Advancing Methods to Study Clinical Variation in Healthcare

Le résumé fourni par la source

Clinical variation (a.k.a. medical practice variation) describes how patient care and outcomes differ across patients, providers, hospitals, geographic regions or other dimensions. By understanding clinical variation in patient care or outcomes that cannot be explained by patient differences, researchers and health systems can (1) highlight overuse, underuse, inefficiencies or inequities impacting patient care, (2) identify the level (e.g., patient, provider, hospital) contributing the most variation and (3) interrogate the reasons that explain the observed variation. Numerous methods exist to understand clinical variation; however, these are not well identified or characterized in the existing body of health services research literature. We aim to provide non-technical methodological guidance to researchers interested in describing and quantifying clinical variation by characterizing and comparing methods suited to this task. We present several plots to display clinical variation, including point and jitter plots, scatterplots, caterpillar plots, box plots and funnel plots. While most plots simply visualize the variation, funnel plots can characterize whether the variation is larger than expected after adjusting for relevant clinical factors that might reasonably explain clinical variation, such as case-mix. We present several basic statistical approaches to measure clinical variation, including variance, standard deviation, coefficient of variation, systematic coefficient of variation and interquartile range. Further, we describe multilevel models that measure clinical variation beyond its magnitude, including the level(s) at which variation occurs and the level-specific factors that explain the variation. We describe several statistics that quantify the clinical variation that exists in a multilevel model, such as the intraclass correlation coefficient and median odds, risk, rate and hazard ratios. We discuss key considerations when describing and quantifying clinical variation, including the appropriate measurement of patient care and outcomes, the factors explaining variation and whether it is warranted and what interventions might help reduce the variation. The objective of this review is to provide an overview of techniques to consider when describing clinical variation, serving as a resource for those examining clinical variation.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.