Our study shows that achieving healthy target levels for CV risk factors, as proposed by the latest ESC guidelines, is associated with lower 10- and 20-year all-cause and CV mortality rates, even in older people. However, these beneficial effects are subject to gender differences: in older women the association between CV risk profile and mortality seems to be stronger than in men and to persist in more advanced age.
In our cohort, around one in five participants had a low or very low CV risk profile, defined as the achievement of target levels for most of the risk factors evaluated. Interestingly, gender differences already appeared on looking at the mean age of men and women by CV risk level. The age of participants increased with the worsening of CV risk in women, while men exhibited the opposite trend, i.e. the lower CV risk categories were associated with the oldest mean age. Although the physio-pathological changes occurring in women with aging can explain the accumulation of risk factors among the oldest old, our data for men may indicate selective survival of those having overcome the high-risk period of middle-age.
In this regard, it should be considered that, firstly, the assessment of some CV risk factors in older individuals, as indicated in the ESC guidelines, may not always reflect their lifelong CV risk profiles. This issue concerns, for example, tobacco exposure, for which only current smoking is assessed, or dietary style and physical activity level. Indeed, individuals with better CV profiles may have either maintained healthy behaviours over their life course or modified their lifestyle and corrected their CV endpoints as a consequence of treatments or preventive actions. However, it is noteworthy that our main findings were confirmed in participants free from CVD. This result is consistent with the hypothesis that the CV risk profiles we assessed were influenced more by the individuals’ lifestyles and functional homeostasis, than by medical interventions.
Secondly, certain CV health factors seemed to be poorly represented in our participants, with the recommended target levels achieved by only a few individuals. This was especially the case for body weight and LDL cholesterol, confirming that the nutritional thresholds for adults may not be fully applicable to older people (Sergi et al. 2005). In particular, it should be borne in mind that, compared with underweight or obesity, mild-to-moderate overweight in the older population has been associated with reduced mortality (Winter et al. 2014; Sergi et al. 2005).
These issues need to be taken into account when interpreting our results and support the adoption for older individuals of risk stratification approaches that would consider life-course risk exposures and age- and sex-specific at-risk thresholds. Nonetheless, overall, this study suggests that achieving the recommended target levels of a greater number of CV risk factors, as indicated by the current ESC guidelines, may gradually reduce all-cause and CV mortality even in older age.
The extent of this protective effect is in line with previous work that has found 50–60% reductions in all-cause mortality for various combinations of healthy lifestyle factors (Knoops et al. 2004). When considering potential gender differences, we found that this association was stronger in women than in men, and that the benefits of a healthy CV risk factor profile for survival seemed to last longer in women, persisting to over the age of 75 years. These findings corroborate the slight gender differences observed in previous studies (Khaw et al. 2008; Li et al. 2018) and lead us to put forward two possible hypotheses: either the mechanisms through which CV health factors influence mortality differ by gender and by age period, or the mechanisms are the same but operate in gender-specific time windows of susceptibility. Regarding the first hypothesis, the effect of risky behaviours on health-related outcomes involves inflammatory and oxidative pathways, which increase the risk of cardiometabolic dysfunction and worsen related outcomes (Rizzuto and Fratiglioni 2014). Avoiding these risk factors and satisfactorily controlling cardiometabolic dysfunctions have demonstrated similar benefits in men and women (Kvaavik 2010; Petersen et al. 2015; Odegaard et al. 2011) and in older age (Li et al. 2018; Rawshani et al. 2018). Such an effect may benefit CV health and prevent other chronic conditions, such as pulmonary, osteoarticular, neurologic diseases, affecting both all-cause and CV mortality. Concerning CVD, in particular, our results on individuals with no history of CVD at baseline support a possible preventive action of healthier risk profiles also in older age in delaying the development and progression of such pathologies.
These data strengthen our second hypothesis, namely that age and gender differences are not to do with the mechanisms, but with the extent to which CV health factors influence survival: the greater effect of these protective factors in older women could be due to their greater vulnerability to CVD in advanced age, unlike men, whose high-risk period is middle-age (Hippisley-Cox et al. 2010). This phenomenon may be influenced by the lower levels of female hormones after menopause and subsequent metabolic changes, and by the greater prevalence in females of conditions such as osteoarthritis and obesity, which increase CVD risk and worsen prognosis (Trevisan et al. 2017; Wilson et al. 2002; Tankó et al. 2005). Accordingly, CVD onset is almost 10 years later in women than in men, and women are more likely to present a higher number of risk factors and comorbidities at CVD diagnosis (Sharma and Gulati 2013). In contrast to the timing hypothesis for hormone replacement therapy (Clarkson et al. 2013), healthy behaviours and greater disease control seem also to benefit the oldest old, suggesting that women’s vulnerability to such factors extends into advanced age. Finally, it is to be noted that generational changes could have also influenced the age and gender differences in the association of each factor with mortality. In this regard, further research comparing the extent to which single risk factors affected mortality between different generations of older men and women will be of high interest.
In addition to the observational nature of the study, one of the limitations is the simple assessment of the number of healthy CV factors without considering their potentially different weights (Rizzuto and Fratiglioni 2014). We chose this approach since we wanted to be consistent with the guidelines recommendations that did not prioritise the achievement of any specific target level. However, the sensitivity analysis exploring the association between each CV risk factor and mortality provided some information in this regard. Secondly, as mentioned above, our evaluation of CV risk factors could be biased by possible socially- or medically-induced changes in the years before the baseline visit, as well as by risk factors variations during the follow-up. Together with the use of self-reported information, this issue could be a source of misclassification. This may concern factors such as tobacco exposure (since never and former smokers were both considered as non-exposed), dietary style, and physical activity level. However, as regards smoking habits, the median number of years since smoking cessation in the former smokers of our sample makes unlikely a substantial impact of previous tobacco exposure on 20-year mortality (Gallucci et al. 2020). Considering dietary patterns, changes in sensory and masticatory functions (Sergi et al. 2017; Tada and Miura 2014) might have slightly influenced the preferences and consistency of the foods in our aging population, but should not have caused relevant variations in dietary style (Tada and Miura 2014). Furthermore, despite the possibility of residual confounding, the median time since diabetes diagnosis in our sample and the epidemiological data on age at disease onset (Sattar et al. 2019) can rule out that a substantial number of individuals might have experienced changes in glycemic control or incident diabetes over the follow-up. Third, we included only Caucasian older adults living in northern Italy; therefore, our results are likely to be generalised to similar Italian and European populations. Moreover, we did not include institutionalised individuals because of the possible influence of co-existing conditions highly prevalent in such population (e.g. frailty, multimorbidity, and disability) on the association between CV risk profile and mortality. Possible variability linked to geographical or ethnic differences and the nursing home setting should be investigated in future studies. On the other hand, our work is strengthened by using reliable administrative data on mortality derived from regional health registers, as well as the 20-year follow-up period and comprehensive data collection in a large cohort of older adults.
In conclusion, our study shows that healthier CV risk profiles in older people are associated with reduced all-cause and CV mortality, suggesting the potential effectiveness of preventative action even in advanced age. Although our data show that the current guidelines are also applicable to older individuals, differences by gender and class of advanced age highlight the need for a personalised and life course approach to delivering care and preventive interventions to older men and women in various settings.