Predicting the Length of Hospital Stay After Stroke Episode in Italy

Abstract:     
Objectives: To assess clinical, demographic and organizational factors influencing length of stay in acute hospitalization for patients with stroke in Italy.

Methods: Data were collected in a prospective, incidence-based, observational multi-centre study. Patients were enrolled in a consecutive manner in 47 hospital centres stratified according to macro-area (North, Centre and South) and admitting organizational unit (Neurology, General Medicine and Stroke Unit). Data regarding demographic, social, clinical variables, time to care (TTC) and length of stay (LOS) were collected through custom designed questionnaires. TTC was defined as the time period between the stroke onset and provided inpatient care. Neurological impairment on baseline was quantified using National Institute of Health Stroke Scale (NIHSS), while physical disability was measured with Barthel Index (BI). The association between demographic characteristics or clinical variables and LOS was examined using multiple regression model.

Results: A total of 503 patients were enrolled across the country (38%, 25% and 37% from the North, Center and South respectively). Patients (59% male) had mean age 69.3±12.8 (range, 19 to 96). The majority (83%) of patients suffered from ischemic stroke, and in only 11% of cases it was recurrent. Atrial fibrillation was present in 19% of patients. Median NIHSS score at admission was 5 (25th to 75th percentile, 3 to 10), while median BI at admission was 52.5 (25th to 75th percentile, 18 to 88). Median TTC was 3h (25th to 75th percentile, 1 to 8h). The multiple regression model identified four significant predictors of LOS: TTC, NIHSS score, geographical area and admitting organisationa unit. One additional hour of delay in providing care (TTC) results in increasing LOS by 0.2% (p=0.021), holding other variables constant. The Baseline NIHSS score also predicts the outcome, with one additional point on NIHSS increasing the LOS by 2% (p<0.0001). Admission to Neurology department increases LOS by 19%, (ß=0.19; p<0.0001), while hospital admissions in northern regions have LOS 20% higher than the other Italian regions (ß=0.20; p<0.0001).

Conclusions: TTC and severity as defined by NIHSS are independent predictor of LOS and, likely, overall use of hospital resources. Reducing TTC may generate both clinical and economic benefit. Both TTC and NIHSS may be used to refine the present system to fund hospital care for stroke patients in Italy.

Keywords: stroke, health resources, outcome

working papers series