A Generalized Estimator of the Attributable Benefit of an Optimal Treatment Regime

摘要: For many diseases where there are several treatment options often is no consensus on the best to give individual patients. In such cases, it may be necessary define a strategy for assignment; that is, an algorithm dictates should receive based their measured characteristics. Such or also referred as regime. The optimal regime would provide most public health benefit by minimizing poor outcomes possible. Using measure generalization of attributable risk (AR) and notions potential outcomes, we derive estimator proportion events could have been prevented had implemented. Traditional AR studies look at added can attributed exposure some contaminant; here will instead study using strategy. We show how regression models used estimate large sample properties this estimator. As motivating example, apply our methods observational 3856 patients treated Duke University Medical Center with prior coronary artery bypass graft surgery further heart-related problems requiring catheterization. either medical therapy alone combination percutaneous intervention without general which