In this cohort of Spanish older adults, the number of positive health behaviors showed an inverse dose-response relation with mortality risk. Older adults with six positive behaviors had an 80% lower mortality risk or the equivalent to a reduction of 14 years in chronological age compared to those with zero to one health behaviors. Importantly, this survival benefit was not completely due to the traditional healthy lifestyles (not smoking, being physically active and adequate diet). In fact, some non-traditional health behaviors, such as adequate sleep duration, avoiding excessive sitting time, and social interaction with friends, were important contributors to improved survival.
Age and sex are the most important predictors of mortality . These and other factors (for example, genetics) cannot, however, be modified. Smoking, physical activity and diet are the three potentially modifiable lifestyles with strongest evidence of their role in mortality . There is also evidence that each health behavior "counts" and that the higher the number of these three health behaviors, the lower the mortality risk in adults [2–7]. Our findings extend previous observations limited to traditional health behaviors, suggesting that health promotion interventions in older adults should also focus on non-traditional healthy lifestyles to obtain full benefit.
A recent systematic review and meta-analysis of prospective studies indicates that both short and long duration of sleep are significant predictors of death . In older adults, long sleep may even be more detrimental on mortality than short sleep . Although the science of sedentary behavior is relatively new, most studies have shown a dose-response association between sitting time and mortality, which is independent of physical activity . Moreover, a poor social network has also been consistently linked to higher mortality , especially in older people .
In the present study, the three non-traditional health behaviors showed protective associations with mortality. In fact, the impact of sitting time on mortality was close to that exerted by smoking, which is the leading preventable cause of death. The combined impact of the number of non-traditional health behaviors was also substantial, as seen in the fact that older adults with these three healthy lifestyles had a 36% lower mortality than those with one or none. Likewise, both liberal (likelihood ratio test) and conservative (c-statistic) statistical methods have indicated that adding these three new lifestyles to the three classical health behaviors may improve mortality prediction. Taken together, these findings support the assertion that in older adults some non-traditional lifestyles may increase longevity beyond what is achieved by the three classical health behaviors.
To our knowledge, only the HALE project has examined the combined effect of health behaviors on mortality in older persons . In this study, Knoops et al. investigated the combined effect of four health behaviors (healthy diet, being physically active, moderate alcohol use and non-smoking) on 10-year mortality among 2,339 individuals aged 70 to 90 years from 11 European countries. Adherence to these four health behaviors was associated with lower mortality compared with those who adhered to only one or no health behaviors. As some researchers have pointed out [3, 30, 31], the HALE project included a highly selected group of older adults (for example, participants had a high educational level, only 36% were women and over 50% of the women drank alcohol regularly), and the consistency of findings was not examined within subpopulations defined by demographic (for example, age and sex) or health conditions (for example, obesity). Also of note is that lifestyles (for example, physical activity) were measured with different methods across the study populations. Our study, while overcoming these limitations, confirmed the results of the HALE project. One further advantage of our study is that the results correspond to a single, relatively homogeneous population showing the usual range of lifestyle variations that may be more realistically achievable and directly relevant to immediate public health . Moreover, in contrast to some complex instruments used in HALE to assess diet or physical activity, our lifestyle score was based on very simple and easy to ask questions, which may be of practical use in clinical practice or for monitoring health in population surveys.
Other relevant findings in our study were the consistency of results across subgroups defined by socio-demographic variables and health conditions, and the improved survival associated with positive health behaviors that was observed since the second year of follow-up. These results suggest that it is never too late - even at age 75 or older - to adopt a healthy lifestyle, and that most older adults (whether or not they have severe disease) may benefit, although further testing in experimental studies is needed to support this contention.
This study has several strengths. It included a representative sample of older adults in Spain, which allows for generalization of results. In fact, in our cohort the annual mortality rates are only slightly lower than in the Spanish population of the same age and sex. This was to be expected because our cohort exclusively included non-institutionalized individuals and participants in epidemiological studies tend to be healthier than in the general population. Likewise, all study variables were obtained by well-trained interviewers using standardized methods. The relatively large sample size made it possible to examine whether the study associations varied according to socio-demographic characteristics and many health conditions. Moreover, analyses were adjusted for a substantial number of confounders. Lastly, to reduce reporting bias linked to disease status, the analyses have been replicated in individuals without severe comorbidity, and also after excluding deaths in the first two years of follow-up.
Our study also had some limitations. First, information on lifestyles was obtained at baseline. Although our analyses assume that lifestyles have certain stability over time, some changes are still possible and would likely have led to an underestimation of the protective impact of health behaviors on mortality. Second, lifestyle was self-reported, which may have led to recall bias, particularly for assessing physical activity and non-traditional health behaviors. However, similar measures of physical activity, sleep duration and sitting time have shown adequate validity compared with objective measures in adults and older adults [23, 32–34]. Third, health behaviors related to physical activity and diet could not be defined according to public health recommendations (for example, 150 minutes/week for physical activity, and at least two servings/day of fruit and vegetables). Moreover, health behaviors were dichotomized; if people who had a high-risk behavior falsely reported having a healthy behavior, it is likely that the hazard ratios may have been underestimated. Also misclassification error may differ between health behaviors, so that our results on the relative influence of each behavior on mortality should be interpreted with caution. Therefore, taken together, these limitations in the measurement of health behaviors have precluded a better understanding of the study associations.
Future research should assess the combined effect of other non-traditional health behaviors (for example, participating in a dancing group, playing a musical instrument, religious activities and so on) on mortality in older adults. It should also examine whether multiple healthy lifestyles might help to maintain a long and disease-free independent life and a good health-related quality of life.