Background and purpose: End point adjudication committees (EPAC) are widely used in large-scale clinical trials to ensure the robustness of diagnosis for end points.
Methods: The Perindopril Protection Against Recurrent Stroke Study (PROGRESS) was a double-blind randomized trial of blood pressure lowering in 6105 participants with pre-existing cerebrovascular disease. Separate estimates of the effects of randomized treatment were determined using Cox regression models that were based on the unadjudicated events initially reported by the investigator and on the final events assigned by the EPAC.
Results: There were 992 strokes initially reported by the investigators and 894 (90%) retained these diagnoses after adjudication by the EPAC. The hazard ratios (95% CIs) for the effect of randomized treatment on stroke were 0.74 (0.64 to 0.85) based on the investigator diagnoses and 0.72 (0.62 to 0.83) based on the EPAC diagnoses (P homogeneity=0.7). For each stroke subtype reported, the corresponding numbers of diagnoses (investigators/EPAC) were ischemic (593/565), hemorrhagic (124/111), and unknown (124/93) with no impact of the EPAC review on the estimates of treatment effects (all P homogeneity >0.3). There was likewise no detectable effect of reclassification of diagnoses for the effect estimates calculated for myocardial infarction or the main causes of death (all P homogeneity >0.5).
Conclusions: The EPAC process had no discernible impact on the trial conclusions. Very large trials powered to detect effects on stroke subtypes might obtain real scientific gain from an EPAC, but in the case of PROGRESS, the value of the EPAC was in the reassurance it provided.