Using ROBINS-I to assess risk of bias in non-randomized studies of interventions

Tags: Workshop
Savovic J1, McAleenan A2, McGuinness L3, Sterne J3, Reeves B3, Higgins J3, The ROBINS-I Development Group
1University of Bristol & NIHR CLAHRC West, 2Population Health Sciences, Bristol Medical School, University of Bristol, 3University of Bristol

Background:

Non-randomized studies of interventions (NRSI) can provide information about effects of interventions that is not available from randomized trials, but their results may be affected by confounding, selection and misclassification biases. Assessment of risk of bias in NRSI included in systematic reviews is essential to ensure the robustness of review findings.

Objectives:

To describe ROBINS-I (Risk Of Bias In Non-randomized Studies – of Interventions), a tool to assess risk of bias in NRSI. To train participants through completion of an example assessment, and describe recent updates, new features and extensions to different types of study design.

Description:

This workshop will describe key features of the ROBINS-I tool:

- specification of a hypothetical pragmatic randomized trial that is free of bias, as a basis for risk of bias assessments;

- preliminary considerations at review protocol stage;

- specification of the effect of interest (that of assignment to intervention, or starting and adhering to intervention);

- assessments of risk of bias within seven domains (confounding; selection of participants into the study; classification of interventions; deviations from intended intervention; missing data; measurement of outcomes; selection of the reported result);

- use of signalling questions to inform 'Risk of bias' judgements;

- derivation of the overall risk of bias in the result.

We will describe recent updates to ROBINS-I, in particular the new algorithms that guide domain-level 'Risk of bias' judgements based on answers to the signalling questions. We will also describe extensions of ROBINS-I to different types of design and analysis of non-randomized studies of interventions.

The workshop will include guided practical sessions in which participants, working in small groups, will complete parts of a risk of bias assessment for one example study. There will be opportunities for discussion of the tool and its role in systematic reviews.

Participants are encouraged to bring their own portable device to access detailed guidance from the internet (www.riskofbias.info). The facilitators are unable to provide printed copies of guidance document.