Comparison of methods to define high use of inpatient services using population-based data

Abstract

BACKGROUND: A variety of methods have been proposed to define “high users” of inpatient services, which may have implications for targeting subgroups for intervention. OBJECTIVE: To compare 3 common definitions of high inpatient service use and their influence on patient capture, outcomes, and inpatient burden. DESIGN, SETTING, AND PATIENTS: Cross-sectional population-level study of 219,106 adults in Alberta, Canada, with ≥ 1 hospitalization from April 1, 2012, to March 31, 2013. MEASUREMENTS: We defined “high use” based on the upper 5th percentile of the population by 3 definitions: (1) number of inpatient episodes (≥ 3 hospitalizations/year), (2) cumulative length of stay (≥ 56 days in hospital/year), and (3) cumulative cost based on hospitalization resource intensity weights (≥ $63,597 Canadian dollars/year). Clinical characteristics, health outcomes, and overall health burden were compared across definitions and stratified by age. RESULTS: Of that population, 10.3% of individuals were common to all definitions. High users based on number of inpatient episodes were more likely to be admitted for acute conditions, with most high users based on length of stay admitted for mental health-related conditions, while those based on costs were more likely to have hospitalizations resulting in death (9.3%). High-episode individuals accounted for 16.6% of all inpatient episodes, high-length of stay individuals for 46.4% of all hospital days, and high-cost individuals for 38.9% of total cost. CONCLUSIONS: Three definitions of high users of inpatient services captured significantly different groups of patients. This has implications for targeting subgroups for intervention and highlights important considerations for selecting the most suitable definition for a given objective.