Quality control in data monitoring of clinical trials.

Eligibility of a patient is based on the patient's status at the time of registration/randomization in the trial and not on something that happens to the patient after entry in the study. All patients entered must be included in the statistical analyses and accounted for in the publication of the results. Stratification done at the time of randomization is used to avoid major imbalances with respect to the distribution of prognostic factors in the different treatment groups. This goal is accomplished only if the information given at randomization is accurate. For this reason, all information given at randomization should be recorded and inserted in the patient's medical file and the randomization done as close as possible to the start of treatment. Protocol violations are to be anticipated but must be kept as low as possible because their percentage will reflect on the quality of the trial. Patients lost to follow up must be kept to a minimum. If patients can no longer attend the consultations at the frequency described in the protocol, every attempt should still be made to follow them for survival. Finally, a clinical trial should be conducted in the time interval foreseen to maintain interest and patient recruitment.