From data to practice change - exploring new territory for atlases of clinical variation.

Despite decades of atlas production and use within multiple healthcare systems, and consistent reporting of geographical differences in the utilisation of services, significant levels of clinical variation persist. Drawing on over forty years of combined experience using atlases of clinical variation, we reflect on why that might be the case and explore the role of atlases have played in efforts to reduce inappropriate overuse, underuse and misuse of healthcare services. We contend that atlases are useful but, on their own, are not enough to drive change in clinical practice and improvement in patient outcomes. Building on four conceptual models we have published since 2017, we argue that atlases, with their focus on measuring healthcare utilisation by residents in different geographies, generally fail to provide sufficient information and statistical analyses to truly assess the nature of the variation and support action for change. They seldom use structures such as hospitals or teams as the unit of analysis to understand variation; they rarely feature the key elements of healthcare performance which underlie variation; they are mostly silent about how to assess whether the variation measured is warranted or truly unwarranted; nor do they identify evidence-based levers for change. This means that a stark choice confronts producers of atlases - to either continue with the current model and more explicitly rely on other players to undertake work to complete the 'data to action' cycle that is necessary to secure improvement; or to refine their offering - including more sophisticated performance measurement approaches, nuanced guides for interpretation of any differences found, support for the selection and application of levers for change that align with local context, and provision of evidence-based options for implementation.