Click on the links below to access all the ArticlePlus for this article.
Please note that ArticlePlus files may launch a viewer application outside of your web browser.
Although physical activity is one of the best documented modifiable factors that can prevent premature mortality in the general population,1–6 it is still unclear how cessation of physical activity or changes in activity level from adolescence to adulthood affect mortality. It has not been convincingly documented whether physical activity also is effective among women under the age of 60 years who have inherently low mortality. In a large population-based cohort study, we attempt to address these issues.
The Women's Lifestyle and Health cohort has been described in detail previously.2,3 Altogether, 57,582 Norwegian women age 34–49 years and 49,259 Swedish women age 30–49 years at recruitment in 1991–1992 were prospectively followed with regard to vital status through year 2003. The source population in Norway was the entire country, whereas in Sweden, the source population was all women residing in the Uppsala Health Care Region. From the initial cohort, we excluded 5 women because of lack of information on vital status, 17 women who had emigrated before the start of follow-up, 405 women who reported a history of cardiovascular disease (myocardial infarction and stroke), and 1539 women who had a prior diagnosis of cancer. We also excluded 5776 women who were missing information on all 3 physical activity variables. Thus, the final analysis was based on data from 99,099 women.
Physical Activity and Other Characteristics
Data were collected in a baseline questionnaire administered at cohort enrollment. Women rated their overall level of physical activity (in the household, occupational, and recreational physical activity) at age 14, age 30, and at enrollment. In Sweden, women ranked their level of physical activity on a 5-point scale with 1 = sedentary (mainly sitting), 3 = moderate physical activity (a few walks a week), and 5 = vigorous physical activity (sports/jogging several times a week). In Norway, women ranked activity on a scale of 1 (sedentary) to 10 (vigorous). We collapsed the 10-level Norwegian scale into 5 levels to be comparable with the Swedish scale. Using the self-reported data, we further categorized individuals on changes in physical activity over time. We also dichotomized physical activity into women who participated in no or low physical activity, and those who participated in moderate, high, or vigorous activity for each time point. Information on demographic characteristics, education, lifelong history of cigarette smoking, alcohol consumption, height, current weight, and medical history (diagnosis of cancer, myocardial infarction, stroke, hypertension, diabetes) was also collected at baseline.
Follow up of the cohort was achieved through linkages with nationwide population registers rather than the death register. Although the former provides no information on causes of death, it is continuously updated and it allowed almost 3 additional years of follow-up compared with the death register. An individually unique national registration number allowed linkage of the data from the cohort with these registers for virtually complete follow-up with respect to death and emigration. The population registers were updated through January 2003 for Norway and through June 2003 for Sweden. The start of follow-up was the date of return of the questionnaire during 1991–1992. Observation time was calculated from date of entry into the cohort until the occurrence of death, censoring on account of emigration, or end of the observation period, whichever occurred first.
The relation between physical activity and total mortality was assessed using time-to-event analyses. The unadjusted cumulative risk of death was estimated using Kaplan-Meier methods. Proportional hazard models were used to estimate the hazard ratio (HR) and 95% confidence interval (CI) controlling for potential confounders. Low levels of physical activity were used as the reference. The following covariates were considered as potential confounders: age (continuous variable), years of education (continuous variable), body mass index, alcohol intake per day (0, <1 drink, 1 or more; in which 1 drink is equivalent to 10 g of ethanol), smoking status (current, former, never), mean number of cigarettes smoked per day while smoking (continuous variable), number of years of smoking (continuous variable), and country of origin (Norway or Sweden). We tried to disentangle the effects of physical activity at baseline from past activity levels by mutually adjusting for physical activity at each time point in one model. Finally, we assessed the linearity of the continuous measures on the log scale.
In the first set of models, we stratified by potential effect modifiers (age, smoking status, education, body mass index [BMI], country of origin). In a second set, we modeled interaction terms and calculated log-likelihood ratio tests. Besides excluding a priori women with a diagnosis of cancer and cardiovascular disease before cohort enrollment, we conducted several subanalyses within the data to assess possible bias associated with preexisting illness.
The responsible ethical committees in both countries approved the study, and all women gave informed consent.
During an average 11.4 years of follow-up, 1313 women died. The cumulative incidence of death in the cohort per 1000 was 13.2 in Norway and 13.3 in Sweden (Table 1). Compared with the Swedish women, the Norwegian women were, on average, more likely to have ever smoked and less likely to be physically active. Self-reported levels of physical activity were higher at age 14 than at either age 30 or enrollment (Fig. 1).
Table 2 compares selected characteristics of the women by physical activity level at cohort enrollment. Less active women tend to have higher BMI, to more likely be smokers (and, among smokers, to smoke more cigarettes per day and to have smoked for a longer period of time), and to be somewhat less educated.
Figure 2 shows the cumulative risk of death by physical activity at enrollment. We found a striking protective effect of physical activity evident even at the moderate activity level. Ten years after enrollment, the cumulative mortality was 1.55% among women with no or low activity at enrollment compared with 0.87% among those with high or vigorous activity. After adjustment for potential confounders, the risk of death decreased monotonically over categories of physical activity at enrollment (Table 3). Compared with women in the lowest category, those in the highest category had half the risk of dying. In contrast, there was no convincing dose response trend for physical activity either at age 14 or age 30.
Illness at enrollment may have affected women's physical activity level. However, excluding women with self-reported fair or poor health, or excluding deaths occurring during the first 2 or 3 years of follow-up, did not change the results. Excluding women with hypertension or diabetes revealed a slightly stronger inverse association compared with analysis of the entire cohort (data available with the online version of this article).
Next we examined the effect of changes in physical activity levels over time (Table 4). The data indicate that physical activity at enrollment is the crucial determinant of subsequent mortality, whereas physical activity at ages 14 and 30 is less relevant except to the extent that it predicts physical activity later in life. Thus, among women who were active at enrollment, it appears irrelevant whether they were active or inactive at ages 14 or 30. Both are consistent with more than a 20% reduction in mortality. These findings are consistent with the data in Table 3, in which controlling for physical activity at enrollment eliminated the association of physical activity at age 14 or at age 30 with mortality during the follow-up period.
The effect of physical activity at enrollment on overall mortality was consistent across age, history of smoking, educational level, and country of origin (P for interaction ranged from 0.27 to 0.97). However, the effect of physical activity was strongest among women with a BMI <18.5 kg/m2 (P for interaction 0.03); among these lean women, mortality was higher for those with no or low activity than for those who were physically active. Among women with a BMI above 29.9 kg/m2, mortality was elevated regardless of physical activity level (Fig. 3).
In this large, population-based, prospective investigation among women below the age of 60, physical activity (as reported at cohort enrollment) substantially reduced overall mortality. This beneficial effect was evident even for those with low physical activity levels compared with sedentary women. When physical activity at enrollment was controlled for, physical activity at ages 14 and 30 was convincingly unrelated to mortality. Hence, our data could indicate that any protective effect of early life physical activity is moderated through its relationship with physical activity in later life. As a corollary, it is apparently beneficial to initiate or increase physical activity at any stage in life without being handicapped by inactivity in the past. Only a few previous investigations, primarily among men, have directly addressed this issue.4–9
These findings extend previous observations, predominantly in men10 and older women.7,11,12 Hence, physical activity has a protective effect that seems consistent across sex, age groups, and populations. Physical activity reduces weight gain,13,14 improves metabolic profiles,14 has beneficial effects on blood pressure,15 increases insulin sensitivity,16,17 and may have a positive effect on the immune system18; all of the these physiological changes are important for the prevention of chronic disease and premature death. Our finding that an increase in physical activity can further reduce overall mortality, even in populations with very low overall mortality rates (such as Norwegian and Swedish women below age 60), underlines the importance of physical activity in the context of primary prevention and indicates that this factor is a largely independent contributor to mortality reduction. However, trials of interventions to increase physical activity are few and hampered by methodologic limitations.18
A weakness of our study is the reliance on self-assessment of physical activity. Elderly women have been shown to overestimate their activity level earlier in life,19 which would result in an underestimation of the effect of physical activity on mortality. Self-assessment can introduce error in exposure ascertainment, but this error is presumably uncorrelated to outcome. Furthermore, the 5-point physical activity scale did not assess frequency, duration, or intensity of total physical activity—information useful for translation into public health recommendations. Finally, given the apparent time-limited effect of physical activity on mortality, our one-time assessment at the start of enrollment may have misclassified person-time over the course of the study. Given that women appear to become more inactive over time, our estimates may reflect an underestimate of physical activity and mortality. However, given the strong correlation between activity levels over time, it is unlikely that the estimates would differ to any great extent.
A more serious concern is that an unascertainable health condition may have introduced confounding by both reducing physical activity and increasing the risk of death. Such a confounding effect is difficult to remove analytically.12 Ascertainable conditions such as cancer and cardiovascular events have been excluded from the study base; exclusion of hypertension and diabetes, self-rated poor health, and deaths during the first few years of follow-up did not substantially affect our results. We adjusted for BMI, but unmeasured variables such as high fruit and vegetable intake, low fat intake, and low lipid levels may have resulted in residual confounding. Nevertheless, given the robustness of our study findings to extensive control of covariates, it is unlikely that our results would be qualitatively different had we been able to control for other factors. In conclusion, physical activity among relatively young women appears to substantially reduce mortality with benefits regardless of exercise habits earlier in life.
1. Hahn RA, Teutsch SM, Rothenberg RB, et al. Excess deaths from nine chronic diseases in the United States, 1986. JAMA
2. Kumle M, Weiderpass E, Braaten T, et al. Use of oral contraceptives and breast cancer risk: the Norwegian–Swedish Women's Lifestyle and Health Cohort Study. Cancer Epidemiol Biomarkers Prev
3. Veierod MB, Weiderpass E, Thorn M, et al. A prospective study of pigmentation, sun exposure, and risk of cutaneous malignant melanoma in women. J Natl Cancer Inst
4. Paffenbarger RS Jr, Hyde RT, Wing AL, et al. The association of changes in physical-activity level and other lifestyle characteristics with mortality among men. N Engl J Med
5. Bijnen FC, Feskens EJ, Caspersen CJ, et al. Baseline and previous physical activity in relation to mortality in elderly men: the Zutphen Elderly Study. Am J Epidemiol
6. Sherman SE, D'Agostino RB, Silbershatz H, et al. Comparison of past versus recent physical activity in the prevention of premature death and coronary artery disease. Am Heart J
7. Gregg EW, Cauley JA, Stone K, et al. Relationship of changes in physical activity and mortality among older women. JAMA
8. Blair SN, Kohl HW 3rd, Barlow CE, et al. Changes in physical fitness and all-cause mortality. A prospective study of healthy and unhealthy men. JAMA
9. Erikssen G, Liestol K, Bjornholt J, et al. Changes in physical fitness and changes in mortality. Lancet
10. Lee IM, Skerrett PJ. Physical activity and all-cause mortality: what is the dose–response relation? Med Sci Sports Exerc
11. Kushi LH, Fee RM, Folsom AR, et al. Physical activity and mortality in postmenopausal women. JAMA
12. Rockhill B, Willett WC, Manson JE, et al. Physical activity and mortality: a prospective study among women. Am J Public Health
13. Fogelholm M, Kukkonen-Harjula K. Does physical activity prevent weight gain—a systematic review. Obes Rev
14. Thune I, Njolstad I, Lochen ML, et al. Physical activity improves the metabolic risk profiles in men and women: the Tromso Study. Arch Intern Med
15. Godsland IF, Leyva F, Walton C, et al. Associations of smoking, alcohol and physical activity with risk factors for coronary heart disease and diabetes in the first follow-up cohort of the Heart Disease and Diabetes Risk Indicators in a Screened Cohort study (HDDRISC-1). J Intern Med
16. Kelley DE, Goodpaster BH. Effects of physical activity on insulin action and glucose tolerance in obesity. Med Sci Sports Exerc
17. van Baak MA, Borghouts LB. Relationships with physical activity. Nutr Rev
18. IARC Handbooks of Cancer Prevention: Weight Control and Physical Activity,
vol 6. Lyon: International Agency for Research on Cancer; 2002.
19. Lissner L, Potischman N, Troiano R, et al. 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