Development and validation of a simplified BRASS index to screen hospital patients needing personalized discharge planning

Abstract

Background: Discharge planning is an important component of hospital care. The Blaylock Risk Assessment Screening Score (BRASS) index is an instrument used to identify patients requiring complex discharge planning. Objectives: (1) Evaluate the ability of the original BRASS index to predict the risk of complex discharge and hospital mortality. (2) Develop and validate a simplified BRASS index by eliminating redundant variables and re-estimating the predictor weights. Design: Prospective cohort study. Participants: Patients admitted at the general internal medicine wards of tertiary referral hospital in Turin, Italy, and screened within 48 h using the BRASS index. Methods: The first phase of the study assessed the performance of the original BRASS index in predicting the risk of complex discharge and hospital mortality, then a simplified score was developed. In the second phase, temporal validation of the simplified BRASS index was performed. The probability of each discharge modality (discharged at home without complications, complex discharge, and dead in hospital) was modeled using polytomous logistic regression. The AUC was used to compare the performance of the different models. Key Results: Among 6044 patients in the first phase of the study, 63% were discharged at home without complications, 31% had complex discharge, and 6% died during the hospital stay. The AUC of the simplified BRASS index, compared with the original index were 0.71 vs. 0.70 for complex discharge and 0.83 vs. 0.80 for hospital mortality. In the validation set (3325 patients), the simplified BRASS index discriminates the outcome categories with an AUC of 0.69 and 0.81 for complex discharge and hospital mortality, respectively. Conclusion: The new, simplified BRASS index showed a slightly better performance in predicting the risk of complex discharge and hospital mortality than the original tool and takes less time to be applied. These results were also confirmed in the validation set.