The problem of attributing deaths of nonadherers: the VA coronary bypass experience.

In the VA Coronary Artery Surgery Study 84 of the 354 patients randomized to medical treatment had subsequent bypass surgery during a 6-year follow-up period. Of these, 18 had left main disease. The remaining 66 of 311 without left main disease crossed over to surgery at a constant rate of about 4% per year. The traditional assumption that patients who crossed over were at greater risk of dying than those who adhered was not substantiated by an analysis of known baseline risk factors. Crossover was not related to number of vessels diseased or other angiographic or noninvasive risk factors but was related to severity of angina which was not a risk factor in this study. These findings lend support to the validity of analyzing our survival data by the crossover method where nonadherers are counted as lost to follow-up at the time of treatment change.

RCT Entities:   Population Interventions Outcomes