Association between trends in hospital length of stay and risk of readmission after surgery

Introduction Medical and surgical innovations have allowed hospitals to shorter length of stay (LOS) under financial constraints. It remains unclear whether hospitals with uncontrolled LOS reduction may expose patients to unsafe care. We aimed to determine if there was an association between trend in hospital LOS and patient's risk of readmission over time. Methods We selected all patients who underwent elective colectomy or urgent hip fracture repair in French hospitals between 2013 and 2016. For each procedure, hospitals were categorized in three groups according to potential variations in their median length of index stays over study period as follows: major decrease, moderate decrease, or no decrease. Three cohorts of patients were then designed by matching these groups of hospitals two-by-two, based on a propensity score approach tacking into account potential confounding factors related to patients and hospital characteristics. In each cohort, outcomes were compared between hospitals groups at 1, 3 and 6 months. Potentially avoidable readmission for severe adverse event was modelled using Fine and Gray's approach with competing risk of death, while number of days in all acute care hospitalizations over study period was analyzed using generalized estimating equations (GEE) Poisson regression. Clustering of patients in matched pairs was taken into account. Results We considered 98,713 patients in 540 hospitals for colectomy and 206,812 patients in 414 hospitals for hip fracture repair before matching. After colectomy, trend in hospital LOS over time was not associated with risk of readmission, except for patients operated in hospitals with a moderate decrease in LOS whose experienced a lower risk at 3 months compared to those operated in hospitals with no decrease (hazard ratio [95% CI]: 0.85 [0.78–0.93]). After hip fracture repair, patients operated in hospitals with major decrease in LOS compared to those operated in hospitals with no decrease had a higher risk of readmission at 1, 3 and 6 months (1.37 [1.18–1.59], 1.23 [1.10–1.37], 1.22 [1.11–1.33], respectively). For the two procedures, major or moderate decrease in hospital LOS over time compared to hospitals with no decrease were associated with fewer number of days in acute care hospitalizations at 1, 3 and 6 months ( P Conclusion Reduction in LOS over time in French hospitals was not associated with worse patient outcomes after colectomy, whereas a higher risk of severe adverse event was observed after hip fracture repair in hospitals with major LOS decrease. For both procedures, clear relationship was found between reduction in LOS and lower consumption of hospitalization days in acute care.