Background: In order to most efficiently improve transition of care, hospitals need to target intensive discharge interventions at those patients at high risk of unplanned readmission. The “HOSPITAL” score, derived previously in the US, is an easy to use prediction model that accurately identifies medical patients at high risk of readmission. It includes the following predictors: Hemoglobin, discharge from an Oncology service, Sodium level, Procedure during the index admission, Index Type of admission (urgent or elective), number of Admission(s) during the last 12 months and Length of stay. This score showed good performance in an international multicenter retrospective study. We aimed to demonstrate in a prospective study the accuracy of the HOSPITAL score to predict high risk for unplanned 30-day readmission and death.
Methods: We prospectively screened all consecutive patients aged ≥50 years admitted to the department of general internal medicine of a large community hospital in Switzerland between April and September 2013. Patients who were transferred to another acute care hospital or rehabilitation center, who refused to sign the inform consent, or who died during hospitalization were excluded. The primary outcome was the first unplanned readmission or death that occurred within 30 days from hospital discharge. We calculated the “HOSPITAL” score for all patients. Since the hospital had no specific oncology division, the variable “Oncology” was replaced by a recent diagnosis of cancer (< 5 years) on admission. The performance of the prediction model was evaluated according to its overall accuracy (Brier score), its discriminatory power (C-statistic), and its calibration (Pearson goodness of fit test).
Results: Among the 346 included patients, 11.6% (n=40) had an unplanned readmission or died within 30-days after discharge. Mean age of the patients was 73.4 years (SD 11.5) and median length of stay 7 days (IQR 4-12). The overall accuracy of the “HOSPITAL” score was very good with a low Brier score (0.09). Its discriminatory power to predict unplanned readmission or death was good with a C-statistic of 0.70 (95% CI 0.61-0.79). Patients were classified into 3 risk categories for the primary outcome: low (67%), intermediate (19%), and high risk (14%). The proportions of unplanned readmission/death for each risk category were 9%, 12%, and 25%, respectively. The calibration was excellent with a P-value of 0.97. A simplified version of the score without the variable “Procedure during the index admission” showed fair performance, with a C-statistic of 0.67 (95% CI 0.58-0.75).
Conclusions: The “HOSPITAL” score identified patients at high risk of 30-day unplanned readmission or death with good discrimination when applied prospectively to a cohort of medical patients in Switzerland. The “HOSPITAL” score has the potential to easily and prospectively identify patients in need of more intensive transitional care interventions.