Evaluation of the treatment integrity as part of the risk of bias assessment. A proposal is needed

Article type
Authors
Yakoub NS1, Wolf M2, Munder T2, López-Alcalde J1, Witt CM1, Barth J1
1Institute for Complementary and Integrative Medicine, University Hospital Zurich and University of Zurich, Zurich
2Institute of Psychology, University of Zurich, Zurich
Abstract
Background: Risk of bias (RoB) assessment is essential in systematic reviews (SRs). The assessment of the risk of performance bias requires to determine the degree to which the interventions were delivered as intended*. The successful implementation of an intervention is also known as treatment integrity (TI)**. The determination of TI is challenging, particularly in studies of face-to-face non-pharmacological interventions, such as psychotherapy, or other complex interventions.

Objectives: Our project has two specific goals. First, to develop a framework for the assessment of TI in each arm of a study evaluating the effects of a face-to-face non-pharmacological psychotherapeutic intervention. Second, to identify core domains of TI that the Cochrane RoB 2 tool should consider.

Methods: Descriptive study of an ongoing project. We generated a list of potential TI domains. We collated these domains from the quality/risk of bias tools used in two sources: a sample of SRs of psychotherapeutic interventions published in Clinical Psychology Review and one SR of quality indicators of psychotherapy outcome studies***. We will share this list of domains and related sub-domains with a small group of experts to check if all relevant domains were considered. This preliminary list will be used in an upcoming survey with approximately 300 invited people (covering psychotherapists, researchers, Cochrane reviewers and patients). The survey will inform about the adequacy of each domain (or sub-domain) and the feasibility to rate each sub-domain. We will conduct a consensus meeting with key stakeholders (preferably Cochrane reviewers, meta-analysts of psychotherapy outcome studies and trialists) to agree on the TI domains and sub-domains that the RoB 2 tool should include.

Results: A total of nineteen different quality/risk of bias tools were identified. Eleven tools (58%) considered TI in at least one domain. Our preliminary analysis suggests that the tools considered heterogenous aspects concerning TI, such as the competence of the provider, the compliance by the patients, or the description of co-interventions.

Conclusions: To assess TI is critical in the risk of bias assessment. However, SRs of face-to-face non-pharmacological psychotherapeutic interventions assessed TI inconsistently. There is an urgent need to define a pragmatic, explicit and reproducible approach to determine TI in SRs.

Patient or healthcare consumer involvement: Valid assessment of the TI is a critical step of the risk of bias assessment. Besides, to assess the TI can also be useful for providers and healthcare consumers to know the acceptability and feasibility of an intervention. Therefore, we plan to integrate psychotherapists and patients’ views in our project.

Bibliography: *Cochrane Handbook 6; **PMID:1808590; ***PMID: 26169720.