Background In order to monitor the safety of patients in clinical trials, data safety monitoring boards (DSMB) are organised. These DSMB contain independent volunteer experts in the medical field of the research (clinician, pharmacologist, methodologist or statistician). They give their advice about the continuation with or without modification or stopping of the study. They are increasingly questioned by competent authorities during the study authorisation. This is an ambitious challenge to improve the organisation of these DSMB which requires time and work, and relatively few studies have looked at this topic.

Purpose On which types of studies should we concentrate our efforts to implicate experts and optimise these DSMB?

Material and methods The study design is an observational retrospective study, based on a register of an academic sponsor. It provides data from August 2011 until September 2017 on 89 clinical trials (investigational medical products, medical devices, other than health products) with DSMB. We have analysed the following parameters: type of study, meeting before patient inclusion, meetings during studies and actions taken following the decisions of the DSMB. The implication of experts after a meeting was measured by the decision of experts for all types of studies. We tested the hypothesis that initial meeting before the start of trials may aid a superior involvement of the DSMB members. A Chi-² test was used in order to compare observed proportions.

Results Seventy-eight per cent of DSMB recommendations were to modify or stop the trial in Phase I or I/II drugs trials against 36% in the medical devices study, 11% in Phase II and III drugs trials, and 8% in trials other than for health products. The establishment of initial meetings has highlighted the percentages of recommendations by DSMB members of 86% compared to 15% without initial meetings (p<0.0001).

Conclusion The major importance of DSMB decisions were for the precocious drugs phase trials and medical device studies. A DSMB is necessary for these study types, unlike others types. The initial meetings before the start of the study are one of the main parts of this challenge. A national survey is needed in order to validate our results and make recommendations.

No conflict of interest