CORRESPONDENCE events. A significantly greater proportion of participants with ED experienced hard events than those without ED (CHD hard events: 3.4% versus 1.4%, P< 0.001; CVD hard events: 6.3% versus 2.6%, P< 0.001). In the unadjusted Cox models, ED was a significant predictor of both hard CHD (hazard ratio, 2.5; 95% confidence interval [CI], 1.3–4.8) and CVD (hazard ratio, 2.6; 95% CI, 1.6–4.1) events. In the fully adjusted models (model 3), ED remained a significant predictor of hard CVD events (hazard ratio, 1.9; 95% CI, 1.1–3.4), whereas hard CHD events became nonsignificant, albeit with a similar point estimate of risk (Figure). In the shifted-time cross-sectional analysis, a significant association was also seen between prior CVD event and ED at visit 5 (odds ratio, 2.1; 95% CI, 1.4–3.2), which remained significant but was attenuated by medication use and depression in the fully adjusted models (odds ratio, 1.7; 95% CI, 1.1–2.6). In an ethnically diverse, community-based cohort, ED was found to be a significant predictor of hard CVD events after adjustment for traditional CVD risk factors, depression, and β-blocker use. To our knowledge, this is the first study of ED and subsequent CVD that adjusted for depression and β-blocker use. Our results suggest that these 2 factors may partially mediate the relationship between prior CVD and subsequent ED, but do not attenuate the prospective association of ED and incident CVD. Our findings strengthen the existing evidence for the independent association between ED and incident CVD, and could have important clinical implications for risk stratification in middle-aged men. We have previously documented increased subclinical atherosclerosis in those who subsequently report ED. 4 In 2017, the UK QRISK score was the first to incorporate ED as an independent risk factor for CVD, 5 yet ED remains absent from US risk prediction guidelines. Our results may justify more aggressive preventive therapy in such patients.

Our study had certain limitations. Although similar to the primary care assessment of ED, the single Massachusetts Male Aging Study question does not distinguish between vascular and nonvascular types of ED, which may have attenuated the association between ED and CVD. In addition, because our follow-up was just 3.8 years, additional 10-year data on the risk predictive value of ED are needed. In conclusion, our study provides some of the strongest evidence to date for the independent predictive value of ED in a modern, multiethnic, well-phenotyped cohort.