Variability in Length of Hospitalization for Stroke

Abstract

Objectives: To measure hospital stay for acute stroke care and to describe health services and demographic factors associated with longer length of stay (LOS). Design: Observational, retrospective consecutive case series. Setting: Large tertiary-care teaching hospital in Massachusetts. Patients: The patient population comprised 745 patients aged 65 years and older admitted with ischemic stroke from 1982 through 1995. Main Outcome Measures: Hospital LOS (1-5, 6-10, and > 10 days) as well as total charges and discharge location. Results: Median LOS was 7 days (range, 1-289 days), and median total charges were $8740 (range, $522-$135 172); LOS explained 62% of the variance in total charges. Insurance status was a major factor in determining LOS: after possible confounders were controlled for, patients enrolled in a health maintenance organization were significantly less likely to have long hospital stays (odds ratio [OR], 0.45; 95% confidence interval, 0.31-0.66) than were conventional Medicare enrollees, while the LOS of patients with other insurance coverage was no different from that of Medicare patients. Longer LOS was significantly associated with greater comorbidity (OR, 1.52 for a Charlson comorbidity index >2), institutionalization prior to hospital admission (OR, 1.83), and unmarried status (OR, 1.37) and was inversely associated with year of admission (OR, 0.30 in years 1991-1995 vs 1982-1986). Age, sex, and race were not associated with LOS. Discharge to a nursing home or inpatient rehabilitation site was not associated with type of insurance coverage (OR, 1.10; 95% confidence interval, 0.72-1.69 for patients in a health maintenance organization vs conventional Medicare patients). Conclusions: There is marked variability in length of hospital stay for ischemic stroke among the elderly, even after underlying patient differences are controlled for. Managed care may result in increased efficiency of in-hospital care and improved discharge planning for these patients; further study of the ultimate clinical outcomes of such care is needed.