Post-Discharge Rebleeding in Splenic Injury: A Predictive Model

Abstract BACKGROUND: This study aimed to develop a predictive model for identifying patients with high rebleeding risk after discharge from treatment without splenectomy and embolization for splenic injury. METHODS: Data were extracted from the 1996–2013 National Health Insurance Research Database. Adult patients aged ≥18 years with splenic injury (ICD-9-CM codes 865.01–865.09) from 2000 to 2012 were included. Patients who died, underwent splenectomy (ICD-9-OP codes 41.5, 41.42,41.43, and 41.95) or transcatheter arterial embolization (TAE) (ICD-9-OP codes 39.79 and 99.29) on the first admission were excluded. The primary endpoint was rebleeding, which was identified if patients underwent splenectomy or TAE at 180 days after discharge. The patients were randomly divided into derivation (80%) and validation (20%) groups for analyze. RESULTS: 6,140 were finally included. After conducting a multivariate analysis on the derivation group (n=4916), six risk factors were identified and assigned with different scores to generate the Delayed Splenic Bleeding System (DSBS) based on the coefficient of variables significantly related to rebleeding: age <54 years (1 point), male sex (1 point), length of hospitalization ≤6 days (1 point), non-traffic accident-induced injury (1 point), ISS ≥16 (1 point), and history of congestive heart failure (2 points). The rebleeding rates at 180 days after discharge was significantly higher with DSBS score >3 (3.4%) than with DSBS score ≤3 (0.8%) ( p< 0.001); similar results were observed in the validation group (1.0% for DSBS score ≤3 vs. 4.2% for DSBS score >3, p =0.002). The areas under the ROC curve were 0.701 ( p <0.001) and 0.675 ( p =0.009) for the derivation and validation groups, respectively. The DSBS showed good calibration in the derivation group (Hosmer-Lemeshow χ2=2.607, p =0.272) and validation group (Hosmer-Lemeshow χ2=3.009, p =0.222). CONCLUSIONS: Our prediction model (i.e., the DSBS) effectively predicted the 180-day rebleeding risk, which could enable stratification in further studies or pre-discharge treatment for blunt splenic injury.