Early termination of clinical trials with prolonged observation of individual patients: a case study

Stat Med. 1987 Dec;6(8):927-37. doi: 10.1002/sim.4780060807.

Abstract

Stopping rules for a placebo controlled clinical trial of anticoagulants after acute myocardial infarction were evaluated by means of computer simulation for the case of five interim analyses. The trial will be terminated and the null hypothesis of no treatment effect rejected when the one-sided P-value (logrank test) is lower than 0.005, 0.005, 0.005, 0.014, and 0.023 at the respective interim analyses, and 0.032 at final evaluation. This implies a total size alpha = 0.05 and a power close to that of fixed sample size testing. The trial will also be terminated, without rejecting the null hypothesis, when the one-sided P-value exceeds 0.95, 0.88, 0.81, 0.74, and 0.67 at the respective interim analyses. This modification hardly affects size and power.

Publication types

  • Clinical Trial
  • Randomized Controlled Trial

MeSH terms

  • Anticoagulants / therapeutic use*
  • Clinical Trials as Topic / methods*
  • Computer Simulation
  • Double-Blind Method
  • Humans
  • Models, Biological
  • Myocardial Infarction / drug therapy*
  • Myocardial Infarction / mortality
  • Random Allocation
  • Research Design*
  • Statistics as Topic

Substances

  • Anticoagulants