In Reply For analyses of both all-cause and cause-specific mortality in our study,1 follow-up was calculated from date of baseline assessment to date of death (from any cause) or end of follow-up. For analyses of cause-specific mortality, we censored at the age of death from any cause. Consequently, the hazard ratio presented in our study for each cause of death is the relative risk of dying from that cause among the risk set of people who are alive. This approach not only ensured consistency with past studies of coffee drinking and mortality,2,3 but also offered an easily understandable quantity—namely the relative risk of dying from a specific cause among people who are alive. Thus, it is the preferred statistical approach for etiologic research questions as described in detail by Noordzij and colleagues.4 We appreciate that Cox proportional hazards regression assumes independent competing risks and that violations of this assumption could be envisioned; however, no data exist to evaluate whether this assumption holds true. Nevertheless, independence of competing events is not required to obtain valid estimates from a cause-specific approach, such as what was used in our analysis.5
Loftfield E, Graubard BI, Freedman ND. The Alleged Health-Protective Effects of Coffee—Reply. JAMA Intern Med. 2018;178(12):1726. doi:10.1001/jamainternmed.2018.6545
Customize your JAMA Network experience by selecting one or more topics from the list below.
Create a personal account or sign in to: