Physical activity is undertaken in different contexts or domains: at home, during work, for transportation, and for sport or exercise. Although a large body of evidence suggests an independent dose-response association between measures of physical activity and mortality (7,13,33), there is considerable heterogeneity in how physical activity was assessed in these studies and in the nature of the populations studied, with some being defined by occupation, thus limiting variation in that specific domain. Two previous European studies (4,5) have investigated the domain-specific associations of physical activity with mortality. However, none of these studies provided data on associations between physical activity undertaken at home and mortality. This information may be important because activities performed in different domains of life are likely to differ between men and women, which may affect the physical activity disease associations. A recent study from China suggested an independent association between domestic physical activity and mortality (23) but only in women.
Recently, we reported a 32% lower all-cause mortality risk among the active compared with the inactive participants in the EPIC-Norfolk study using a simple overall index of physical activity that was calculated from a questionnaire administered at baseline (19). In a later phase of the EPIC-Norfolk project, we collected more detailed information on domain-specific activity (34), which has now allowed us to extend our previous observations by examining the separate associations of each domain of physical activity with all-cause and cardiovascular mortality.
Participants and methods.
The European Prospective Investigation into Cancer (EPIC) study is a prospective cohort study designed to investigate the etiology of major chronic diseases. From 1993 to 1997, EPIC-Norfolk (UK) recruited a population-based cohort of 25,639 men and women aged 45-79 yr, identified from participating general practice lists. Between January 1998 and October 2000, 14,905 of these attended the second health check and completed the EPIC Physical Activity Questionnaire (EPAQ2) (31). Written informed consents were obtained from the participants, and The Norwich Local Research Ethics Committee approved the study. Detailed descriptions of the recruitment and study methodology have been reported elsewhere (10).
Physical activity assessment.
Physical activity was assessed using the self-completed EPAQ2 questionnaire that collects data on past year's physical activity behaviors in a disaggregated way so that the information may be reaggregated according to the dimension of physical activity of interest (34). The questionnaire consists of four sections: activity in and around the home, during work, commuting to work, and recreational physical activity. All commuting and some domestic questions were designed specifically for the study, whereas the questions on occupational activity were derived from the validated Modified Tecumseh Occupational Activity Questionnaire (2). The recreational section of the EPAQ2 was derived from the Minnesota Leisure Time Activity Questionnaire (30), with 30 predetermined sports selected according to their frequency and duration in a UK population (The Sports Council and The Health Education Authority, 1992) and six nonsportive activities, such as mowing the lawn, watering the lawn, digging, weeding, DIY (do it yourself), and playing music, which are considered activities undertaken in or around the home. The frame of reference for EPAQ2 is the past year. EPAQ2 was validated against repeated measures of free-living energy expenditure estimated from a 4-d individually calibrated minute-by-minute heart rate monitoring throughout a year suggesting that the questionnaire is valid for ranking individuals (34).
A score for physical activity at home was calculated by summing energy expenditure (MET·h·wk−1) derived from questions on the number of flights of stairs climbed at home and the frequency and duration of participation in activities undertaken in or around the home, together with the energy costs of these activities (1). The energy costs of sport and exercise activities and a total sport physical activity score (MET·h·wk−1) were similarly calculated from the sum of these individual activities. A score for transportation physical activity was calculated by summing energy expenditure (MET·h·wk−1) derived from questions on how people commute from home to work and how they usually get about excluding work-related commuting. On the basis of the same principle, an occupational physical activity score (MET·h·wk−1) was calculated using the reported duration of work and self-categorization of activity level at work. The different subdomains of physical activity being mutually exclusive, a total physical activity score was calculated by summing energy expenditure at home, work, transportation, and sport or exercise. The participants werethen divided into four categories (inactive: <60 MET·h·wk−1; moderately inactive: between 60 and 90MET·h·wk−1; moderately active: between 90 and 120 MET·h·wk−1; active: >125 MET·h·wk−1) on the basis of the quartiles of the total physical activity scores. Similarly, within each domain but with the exception of transportation, quartiles of physical activity were computed. For the transportation domain, activity categories were based on tertiles of scores among participants reporting some form of active commuting, and the lowest category was composed of participants reporting no such activity. In addition, we examined cycling and walking for transportation separately; the lowest category was composed of participants not reporting any cycling or walking activity, respectively, and the remaining participants were divided on the basis of the median of weekly duration of the given activity.
Other exposure assessments.
Trained nurses carried out a health examination during a clinic visit. Height and weight were measured in light clothes and without shoes following standard clinical procedures. Body mass index (BMI) was calculated as weight in kilograms divided by meters squared (kg·m−2). Blood pressure was measured using a sphygmomanometer (Accutorr) after 3 min of seated rest. Two measurements of blood pressure were taken, and the mean of the readings was used in analysis. Serum levels of total cholesterol were measured on fresh samples with RA 100 (Bayer Diagnostics, Basingstoke, United Kingdom). Information on history of heart attack, stroke, cancer, and diabetes was obtained by self-report. Smoking status was classified into three categories (current smoker, former smoker, or never smoker); alcohol consumption was quantified in grams per day (g·d−1); and social classes were categorized into four categories on the basis of the last reported occupation (1 = professional occupations, 2 = managerial and technical jobs, 3 = skilled and partly skilled labor, and 4 = unskilled labor) as previously described (19).
Individuals were flagged for death certification at the UK Office of National Statistics (ONS), with vital status ascertained for the whole cohort. The ONS reports deaths in the cohort via a regular record linkage system. Causes of deaths were classified as follows: (i) death due to all causes; (ii) death due to underlying cardiovascular disease; (iii) death due to cancer; and (iv) death due to all other causes (i.e., deaths not related to cardiovascular disease or cancer). For our analyses, each participant contributed person-time from the date of their second health check until the date of death or the end of follow-up at March 31, 2006. Of 14,905 participants who completed the EPAQ2, two participants were lost to follow-up, reducing the cohort of the present analysis to 14,903 participants and a total of 102,964 person-years.
The analysis of the association between occupational physical activity and mortality being restricted to the subsample of working participants, the main variables of interest have been described by working status. The associations between the different levels of physical activity and mortality were analyzed using Cox regression analyses in which exit time was the earlier of date of death or the end of follow-up. All regression models were adjusted for baseline values of age, sex, smoking status, alcohol consumption, social class, history of cancer (yes or no), history of cardiovascular disease or stroke (yes or no), and history of diabetes (yes or no). When examining the domain-specific association of physical activity with mortality, all analyses were additionally adjusted for the other domains of activity. The inactive category was coded as the reference category for all analysis. Linear trend tests across levels of physical activity were performed using the continuous variables in nonparametric multivariate Cox models. Risk modification by sex, age, and BMI as well as the interaction between the different domains of physical activity were tested by adding the respective interaction terms to the Cox model, and their significance was tested by the likelihood ratio statistic.
Finally, we examined the effect of further adjustment for BMI, systolic blood pressure, and total cholesterol on the association between physical activity and mortality. All analyses were repeated after excluding 187 participants who died within 2 yr of follow-up and 1458 participants who had a history of heart disease, stroke, or cancer. Departure from the proportional hazards assumption was evaluated by Schoenfeld residuals. Analyses were conducted using Statistical Analysis System software version 9.1 (26).
The correlation coefficients between the scores from different subdomains of activity are depicted in Table 1. Physical activity at home was inversely correlated with physical activity undertaken at work, and physical activity during exercise was positively correlated with the physical activity performed for transportation. None of the other domains of activity were correlated to each other.
The characteristics of the participants stratified by working status are shown in Table 2. Nonworking participants were 11 yr older, and 77% of the deaths occurred in this group. The history of known cancer, cardiovascular disease, and type 2 diabetes at baseline was twice as high in the nonworking group compared with the working participants. Among the nonworking participants, 26.4% were categorized as being active or moderately active compared with 71.8% in the working group. Differences in total physical activity between the two groups were largely explained by differences in occupational activity because the levels of sport- and transport-related activity were similar. The level of activity undertaken at home was higher in the nonworking participants. A higher proportion of the working participants were male (46.1%) compared with the nonworking group (41.4%). There were no significant differences between the two groups for all other variables.
Total physical activity was significantly inversely related to all-cause and cardiovascular mortality (Table 3). The home and sport domains of physical activity were significantly associated with lower all-cause and cardiovascular mortality, independently of each other and of the activity undertaken in the other domains. After excluding the more intense activities undertaken at home (MET ≥ 5), namely, mowing the lawn, digging, and stair climbing, all-cause mortality was no longer associated with duration of physical activity undertaken at home (P for trend = 0.93; data not shown). The duration of the most intense physical activity undertaken at home (i.e., climbing stairs, mowing the lawn, and digging) was associated with a reduced risk of all-cause mortality (P for trend = 0.03; data not shown). Neither cycling nor walking for transportation, analyzed separately and together, nor activity at work was significantly associated with all-cause and cardiovascular mortality. Total and domain-specific activity was not related to cancer mortality (data not shown). Domain-specific results were similar with and without adjusting for the other domains. The association between total activity and all-cause mortality was not modified by sex (P for interaction = 0.90) or BMI (P for interaction = 0.81). Domain-specific results observed among men were also similar to those observed among women. When comparing the most active to the least active participants, the relative risks of all-cause mortality (95% confidence interval) for physical activity at home and for sport or exercise were 0.84 (0.65-1.10) and 0.63 (0.49-0.81) among men and 0.70 (0.48-1.03) and 0.73 (0.53-1.02) among women, respectively. However, effect modification by age was observed (P = 0.02). In participants older than 63 yr (median), total activity was strongly and inversely related to all-cause mortality (P < 0.0001), whereas no relationship was observed in younger participants (P = 0.57). Effect modification by age was also observed for activity at home (P = 0.002) but not for the other domains of activity.
After excluding participants who died within 2 yr of follow-up and those who had a history of heart disease, stroke, or cancer, total activity (P for trend = 0.006), activity at home (P for trend = 0.01), and during sport or exercise (P for trend = 0.01) were inversely related to all-cause mortality, after additional adjustment for BMI, systolic blood pressure, and cholesterol, whereas no association was observed for transport- or work-related activity.
We observed a strong and highly significant interaction between activity at home and sport or exercise activity on all-cause mortality (P = 0.002). Physical activity at home was inversely related to all-cause mortality only when participants were categorized as being inactive or moderately inactive for sport or exercise (Fig. 1A). Reciprocally, activity for sport or exercise was inversely related to all-cause mortality only when the participants were inactive or moderately inactive at home (Fig. 1B).
The results from this population-based cohort study suggest that the inverse association between total physical activity and all-cause mortality is predominantly driven by sport, exercise, and the most intense activities carried out in and around the home. We found little evidence of an association between occupational or transport-related activity with mortality. Although the results for physical activity during sport or exercise are compatible with previous reports, few studies have been able to examine the association of the other domains of activity, particularly activity at home. Participants who reported being most active at home had a 19% lower risk of all-cause mortality compared with the least active group even after adjustment for potential confounders and physical activity in other domains. Our results also suggest that the reduced risk of mortality associated with the exercise and home domains of activity was not additive, indicating a possible ceiling effect of activity on all-cause mortality.
Others have reported an inverse association between total activity and all-cause mortality similar to the observations from this study, but most of the previous analyses were restricted to leisure time activity or did not combine all domains of activity (7,13,33). To our knowledge, only two previous studies, one from Puerto Rico (8) and one from China (23), have examined the associations between physical activity from all domains with mortality, both suggesting an inverse association between total physical activity with all-cause mortality.
We observed an effect modification by age on the association between activity and mortality, which is supported by others (4). However, in a recent review (18), an inverse association between vigorous activity or exercise and all-cause mortality was reported in all studies reviewed, regardless of the age of the participants. The absence of association between activity and mortality among young participants in this study may be explained by lack of power within this age group (n = 198 deaths).
Our findings of a reduced risk of all-cause mortality associated with activity for sport and exercise are consistent with recent reviews reporting an inverse association between exercise or vigorous leisure time activity and all-cause mortality (7,13,33). The inverse relation observed in our study between intense activities undertaken at home and all-cause and cardiovascular mortality independently of activity in other domains is consistent with the most recent recommendations emphasizing the importance of moderate- to vigorous-intensity physical activity performed as part of daily living (13). In our study, these intense domestic activities corresponded to heavy gardening and climbing stairs at home. One recent study in women suggested an inverse relationship between domestic physical activity and all-cause mortality independent of activity from other domains (23). Our results extend these observations to both men and women. Three previous studies included a few domestic activities in their leisure time activity categorization (4,5,22). In two of these studies (4,5), low or moderately intense leisure time activity was associated with a reduced risk of all-cause mortality. The study that failed to show an inverse relationship between nonvigorous leisure activity and all-cause mortality was restricted to men (22). The negative and significant interaction between activity at home and during sport and exercise suggests a ceiling effect of activity on mortality. The amount of activity during sport or exercise associated with an inverse relation in this study (i.e., moving from moderately inactive to moderately active) is equal to approximately 30 min of moderate-intensity activity five times per week, which is compatible with official recommendations (25). In contrast, 3 h of activity at home is associated with the same reduced risk. This difference in the magnitude of the association between activity at home and during sport might be due to a higher intensity during leisure time and/or different degree of measurement error from these two domains.
We found no evidence of an association between transport-related physical activity and all-cause mortality. This was also the case when analyzing cycling and walking separately. This is in accordance with two previous studies (5,6). In contrast, two studies from Denmark (4) and China (23) observed a reduced risk of all-cause mortality by increasing levels of transport-related activity. However, these results were mainly explained by cycling as a mode of transportation, an activity more prevalent in these countries than in the UK (4,23). Neither did we observe an association between activity at work and all-cause mortality. These findings are consistent with some (4,11,20,21) but not all (5,12,14,28) previous studies.
The inverse associations between activity and cardiovascular mortality are consistent with the literature (7,13,33). Apart from improving plasma lipid profile, reducing body weight, and lowering blood pressure for which our post hoc analyses were adjusted, activity may act through many other biological pathways such as reducing platelet aggregation, increasing fibrinolytic activity, improving insulin sensitivity and glucose tolerance, improving cardiac function, improving cardiorespiratory fitness, and lowering the resting heart rate (16). In contrast, we did not observe any association between activity and cancer mortality. As noticed in a recent review (33), physical activity is only associated with risk reduction of some specific cancers. Most of the studies investigating on overall cancer mortality reported little evidence of reduced risk due to activity, except some with follow-up longer than 10 yr and/or on the basis of leisure activity (3,9,15,17,24,27,29,32). It suggests that if an association exists between physical activity and overall cancer mortality, the magnitude of the association is low, or alternatively, the association occurs at least 10 yr after the activity has been undertaken.
Our study has several methodological strengths, including its prospective design and the high proportion of individuals followed up (99.99%). We assessed all domains of daily activity, including activity at home, at work, for transportation, and for sport or exercise. We mutually adjusted our analyses for activity in the various domains and controlled for a wide range of potential confounding factors collected at baseline. Analyses performed after exclusion of the participants with history of heart disease, stroke, and cancer at baseline as well as those deceased within 2 yr of follow-up showed similar results as analyses adjusting for these factors. Therefore, it is unlikely that our findings might be due to the underlying pathological process underlying the decreasing physical activity before mortality. In addition, the population-based sample of our study is heterogeneous in respect to several sociodemographic and anthropometric variables that make the results more generalizable compared with studies that have focused on defined groups (e.g., specific occupations).
Nonetheless, our results should be interpreted considering the following potential limitations. Physical activity was assessed by self-report. Although the questionnaire has been validated previously (34), the level of measurement error was considerable. In particular, reported housework was not correlated to energy expenditure assessed by the flex heart rate method, whereas occupational and recreational activities were. It should be noted, however, that the flex heart rate method may not be a perfect measure of the light-intensity activity that typically takes place in the home. In the context of this study, that measurement error is likely to be nondifferential because it would not be associated with the prognosis of the participants, and thus, its effect would be to attenuate the true association. Possibly, the different subdimensions of activity are recalled with different degrees of measurement error, a fact that would complicate judgment of their relative importance. In addition, physical activity was assessed at baseline, and possible changes in physical activity during follow-up have not been taken into account. Although the effect of confounding has been diminished by adjustment for potential confounders, residual confounding may still exist. Furthermore, our results may be explained by unmeasured or unknown confounding factors. Finally, the analyses of the effect of work related on mortality were based on a restricted sample of participants, who were younger and therefore presented fewer events that limit the power to detect an association.
Engaging in sport or exercise and intense activities at home were independently associated with mortality risk reductions of 19% and 34%, respectively, compared with being physically inactive. The reduced risks of mortality associated with physical activity from these two domains were not additive, suggesting a ceiling effect. In contrast, physical activity at work and for transportation did not confer a risk reduction for all-cause mortality.
The authors thank the staff of EPIC for their invaluable contributions. The cohort of EPIC-Norfolk is funded by the Cancer Research Campaign; the Medical Research Council; the Stroke Association; the British Heart Foundation; the Department of Health; the Europe Against Cancer Programme Commission of the European Union and the Ministry of Agriculture; Fisheries and Food. The results of the present study do not constitute endorsement by ACSM.
1. 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.
2. Ainsworth BE, Jacobs DR Jr, Leon AS, Richardson MT, Montoye HJ. Assessment of the accuracy of physical activity questionnaire occupational data. J Occup Med
3. Albanes D, Blair A, Taylor PR. Physical activity and risk of cancer in the NHANES I population. Am J Public Health
4. Andersen LB, Schnohr P, Schroll M, Hein HO. All-cause mortality associated with physical activity during leisure time, work, sports, and cycling to work. Arch Intern Med
5. Barengo NC, Hu G, Lakka TA, Pekkarinen H, Nissinen A, Tuomilehto J. Low physical activity as a predictor for total and cardiovascular disease mortality in middle-aged men and women in Finland. Eur Heart J
6. Batty GD, Shipley MJ, Marmot M, Davey-Smith G. Physical activity and cause-specific mortality in men: further evidence from the Whitehall study. Eur J Epidemiol
7. Bauman AE. Updating the evidence that physical activity is good for health: an epidemiological review 2000-2003. J Sci Med Sport
. 2004;7(1 Suppl):6-19.
8. Crespo CJ, Palmieri MR, Perdomo RP, et al. The relationship of physical activity and body weight with all-cause mortality: results from the Puerto Rico Heart Health Program. Ann Epidemiol
9. Davey-Smith G, Shipley MJ, Batty GD, Morris JN, Marmot M. Physical activity and cause-specific mortality in the Whitehall study. Public Health
10. Day N, Oakes S, Luben R, et al. EPIC-Norfolk
: study design and characteristics of the cohort. European Prospective Investigation of Cancer. Br J Cancer
. 1999;80 Suppl 1:95-103.
11. Eaton CB, Medalie JH, Flocke SA, Zyzanski SJ, Yaari S, Goldbourt U. Self-reported physical activity predicts long-term coronary heart disease and all-cause mortalities. Twenty-one-year follow-up of the Israeli Ischemic Heart Disease Study. Arch Fam Med
12. Graff-Iversen S, Selmer R, Sorensen M, Skurtveit S. Occupational physical activity, overweight, and mortality: a follow-up study of 47,405 Norwegian women and men. Res Q Exerc Sport
13. 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. Circulation
14. Hu G, Eriksson J, Barengo NC, et al. Occupational, commuting, and leisure-time physical activity in relation to total and cardiovascular mortality among Finnish subjects with type 2 diabetes. Circulation
15. Hu G, Tuomilehto J, Silventoinen K, Barengo NC, Peltonen M, Jousilahti P. The effects of physical activity and body mass index on cardiovascular, cancer and all-cause mortality among 47 212 middle-aged Finnish men and women. Int J Obes (Lond)
16. Ignarro LJ, Balestrieri ML, Napoli C. Nutrition, physical activity, and cardiovascular disease: an update. Cardiovasc Res
17. Kampert JB, Blair SN, Barlow CE, Kohl HW III. Physical activity, physical fitness, and all-cause and cancer mortality: a prospective study of men and women. Ann Epidemiol
18. Katzmarzyk PT, Janssen I, Ardern CI. Physical inactivity, excess adiposity and premature mortality. Obes Rev
19. Khaw KT, Jakes R, Bingham S, et al. Work and leisure time physical activity assessed using a simple, pragmatic, validated questionnaire and incident cardiovascular disease and all-cause mortality in men and women: The European Prospective Investigation into Cancer in Norfolk prospective population study. Int J Epidemiol
20. Kristal-Boneh E, Harari G, Melamed S, Froom P. Association of physical activity at work with mortality in Israeli industrial employees: the CORDIS study. J Occup Environ Med
21. Lapidus L, Bengtsson C. Socioeconomic factors and physical activity in relation to cardiovascular disease and death. A 12 year follow up of participants in a population study of women in Gothenburg, Sweden. Br Heart J
22. Lee IM, Paffenbarger RS Jr. Associations of light, moderate, and vigorous intensity physical activity with longevity. The Harvard Alumni Health Study. Am J Epidemiol
23. Matthews CE, Jurj AL, Shu XO, et al. Influence of exercise, walking, cycling, and overall nonexercise physical activity on mortality in Chinese women. Am J Epidemiol
24. Paffenbarger RS Jr, Hyde RT, Wing AL. Physical activity and incidence of cancer in diverse populations: a preliminary report. Am J Clin Nutr
. 1987;45(1 Suppl):312-7.
25. 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
26. SAS Institute, Inc. What' New in SAS® 9.0, 9.1, 9.1.2, and 9.1.3 Copyright ©
. Cary (NC); SAS Institute, Inc.; 2004. 399 p
27. Schnohr P, Gronbaek M, Petersen L, Hein HO, Sorensen TI. Physical activity in leisure-time and risk of cancer: 14-year follow-up of 28,000 Danish men and women. Scand J Public Health
28. Seccareccia F, Menotti A. Physical activity, physical fitness and mortality in a sample of middle aged men followed-up 25 years. J Sports Med Phys Fitness
29. Steenland K, Nowlin S, Palu S. Cancer incidence in the National Health and Nutrition Survey I. Follow-up data: diabetes, cholesterol, pulse and physical activity. Cancer Epidemiol Biomarkers Prev
30. Taylor HL, Jacobs DR Jr, Schucker B, Knudsen J, Leon AS, DeBacker G. A questionnaire for the assessment of leisure time physical activities. J Chronic Dis
31. The EPIC-Norfolk
Study Web site [Internet]. Cambridge (UK): The EPIC-Norfolk
Study; [cited 2008 Feb 8]. Available from: http://www.srl.cam.ac.uk/epic/
32. Wannamethee G, Shaper AG, Macfarlane PW. Heart rate, physical activity, and mortality from cancer and other noncardiovascular diseases. Am J Epidemiol
33. Warburton DE, Nicol CW, Bredin SS. Health benefits of physical activity: the evidence. CMAJ
34. Wareham NJ, Jakes RW, Rennie KL, Mitchell J, Hennings S, Day NE. Validity and repeatability of the EPIC-Norfolk
Physical Activity Questionnaire. Int J Epidemiol