Reducing the risk of bias in health behaviour change trials: Improving trial design, reporting or bias assessment criteria? A review and case study

Abstract

Objective: It has been suggested that randomised controlled trials (RCTs) of health behaviour change (HBC) interventions are less rigorously designed than – for example– drug trials. This study presents an approach to clarifying whether this is due to poor trial design, incomplete trial reporting and/or the inappropriateness of commonly applied risk of bias assessment criteria.

Design: First, a framework of key sources of bias and common strategies for reducing bias risk is developed based on a literature review. Second, we describe the design of a multi-site RCT evaluating the cost-effectiveness of an HIV-treatment adherence intervention (case study). The choices made by the multidisciplinary team trying to minimise the risk of bias are compared against the risk of bias framework.

Main outcome measures: Implementation of common strategies for reducing the risk of bias in the case study; alternative or additional strategies applied; a justification for each deviation from the risk of bias framework.

Results: Most of the common strategies for reducing the risk of bias could be implemented. Alternative strategies were developed for minimising the risk of performance bias and contamination. Several additional, domain-specific risk of bias strategies were implemented.

Conclusions: The literature provides useful guidance for reducing the risk of bias in HBC trials. Yet, the case study suggests that HBC trial designers may face specific challenges that require alternative/additional measures for reducing the risk of bias. Using the risk of bias justification table (RATIONALE) could lead to better-designed HBC trials, more comprehensive trial reports and the data necessary for evaluating the appropriateness of commonly applied risk of bias assessment criteria to HBC trials.

Keywords:

• risk of bias
• randomised controlled trial
• behaviour change
• intervention
• methodology
• Cochrane

Acknowledgements

The authors would like to express their gratitude to the physicians, nurses and patients from the participation of HIV clinics: Academic Medical Center Amsterdam, Leiden University Medical Center Leiden, Erasmus Medical Center Rotterdam, Slotervaart hospital Amsterdam, St. Lucas-Andreas hospital Amsterdam, Isala clinics Zwolle and HAGA hospital in the Hague. This study is approved by the ethics committee of the Academic Medical Centre Amsterdam, and of each participating centre.

Additional information

Funding

Funding. The AIMS-study is funded by the Netherlands Organization for Health Research and Innovation (ZonMw) as part of the Efficiency (Doelmatigheid) [grant number 171002208].