A model for the interim analysis process: a case study

Abstract

To evaluate data from a clinical trial before its completion, researchers routinely perform interim analyses. However, if not performed carefully, interim analyses can compromise the integrity of a clinical trial. In the last 10-15 years, regulatory authorities and the pharmaceutical industry have developed procedures and guidelines to allow trial sponsors access to unblinded data in an ongoing clinical trial without affecting the outcome. In December 1996, Abbott Laboratories was codeveloping a drug for treatment of an autoimmune disease. The pivotal phase II/III trial for the new drug application was very expensive, large, long term, and slow-accruing. The trial was initiated with a great deal of uncertainty concerning the safety and efficacy of the proposed treatment. An interim analysis was a logical part of the trial design. Two interim analyses were performed; the second analysis resulted in early termination of the trial. This article describes the interim analysis, including the process used for planning and execution and the lessons learned from the experience. In addition, the methodology for performing an interim analysis and the roles and responsibilities of involved members are discussed.