A preoperative predictive model-based scoring system for assessing postoperative bleeding after percutaneous liver biopsy

OBJECTIVE: Percutaneous liver biopsy (PLB) is the gold standard for diagnosing liver diseases, yet postoperative bleeding remains the most common and severe complication, constraining its clinical application. Accurate prediction of postoperative bleeding risk is essential to enhance PLB safety. METHODS: This study first used multivariate regression analysis in a retrospective cohort to identify independent risk factors associated with postoperative bleeding after PLB. Based on these factors, a preoperative bleeding risk scoring system was further developed, and its performance was analyzed across subgroups defined by different clinical indications. Finally, the model and scoring system were prospectively validated in an external cohort to assess generalizability. RESULTS: Multivariable analysis identified lesion type, portosystemic shunt (PSS), and total bilirubin as independent risk factors, and a significant interaction between lesion type and PSS status was observed. The bleeding-prediction model was: logit( P ) = -3.5 + (1.223 × lesion type) + (1.018 × PSS) + (0.454 × total bilirubin) + (1.523 × lesion type × PSS). The scoring system derived from these factors showed a marked increase in postoperative bleeding rate with rising scores and demonstrated good discrimination in both the standard-indication group (the area under the receiver operating characteristic curve = 0.892) and the super-indication group (the area under the receiver operating characteristic curve = 0.846). External validation further confirmed robust generalizability across populations. CONCLUSION: The preoperative bleeding-prediction model and risk scoring system developed in this study can accurately predict postoperative bleeding after PLB, enhance PLB safety, and support optimization of preoperative assessment and postoperative management in clinical practice for diverse patient groups effectively.