In this cohort of 11,340 community dwelling healthy older Australians, we examined the association between a healthy lifestyle score and all-cause mortality, cancer-related mortality, CVD-related mortality and ‘other’ causes over a median follow-up time of 6.8 years (IQR: 5.7, 7.9). We found that a healthy lifestyle score at baseline, comprising of four common and potentially modifiable lifestyle factors (non-smoking, moderate alcohol consumption, a healthy dietary pattern and physical activity) was associated with prolonged lifespan in a dose-response relationship, such that each additional lifestyle factor was associated with a 16% lower risk of all-cause mortality, 25% lower risk of CVD mortality and 22% lower risk of ‘Other’ mortality. The current data found no association between lifestyle groups and cancer-related mortality. The benefits of a healthy lifestyle on all-cause mortality had larger effect sizes among males, those ≤ 73 years old and among those in the aspirin treatment group, although interactions were not statistically significant.
This is one of the largest and most comprehensive studies, conducted exclusively within community dwelling older people, reporting the association between a lifestyle composite score based on adherence to international health behaviour recommendations and all-cause plus cause-specific mortality. The results are largely in agreement with previous reports investigating different combinations of healthy lifestyle characteristics. A similar study of older Chinese people (n = 11,224, aged 65–90 years) reported that, compared to those without any unhealthy factors, those who had a high BMI, poor sleep, unhealthy diet, no physical activity, consumed alcohol and currently smoked were 1.34 (95% CI 1.02, 1.76) times more likely to die from any cause over a ten-year follow-up period (14). Another 18-year follow-up study of Swedish older adults (75 + years of age; n = 1,810) reported a median survival of 5.4 years longer among those who had a healthy BMI, never smoked or drank alcohol, engaged in leisure activities and moderate levels of physical activity versus those who did not (13). “The healthy aging: a longitudinal study in Europe” (HALE) study (n = 1,507) reported that 70 to 90-year-old community-dwelling people who did not smoke, consumed a Mediterranean diet, reported moderate alcohol consumption and was physically active had a 50% lower rate of all-cause and cause-specific mortality over 10-years, including CVD and cancer-related mortality (11). Similar protective effects of a composite lifestyle score, typically including at least diet, physical activity, smoking and alcohol, have been reported among middle-aged cohorts from different counties including Japan (30), China (12), United States of America (USA) (2), Australia (31) and the United Kingdom (UK) (32).
Results reported here and previously, provide compelling evidence to suggest that individuals reporting a healthy lifestyle in older age have a significantly reduced risk of earlier mortality. The results also demonstrate that current international recommendations for moderate physical activity, no smoking, a healthy dietary pattern and moderate alcohol consumption may still provide a useful predictor of longevity among this older aged cohort.
We found no relationship between a healthy lifestyle on risk of overall cancer-related death, which is contrary to findings previously reported in the HALE study as well as among studies of younger cohorts (11). Although, it is possible that the lack of a broader link with cancer reflects the very small percentage of current smokers in ASPREE. The HALE study was conducted via survey only and among a demographic born up to 40-years earlier than ASPREE participants.
Some methodological points may impact the conclusions of this study. In order to construct the healthy lifestyle score, we dichotomised each lifestyle factor according to pre-defined cut-off points. Different threshold values may have resulted in different risk estimates. However, the choice of cut-off was largely based on national and international public health recommendations (25–28). In sensitivity analyses we trialled different cut-offs and multi-levels but the results remained largely unchanged (Tables S10, S11 and supplement p 14). The approach of designating compliance versus non-compliance allows a simple objective classification to assess the health impact of lifestyle and can inform a clear public health message. Future modifications of this approach may involve differential weighting of the health impact of each lifestyle measure.
Owing to the absence of baseline dietary data, we utilised year-three dietary data as an alternative replacement within the baseline lifestyle score. It is not certain whether dietary behavior had significantly changed over this three-year period. Dietary changes can occur in older people due to factors such as oral health, income, marital status, medication or change of residence (33). Nonetheless, as healthy lifestyle habits are characteristic of a person’s way of living, and given ASPREE is an especially healthy cohort, majority of participants were unlikely to show a substantial change in general dietary habits over a three-year period (34). Furthermore, the year-three lifestyle score was associated with all-cause and cause-specific mortality with similar effect size to associations between baseline lifestyle and mortality, sanctioning this assumption.
Finally, given we do not have detailed information about mid-life lifestyle behaviour in the ASPREE cohort, we cannot confirm whether observed associations are not driven by behaviour earlier in life. Healthy lifestyle behaviours in older age may reflect a long-standing approach to healthy living which, in turn, may be driving these observations. Our results still, however, highlight the benefits of identifying healthy lifestyle factors as predictors of likely future mortality, even among already healthy older people.
Strengths and limitations
There are several strengths of our study. ASPREE is a well characterised, large and contemporary cohort of older people who had reached age 70-years or more in relatively good health (18). Furthermore, rigorous methods for the ascertainment of cause-specific mortality ensured highly accurate endpoints. The investigation of not only all-cause but cause-specific mortality is a further strength.
There are also several potential limitations. First, the ASPREE cohort is comprised of initially healthy volunteers for a clinical trial who are more likely to be attentive to maintaining a healthy lifestyle, hence, may represent a healthier sample of older people compared with the general population. Second, the cohort is largely Caucasian, educated and drawn from communities with access to universal healthcare as reflected by the extensive use of preventive medications including statins (in 30%) and antihypertensive agents (in 51%). Therefore, our results may not be applicable among other socioeconomic and ethnic groups as well as among those residing in lower- to middle-income countries. Third, due to the progressive nature of noncommunicable disease leading to death, with declining function often preceding and possibly influencing lifestyle behaviour, we cannot rule out reverse causality as a partial explanation for these observations. Nonetheless, although survivor bias is a common limitation in healthy cohort studies, our censoring of death events at three-years may also help to mitigate reverse causality.
Finally, although potential confounders were considered in multivariable analyses, residual confounding cannot be ruled out. Furthermore, other unmeasured lifestyle and environmental factors may also play a role in determining risk of death. However, demonstrating that these four common lifestyle behaviors are associated with prolongation of an individual’s lifespan remains an important public health message.