Étude PCPT : les biais attendus et leur gestion

Abstract

The aim of the PCPT trial was to compare the incidence of prostate cancer in men treated with finasteride or placebo. The study protocol contains some methodological difficulties. Solutions were adopted designed to minimise damaging bias which could have incorrectly suggested that finasteride was beneficial, or masked a benefit of the drug. As finastéride reduced PSA by an average of 50% the PSA values were interpreted against a new threshold to provide as many biopsies in the finasteride arm as in the placebo arm. In view of the reduction in gland volume, the biopsy protocol was modified to prioritise lateral peripheral cores. Patient adherence to treatment in the finasteride arm and contamination of the placebo arm by people taking finasteride retrospectively after randomisation were taken into account. The increase in sensitivity of rectal examination and in the number of prostatic resections in the placebo arm were also two sources of bias in favour of finasteride. The different envisaged sources of bias were analysed and plans were adopted to manage these wherever possible. Analysis of the sources of bias led to the conclusion that systematic end of trial biopsies constituted the most robust end point.