DOI: 10.12688/wellcomeopenres.26912.1 ISSN: 2398-502X

Instrumental variable analysis applied to randomised controlled trials of glucagon-like-peptide-1 receptor agonists and bariatric surgery: A study protocol

Winfred Gatua, Yi Liu, Deborah A Lawlor, Julian PT Higgins, Tom R Gaunt
Show PDF Cite
Background Randomised controlled trials (RCTs) can be used to determine causal effect of an exposure (e.g. body mass index (BMI)) on an outcome (e.g. blood pressure (BP)) using randomisation as an instrumental variable (IV). The increasing number of weight-loss trials provides an opportunity to estimate the effect of BMI on BP using IV analysis. Methods To identify RCTs to be included in the IV analysis, we will search Cochrane Library, Embase and MEDLINE for systematic reviews of a) glucagon-like-peptide-1 receptor agonists and b) bariatric surgery trials. One reviewer will evaluate database search output to identify two high-quality systematic reviews and extract RCT studies from each review. Two reviewers will independently evaluate each RCT study for inclusion in the IV analysis. We will extract data from included RCTs using Microsoft Excel. We will develop a tool for assessing validity of randomisation as an IV. We will adapt and enhance Risk of bias for Non-randomised studies of Intervention (ROBINS-I) and Risk of bias for Non-randomised studies of exposures (ROBINS-E) for risk-of-bias assessment. We will calculate IV estimates for each BMI-BP pair post-randomisation and pool the estimates using meta-analysis where possible. We will explore potential sources of between-study heterogeneity. We will assess certainty of evidence using the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) framework. Results This study will identify RCTs suitable for IV analysis and generate effect estimates of BMI on BP using randomisation as an IV. We will develop tools for testing validity of randomisation and assessing risk-of-bias in studies using RCT-IV analysis that could be used by others. Conclusion This protocol provides a systematic approach for using randomisation as an IV to estimate the causal effect of BMI on BP. This study will contribute to methods for assessing IV validity and risk-of-bias assessments in RCT-IV analysis.

More from our Archive