Welcome!

On these pages, you can find the latest version of the "Metapsy RoB specifications for the use in psychotherapy trials" extended manual, which is the basis of the Metapsy risk of bias tool.

Preliminary considerations

Metapsy RoB comprises 24 items that assist the conduct of risk of bias assessments in the context of psychological interventions and large meta-analytic projects. All items are formulated in the same direction, in which “Yes/PY” answers are indicative of lower risk of bias.

The current recommendations of this guidance document are applicable for individually randomized parallel-group trials. Some of the core specifications can be applied for cluster randomized trials or crossover trials, however, these types of designs pose specific challenges that should be further evaluated. Further guidance for these types of designs is available in Cochrane’s RoB 2 documentation (Higgins et al., 2019).

As recommended in the risk of bias literature (e.g., Cochrane’s RoB 2 guidelines), assessment of risk of bias should be specific to a particular result (e.g., standardized mean difference between psychotherapy and control condition) for a particular outcome (e.g., depression severity) and measured at a particular time (e.g., post-test), although some biases may apply to the whole study.

The recommendations of this document are aimed at examining the effect of assignment to intervention, given that this is the type of assessment of primary choice by clinical guideline developers and policy makers, and also typically of Metapsy databases. However, if reviewers wish to assess the effect of adhering to intervention, Cochrane’s RoB 2 documentation contains a section specific to that aim.

"createRobRatings"

Variables marked with the tag createRobRatings↗ do not need to be answered manually by the reviewers if there is a Metapsy formatted dataset that has already collected required variables for these items: rating, attrition in each arm, and randomization ratio.

PDF Version

A PDF version of this handbook can be found here.