Bayesian analysis from phase III trials was underused and poorly reported: a systematic review

Area of Science:

• Clinical Trials Methodology
• Biostatistics
• Health Research Reporting

Background:

• Bayesian methods offer advantages in clinical trial analysis but their adoption in phase III trials remains limited.
• Standardized guidelines like BayesWatch and ROBUST aim to improve the reporting of Bayesian analyses.
• Assessing the adherence to these guidelines is crucial for ensuring the transparency and replicability of trial results.

Purpose of the Study:

• To evaluate the reporting quality of Bayesian analyses in phase III trials using the BayesWatch and Reporting Of Bayes Used in clinical STudies (ROBUST) guidelines.
• To identify specific areas where reporting of Bayesian methods is inadequate.

Main Methods:

• A systematic literature search was conducted across Cochrane CENTRAL, Ovid MEDLINE, and EMBASE to identify phase III trials employing Bayesian methods.
• Two independent reviewers screened the retrieved articles, with 49 trial reports ultimately included in the analysis.
• Reporting items were assessed against the BayesWatch and ROBUST guidelines.

Main Results:

• The median reporting score was 4/7 for ROBUST and 7/10 for BayesWatch guidelines.
• Specific Bayesian elements, including analytic techniques and prior specifications, were poorly reported.
• Priors were reported in two-thirds of studies but justified in only 22%; sensitivity analyses were reported in 57% of trials.

Conclusions:

• Bayesian methods are underutilized in phase III trials.
• Current reporting practices frequently omit essential methodological details of Bayesian analyses, particularly prior specifications.
• Improved reporting is critical for the accurate interpretation and replicability of Bayesian clinical trial results.