Publication bias: the case for an international registry of clinical trials.

Abstract

A problem in evaluating different therapies from a review of clinical trials is that the published clinical trial literature may be biased in favor of positive or promising results. In this report, a model is proposed for reviewing clinical trial results which is free from publication bias based on the selection of trials registered in advance in a registry. The value of a registry is illustrated by comparing a review of published clinical trials located by a literature search with a review of registered trials contained in a cancer trials registry. Two therapeutic questions are examined: the survival impact of initial alkylating agent (AA) v combination chemotherapy (CC) in advanced ovarian cancer, and the survival impact of AA/prednisone v CC in multiple myeloma. In advanced ovarian cancer, a pooled analysis of published clinical trials demonstrates a significant survival advantage for combination chemotherapy (median survival ratio of CC to AA, 1.16; P = .02). However, no significant difference in survival is demonstrated based on a pooled analysis of registered trials (median survival ratio, 1.05; P = .25). For multiple myeloma, a pooled analysis of published trials also demonstrates a significant survival advantage for CC (median survival ratio, 1.26; P = 04), especially for poor risk patients (ratio, 1.66; P = .002). A pooled analysis of registered trials also shows a survival benefit for patients receiving combination chemotherapy (all patients, P = .06; poor risk, P = .03), but the estimated magnitude of the benefit is reduced (all patients: ratio, 1.11; poor risk: ratio, 1.22). These examples illustrate an approach to reviewing the clinical trial literature, which is free from publication bias, and demonstrate the value and importance of an international registry of all clinical trials.