In this large cohort, we observed that men and women with favorable risk factor profiles in middle age have a low remaining lifetime risk for CVD death, and prolonged survival. In contrast, greater risk factor burden in middle age is associated with higher risk for both CVD and non-CVD death. Despite the higher competing risk for non-CVD death in those with greater CVD risk factor burden in middle age, remaining lifetime risk for CVD death is substantially higher and median overall survival is dramatically lower compared to individuals with favorable profiles.
Our findings suggest that public health and clinical prevention efforts need to focus on individuals well before middle age, since even the presence of a single elevated risk factor in middle age is associated with substantially increased lifetime risk for CVD death and shorter survival. Of note, all of the risk factors that we examined are potentially preventable through appropriate diet and lifestyle choices at younger ages. Among individuals who already have ≥1 elevated risk factors in middle age, our data suggest that intensive global risk factor modification should be considered, given the associated high lifetime risks for CVD death and decreased longevity.
The large differences in lifetime risk for CVD death and median survival between participants with favorable risk factor profiles and those with ≥3 elevated risk factors were striking. Increasing interest is being focused on individuals with optimal or low levels of traditional risk factors. In addition to substantially lower risks for cardiovascular disease, cardiovascular death, and total mortality,4,21,22
individuals with low risk factor levels in middle age also appear to have fewer comorbidities at older ages. Data from the CHA cohort indicate that individuals with favorable risk profiles in middle age who survive to older age have better health-related quality of life 25 years later, compared with those who have elevated risk factors in middle age.23
These data underscore the importance of preventing the development of traditional risk factors at younger ages in order to increase healthy longevity in older adults.
The observation that CVD risk factors are associated with markedly increased risk for non-
CVD death may not be novel, but this association is not widely appreciated. Examination of indicates that the incidence of non-CVD death exceeds that for CVD death at all levels of CVD risk factor burden. Indeed, even among those with the greatest CVD risk factor burden, the risk for non-CVD mortality exceeds the risk for CVD mortality (although the curves in Figure are not adjusted for age). That lifetime risk for CVD death still increases so dramatically with higher risk factor burden, despite the increased competing risk from non-CVD death, indicates the importance of these risk factors. Nonetheless, our findings suggest the possibility that it may be difficult to improve CVD risk estimation using current multivariable equations11,24
by a substantial amount, given the effect of competing risks from non-CVD death associated with the same risk factors that are used to estimate multivariable CVD risk. Current risk equations11,24
do not account for these competing risks.
The present findings confirm and extend the results of a similar analysis from the Framingham Heart Study.2
In that study, Framingham participants were stratified according to risk factor burden at age 50 and lifetime risks for atherosclerotic CVD were estimated. Compared with participants with ≥2 major risk factors, those with low risk factor levels had substantially lower lifetime risks (5.2% vs. 68.9% in men, 8.2% vs. 50.2% in women) and markedly longer median survivals (>39 vs. 28 years in men, >39 vs. 31 years in women).2
The differences in lifetime risks between that study and the present one for those with high risk factor burden can be attributed to the inclusion of non-fatal as well as fatal events in the Framingham analysis.
To date, only a limited number of studies performed in exclusively white cohorts have examined lifetime risks for CVD events.25–29
To our knowledge, no published data are available regarding lifetime risks for CVD among other race/ethnic groups. The number of blacks in our study sample was limited, precluding our ability to compare outcomes directly by race. We had limited power to generate robust lifetime risk estimates for individuals with favorable risk factor profiles in middle age, resulting in wide confidence intervals. Pooling of multiple cohorts would be useful to include enough men and women with low risk factor burden to allow more precise estimates. Our findings may have been subject to some misclassification, since death certificates tend to over-diagnose coronary and/or cardiovascular disease as underlying causes of death.30
Because of the nature of follow up in the CHA cohorts, we did not ascertain non-fatal CVD events. Had we been able to do so, the lifetime risks for all CVD (not just fatal CVD) among those with elevated risk factors would likely have been substantially higher. A limitation of Kaplan-Meier and lifetime risk estimation methods is that they allow only for single assignment into risk factor strata at baseline. Thus, our estimates for each stratum represent averages. Some individuals with low risk factor burden undoubtedly developed new risk factors as they aged, and some with high risk factors undoubtedly received treatment or modified their risk factors through lifestyle changes. Nonetheless, the estimates provided by our results represent useful clinical information, since the risk factor data are what would be available to a clinician and a patient at a given point in time (middle age), when future changes in risk factors or treatment would be difficult to predict. Likewise, our methods do not allow for adjustment for age or other covariates. Lifetime risk estimates may be somewhat subject to birth cohort effects, but inclusion of broad age ranges at baseline and use of long-term follow up tend to reduce birth cohort effects, since multiple individuals from different birth cohorts contribute to any given age-specific incidence of disease or non-disease mortality.