Assessment of study quality for systematic reviews: a comparison of the Cochrane Collaboration Risk of Bias Tool and the Effective Public Health Practice Project Quality Assessment Tool: methodological research

TLDR

The Cochrane Collaboration promotes the newly developed Cochrane Collaboration Risk of Bias Tool (CCRBT), yet its psychometric properties remain uncharacterised. This study aimed to evaluate the CCRBT’s inter‑rater reliability and concurrent validity against the Effective Public Health Practice Project Quality Assessment Tool (EPHPP) by assessing 20 RCTs on cancer‑pain knowledge‑translation interventions, and to highlight the need for broader validation. Two reviewers independently applied both tools to each trial, and the resulting domain scores and overall grades were analysed for inter‑rater agreement. EPHPP showed fair domain agreement and excellent final‑grade agreement, whereas CCRBT exhibited only slight domain agreement and fair final‑grade agreement, with no concordance between tools’ overall grades, indicating that the two instruments assess distinct constructs and that CCRBT tends to assign higher risk of bias.

Abstract

The Cochrane Collaboration is strongly encouraging the use of a newly developed tool, the Cochrane Collaboration Risk of Bias Tool (CCRBT), for all review groups. However, the psychometric properties of this tool to date have yet to be described. Thus, the objective of this study was to add information about psychometric properties of the CCRBT including inter-rater reliability and concurrent validity, in comparison with the Effective Public Health Practice Project Quality Assessment Tool (EPHPP).Both tools were used to assess the methodological quality of 20 randomized controlled trials included in our systematic review of the effectiveness of knowledge translation interventions to improve the management of cancer pain. Each study assessment was completed independently by two reviewers using each tool. We analysed the inter-rater reliability of each tool's individual domains, as well as final grade assigned to each study.The EPHPP had fair inter-rater agreement for individual domains and excellent agreement for the final grade. In contrast, the CCRBT had slight inter-rater agreement for individual domains and fair inter-rater agreement for final grade. Of interest, no agreement between the two tools was evident in their final grade assigned to each study. Although both tools were developed to assess 'quality of the evidence', they appear to measure different constructs.Both tools performed quite differently when evaluating the risk of bias or methodological quality of studies in knowledge translation interventions for cancer pain. The newly introduced CCRBT assigned these studies a higher risk of bias. Its psychometric properties need to be more thoroughly validated, in a range of research fields, to understand fully how to interpret results from its application.