Do prospective randomized controlled trials comply with filed protocols? Spin study of 206 trials from 2010 to 2023 - 04/10/24

Doi : 10.1016/j.otsr.2024.104013

Roger Erivan ^a,⁎ , Bastien Michon ^b, Guillaume Villatte ^a, Stéphane Descamps ^a, Stéphane Boisgard ^a, Pierre Martz ^b,^c
^a Université Clermont Auvergne, CHU Clermont-Ferrand, CNRS, SIGMA Clermont, ICCF, F-63000 Clermont–Ferrand, France
^b Service de Chirurgie Orthopédique et Traumatologique, Hôpital François Mitterrand, CHU, 21079 Dijon Cedex, France
^c INSERM, UMR1093-CAPS, Université Bourgogne Franche-Comté, UB, 21000, Dijon, France

^⁎Corresponding author.

Sous presse. Épreuves corrigées par l'auteur. Disponible en ligne depuis le Friday 04 October 2024

Abstract

Introduction

Prospective randomized controlled trials (RCT) have a robust methodology, but some distortions may occur during the course of study. Some protocols may not be available at the time an article is reading. Some authors may change the methodology between the time the protocol was submitted and when the trial results are actually published. Others may distort results to favor more attractive findings and draw conclusions that support prior hypotheses. This has rarely been investigated and none explored the RCTs published in the Journal of Bone and Joint Surgery (JBJS). Therefore, we did a retrospective investigation aiming to determine: (1) the proportion of trials with a protocol deposited and accessible to the reader, (2) whether the trials scrupulously followed the filed protocols.

Hypothesis

Protocols were available in over 80% of cases, and these protocols were followed in over 80% of trials for the primary endpoint.

Patients and methods

This was a retrospective study of articles published in the JBJS between January 2010 and November 2023. Differences in primary and secondary endpoints between protocols and articles were sought.

Results

Of the 206 RCT articles studied, 113 (54.9%) described clear and identifiable endpoints, and 93 (45.1%) were not identifiable and were inferred in the results; 184 (89.3%) articles identified a trial protocol. For the 184 articles (89.3%) declaring a trial protocol in the text, 23 (11.1%) protocols were not accessible. In all, 45 articles (21.8%) thus had no protocol available on the Internet (i.e., not available to the reader either because it was not cited in the text or because it was not accessible) so we analyzed 161 articles. The primary endpoint remained unchanged in 97 articles (60.2%) out of the 161 studied, was changed in 64 articles (39.8%), and was lacking (protocol not accessible) in 45 articles (21.8% of all articles). The secondary endpoints of the articles were unchanged in 61 articles (37.9%) out of the 161 studied.

Discussion

Like other leading journals, JBJS publishes RCT articles containing a significant proportion of inconsistencies between preoperative trial protocols and the methods actually used in the research.