading cause of mortality among older women in the United States (1). Increasing physical activity levels can reduce the risk of developing chronic illness including cardiovascular disease and diabetes (14). Middle-aged women (age 45-64 yr) are among the least active subgroups in the United States, with more than 38% of women in this age group reporting being physically inactive during their leisure time (15). Understanding more about patterns of physical activity levels in postmenopausal women will guide the field in developing better methods for physical activity assessment and intervention in this unique and currently underserved population.
Environmental changes in temperature, precipitation, and number of daylight hours are thought to provoke seasonal changes in physical activity levels (7). In parts of the United States, where weather patterns fluctuate with the changing seasons, it appears that activity levels may vary by season. For example, in the southwestern US, physical activity levels may drop during the summer months due to the hot and dry climate. In contrast, in the northeastern US, physical activity may decline in the cold, snowy, winter months. Several previous investigations, set in geographical areas with distinct seasons, found that individuals reported peak physical activity levels during summer months, during which the weather was more conducive to being physically active (7,11,13). The issue of seasonality in physical activity research is important because it may relate to health-related outcomes. For example, findings from the Seasonal Variation of Blood Cholesterol study showed combined moderate-intensity household and doubled leisure time activity (i.e., increase of 2.0-2.4 MET·h·d−1) during the summer in comparison with winter in men and women ages 20-70 yr (7). In the Seasonal Variation of Blood Cholesterol study, the issue of seasonality was not limited to changes in physical activity levels; seasonal differences were also observed in relation to blood lipid levels. Average total cholesterol peaked in men during the month of December and in women during the month of January when physical activity levels were lower, which suggests that fluctuations in physical activity levels across seasons may positively or negatively influence health-related outcomes (8). Although all of the above-mentioned investigations used questionnaires to assess physical activity levels, two also used objective measures (7,11) to capture physical activity levels but had relatively small sample sizes (n = 62 and 23). Therefore, there is a lack of valid information regarding how seasonal variations in physical activity affect postmenopausal middle-aged women.
The purpose of the current report is to determine whether variations occur in leisure physical activity levels across the year in a large sample of postmenopausal women (N = 508). Because we are hypothesizing that this will be the case, this study will also allow us the opportunity to examine the influence of a lifestyle intervention on this variation in activity in postmenopausal women. Physical activity levels will be examined both at baseline (before randomization) and after 18 months in postmenopausal women randomized to either a lifestyle change group or a health education group. It is hypothesized that there will be fluctuations in baseline physical activity levels throughout the year. We also hypothesize that physical activity levels among women randomized to the lifestyle change group will be less influenced by month-to-month climate changes after 18 months of a structured lifestyle intervention than the health education group.
Five hundred and eight postmenopausal women, aged 52-62 yr, were recruited for the Women on the Move through Activity and Nutrition (WOMAN) study, primarily through direct mailing from selected zip codes in Allegheny County, PA, from April 2002 to October 2003. Eligibility criteria for enrollment into the study included waist circumference >80 cm, body mass index (BMI) between 25 and 39.9 kg·m−2, not currently taking lipid-lowering drugs and having a low-density lipoprotein (LDL-C) level between 100 and 160 mg·dL−1, no physical limitations that would preclude walking, no known diabetes, and no diagnosed psychotic disorder or depression. All participants provided written informed consent, and all protocols were approved by the institutional review board at the University of Pittsburgh. Results from the current investigation were generated from data collected at the baseline and at the 18-month follow-up visit.
Group randomization: intervention design.
Eligible women were randomized to a health education comparison group or a lifestyle change group using a block randomized design prepared by a statistician (6). The healthy lifestyle behaviors promoted in the lifestyle change group included 150 min·wk−1 of moderate-intensity physical activity similar to brisk walking and caloric intake of 1300-1500 cal·d−1, emphasizing an eating pattern low in total fat (<17%) and saturated fat (<7%). Following these guidelines, women randomized to the lifestyle change group were asked to lose ≥10% of their initial body weight. The health education group included a core educational series of six courses offered throughout the first year.
Physical activity measures.
Physical activity levels were assessed using both subjective and objective measurement tools. The Modifiable Activity Questionnaire, an interviewer-administered questionnaire, assesses current leisure and occupational activities over the past year (9). Although this questionnaire assesses leisure and occupational activities, we presented only the leisure activity because there was little reported variation in occupational activity in our study population. Before administering the Modifiable Activity Questionnaire to study participants, each interviewer was thoroughly trained by experienced study investigators. Physical activity levels were calculated as the product of the duration and the frequency of each activity (h·wk−1), weighted by an estimate of MET of that activity and summed for all activities performed, and data were expressed as MET-hours per week (MET·h·wk−1). The Modifiable Activity Questionnaire has been shown to be both a reliable (5,9) and a valid (5,9,10) assessment tool. Leisure physical activity was also measured using a past-week version of the Modifiable Activity Questionnaire to obtain an acute estimate of physical activity levels. Study participants were asked to record leisure activities, which were done for at least 10 min each time, during the 7 d before the clinic assessment. The past-week leisure activity estimate was calculated similar to that used for the past-year Modifiable Activity Questionnaire but covered a shorter time frame.
Objective assessments of physical activity were obtained using the Accusplit AE120 (Accusplit Inc., Pleasanton, CA) pedometer in a convenient sample of WOMAN study participants at the baseline visit. Pedometer data were obtained at baseline on the first 15 women each month that completed a clinic visit due to the limited number of pedometers available. At baseline, those women who wore the pedometer did not significantly differ by age, waist circumference, BMI, or physical activity level when compared with those who did not wear the pedometer (data not shown). However, at the 18-month clinic visit, pedometer data were collected in all women. Pedometers are both a valid and a reliable way to measure physical activity (2). The participants were instructed to wear the pedometer clipped to their waistband over the dominant hip for 1 wk. Participants were provided with an activity diary and were asked to record the time the pedometer was put on in the morning as well as, at the end of the day, the time that the monitor was taken off and the number of steps taken. At the end of the week, the participant returned the activity diary to the investigator. The daily step counts recorded in the diary for each of the 7 d were averaged for the week to obtain a 7-d average of the number of steps taken per day.
For the purposes of this report, seasons were defined as summer (June, July, and August), fall (September, October, and November), winter (December, January, and February), and spring (March, April, and May). These cut points were determined in an effort to accurately capture the months in Pittsburgh, PA, in which the weather patterns most closely resemble each other during that year.
Descriptive statistics were used to describe demographic and anthropometric data. Normally distributed variables were reported as mean ± SD; nonnormally distributed variables were reported as median with 25th and 75th percentiles and categorical data as proportions.
At 18 months, descriptive statistics were also used to describe demographic data, anthropometric measures, and physical activity levels in the entire cohort and were stratified by randomized group assignment (health education vs lifestyle change). Depending upon the characteristics of the variable, t-tests, Wilcoxon rank sum, or chi-square tests were used to compare descriptive statistics between randomized groups. Finally, physical activity levels were compared between the four defined seasons using a Kruskal-Wallis test.
At baseline, 508 women were randomized into the WOMAN study. Of those 508 women, pedometer data were collected at baseline on a sample of 170 women (33.5% of total study participants), whereas past-year and past-week leisure physical activity estimates were available for 500 women. At 18 months, 455 women (90%) completed the follow-up visit. Of these 455 women, complete pedometer data were collected in 318 women (70%) during this clinic visit. Descriptive statistics for the WOMAN study baseline cohort are presented in Table 1. At baseline, the median physical activity level reported on the past-week Modifiable Activity Questionnaire was 11.4 MET·h·wk−1, whereas self-reported past-year leisure physical activity at baseline was 12.2 MET·h·wk−1. The leisure physical activity estimate obtained from the Modifiable Activity Questionnaire was validated in this specific population of postmenopausal women against a pedometer step counter (P < 0.0001, ρ = 0.30) (6). Median pedometer step counts at baseline were 6447.
Figure 1 examines the seasonal differences of step counts for the sample of 170 women who wore the pedometer at baseline before randomization. It is important to note that each woman is represented once in this figure, depending on the date of her baseline visit. At baseline, participants did not differ significantly by age, body mass index (BMI), or waist circumference by season; however, there were notable difference in physical activity levels. Pedometer steps were highest in the summer months, lower in the fall, lowest in winter, and then rebounded in the spring. Pedometer step counts at baseline were higher in the summer months when compared with the winter months (P = 0.06).
Figure 2 illustrates leisure physical activity levels, as assessed by the past-week version of the Modifiable Activity Questionnaire, by month over a 1.5-yr duration. Due to the normal flow of the clinic, it took approximately 1.5 yr to get all 508 women in for any one clinic visit. Similar to the objective pedometer data, self-reported leisure activity levels also fluctuated throughout the year and were lowest in the winter months.
Descriptive statistics for women with complete physical activity data at 18 months is presented in Table 2, stratified by randomized group assignment. When compared with the health education group, the women in the lifestyle intervention group had significantly lower BMI, waist circumference, and body weight (all P values <0.0001). The lifestyle change group had significantly higher physical activity levels at 18 months when compared with the health education group, which was confirmed by both subjective and objective measures of physical activity (P = 0.003 and 0.0001, respectively).
Figure 3 presents the pedometer step counts at the 18-month follow-up visit by month and randomized group assignment. When examining pedometer step counts over 18 months by randomized group assignment, the lifestyle change group was significantly more active, averaging over 2000 steps more per week when compared with the health education group (P = 0.0001). Furthermore, in an attempt to quantify the month-to-month variability, the sum of the absolute value around the mean across the months of the intervention was determined for each randomized group and was found to be less than half in the lifestyle change group when compared with the health education group (8089 vs 16,813 steps). These findings suggest that the lifestyle intervention not only improved step counts in the women randomized to that group but also appeared to have attenuated the variation in physical activity levels that are commonly observed over the course of the year.
In this large cohort of postmenopausal women, fluctuations in physical activity levels were examined at the baseline and at the 18-month follow-up visit. Seasonal variation in physical activity, likely due to weather, has been suggested in the literature (7,13) but has yet to be examined in postmenopausal women. Seasonal variation was found at baseline before any intervention with differences noted between summer and winter. This is in line with findings from the Behavioral Risk Factor Surveillance System study that showed a high prevalence of physical inactivity in the winter months and a low prevalence of physical inactivity during the summer months (3).
This cohort of postmenopausal women at baseline had a median step count of 6447 (4823-8722) steps per day, which would be considered relatively "low active" (5000-7499 steps per day) and not "sedentary" (<5000 steps per day) according to established pedometer step-count criteria (12). This may be due, in part, to a "volunteer effect" because the women in this study had committed to being part of a 5-yr clinical trial. Kriska et al. (4) described this possible "volunteer effect" when comparing physical inactivity data from participants in the Diabetes Prevention Program (DPP), a multicenter randomized clinical trial that involved an intensive lifestyle component, with a subgroup of people with impaired glucose tolerance from the Third National Health and Nutrition Examination Survey (NHANES III) (4). The DPP cohort represents a volunteer sample of individuals committed to a long-term intervention. In contrast, the NHANES III subgroup represents a random sample of the US population comprised of individuals from NHANES III that would have met DPP entry criteria because both groups completed the same physical activity questionnaire used in NHANES III. The authors found that physical inactivity in the DPP cohort was significantly less than that reported in the NHANES III subgroup for every age, gender, and ethnic group.
The goal of any physical activity intervention is to use physical activity as a means to slow, prevent, or even reverse a disease process. This can only be achieved, in the long term, by making physical activity a regular part of a healthy lifestyle (14). When examining median weekly pedometer step counts 18 months into the physical activity intervention, women randomized to the lifestyle change group were not only significantly more physically active but their physical activity levels also appeared to be more consistent throughout the year when compared with women in the health education group. The implications of these findings suggest that lifestyle intervention may not only serve to increase participant's physical activity levels but may also promote life-long behavior change, which may help participants to attain regular physical activity levels year-round.
The monthly fluctuations in physical activity levels across the year observed in this study suggest the need to assess physical activity levels several times per year, in areas with changing seasons or changing activity demands, to achieve the most accurate picture of usual activity. For example, over that year, if physical activity levels were only assessed once per year and that happened to fall sometime during harsh weather, physical activity levels could be uncharacteristically low; the converse could be true if assessments were only made during one of the more pleasant times of the year. Therefore, the issue of seasonality in physical activity assessment has important public health implications and if not considered has the potential to compromise study results.
A potential limitation that should be considered when interpreting these results is that this investigation did not capture 12 months of physical activity data on the same individuals but represented different women assessed over the course of the year. However, we do not believe that this compromises the overall findings because there were no significant differences by month in the average age, BMI, or waist circumference of the women at baseline. Future research could take the approach of examining a cohort of women every month throughout the year. A second potential limitation of this study is that the pedometer was used as both a physical activity assessment and an intervention tool. We do not believe that this will have a major impact on study results because both the health education and the lifestyle change groups were exposed to the pedometer during the study.
In conclusion, results of the present investigation suggest that physical activity levels, measured objectively by pedometer and subjectively by questionnaire, fluctuate throughout the year in this group of postmenopausal women measured at baseline. The results of this study also indicate that those participants randomized to the lifestyle modification group not only significantly increased their physical activity levels but also appeared to be less prone to monthly fluctuations in physical activity levels after 18 months of physical activity intervention as opposed to women randomized to the health education group.
The authors would like to acknowledge the 508 dedicated WOMAN study participants and the contributions of the WOMAN study staff, including Alhaji Buhari, Eileen Cole, Phyllis Jones, Laura Kinzel, Barbara Kolodziej, Wm. Scott Pappert, Darcy Underwood, and Dr. Laurey Simkin-Silverman. The authors would also like to acknowledge Dr. Kenneth J. Jaros for his expert review of this manuscript. This research was funded by National Heart, Lung, and Blood Institute grant R01-HL-6646. The results of the present study do not constitute endorsement by the American College of Sports Medicine.
1. American Heart Association. Heart Disease and Strok Statistics-2007 Update at-a-Glance Version
. Dallas (TX): American Heart Association; 2007. p. 43.
2. Bassett DR. Validity and reliability issues in objective monitoring of physical activity. Res Q Exerc Sport
. 2000;71(suppl 2):S30-6.
3. Centers for Disease Control and Prevention (CDC). Monthly estimates of leisure-time physical inactivity-United States, 1994. Morb Mortal Wkly Rep
4. Kriska AM, Edelstein SL, Hamman RF, et al. Physical activity in individuals at risk for diabetes: Diabetes Prevention Program. Med Sci Sports Exerc
5. Kriska AM, Knowler WC, LaPorte RE. Development of a questionnaire to examine the relationship of physical activity and diabetes in Pima Indians. Diabetes Care
6. Kuller LH, Kinzel LS, Pettee KK, et al. Lifestyle intervention and coronary heart disease risk factor changes over 18 months in postmenopausal women: the Women on the Move through Activity and Nutrition (WOMAN study) clinical trial. J Womens Health
7. Matthews CE, Freedson PS, Hebert JR, et al. Seasonal variation in household, occupational, and leisure time physical activity: longitudinal analyses from the Seasonal Variation of Blood Cholesterol study. Am J Epidemiol
8. Ockene IS, Chiriboga DE, Stanek EJ III, et al. Seasonal variation in serum cholesterol levels. Arch Intern Med
9. Pereira MA, FitzerGerald SJ, Gregg EW, et al. A collection of physical activity questionnaires for health-related research. Med Sci Sports Exerc
. 1997;29(6 suppl):S1-205.
10. Schulz LO, Harper IT, Smith CJ, Kriska AM, Ravussin E. Energy intake and physical activity in Pima Indians: comparison with energy expenditure measured by doubly-labeled water. Obes Res
11. Tudor-Locke C, Bassett DR, Swartz AM, et al. A preliminary study of one year of pedometer self-monitoring. Ann Behav Med
12. Tudor-Locke C, Bassett DR. How many steps/day are enough? Preliminary pedometer indices for public health. Sports Med
13. Uitenbroek DG. Seasonal variation in leisure time physical activity. Med Sci Sports Exerc
14. US Department of Health and Human Services. Healthy People 2010. 2nd ed. Understanding and Improving Health and Objectives for Improving Health 2 vols
. Washington (DC): US Department of Health and Human Services; 2000 Nov. 664 p. Available from: US GPO, Washington.
15. US Department of Health and Human Services. Health Behaviors of Adults: United States, 2002-04. Vital and Health Statistics
. Hyattsville (MD): US Department of Health and Human Services; 2006 Sept. 151 p. Available from: US GPO, Washington.