Simple risk score (MV-SoLdHARP) for intensive care unit readmission or in-hospital death among surgical intensive care unit patients: Derivation and internal validation using a Japanese nationwide inpatient database.

Abstract

Intensive care unit readmissions or in-hospital death following intensive care unit discharge are critical issues in postoperative management. However, existing predictive models have suboptimal performance and lack generalizability. We aimed to develop and internally validate a simplified risk score to identify surgical patients at high risk of adverse events (intensive care unit readmission or in-hospital death) within 7 days after intensive care unit discharge. The score was intended to be used at the time of intensive care unit discharge by bedside clinicians and intensive care unit bed managers. We conducted a retrospective cohort study using inpatient claims database, enrolling adult surgical patients restricted to intensive care unit stays >48 hours between April 2018 and September 2021. Variable selection was performed using least absolute shrinkage and selection operator logistic regression, targeting adverse events within 7 days after intensive care unit discharge. We assessed discrimination by the c-statistic and calibration by intercept/slope; bootstrap internal validation quantified and corrected optimism. Of 9,392 patients, 99 (1.1%) had adverse events. The MV-SoLdHARP score comprises mechanical ventilation, Sequential Organ Failure Assessment score at intensive care unit discharge, lactate dehydrogenase >500 U/L, hematocrit <25%, emergency admission, renal disease, and serum potassium >4 mEq/L. The optimism-corrected c-statistic was 0.82, with calibration intercept/slope 0.20/1.04. At the moderate-or-higher risk threshold (score ≥5), sensitivity was 12.1%, specificity 99.4%, positive predictive value 18.8%, and negative predictive value 99.1%. The risk score may help identify surgical patients at high risk of early adverse events after intensive care unit discharge. The findings may guide intensive care unit discharge decisions and resource allocation. External validation is needed to confirm stability and transportability.