Author + information
- Jarett D. Berry, MD, MS⁎ (, )
- Benjamin Willis, MD, MPH and
- Donald M. Lloyd-Jones, MD, MSc
- ↵⁎University of Texas Southwestern Medical Center Cardiology, 5323 Harry Hines Boulevard, Dallas, Texas 75390
We appreciate the thoughtful comments from Dr. Kuller on our recent study (1) demonstrating the contribution of measured fitness in midlife to the lifetime risk for cardiovascular disease (CVD) mortality. He raises important questions regarding: 1) the contribution of fitness to both CVD and non-CVD mortality; 2) the generalizability of the findings from the CCLS (Cooper Center Longitudinal Study); and 3) the consistency of the findings across individuals with an abnormal electrocardiography (ECG) response.
First, as Dr. Kuller points out, fitness is associated with both CVD and non-CVD mortality. In the present study, compared with low fitness in mid-life, high fitness was associated with both lower CVD mortality (13.1% vs. 33.1%) and non-CVD mortality (22.3% vs. 34.9%) using the standard Kaplan-Meier cumulative incidence estimate that ignores competing risks. The very purpose of the present study was to determine the association between fitness and CVD mortality after taking into account competing risks from non-CVD mortality. After adjustment for the competing risks in our methods, we see that the association between low fitness and CVD death is attenuated by nearly 20% (adjusted cumulative incidence 27% vs. unadjusted cumulative incidence of 33.1%). Therefore, the adjustment for competing risks from non-CVD death provides a more conservative and more realistic estimate of the association between fitness in mid-life and long-term risk for CVD death.
Second, we believe that the estimates reported in the present study are representative of the general population, particularly for men at ages 55 and 65 years. As we reported in Table 2 in our paper (1), the association between the burden of traditional risk factors and lifetime risk for CVD mortality in the CCLS was strikingly similar to those observed in the Lifetime Risk Pooling Project, a combined analysis from 16 representative cohorts. Thus, although the burden of risk factors is lower in the CCLS, the effect of these risk factors is quite similar to other, more representative cohorts.
Finally, in the present study, we sought to extend our prior lifetime risk work and therefore did not exclude the small number of participants with an abnormal ECG response to exercise (7%). Nevertheless, the contribution of abnormal ECG response to fitness in this dataset is limited. Recently, we reported a systematic analysis of the contribution of fitness to CVD risk prediction (2). In this study, we observed that the incremental contribution of fitness to the net reclassification index was similar across all subgroups, including those with and without an abnormal ECG response.
- American College of Cardiology Foundation
- Berry J.D.,
- Willis B.,
- Gupta S.,
- et al.
- Gupta S.,
- Rohatgi A.,
- Ayers C.R.,
- et al.