Physical activity is known to have a positive effect on health and contributes to a high quality of life. Regular exercise reduces the risk of overall mortality and the risk of several chronic diseases (14,23). Despite this common knowledge, a large part of the population from developed countries does not adhere to the international physical activity recommendation (14) of being moderately active for 30 min on at least five and preferably all days of the week. In the United States, 50% of the adult population is sufficiently active (4) compared with 56% of the Dutch population above 12 yr of age (20). Actually, the difference between these countries is larger because moderate-intensity activities are defined as 3-6 METs for the total population in the United States (14), whereas in The Netherlands, it is defined as 4-6.5 METs for adults between 18 and 55 yr and 3-5 METs for the population 55 yr and older (6).
Public health strategies attempt to stimulate participation in physical activity. To succeed, knowledge about the determinants of an active lifestyle is necessary. Earlier research identified several demographic, psychosocial, physiological, and environmental factors that are associated with (change in) physical activity behavior (7,10,17-19,22). However, most studies reported only the determinants of the short-term adoption of physical activity during an intervention study, whereas from a public health perspective, permanent changes in physical activity behavior are relevant (9,11,15).
The Sex Hormones and Physical Exercise (SHAPE) study is a randomized controlled trial designed to assess the effect of a 1-yr exercise program on sex hormone levels in low-active postmenopausal women (12). One year after finishing this study, all participants were recontacted. First, we investigated the effect of participation in an exercise intervention study on their current physical activity behavior. Second, we investigated which factors predict a higher level of physical activity in women in the intervention group 1 yr after finishing the intervention.
An extensive description of the SHAPE study has been published previously (12). In short, this study included 189 healthy low-active postmenopausal women (50-69 yr) who were volunteers. Low active was defined as participating less than 2 h·wk−1 in moderate-intensity sports and recreational activities (e.g., tennis, swimming, running, aerobics, fitness, and volleyball) and not adherent to the international physical activity recommendation. Because bicycling is a very common form of transportation in The Netherlands, also practiced by people not engaged in any other activities, bicycling for transportation in their hometown was not considered as a moderate-intensity activity when screening women for eligibility. Women were randomly allocated to an exercise intervention group (n = 96) or control group (n = 93), stratified by waist circumference of <92 and ≥92 cm. The intervention consisted of a combined endurance and strength training program over a period of 12 months. Twice a week, participants of the exercise group gathered for a group session of 1 h, supervised by a qualified sports instructor. Furthermore, once a week, participants conducted an individual session of brisk walking or cycling for at least 30 min, for which they received instructions during the group sessions.
The control group was requested to retain their habitual physical activity pattern. Both intervention and control groups were asked to maintain their usual food intake. The study was approved by the ethical committee of the University Medical Center of Utrecht, and all subjects gave their written informed consent.
Results of the SHAPE study have been published elsewhere (13).
To study whether participation in the SHAPE study led to sustained behavioral changes and which characteristics predict maintenance of physical activity level, we recontacted participants 1 yr after finishing the program. This follow-up measurement was not announced during the study period and therefore could not influence behavior.
Of the 96 women allocated to the physical exercise program, 95 completed the study and 78 gave consent for the follow-up interview; of the 93 women allocated to the control group, 88 completed the study and 68 gave consent for the follow-up interview (Fig. 1).
First research question: sustained changes of physical activity level.
Activity levels of women in the previous intervention and control groups were compared to measure the effect of participation in the SHAPE study 1 yr after completion of the program. Habitual exercise was measured with the Modified Baecke Questionnaire at the start of the SHAPE study, at 12 months (end of the SHAPE study), and at 24 months. The Modified Baecke Questionnaire was designed by Voorrips et al. (24) to measure habitual physical activity in the elderly. This questionnaire was chosen because it reflects activities of women in the age group of the SHAPE study, for example, by extensively addressing household activities. The questionnaire includes questions about household activities, sports, and leisure time activities over the past year. All items result in a separate score that incorporates activity duration, frequency, and intensity code on the basis of energy expenditure. Summing the household score, sport score, and leisure time activity score results in a continuous overall unitless activity score.
In addition, we calculated the MET-hours per week spent on at least moderate-intensity activities by coding the sports and the leisure time activities reported in the Modified Baecke Questionnaire according to the Ainsworth compendium of physical activities (1,2). We decided that moderate-intensity activities included all sports and leisure time activities of at least 4 METs. The value of 4 METs was chosen on the basis of the Dutch standard, which implies that moderate-intensity activities lie between 4 and 6.5 METs (6). To measure the effect of participation in the SHAPE study, we classified women as either compliant or noncompliant to the international physical activity recommendation of being moderately active for 30 min on at least 5 d of the week, which equals a minimum of 10 MET·h·wk−1 (4 METs × 0.5 h × 5 d).
Second research question: predictors of a sustained physical activity level in the intervention group.
We investigated which characteristics predict maintenance of physical activity level 1 yr after participation in an exercise program of limited duration. For this second part of the analysis, we included data of the intervention group only (n = 76). A limited set of potential predictors of physical activity level were available. Age, baseline activity, body mass index (BMI) change during the intervention, employment, attendance to group sessions, and family history of breast or colorectal cancer were considered as potential predictors of the physical activity level 1 yr after the intervention. BMI was calculated as weight (kg)/height (m)2. BMI change during the intervention study was measured as the percentage change compared with baseline. Hours of employment per week, including volunteer work, was assessed by the Physical Activity Scale for the Elderly at 12 months (25,26). Attendance in the exercise group was calculated as the percentage of presence at the group sessions relative to the total amount of group sessions. Family history of breast or colorectal cancer was assessed by a questionnaire at the start of the SHAPE study. A positive family history may indicate motivation for participation in an exercise study such as the SHAPE study.
Baseline characteristics are reported as means, medians, and percentages.
The effect of participation in an exercise intervention study was assessed 1 yr after finishing the study by comparing the intervention and the control groups by a) the Modified Baecke Questionnaire score, b) the number of MET-hours per week spent on at least moderate-intensity activities, c) the percentage of women compliant to the international physical activity recommendation, and d) the physical activity patterns incorporated in daily life. Differences between the groups were tested with the Mann-Whitney U test for two independent samples in case of continuous nonparametric data or the chi-square test in case of percentages of categorical or dichotomous data. Confidence intervals for nonparametric data were computed by bootstrapping techniques (5).
To assess which characteristics best predict future maintenance of activity levels, we applied univariable and multivariable linear regression analyses in the intervention group. The number of MET-hours per week spent on at least moderate-intensity activities was considered the main outcome (dependent continuous variable). Linearity of the univariable association between a continuous predictor and the outcome was examined with restricted cubic spline functions. Nonlinear associations according to the spline function were approximated with a simple transformation, such as the square or the natural logarithm. As a result, baseline activity, BMI change, employment, and group session attendance were included as a linear function for age a square transformation was used.
Predictive factors were selected with backward stepwise selection using a P value of 0.20. Selection of characteristics with a liberal P value (larger than 0.05) intends to select also weaker predictors at the cost of including "noise" variables. These models usually perform better in a new (external) data set (21).
At baseline of the SHAPE study, women in the intervention and control groups were comparable with respect to age (59 yr) and habitual physical activity levels. The Modified Baecke Questionnaire score was 8.3 and 8.8 for the intervention and control groups, respectively, whereas the number of MET-hours per week spent on at least moderate-intensity activities was 4.9 and 4.3, respectively (Table 1).
Of the 189 participants of the SHAPE study, 168 (89%) were recontacted 1 yr after the study, and 142 (75%) of them completed all follow-up questionnaires (Fig. 1). Baseline characteristics of the responders are shown in Table 1. Reasons to refuse participation in the follow-up were mostly lack of motivation, time constraints, medical problems, or personal problems. Comparing the baseline characteristics of women participating in the follow-up and women who did not participate in the follow-up, the follow-up participants had a significantly lower BMI at baseline (26.6 vs 28.0 kg·m−2; intervention and control groups combined) and had a significantly higher Modified Baecke Questionnaire score (8.9 vs 6.8; intervention and control groups combined) and leisure time subscore (5.9 vs 4.2; intervention and control groups combined). This trend was similar within the intervention and the control groups. Baseline characteristics of nonparticipants did not differ between the intervention and the control subjects (data not shown).
Table 2 shows the median values of physical activity measures over time. After 12 months of exercise, the intervention group had a significantly higher Modified Baecke Questionnaire total score (15.2 vs 10.3) and sport score (4.3 vs 0.0) and spent more MET-hours per week on at least moderate-intensity activities compared with the control group (19.8 vs 5.8). Although these differences declined during the 1-yr period after finishing the exercise program, the sport score and the MET-hours per week spent on at least moderate-intensity activities remained significantly higher for the intervention group compared with the control group (sport score: difference of 0.7, 95% CI = 0.0-2.9; MET-hours per week spent on at least moderate-intensity activities: difference of 4.1, 95% CI = 0.3-8.3). Despite the request to maintain their usual activity pattern, the control group showed a slight increase of the Modified Baecke Questionnaire score during the study period because of an increase of time spent on leisure time activities. After finishing the study, the control group further increased its physical activity level, but they remained less active than the intervention group.
Changes in overall physical activity levels between baseline and 12 months are not significantly different between responders and nonresponders. Only the sport score in the intervention group changed significantly more in the group of responders than that in the group of nonresponders (+4.3 vs +2.6; data not shown).
Figure 2A shows the percentage of compliance to the international physical activity recommendation over time for both the intervention and the control groups. One year after the study, the compliance rate in the intervention group was still significantly higher than that in the control group (difference of 23.6%, 95% CI = 6.1-41.0).
Participation in the SHAPE study did not only result in an increase of time spent on sport activities but also led to behavioral changes in everyday life. Figure 2B shows the percentage of women who most often used bicycling or walking as type of transportation in their hometown instead of motorized transportation. The intervention and the control groups show a similar increase after 12 months compared with baseline. In the intervention group, the increase continued to 24 months, whereas in the control group, it leveled off. However, the difference between the groups at 24 months was not statistically significant.
Table 3 shows the characteristics of the women in the intervention group related to their physical activity level 1 yr after finishing the study. The median MET-hours per week spent on at least moderate-intensity activities was high (21.0) among women with higher physical activity levels at baseline. Women who attended less than 50% of the group sessions had the lowest number of MET-hours per week (7.2).
The univariable and the multivariable linear regression analyses are shown in Table 4. The strongest predictors were age, baseline activity, and employment. The level of baseline activity and the hours of employment per week were positively associated with a higher level of physical activity. Age showed a U-shaped association with a minimum around 61 yr. We could therefore describe the association of age as (age − 61)2. The percentage-explained variance of the model was 38.9%, which means that 38.9% of the variation in physical activity level can be explained by this prediction model.
Participation in the SHAPE exercise intervention study resulted in behavioral change in the intervention as well as in the control group 1 yr after finishing the study. Women in the intervention group substantially increased their physical activity level compared with baseline, although they were not able to maintain the physical activity level as high as during the study. The control group slightly increased its level of physical activity both during and after the study but not to the same extent as the intervention group. Besides an increase in time spent on sport activities, behavioral changes in everyday life were also observed. Age, baseline activity, and employment were the strongest predictors of a high level of physical activity in the intervention group 1 yr after finishing the study.
The long-term effects of participation in an exercise study beyond completion of the program have been studied rarely. Two studies from the United States confirm our finding that an exercise program of limited duration can lead to prolonged behavior change. Pereira et al. (15) recontacted postmenopausal women who participated in a randomized walking intervention trial 10 yr after finishing the study. They demonstrated that women originally randomized to the walking intervention maintained higher levels of physical activity than the control group. McAuley et al. (11) conducted a follow-up study 2 and 5 yr after completion of a randomized controlled exercise trial in sedentary older adults aged 60-75 yr. Although physical activity levels were still elevated compared with baseline after 2 yr of follow-up, these levels declined in the next 3-yr period. Participants with a higher physical activity level, a higher self-efficacy, and a more positive effect 2 yr beyond baseline were more likely to remain active after 5 yr. Because of the older-aged population, the authors conclude that participation in an exercise program may postpone the age-related decline of physical activity level.
Knowledge about determinants and barriers to an active lifestyle can be used to design effective intervention programs and to identify potential poor adherers. Earlier research focused on the initiation of exercise rather than on the maintenance after completion of a structured program. In this study, we found that age, baseline activity, and employment are the strongest predictors of exercise level 1 yr after finishing the program. The proportion of explained variance (38.9%) indicates that a substantial part of the variance could be explained by the predictors in the model. Because of a lack of power, we were not able to perform subgroup analysis to study potential effect modification. However, one can imagine that predictors of exercise level might not be the same in subgroups with for instance a different BMI at baseline.
A limitation of this study was that not all previous participants could be included in the follow-up study. There were more nonresponders within the control group, and because nonresponders in this study were less active at baseline, the activity level at 24 months was probably more overestimated in the control group. This could have resulted in a slightly underestimated intervention effect.
The women participating in the SHAPE study were not completely sedentary at baseline. Because bicycling is a common form of transportation in The Netherlands, it is not possible to find a group of women with a very low level of physical activity within this age range. Therefore, bicycling for transportation in their hometown was not taken into account when screening women for eligibility. This should be taken into account when generalizing the results to a non-Dutch population.
The study population for linear regression analysis was relatively small, and information on some potential predictors was lacking. Exercise stage of change, on the basis of the transtheoretical model (16), and self-efficacy, a key construct in social cognitive theory (3), have also been found to be related to the maintenance of physical activity (8,27) beyond termination of a structured exercise program (11). However, measurements of these psychosocial variables were not available at baseline.
Employment, which included volunteer work, should be handled with caution in this population. Because the age range of 50-69 yr also included retirees, participation in employment or volunteer work may be based on different motives and therefore reflect different subject characteristics. Furthermore, we would like to have included time spent on taking care of others (e.g., babysitting, informal care) in the prediction model as another inverse surrogate for spare time. However, the available information about this potential predictor was too limited because the Physical Activity Scale for the Elderly questionnaire did not provide information about duration and frequency of these activities.
The strength of this study was that the follow-up measurement was not announced during the study period. Therefore, behavior could not have been influenced by the upcoming measurements. On the basis of the transtheoretical model (16), maintenance is defined as implementing behavior change for at least 6 months. Therefore, the higher level of physical activity compared with baseline 1 yr after completion of the program suggests that sustained lifestyle changes have been reached. From a public health perspective, it is important to know that offering an intensive exercise program of limited duration to motivated women could lead to sustained behavioral change. The increased level of physical activity in the control group shows that motivation and awareness about the beneficial effects of physical activity could also induce some behavioral change but not to the same extent as in the intervention group.
In conclusion, participation in the SHAPE exercise study resulted in sustained changes of physical activity behavior 1 yr after completion of the study. Age, baseline activity, and employment were the strongest predictors of physical activity level in the intervention group.
This work was supported by the Dutch Cancer Society (grant no. UU 2003-2793).
The results of the present study do not constitute endorsement by the American College of Sports Medicine.
1. Ainsworth BE, Haskell WL, Leon AS, et al. Compendium of physical activities: classification of energy costs of human physical activities. Med Sci Sports Exerc
2. Ainsworth BE, Haskell WL, Whitt MC, et al. Compendium of physical activities: an update of activity codes and MET intensities. Med Sci Sports Exerc
. 2000;32(9 suppl):S498-516.
3. Bandura A. Self-efficacy: toward a unifying theory of behavioral change. Psychol Rev
4. Behavioral Risk Factor Surveillance System [Internet]. Atlanta (GA): U.S. Department of Health and Human Services, Centers for Disease Control and Prevention; [cited 2008 Oct 10]. Available from: http://www.cdc.gov/brfss/
5. Efron B, Tibshirani R. An Introduction to the Bootstrap
. London (UK): Chapman & Hall; 1993. p. 178-201.
6. Kemper HCG, Ooijendijk WTM, Stiggelbout M. Consensus about the Dutch recommendation for physical activity
to promote health (Consensus over de Nederlandse norm gezond bewegen). Tijdschr Gezondheidswet
7. King AC. Interventions to promote physical activity
by older adults. J Gerontol A Biol Sci Med Sci
. 2001;56(Spec No. 2):36-46.
8. Litt MD, Kleppinger A, Judge JO. Initiation and maintenance of exercise behavior in older women: predictors from the social learning model. J Behav Med
9. Marcus BH, Dubbert PM, Forsyth LH, et al. Physical activity
behavior change: issues in adoption and maintenance. Health Psychol
. 2000;19(1 suppl):32-41.
10. Martin KA, Sinden AR. Who will stay and who will go? A review of older adults' adherence to randomized controlled trials of exercise. JAPA
11. McAuley E, Morris KS, Motl RW, Hu L, Konopack JF, Elavsky S. Long-term follow-up
of physical activity
behavior in older adults. Health Psychol
12. Monninkhof EM, Peeters PH, Schuit AJ. Design of the Sex Hormones and Physical Exercise (SHAPE) study. BMC Public Health
13. Monninkhof EM, Velthuis MJ, Peeters PH, Twisk JW, Schuit AJ. Effect of exercise on postmenopausal sex hormone levels and role of body fat: a randomized controlled trial
. J Clin Oncol
14. 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
15. Pereira MA, Kriska AM, Day RD, Cauley JA, LaPorte RE, Kuller LH. A randomized walking trial in postmenopausal women: effects on physical activity
and health 10 years later. Arch Intern Med
16. Prochaska JO, DiClemente CC. Stages and processes of self-change of smoking: toward an integrative model of change. J Consult Clin Psychol
17. Rhodes RE, Martin AD, Taunton JE, Rhodes EC, Donnelly M, Elliot J. Factors associated with exercise adherence among older adults. An individual perspective. Sports Med
18. Schutzer KA, Graves BS. Barriers and motivations to exercise in older adults. Prev Med
19. Seefeldt V, Malina RM, Clark MA. Factors affecting levels of physical activity
in adults. Sports Med
20. StatLine [Internet]. Den Haag/Heerlen (The Netherlands): Statistics Netherlands; [cited 2008 Oct 10]. Available from: http://statline.cbs.nl/
21. Steyerberg EW, Eijkemans MJ, Harrell FE Jr, Habbema JD. Prognostic modeling with logistic regression analysis: in search of a sensible strategy in small data sets. Med Decis Making
22. Trost SG, Owen N, Bauman AE, Sallis JF, Brown W. Correlates of adults' participation in physical activity
: review and update. Med Sci Sports Exerc
23. U.S. Department of Health and Human Services. Physical Activity and Health: A Report of the Surgeon General
. Atlanta (GA): U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion; 1996. p. 3-8.
24. Voorrips LE, Ravelli AC, Dongelmans PC, Deurenberg P, van Staveren WA. A physical activity
questionnaire for the elderly. Med Sci Sports Exerc
25. Washburn RA, McAuley E, Katula J, Mihalko SL, Boileau RA. The physical activity
scale for the elderly (PASE): evidence for validity. J Clin Epidemiol
26. Washburn RA, Smith KW, Jette AM, Janney CA. The Physical Activity
Scale for the Elderly (PASE): development and evaluation. J Clin Epidemiol
27. Williams DM, Lewis BA, Dunsiger S, et al. Comparing psychosocial predictors of physical activity
adoption and maintenance. Ann Behav Med