Reply to Comment on Effect of Resting Heart Rate on All-Cause Mortality and Cardiovascular Events According to Age.

Abstract

To the Editor: We are grateful to Saeid Safiri and his colleagues1 for their comments on our work and providing such wonderful suggestions in terms of methodological issues, which may make our findings, if addressing these issues, more convincing. Indeed, given the survival plots of all-cause mortality and cardiovascular events among people aged <60 years are not completely parallel in our study,2 one may argue that the Cox proportional hazard regression model used in our study is inappropriate. To assess the proportional hazards (PH) assumption, we fitted extended multivariate Cox regression models by adding a time-dependent covariate (resting heart rate [RHR] X time) in the Cox regression model as they suggested and found that all PH assumptions were appropriate regardless of among people aged <60 years or ≥60 years, which suggests the Cox regression models we used in our original paper were appropriate. In general, we should be concerned about the PH assumption whenever the lines run at different angles, particularly when they cross in terms of the survival plot. However, lines can cross due to unusual values; and sampling variation can produce more subtle perturbations from parallel relations. Accordingly, the survival plot has its limitations as a diagnostic tool to judge whether PH is violated in a study; it would be better to combine it with a few other methods,3 like the extended Cox regression model that we used in the aforementioned section. With regard to the second issues they mentioned, given the aim of our study was to explore an effect modification of age on the association of RHR with risks of CVD events and mortality but not to assess an exact association of heart rate with cause-specific mortality, we used all-cause mortality as a compound outcome instead of cause-specific mortality. The results in our study indicate the utility of RHR for risk stratification may vary according to age. Of course, strength and direction of association of RHR with cause-specific mortality may be not homogenous as in previous studies;4 however, in our opinion it might not affect the effect modification of age on the potential risks of RHR for public health when we used aggregated outcomes in our study. Moreover, there is not anything inherently wrong with analyses of broader outcome groups constructed by aggregating several outcome types. Broad groupings of outcomes may be of interest from a public health perspective.5 Conflict of Interest: The authors declare no financial, personal, or potential conflicts of interest. Authors Contributions: Study concept, drafting of the manuscript, and design: Kuibao Li, Chonghua Yao, and Lei Dong. Critical revision of the manuscript: Chonghua Yao and Xinchun Yang. All of the authors gave final approval of the version to be published. Sponsor's Role: None.