A Comparison of Methods for Estimating the Causal Effect of a Treatment in Randomized Clinical Trials Subject to Noncompliance

Abstract

We consider the analysis of clinical trials that involve randomization to an active treatment (T = 1) or a control treatment (T = 0), when the active treatment is subject to all-or-nothing compliance. We compare three approaches to estimating treatment efficacy in this situation: as-treated analysis, per-protocol analysis, and instrumental variable (IV) estimation, where the treatment effect is estimated using the randomization indicator as an IV. Both model- and method-of-moment based IV estimators are considered. The assumptions underlying these estimators are assessed, standard errors and mean squared errors of the estimates are compared, and design implications of the three methods are examined. Extensions of the methods to include observed covariates are then discussed, emphasizing the role of compliance propensity methods and the contrasting role of covariates in these extensions. Methods are illustrated on data from the Women Take Pride study, an assessment of behavioral treatments for women with heart disease.

Impact of Collaborative Care for Underserved Patients with PTSD in Primary Care: a Randomized Controlled Trial

“…To correct for the bias due to patients' engagement in the PCM, we used the inverse propensity score (PS) weighting method. 40,41 The PS was a conditional probability of engaging in PCM, which is equal to the product of the probability of being randomized to the PCM condition and the conditional probability of engaging in the PCM condition given observed patient characteristics and baseline outcomes. We weighted patients who engaged in PCM by the inverse of their PS, and all other patients by the inverse of one minus their PS.…”

Noncompliance in lifestyle intervention studies: the instrumental variable method provides insight into the bias

“…The approach often used to estimate the maximal treatment effect is the as-treated or perprotocol analysis. However, these methods are prone to selection bias because compliers and noncompliers may have different characteristics [32].…”

Principal Stratification -- Uses and Limitations

“…Numerous studies have fruitfully employed the idea (e.g. Angrist et al, 1996;Imbens and Rubin, 1997;Balke and Pearl, 1997;Cheng and Small, 2006;Cuzick et al, 2007;Little et al, 2009). The central insight was that in a randomized trial with non-compliance in which the group assigned the placebo had no access to treatment (i.e.…”