Purpose: Few epidemiological studies have been conducted among middle-aged women on long-term total and specific physical activity (PA) trends. We studied in a cross-sectional setting the relationship of self-reported total daily PA with age and calendar time.
Methods: In a population-based cohort of 38,988 women aged 49-83 yr in central Sweden, information was collected on physical activity, such as work or occupation, household work, walking or bicycling, exercise, watching TV or reading, and other lifestyle factors through a self-administered questionnaire. Total and specific daily PA levels at ages 15, 30, and 50 yr were recalled retrospectively and measured as metabolic equivalents (MET·h·d−1).
Results: Total PA level linearly decreased with calendar time in all three age groups (slope for 5-yr change in calendar time among those 15 yr of age = −0.82; 95% confidence interval (CI), −0.86 to −0.78; among those 30 yr of age=−0.42; 95% CI, −0.45 to −0.38; and among those 50 yr of age = −0.62; 95% CI, −0.66 to −0.58). High-intensity activities such as walking or bicycling decreased by 0.21 MET·h·d−1 (95% CI, −0.22 to −0.20) every 5-calendar-year change among adolescents between the 1930s and 1960s. Total activity level decreased in all age groups by an average of approximately 3MET·h−1·d−1, corresponding to approximately 45 min of brisk walking.
Conclusions: Our results suggest that intervention efforts aimed at engaging in healthful amounts of physical activity are needed throughout the life cycle.
1Division of Nutritional Epidemiology, Institute of Environmental Medicine, Karolinska Institutet, Stockholm, SWEDEN; 2Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm, SWEDEN; 3Department of Epidemiology and Biostatistics, Arnold School of Public Health, University of South Carolina, Columbia, SC; and 4Department of Biostatistics, Harvard School of Public Health, Boston, MA
Address for correspondence: Nicola Orsini, Division of Nutritional Epidemiology, Institute of Environmental Medicine, Karolinska Institutet, P.O. Box, SE-171 77, Stockholm, Sweden; E-mail: firstname.lastname@example.org.
Submitted for publication May 2005.
Accepted for publication August 2005.
Regularly performed physical activity (PA) is associated with a reduced risk of coronary heart disease (CHD), stroke, type 2 diabetes, some types of cancer, osteoporosis, and all-cause mortality (7,17,25,27). Recently, interest has increased in the trends of leisure-time PA (3,4,10,12,19,22,31), yet trend information on total PA is lacking in most countries. Few studies have examined participation in different types of PA across multiple time points among middle-aged women (11,13). The Women's Health Initiative cohort study assessed retrospectively, for ages 18, 35, and 50 yr, moderate and vigorous leisure activity (13).
The omission of work or occupational and household activity, for women in particular, may underestimate total PA and result in women misclassified as physically not very active when such was not true (11). The present study, therefore, provides further data on total PA and specific types of PA, such as household, work or occupation, exercise, walking, and inactivity (sitting, watching TV or reading), recalled at various times in the life span.
The objective of this study was to analyze the relationship of self-reported total daily PA with age and calendar time (the last 60 yr of the 20th century), in a large population-based study of Swedish women born during the period from 1914 to 1948.
MATERIALS AND METHODS
The Swedish Mammography Cohort (SMC87) was established by introduction of a population-based mammography screening program in central Sweden, 1987-1990. In Västmanland (41,786) and Uppsala (48,517) counties all women aged 40-75 yr received a mailed invitation to have a mammography, together with a six-page food frequency questionnaire. The response rate was 74%. Women were excluded because of missing or wrong identification, lack of date of answering the questionnaire, missing date for moving out of the study area, or missing date of death; prevalent cancer restricted the baseline cohort to 61,433 subjects. In 1997, a second extended questionnaire (SMC97) asked for a more detailed food frequency pattern and information on weight history, height, smoking, alcohol consumption, medical history, educational level, and employment status. Furthermore, detailed questions about physical activity were included. The questionnaire was mailed to 56,054 women who participated in the first wave and who were alive and did not move out from the study area. The response rate was 70% in the SMC97.
Assessment of Physical Activity
Measurement of PA was based on a self-administered questionnaire. Five types of activities (e.g., home or household work, walking or bicycling, work or occupation, TV or reading) and exercise were recalled retrospectively at ages 15, 30, and 50 yr. Six predefined activity time categories were listed for household work (from <1 h·d−1 to >8 h·d−1) and for walking or bicycling (from hardly ever to >1.5 h·d−1). For work or occupation, activity level was estimated by six types of work (from mostly sitting to heavy manual labor). Leisure time inactivity (e.g., watching TV or reading) was measured by five prespecified time categories (from <1 h·d−1 to >6 h·d−1). Leisure time activity exercise was measured by five prespecified time categories (from <1 h·wk−1 to >5 h·wk−1). The intensity of activities used for calculations was defined as multiples of the metabolic equivalent (MET, kcal·kg−1·h−1) of sitting quietly (resting metabolic rate (RMR)) for 1 h, corresponding to specific activities (1,2). Daily PA (MET·h·d−1) for specific types of activities was estimated by multiplying MET·h·d−1 by self-reported time (h) (Table 1). We estimated a total daily PA score at ages 15, 30, and 50 yr by adding the specific activities together. In addition, we estimated nonoccupational PA score by excluding work or occupational activity level from the total PA score. The PA score at baseline has been validated against a 14-d activity diary in a group of Swedish men aged 44-78 yr and was shown to estimate total present physical activity satisfactorily (23). The study was approved by the ethics committees at the Uppsala University and the Karolinska Institutet. Obtaining written information about the study and completion of the questionnaire were considered to imply informed consent.
Changes in PA of subjects were analyzed at ages 15, 30,and 50 yr. The following characteristics (at baseline, 1997) were included as covariates: age, body mass index (BMI = weight (kg) per height (m2)), smoking, drinking status, and educational level. Current age was grouped into seven categories (49-54, 55-59, 60-64, 65-69, 70-74, 75-79, and 80-83 yr), whereas BMI was grouped in three classes (<25, 25-29, and ≥ 30 kg·m−2). Women were also categorized according to their smoking and drinking status (never, former, and current) and educational level (primary, high school, and university). All covariates were categorical and modeled using dummy variables with the lowest level as referent group.
Each person contributed self-reported PA up to three time periods. A random-intercept was introduced for the subject-specific error (18). The random-effects models were compared with generalized estimating equations methods, testing for different possible structures of the within-subject correlation.
Temporal trends in changes in total, occupational and nonoccupational, and specific daily PA were analyzed separately for women at ages 15, 30, and 50 yr using multivariate linear regression models, adjusting for BMI, smoking, drinking status, and educational level. All statistical analyses and graphs were performed with Stata, version 9.0 (30).
Physical Activity by Age Group
Our study cohort SMC97 included 38,988 women from central Sweden born between 1914 and 1948, mean age 62yr (SD ± 9). The proportions of younger (<52 yr), middle-aged (52-62 yr), and older (>62 yr) women were 14, 42, and 44%, respectively. Crude means of total daily PA scores estimated at ages 15, 30, and 50 yr were 41.39 (95% CI, 41.32-41.45), 45.03 (95% CI, 44.97-45.09), and 44.37 (95% CI, 44.31-44.43) MET·h·d−1, respectively. Means of self-reported BMI at ages 20, 30, and 50 yr were 20 (SD ± 3), 21 (SD ± 3), and 23 (SD ± 4) k·m−2, respectively. After taking into account the correlation between observations (at ages 15, 30, and 50 yr) for the same woman and adjusting for BMI, smoking, drinking status, educational level, and age at baseline, the mean total activity scores were 42.17 (95% CI, 41.98-42.35), 45.74 (95% CI, 45.56-45.93), and 44.87 (95% CI, 44.69-45.05) MET·h·d−1, respectively. The adjusted means of nonoccupational PA score estimated at ages 15, 30, and 50 yr were 30.08 (95% CI, 30.00-30.15), 31.36 (95% CI, 31.29-31.44), and 31.37 (95% CI, 31.30-31.45) MET·h·d−1, respectively.
Temporal Trends of Physical Activity
The crude means of total PA score in MET-hours per day were 13.6% lower in 1962-1966 compared with 1929-1936, 44.9 (SD, 8.6), for those 15 yr of age; 7.5% lower in 1977-1981 compared with 1947-1951, 47 (SD, 6.3) for women 30 yr of age; 8.8% lower 1992-1997 compared with 1967-1971, 46.9 (SD, 7), for women 50 yr of age.
Table 2 shows the estimates of changes over calendar time of adjusted means of total PA, work or occupation, household work, walking or bicycling, and exercise by age group. Total PA level decreased by 10% (4.7 MET·h·d−1) at age 15 yr between 1930 and 1960, 5% (2.4 MET·h·d−1) at age 30 yr between 1950 and 1980, and 6% (2.9 MET·h·d−1) at age 50 yr between 1970 and 1990; and decreased in all age groups by an average of approximately 3 MET·h·d−1. Among women aged 15 yr, the highest decrease observed was in MET-hours per day for household work (−49%), exercise (−34%), walking or bicycling (−29%), and work or occupation (−20%). Among women aged 30 yr, the highest decrease observed was in MET-hours per day, for walking or bicycling (−45%), exercise (−24%), household work (−17%), and work or occupation (−7%). Among women aged 50 yr, the highest decrease observed was in MET-hours per day, for household work (−36%), walking or bicycling (−29%), exercise (−21%), and work or occupation (−9%). Smooth temporal trends for specific types of PA at ages 15, 30, and 50 yr are shown in Figure 1. Occupational and household work activities account for the largest part of the total daily PA at all ages.
Table 3 reports slopes for 5-calendar-year change of total and specific PA by age group. Total PA levels were decreasing gradually over calendar time at age 15 yr (slope for 5-yr change = −0.82), at age 30 yr (−0.42), and at age 50 yr (−0.62).
Home or household work PA was decreasing, on average, 0.95 MET·h·d−1 at age 15 yr, and 1.21 MET·h·d−1 at age 50 yr, for every 5-yr calendar period. High-intensity activities among adolescents (e.g., walking or bicycling) decreased by 0.21 MET·h·d−1 for every 5-yr period between 1930 and 1960.
From the 1950s, physical inactivity during leisure time (e.g., watching TV or reading) increased among those 15 yr (6%) and 30 yr of age (26.5%) when TV was introduced (Fig. 1).
In this large population-based study of middle-aged and elderly women, we observed a decreasing trend of total PA by calendar time during the last 60 yr of the 20th century. The most striking decreases in total PA were observed in those 15 yr of age between 1930 and 1960 and in the age group of women 50 yr of age between 1960 and 1990. Responsible for these decreases in both groups was home or household work, followed by work or occupation, walking or bicycling, and exercise.
We also observed that, on average, during the transition period from adolescence (age 15 yr) to young adulthood (age 30 yr) women were increasing total PA, and this change was largely explained by an increase in home or household activity.
The strength of this investigation is the large size of our population-based study that well represents the Swedish female population of corresponding age range, regarding distribution of education and relative weight (20). We assume, therefore, that a potential self-selection according to sociodemographic variables should have a negligible, if any, impact on the estimated means and trends of PA. The quantitatively estimated total daily PA is based on a validated instrument that was shown to estimate total PA satisfactorily as measured by correlation with activity records and the reproducibility of total PA scores at ages 15, 30, and 50 yr was relatively high (23).
A limitation of our estimates is that we assumed that all individuals perform the same type of activities at the same intensity level. A range of intensity levels exists for each type of PA, however, which could change over time. It should also be noticed, however, that the guidelines from the compendium of physical activities by Ainsworth et al. (2) were not developed to determine precise energy cost of PA in individuals, but instead to provide an activity classification system that standardize the MET intensities used in research.
In the interpretation of our results, we need to consider the possibility of recall bias in the estimate of the distant past activities, some 30-60 yr ago. The retrospective information in our data is based on the participants' ability to remember daily activities at ages 15, 30, and 50 yr. Results from two prospective studies, assessing recall bias of PA in the distant past (32-35 yr ago) are not consistent (14,21).Falkner et al. (14) found that distant leisure-time activities were recalled at a nonsignificantly higher level by middle-aged and older women (≥52 yr) and at a significantly lower level by younger women (<52 yr). In our study, however, we had only 14% of women younger than 52 yr. Lissner et al. (21) found that 44% of the Swedish women aged 70-92 yr had good recall, and 49% were more likely to overestimate their leisure-time PA 32 yr ago. In American women of a similar age as those in our study, the recall of past weekday activities (occupational and home or household work) after 32-35 yr was satisfactory (intraclass correlation with the original reports was 0.43-0.45) and age did not influence the quality of recall (15). These findings, if also true in our study population, would suggest that our PA estimates for the two major contributors of total PA (occupational and household PA) were also estimated with a satisfactory quality and were not likely to be susceptible to recall bias. Other studies assessing recall of activity levels in the distant past have been based on shorter intervals, 1-10 yr, (6) and 2-3 yr (29). A prospective collection of data across six decades (between 1930 and 1990) captured in our study would be difficult to obtain prospectively. Although a prospective design is desirable, retrospective study can nonetheless contribute useful and reliable information in this field (6,8,32).
The steep decrease in home or household work at age 15 yr between 1930 and 1960 may be attributed to an increased proportion of girls attending school and to a decrease of their help in household activities. Decreasing trends of household work at ages 30 and 50 yr may be explained by systematic introduction over time of household appliances and equipment for indoor and outdoor activities. The decrease in work or occupational activity is likely to be explained by the change in the work profile; many manual jobs became less demanding and sedentary jobs increased. Decreased level of walking or bicycling likely can be attributed to motorization of daily transports. Introduction of television during the 1950s partially explains the higher levels of inactive leisure time. Our findings are in agreement with studies that indicate decreasing trend of moderate and vigorous leisure-time physical activities in the life transition from adolescence to young adulthood (9,13). The overall impact of life events (e.g., marriage, having one or more children) on total PA and in particular home or household activity, however, is not yet understood (9).
Temporal trends of leisure-time PA among women have been investigated previously (3,10,16,19,22), but very few studies have investigated total and specific daily PA levels recalled in the distant past (4,28). Furthermore, because of different types of measures and definitions of PA, it is difficult to compare results. A study of a population age 30-59 yr in eastern Finland showed trends for both leisure-time PA and for occupational PA and commuting PA between 1972 and 1997 (4). In agreement with our findings, the authors reported a decrease of high occupational PA because of change in the work profile, namely to an increased proportion of white collar workers. Commuting PA continually decreased since 1972 as well (4). Occupational and leisure-time PA were assessed during the years 1964-1992 in the general Danish population, showing a decreasing trend of physical activity at work in both genders and all age groups (28). A study on secular trends in leisure-time PA, from 1958 to 1998 among men and women of the Baltimore Longitudinal Study of Aging (31) shows only a modest increase, limited to men, in self-reported leisure-time PA over the past four decades, despite numerous public recommendations to promote PA.
Our study can be compared with a large population-based study of Swedish men aged 45-79 yr, which also shows negative trends in total PA observed by age and calendar time, from 1930 to 1990 (24). Our finding in women that total PA was increasing in young adulthood between ages 15 and 30 yr was not observed among men. This may be explained by gender differences regarding home or household work PA (5).
The PA recommendations from the Centers for Disease Control and Prevention and the American College of Sports Medicine emphasize the importance of the total amount of activity performed daily rather than the specific manner in which it is performed (26). In our data, total daily activity level decreased at age 15 (from 1930 to 1960), 30 (from 1950 to 1970), and 50 yr (from 1960 to 1990) by an average of approximately 3 MET·h·d−1 corresponding to approximately 45 min of brisk walking or gardening. Therefore, intervention efforts aimed at engaging or maintaining healthful amounts of physical activity are needed throughout the life cycle. Furthermore, despite the difficulties in assessing long-term PA patterns with precision, these patterns may be important in chronic disease research (6).
In summary, this study provides a useful description of temporal changes in total and specific daily PA among Swedish women. We observed a decreasing trend of total daily PA, because of decreases in occupational and household activities as well as of walking or bicycling and exercise in all age groups by calendar time during the last 60 yr of the 20th century.
This study was supported by the Swedish Research Council, longitudinal studies, the Swedish Cancer Society, and the Swedish Foundation for International Cooperation in Research and Higher Education (STINT).
1. Ainsworth, B. E., W. L. Haskell, A. S. Leon, et al. Compendium of physical activities: classification of energy costs of human physical activities. Med. Sci. Sports Exerc.
2. Ainsworth, B. E., W. L. Haskell, M. C. Whitt, et al. Compendium of physical activities: an update of activity codes and MET intensities. Med. Sci. Sports Exerc.
3. Anderssen, N., D. R. Jacobs, Jr., S. Sidney, et al. Change and secular trends in physical activity patterns in young adults: a seven-year longitudinal follow-up in the Coronary Artery Risk Development in Young Adults Study (CARDIA). Am. J. Epidemiol.
4. Barengo, N. C., A. Nissinen, J. Tuomilehto, and H. Pekkarinen. Twenty-five-year trends in physical activity of 30- to 59-year-old populations in eastern Finland. Med. Sci. Sports Exerc.
34: 1302-1307, 2002.
5. Bennett, K. M. Gender and longitudinal changes in physical activities in later life. Age Ageing
27(Suppl. 3):24-28, 1998.
6. Blair, S. N., M. Dowda, R. R. Pate, et al. Reliability of long-term recall of participation in physical activity by middle-aged men and women. Am. J. Epidemiol.
7. Blair, S. N., H. W. Kohl, 3rd, R. S. Paffenbarger, Jr., D. G. Clark, K. H. Cooper, and L. W. Gibbons. Physical fitness and all-cause mortality. A prospective study of healthy men and women. JAMA
8. Bowles, H. R., S. J. FitzGerald, J. R. Morrow, Jr., A. W. Jackson, and S. N. Blair. Construct validity of self-reported historical physical activity. Am. J. Epidemiol.
9. Brown, W. J., and S. G. Trost. Life transitions and changing physical activity patterns in young women. Am. J. Prev. Med.
10. Caspersen, C. J., and R. K. Merritt. Physical activity trends among 26 states, 1986-1990. Med. Sci. Sports Exerc.
11. Chasan-Taber, L., J. B. Erickson, J. W. McBride, P. C. Nasca, S. Chasan-Taber, and P. S. Freedson. Reproducibility of a self-administered lifetime physical activity questionnaire among female college alumnae. Am. J. Epidemiol.
12. DuBose, K. D., K. A. Kirtland, S. P. Hooker, and R. M. Fields. Physical activity trends in South Carolina, 1994-2000. South Med. J.
13. Evenson, K. R., S. Wilcox, M. Pettinger, R. Brunner, A. C. King, and A. McTiernan. Vigorous leisure activity through women's adult life: the Women's Health Initiative Observational Cohort Study. Am. J. Epidemiol.
14. Falkner, K. L., S. E. McCann, and M. Trevisan. Participant characteristics and quality of recall of physical activity in the distant past. Am. J. Epidemiol.
15. Falkner, K. L., M. Trevisan, and S. E. McCann. Reliability of recall of physical activity in the distant past. Am. J. Epidemiol.
16. Jacobs, Jr., D. R., L. P. Hahn, A. R. Folsom, P. J. Hannan, J. M. Sprafka, and G. L. Burke. Time trends in leisure-time physical activity in the upper midwest 1957-1987: University of Minnesota studies. Epidemiology
17. Kushi, L. H., R. M. Fee, A. R. Folsom, P. J. Mink, K. E. Anderson, and T. A. Sellers. Physical activity and mortality in postmenopausal women. JAMA
18. Laird, N. M., and J. H. Ware. Random-effects models for longitudinal data. Biometrics
19. Lee, I. M., R. S. Paffenbarger, Jr., and C. C. Hsieh. Time trends in physical activity among college alumni, 1962-1988. Am. J. Epidemiol.
20. Lissner, L., S. E. Johansson, J. Qvist, S. Rossner, and A. Wolk. Social mapping of the obesity epidemic in Sweden. Int. J. Obes. Relat. Metab. Disord.
21. Lissner, L., N. Potischman, R. Troiano, and C. Bengtsson. Recall of physical activity in the distant past: the 32-year follow-up of the Prospective Population Study of Women in Goteborg, Sweden. Am. J. Epidemiol.
22. Marti, B., J. T. Salonen, J. Tuomilehto, and P. Puska. 10-year trends in physical activity in the eastern Finnish adult population: relationship to socioeconomic and lifestyle characteristics. Acta Med. Scand.
23. Norman, A., R. Bellocco, A. Bergstrom, and A. Wolk. Validity and reproducibility of self-reported total physical activity-differences by relative weight. Int. J. Obes. Relat. Metab. Disord.
24. Norman, A., R. Bellocco, F. Vaida, and A. Wolk. Age and temporal trends of total physical activity in Swedish men. Med. Sci. Sports Exerc.
25. Paffenbarger, Jr., R. S., and I. M. Lee. Intensity of physical activity related to incidence of hypertension and all-cause mortality: an epidemiological view. Blood Press. Monit.
26. Pate, R. R., M. Pratt, S. N. Blair, 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
27. Rockhill, B., W. C. Willett, J. E. Manson, et al. Physical activity and mortality: a prospective study among women. Am. J. Public Health
28. Sjol, A., K. K. Thomsen, M. Schroll, and L. B. Andersen. Secular trends in acute myocardial infarction in relation to physical activity in the general Danish population. Scand. J. Med. Sci. Sports
29. Slattery, M. L., and D. R. Jacobs, Jr. Assessment of ability to recall physical activity of several years ago. Ann. Epidemiol.
30. StataCorp. Stata Statistical Software: Release 9.
College Station, TX, 2005.
31. Talbot, L. A., J. L. Fleg, and E. J. Metter. Secular trends in leisure-time physical activity in men and women across four decades. Prev. Med.
32. Winters-Hart, C. S., J. S. Brach, K. L. Storti, J. M. Trauth, and A. M. Kriska. Validity of a questionnaire to assess historical physical activity in older women. Med. Sci. Sports Exerc.
Keywords:©2006The American College of Sports Medicine
EXERCISE; HOUSEHOLD; CROSS-SECTIONAL; LONG-TERM; CALENDAR TIME