When selecting clinical trial end points, some investigators prefer cardiovascular death (CVD) while others believe that all-cause mortality is more relevant. However, the relative contribution of CVD to 1-year mortality after contemporary percutaneous coronary intervention (PCI) is not known.

We evaluated the mode of death (MOD) in EVENT, a prospective PCI registry at 55 US hospitals. Vital status was assessed at 6 and 12 months as part of the study protocol, and MOD was independently reviewed in blinded fashion.

Between 2004 and 2007, EVENT enrolled 10,144 patients of whom 295 (2.9%) died within the first year: 51 (17%) ≤30 days; and 244 (83%) between 31 and 365 days after index PCI. Overall, CVD accounted for 42% of deaths, and no clear cause could be identified in a substantial subgroup (25% of deaths). Among patients who died ≤30 days, the MOD was more likely to be CVD (odds ratio 3.96, 95% CI 2.08-7.55), whereas the incidence of CVD and non-CVD was similar after 30 days. Findings were similar after a series of sensitivity analyses including reassignment of unknown MOD to the CVD category, using multiple imputation modeling, or when evaluating MOD in prespecified subgroups of patients with diabetes, acute coronary syndromes, or left ventricular dysfunction.

Among unselected PCI patients, 1-year mortality is approximately 3%, and CVD is confirmed in <50% of all deaths. Regardless of analytic approach, CVD is the primary contributor to overall mortality during the first 30 days after PCI, whereas rates of CVD and non-CVD are remarkably similar after the first month after PCI.