Article Text

PDF
FRI0192 Too many patients in randomized controlled trials (RCTS) of ra? a critical reassesment of power and the actual numbers of patients enrolled
  1. S. Celik1,
  2. H. Yazici2,
  3. Y. Yazici1
  1. 1Department of Medicine, Rheumatology Division, New York University, Hospital for Joint Diseases, New York, United States
  2. 2Rheumatology, Medicine, Cerrahpasa Medical Faculty, University of Istanbul, Istanbul, Turkey

Abstract

Background A proper power calculation and adequate sample size are critical for RCTs to avoid enrolling more than the needed number of patients leading to unnecessary exposure and adverse events (AE). On the other hand underpowered studies will fail to show any significant findings. Since in the real world ideal power calculations are not always possible, it would be expected that under and over enrolling would occur in roughly similar proportions in reports of RCTs. Nevertheless we had the impression that RCTs were very frequently over enrolling patients.

Objectives To review power calculations of RCT in RA and determine if there is over-, under- enrollment of patients.

Methods A Pubmed search was conducted for RCTs of infliximab, etanercept, adalimumab, abatacept, rituximab and tocilizumab in RA patients (n=291). Only original initial studies where the primary outcome was efficacy were analyzed (n=40). As a first analysis, based on a two equal arms study with an alpha error of 0.05 and a power of 80 %, recalculation of the sample size was done for each reported study by using the primary outcome results. Next the same was done only in those 2 arm studies, with equal number of patients in both arms, also in which all elements of the power calculation were adequately given in the manuscript.

Results 40 studies were analyzed. In the first analysis, the recalculated mean sample size was 62 ± 27 (95% CI) for either arm while mean for the number of patients actually enrolled was 154 ± 46 (CI). There were more than the required number of patients in 32/40 studies (80 %) with a mean 131 ± 46 (CI) extra patients in the treatment groups of per study. Out of the 29 studies in which power was given 11 studies qualified for the second analysis. The mean power for sample size in these 11 studies was 0.88. The recalculated mean sample size was 76 ± 32 (CI) for the treatment arm while the mean for the actually enrolled patients, again in the treatment arm was 115 ± 49 (CI). There were still more patients with a mean 48 ± 19 extra patients enrolled in this group.

Conclusions Most RCTs in RA enroll more patients than needed. While this practice is a misuse of the RCT, still a very important research tool, it also has the immediate consequence of exposing more than needed number of patients to potential AEs.

Disclosure of Interest S. Celik: None Declared, H. Yazici: None Declared, Y. Yazici Consultant for: Abbott, BMS, celgene, genentech, Horizon, UCB, Pfizer

Statistics from Altmetric.com

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.