Nursing Home Length of Stay in 3 Canadian Health Regions: Temporal Trends, Jurisdictional Differences, and Associated Factors

Objectives: To assess (1) temporal changes (2008–2015) in nursing home (NH) length of stay (LoS) in 3 Canadian health jurisdictions (Edmonton, Calgary, Winnipeg), (2) resident admission characteristics associated with LoS, and (3) temporal changes of admission characteristics in each of the 3 jurisdictions. Design: Retrospective cohort study using data previously collected in Translating Research in Elder Care (TREC), a longitudinal program of applied health services research in Canadian NHs. Setting and participants: 7817 residents admitted between January 2008 and December 2015 to a stable cohort of 18 NHs that have consistently participated in TREC since 2007. Methods: LoS was defined as time between a resident's first NH admission and final discharge from the NH sector. Analyses included descriptive statistics, Kaplan Meier estimates (unadjusted LoS), and Cox proportional hazard regressions (adjusted LoS), adjusted for resident characteristics (eg, age, cognitive performance, and health instability). We also controlled for NH size and ownership. Results: In jurisdictions with increasing care needs, unadjusted median LoS [95% confidence interval (CI)] decreased over time (2008 and 2009 vs 2014 and 2015 admissions); in Calgary from 1.837 (95% CI 1.618, 2.275) to 1.328 (95% CI 1.185, 1.489) years and in Edmonton from 1.927 (95% CI 1.725, 2.188) to 1.073 (95% CI 0.936, 1.248) years. In Winnipeg, care needs and LoS remained constant (2.163, 95% CI 1.867, 2.494, vs 2.459, 95% CI 2.155, 2.883, years). Resident characteristics including higher physical dependency [hazard ratio (HR) 1.205, 95% CI 1.133, 1.282], higher cognitive impairment (HR 1.112, 95% CI 1.042, 1.187), or higher health instability (HR 1.333, 95% CI 1.224, 1.452) were associated with lower LoS. Adjustment for resident characteristics reduced jurisdictional LoS differences and rendered temporal LoS differences within jurisdictions statistically nonsignificant. Conclusions/Implications: In jurisdictions where care needs at admission have increased since 2008, resident LoS has decreased. Jurisdictional differences in care needs and LoS indicate that health policies may affect these outcomes. Variations of resident outcomes by policy environment require additional scrutiny.