Background Regular physical activity is associated with reduced risk of mortality in middle-aged adults; however, associations between physical activity and mortality in older people have been less well studied. The objective of this study was to compare relationships between physical activity and mortality in older women and men.
Methods The prospective cohort design involved 7080 women aged 70–75 years and 11 668 men aged 65–83 years at baseline, from two Australian cohorts – the Australian Longitudinal Study on Women's Health and the Health in Men Study. Self-reported low, moderate and vigorous intensity physical activity, socio-demographic, behavioural and health characteristics were assessed in relation to all-cause mortality from the National Death Index from 1996 to 2009; the median follow-up of 10.4 (women) and 11.5 (men) years.
Results There were 1807 (25.5%) and 4705 (40.3%) deaths in women and men, respectively. After adjustment for behavioural risk factors, demographic variables and self-reported health at baseline, there was an inverse dose – response relationship between physical activity and all-cause mortality. Compared with women and men who reported no activity, there were statistically significant lower hazard ratios for women who reported any activity and for men who reported activities equivalent to at least 300 metabolic equivalent.min/week. Risk reductions were 30–50% greater in women than in men in every physical activity category.
Conclusions Physical activity is inversely associated with all-cause mortality in older men and women. The relationship is stronger in women than in men, and there are benefits from even low levels of activity.
Statistics from Altmetric.com
Three recent systematic reviews and meta- analyses of observational studies have confirmed that physical activity is inversely associated with all-cause mortality in men and women, after adjustment for other demographic and behavioural risk factors.1,–,3 However, as most of the studies included in these reviews involved people whose physical activity was first measured between the ages of 20 and 65 years, it is unclear whether these findings are generalisable to older people.
There have been relatively few studies of people aged 65 years or more. Some of these have suggested that low levels of physical activity (below the current recommendation of 30 min of moderate intensity activity on most days each week)4 ,5 may be effective in lowering the risk of all-cause mortality in older people.6 ,7 Three others have suggested that the reduction in mortality risk in older people (>65 years at baseline) may be greater than those reported in the studies of mid-aged adults;8,–,10 however, these three studies had small samples and included only men.
The aims of this study were to compare the nature and magnitude of the relationship between physical activity and all-cause mortality in two large Australian cohorts of older women and men, and to address the following questions: (1) Is there a significant benefit (in terms of survival) for older people who report lower levels of activity than are currently recommended in public health guidelines? (2) Does the mortality risk reduction associated with physical activity differ in older women and older men?
We used data from the 1921–26 birth cohort of the Australian Longitudinal Study on Women's Health (ALSWH) and from the Health in Men Study (HIMS). The methods for both studies have been described in detail elsewhere.11 ,12
The ALSWH women were randomly selected from all women aged 70–75 years listed in the Australian National Health Insurance Database (Medicare), which includes all citizens and permanent residents, with intentional over-sampling of women from rural and remote areas.11 A total of 12 432 women formed the 1921–26 ALSWH cohort when recruited in 1996.13 The project uses mailed questionnaires to collect self-report data on health and related variables every 3 years. Because the physical activity questions in the 1996 survey differed from those in the HIMS survey, baseline data for the current analyses were drawn from survey 2 (in 1999), to which 10 430 women responded (as 529 had died, the follow-up rate was 87.6% of survivors). The ethics committees of the University of Newcastle and the University of Queensland approved the research protocol and all participants provided informed written consent.
The HIMS cohort includes 12 203 men who were originally recruited for a screening trial between 1996 and 1999 (referred to here as baseline). They were randomly selected from all men aged 65–83 years listed on the electoral roll (voting is compulsory in Australia) who resided in Perth (the capital of Western Australia). The HIMS research protocol was approved by the ethics committee of the University of Western Australia and all participants provided informed written consent.
A postal questionnaire was used to collect data from the women, and all variables, including height and weight and smoking status, were self-reported. Data from the men were obtained using a postal questionnaire that was reviewed during a face-to-face interview by a research nurse. During this interview, the physical measures including height and weight were recorded, and participants were asked about their current smoking status. For all ALSWH participants, identification information was probabilistically matched to the Australian National Death Index to identify deaths.14 For HIMS participants, death information was obtained through the Western Australian Data Linkage System,15 which provides electronic linkage to the state's population health data collections and includes all records from the death register. Survival time was calculated as the number of days between the return of the survey (in 1999 for ALSWH and between 1996 and 1999 for HIMS) and the date of death or survival to 1 October 2009.
In both cohorts, participants reported the duration of time spent in vigorous leisure activity/exercise (that makes you breathe harder or puff and pant) in the last week (women) or in a usual week (men). A metabolic equivalent (MET) value of 5 was applied to responses to these questions, in line with values suggested for people aged 65–79 years in the US Surgeon General's Report on Physical Activity and Health.16 In addition, the women were asked to report the time spent walking briskly (for recreation or exercise, or to get from place to place) and in moderate leisure activity (like golf, bowls, social tennis, moderate exercise classes), while the men were asked to report on non-vigorous exercise for recreation or health and fitness (eg, slow walking, slow cycling, Tai Chi, yoga, etc). Responses to the women's walking and moderate activity questions were combined to create time in non-vigorous exercise and a MET value of 3 was applied to the responses for non-vigorous activity from men and women.16 A physical activity score was calculated from (non-vigorous min per week×3 MET)+(vigorous min per week×5 MET). Scores were categorised as ‘none’ (0 MET.min/week), ‘very low’ (1–<300), ‘low’ (300–<450), ‘moderate’ (450–<600), ‘moderate-high’ (600–<1050), ‘high’ (1050–<1500) or ‘very high’ (≥1500). This physical activity measure has been shown to have an acceptable reliability and validity.17 ,18
Participants were asked to report their highest level of education, which was categorised as ‘low’ (no school or primary school), ‘middle’ (high school) or ‘high’ (postschool qualification). Marital status was categorised as ‘partnered’ (married or de facto), ‘widowed’ or ‘not partnered’ (never married, separated or divorced). Smoking status was categorised as ‘never smoked’, ‘ex-smoker’ or ‘current smoker’.
For alcohol intake, the women were asked about the frequency of alcohol use and the quantity of alcohol consumed on a usual day, while the men were asked if they had ever drunk alcohol, and if so how many standard drinks they consumed on each day of a usual week. In Australia, a standard drink is defined as 10 g of alcohol.19 Consistent with earlier analyses of alcohol consumption in this cohort,20 the participants were classified as ‘never drink’, ‘rarely drink’ (drank alcohol, but not in the past year), ‘less than once a week’ (drank alcohol, but zero consumption on each day of a usual week) or ‘weekly’ (drank alcohol at least once per week). The body mass index (BMI) was calculated from height and weight (kg/m2) and categorised according to WHO recommendations21 as ‘underweight’ (<18.5), ‘normal’ (18.5 to <25), ‘overweight’ (25 to <30) and ‘obese’ (≥30). Participants reported if they had ever been diagnosed (yes/no) with any of the following chronic conditions: heart disease, diabetes, hypertension, stroke or chronic obstructive pulmonary disease.
Due to the differences in data collection methods, missing data were more common from the women than from the men. All participants with missing data were excluded from the analyses. The missing data were as follows:
▶ for physical activity, women N=1045, men N=38
▶ for education, women N=529, men N=7
▶ for marital status, women N=41, men N=3
▶ for smoking, women N=781, men N=0
▶ for alcohol, women N=1294, men N=482
▶ for BMI, women N=1389, men N=9
▶ for chronic conditions, women N=226, men N=458
with some people having missing data for more than one of these variables. The resulting dataset available for the analysis was for N=7080 women and N=11 668 men.
The form of the association between physical activity and mortality risk was investigated using a Cox proportional hazards model of all-cause mortality with sex, physical activity and a sex×physical activity interaction in the model. Continuous physical activity data were centred at 495 MET.min/week and a three-knot restricted cubic spline was used to fit the raw MET values. Proportional hazard models were also used to estimate the HRs for mortality in women and men in each physical activity category, compared with those in the respective ‘none’ groups. These models were adjusted for age (in years), marital status, level of education, smoking status, alcohol consumption, BMI category and the presence/absence of each of the chronic conditions as covariates, and the proportional hazard assumption was tested by the examination of Kaplan–Meier curves. Separate models were first fitted in women and men; then an analysis for both genders together was conducted with a gender interaction term to estimate the relative HRs for men compared with women. All statistical analyses were conducted in SAS 9.2.
Sensitivity analyses were conducted: (1) after omitting deaths within the first 2 years; (2) using different MET weights from those specified above, to reflect the possibility that vigorous activity may be more intense (ie, 6 or 7 METs) in men than in women, or that the MET value for moderate intensity activities should be higher in men (eg, 4 METs); (3) for women and men with and without any of the chronic conditions and (4) using interaction terms for physical activity and BMI categories.
The characteristics of the men and women at baseline are shown in table 1. On average, the men were slightly younger than the women and were more likely to report moderate, high or very high levels of physical activity, to have higher levels of education, to be partnered and to be current or ex-smokers. The men were also more likely to report more frequent alcohol use and to be overweight or obese. There were small but statistically significant gender differences in chronic conditions; the most marked was for hypertension which was reported by more than half the women (table 1).
The numbers of deaths between the date of baseline assessment and 1 October 2009, for women and men were 1807 (25.5%) and 4705 (40.3%), respectively. Of these, 151 (2.1%) women and 491 (4.2%) men died within the first 2 years of follow-up. Median follow-up time (to death or 1 October 2009) was 10.4 years for the women and 11.5 years for the men.
Relationships between the continuous physical activity data and mortality in women and men are shown in figure 1. There was a more marked decline in mortality risk with increasing physical activity in women than in men; with risk decreasing in women up to about 1000 MET.min/week (5.6 h/week at moderate (3 MET) intensity).
HRs for all-cause mortality by categories of physical activity (with ‘none’ as the referent), adjusted for age (years), highest level of education, marital status, smoking status, alcohol use, BMI category and the presence/absence of each chronic condition are shown in table 2. The HRs were slightly attenuated when deaths during the first 2 years were excluded (see the right panel of table 2). For women and men, a dose-response trend was evident, with higher levels of physical activity associated with lower mortality risk (p<0.001). The magnitude of the risk reduction was greater in the women; in every physical activity category, the HRs were between 1.1 and 1.5 times higher for the men than for the women (table 2). Using different MET values for the various response categories in men and women did not change the estimates of the HRs for each gender, or the ratios of these estimates. There was no evidence of any effect of a BMI by physical activity interaction, and the relationships were very similar for the groups with or without any of the chronic conditions.
The aim of this study was to use data from our two cohort studies of older Australians to address the following questions.
Is there a significant benefit (in terms of survival) for older people who report lower levels of activity than are currently recommended in public health guidelines?
Our results suggest that very low levels of activity in women, and low levels in men, are associated with significant reductions in all-cause mortality. Current guidelines for the general population4 and for older people5 suggest that a minimum of 30 min of moderate intensity activity on 5 days each week (450 MET.min/week using our value of three for moderate activity) will confer health benefits, with modest additional benefit gained from higher volumes (increased duration or intensity) of activity.4 ,5 Our data suggest that there may be survival benefits for older people (especially women) who report less than these recommended levels.
These data are consistent with the results from several earlier epidemiological studies of physical activity, which also showed that low levels of activity were associated with reduced mortality risk in cohorts of younger men22 ,23 and in post- menopausal women.24 Similar findings have been reported in more recent studies of older cohorts. For example, in the Chinese older cohort study, there was a 13% reduction in mortality risk with low levels (<30 min/day) of physical activity;7 data were not analysed separately for women and men. Another study of almost 10 000 white US women aged 65 years or older found a 22–29% reduction in the risk of mortality over 12.5 years, among women who reported walking, with estimated energy expenditure of 187–897 kcal/week.6 This ‘dose’ is lower than that suggested by minimal adherence to the current physical activity guidelines, 30 min/day of moderate intensity activity, which equates with about 1000 kcal or 4200 kJ/week).25 Another study of about 3200 Swedish people over the age of 65 years found that as little as ‘weekly physical activity’ was associated with a 40% reduction in mortality risk.26
Does the mortality risk reduction associated with physical activity differ in older women and older men?
Our results clearly suggest that, relative to the most inactive participants in each gender group, there may be greater survival benefits across the range of physical activity categories for women than for men. In the fully adjusted models, men who reported low levels of weekly physical activity (300–<450 MET.min/week) had a 14% reduction in mortality risk, (relative to the least active men) compared with a 41% reduction in women in this activity group. Men who reported very high levels of activity (≥1500 MET.min/week) had a 27% reduction in risk (relative to inactive men), while there was a 48% reduction in the high active women.
Two of the recent meta-analyses also reported that reductions in mortality risks (HRs) attributable to physical activity were around 10–12% lower women than men.1 ,2 However, in contrast with our findings, the gender differences in these analyses of largely younger cohorts were not statistically significant.
There are two explanations for the apparently greater benefit observed here for women. The first relates to the potential measurement error arising from the slightly different questions asked in the two surveys. We tried ascribing different MET values for walking and moderate activity for the women and higher MET values for the men for moderate and vigorous activities. Using all possible combinations of these differing values for the intensity of physical activity made little difference to the results. We could not, however, judge whether there was differential reporting of the duration of activities. Levels of activity reported by men were higher, and it is possible that the men may have overestimated, or the women may have underestimated, their physical activity. The net effect of this would, however, have been to increase the estimated gender difference in the effect of physical activity intensity on mortality.
The second possible explanation is that the women's reported physical activity may have occurred against a higher ‘background’ level of activity, in terms of routine household tasks (eg, preparing meals, doing laundry and light housework) and caring duties. Australian time-use data show that women aged 65–74 years spend more than 1 h a day more in domestic and volunteer caring work than men of the same age. In contrast, older men spend almost an hour more each day in ‘audiovisual’ activities (including watching TV) than older women.27 It is possible that these differing ‘background’ levels of light activity and sedentary behaviour may be important in terms of long-term health outcomes.28 A recent Japanese study involving more than 10 000 ‘older’ participants (aged 65–84 years) has also reported a much more pronounced decrease in the mortality hazard for women (62%) than for men (46%) when looking at the exposure of ‘non-exercise’ (ie, domestic and caring) activity.29
Strengths and limitations
The strengths of this study are its large size, excellent identification of deaths and the ascertainment of many potentially confounding variables such as BMI, previous chronic health conditions and smoking status. This is also one of the few studies to directly compare mortality using the same reference and other categories of physical activity in women and men who were already 65–83 years old at baseline.
The potential limitations include the use of self-reported measures of physical activity, and differences in the data collection methods and questions used to assess physical activity in the men and women. We have, however, demonstrated the validity of our physical activity measure in community-dwelling older adults,18 and we conducted sensitivity analyses using different MET values to reflect the potential different interpretation of questions by the men and women. We also explored the possibility of reverse causality, that is, that unhealthy people are less able to participate in physical activity. However, even after excluding deaths in the first 2 years, and stratifying for five chronic health conditions, the effect estimates for physical activity remained largely unchanged. As other researchers have suggested that the effect of physical activity on mortality is largely mediated by cardiorespiratory fitness,30 another limitation is that we had no way of assessing the differences in fitness in the two cohorts.
What is already known on this topic
▶ Regular physical activity is associated with reduced risk of mortality in numerous studies that first assessed the activity when the participants were mid-aged.
▶ Associations between physical activity and mortality in older people have been less well studied, though suggest that even low levels of physical activity may be effective in lowering the risk of all-cause mortality.
What this study adds
▶ A dose-response reduction in the risk of mortality was confirmed in older women and men after adjustment for many potential confounding variables.
▶ For any given amount of physical activity, mortality risk reduction was greater in older women than in older men.
▶ There are survival benefits from very low levels of physical activity in women.
In conclusion, we found dose-response relationships between physical activity and all-cause mortality, with benefits from low levels of activity in older women and older men, but with risk reduction for any given amount of activity greater in women than in men. This study has implications for the advice provided to older women and men: sedentary older people should be encouraged to become more active without emphasising the high levels specified in the various guidelines; it is never too late to begin, and there are benefits from even low levels of activity, especially for women.
This work was conducted as part of the Men, Women and Ageing project which is supported by a National Health and Medical Research Council of Australia/Australian Research Council Ageing Well, Ageing Productively Strategic Award (409953). The Australian Longitudinal Study on Women's Health is supported by the Australian Government Department of Health and Ageing. The Health in Men Study is supported by grants from the National Health and Medical Research Council of Australia (279408, 379600, 403963).
Funding National Health and Medical Research Council of Australia/Australian Research Council Ageing Well, Ageing Productively Strategic Award; Australian Government Department of Health and Ageing; and National Health and Medical Research Council of Australia.
Competing interests None.
Ethics approval Ethics committees of the University of Newcastle, University of Queensland and University of Western Australia.
Provenance and peer Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.