A multiple imputation strategy for clinical trials with truncation of patient data

Stat Med. 1995 Sep 15;14(17):1913-25. doi: 10.1002/sim.4780141707.


Clinical trials of drug treatments for psychiatric disorders commonly employ the parallel groups, placebo-controlled, repeated measure randomized comparison. When patients stop adhering to their originally assigned treatment, investigators often abandon data collection. Thus, non-adherence produces a monotone pattern of unit-level missing data, disabling the analysis by intent-to-treat. We propose an approach based on multiple imputation of the missing responses, using the approximate Bayesian bootstrap to draw ignorable repeated imputations from the posterior predictive distribution of the missing data, stratifying by a balancing score for the observed responses prior to withdrawal. We apply the method and some variations to data from a large randomized trial of treatments for panic disorder, and compare the results to those obtained by the original analysis that used the standard (endpoint) method.

Publication types

  • Clinical Trial
  • Multicenter Study
  • Randomized Controlled Trial

MeSH terms

  • Adult
  • Alprazolam / therapeutic use*
  • Ambulatory Care
  • Anti-Anxiety Agents / therapeutic use*
  • Antidepressive Agents, Tricyclic / therapeutic use*
  • Bayes Theorem
  • Bias
  • Data Interpretation, Statistical*
  • Double-Blind Method
  • Female
  • Follow-Up Studies
  • Humans
  • Imipramine / therapeutic use*
  • Male
  • Middle Aged
  • Panic Disorder / drug therapy*
  • Treatment Outcome


  • Anti-Anxiety Agents
  • Antidepressive Agents, Tricyclic
  • Imipramine
  • Alprazolam