Hospital‐level variation in risk‐standardized admission rates for emergency care–sensitive conditions among older and younger Veterans

Abstract

AbstractObjectivesResearch examining emergency department (ED) admission practices within the Department of Veterans Affairs (VA) is limited. This study investigates facility‐level variation in risk‐standardized admission rates (RSARs) for emergency care–sensitive conditions (ECSCs) among older (≥65 years) and younger (<65 years) Veterans across VA EDs.MethodsVeterans presenting to a VA ED for an ECSC between October 1, 2016 and September 30, 2019 were identified and the 10 most common ECSCs established. ECSC‐specific RSARs were calculated using hierarchical generalized linear models, adjusting for Veteran and encounter characteristics. The interquartile range ratio (IQR ratio) and coefficient of variation were measures of dispersion for each condition and were stratified by age group. Associations with facility characteristics were also examined in condition‐specific multivariable models.ResultsThe overall cohort included 651,336 ED visits across 110 VA facilities for the 10 most common ECSCs—chronic obstructive pulmonary disease (COPD), heart failure, pneumonia, volume depletion, tachyarrhythmias, acute diabetes mellitus, gastrointestinal (GI) bleeding, asthma, sepsis, and myocardial infarction (MI). After adjusting for case mix, the ECSCs with the greatest variation (IQR ratio, coefficient of variation) in RSARs were asthma (1.43, 32.12), COPD (1.39, 24.64), volume depletion (1.38, 23.67), and acute diabetes mellitus (1.28, 17.52), whereas those with the least variation were MI (1.01, 0.87) and sepsis (1.02, 2.41). Condition‐specific RSARs were not qualitatively different between age subgroups. Association with facility characteristics varied across ECSCs and within condition‐specific age subgroups.ConclusionsWe identified unexplained facility‐level variation in RSARs for Veterans presenting with the 10 most common ECSCs to VA EDs. The magnitude of variation did not appear to be qualitatively different between older and younger Veteran subgroups. Variation in RSARs for ECSCs may be an important target for systems‐based levers to improve value in VA emergency care.