Only 26% of collected data are published in the main manuscript of randomized clinical trials: could we do better? A retrospective methodological analysis of two French academic centres

Case report form (CRF) conception is an important step in research methodology. An excessive amount of data to collect may however negatively affect data management and resource utilization. The aim of this work was to evaluate the percentage of data used in published manuscripts out of the total number of data collected. Secondary objectives included the estimation of percentage of data used by variable categories and the cost of unused but collected data. A retrospective methodological analysis was conducted using data from 8 randomized clinical trials in anesthesia and intensive care published in high impact factor journals from 2 University Hospital Centres (UHCs) of Clermont-Ferrand and Montpellier, France. We constructed a counting method based on the overriding principle that any data field that required an investigator or delegate to record data was considered a single data item. Two independent data managers performed the count. Inconsistencies were resolved with the statistician for each trial. Data were sorted in seven categories. The relative cost of data items was estimated according to the time spent for data management of research members of each study team with standard salary scales for personnel. Study sample size ranged between from 298 to 1000 participants with a median at 623. The rate of data used in manuscript is around 26% (3212/12197) ranging between 10% (309/2959) and 53% (557/1042). There was no significant difference between-center (28% vs 26%). There was significant heterogeneity according to variable categories ranging from more than 50% (demography 100% and medical history 71%) to less than 30% (biology 28%, concomitant medications 23%, study treatment 21%). The costs calculated by taking only the variables used (useful) per study show a cost saving ranging from 12,489 euros to 127,275 euros, i.e. a cost saving of 28.72 to 411.90 euros per variable. Findings from this first multicentre study, conducted with a robust and reproducible methodology, highlight that a significant rate of collected data are not reported in final publication. By addressing these issues, we aim to provide evidence supporting more rational CRF design and contribute to the growing movement toward optimizing research efficiency.