The design of studies testing the effectiveness of risk-guided care has many challenges: a scoping review addressing key considerations

Abstract

Objectives: Studies evaluating the effectiveness of care based on patients' risk of adverse outcomes (risk-guided care) use a variety of study designs. In this scoping review, using examples, we review characteristics of relevant studies and present key design features to optimize the trustworthiness of results.

Study design and setting: We searched five online databases for studies evaluating the effect of risk-guided care among adults on clinical outcomes, process, or cost. Pairs of reviewers independently performed screening and data abstraction. We descriptively summarized the study design and characteristics.

Results: Among 14,561 hits, we identified 116 eligible studies. Study designs included randomized controlled trials (RCTs), post hoc analysis of RCTs, and retrospective or prospective cohort studies. Challenges and sources of bias in the design included limited performance of predictive models, contamination, inadequacy to address the credibility of subgroup effects, absence of differences in care across risk strata, reporting only process measures as opposed to clinical outcomes, and failure to report benefits and harms.

Conclusion: To assess the benefit of risk-guided care, RCTs provide the most trustworthy evidence. Observational studies offer an alternative but are hampered by confounding and other limitations. Reaching valid conclusions when testing risk-guided care requires addressing the challenges identified in our review.

Keywords: Bias; Methodological challenges; Risk-guided care; Scoping review; Stratified medicine; Study design.