Coverage uncertainty range: A new method for calculating uncertainty around summary statistics in healthcare quality indicators

Abstract

Data from clinical health registries, such as medical quality registries, are often used as basis for healthcarequality indicators (QI). To aid the interpretation of quality indicators and support decisions, it is importantto quantify the uncertainty around the QI summary statistics. In this paper we suggest a novel method forquantifying such uncertainty: the Coverage uncertainty range. The method is based on the size of thepopulation present in the register relative to the total relevant population and does not make any assumptionsabout the sampling strategy or the value of the summary statistic. Furthermore, using both simulated dataand real-life data from a Norwegian medical quality registry, we illustrate why using confidence intervalswhen presenting healthcare quality indicators may lead to erroneous conclusions.