Assessment of methodological quality of adaptive clinical trials

Article type
Authors
He R1, Jackson R2, Reljic T3, Mhaskar R3, Kumar A4, Djulbegovic B4, Miladinovic B3
1University of South Florida College of Public Health, Department of Epidemiology and Biostatistics, USA
2University of South Florida College of Medicine, Department of Otolaryngology Head and Neck Surgery, USA
3University of South Florida College of Medicine, Department of Internal Medicine , USA
4University of South Florida College of Medicine, Department of Internal Medicine , USA
Abstract
Background:
Adaptive clinical trials have become increasingly popular in recent years and about 20% of clinical trials currently underway already use adaptive designs. Unlike traditional clinical trials, they allow for a review of accumulating trial information and modification of trial characteristics. Therefore, the methodological and design standards are the key issues in their reporting.

Objective:
To assess the methodological quality of reporting of adaptive clinical trials.

Methods:
MEDLINE and The Cochrane Library were searched from the inception to 4 November 2013. Data were abstracted independently on demographics, design, and analysis.

Results:
The search yielded 2136 citations, from which 95 adaptive clinical trials were included in the final analysis (Figure 1A). Fifty-two per cent of the trials were industry sponsored and 93% of the publications were full manuscripts. The median sample size was 124 (interquartile range 57 to 297). Only 42% of the trials indicated the type of design in the title and 82% in the abstract; 44% presented the flow diagram of the participants in the study. Most studies were dose-related (35%), followed by adaptive group sequential (27%), adaptive randomization (11%), and multiple design (10%). Most trials were frequentist (68%), followed by Bayesian (22%) and hybrid (10%). In 65% the statistical model used was pre-specified. Only 42% reported stopping rules in the manuscript and 60% described the interim analyses (Figure 1B). In the frequentist or hybrid trials, 63% presented P value and confidence intervals allowing for summary data to be used in a meta-analysis; 92% of all trials reported statistical methods used to evaluate primary outcomes and 79% to evaluate secondary outcomes.

Conclusions:
Reports of adaptive clinical trials often do not publish important issues related to the design, conduct, analysis, and data presentation. To establish definitive criteria for the conduct and reporting of adaptive clinical trials, a new set of guidelines should be proposed and implemented.