妊娠期急性脂肪肝围产期不良结局与术中输血关联的探索性分析及风险预测模型构建OA

Objective To analyze the association between intraoperative blood transfusion and adverse perinatal out-comes(a composite of maternal death,hepatic encephalopathy,etc)in patients with acute fatty liver of pregnancy(AFLP),identify independent risk factors for these outcomes,and develop and validate a risk prediction model for early i-dentification of high-risk patients.Methods Clinical data of 56 AFLP patients admitted to our hospital from January 2008 to January 2025 were retrospectively analyzed.The chi-square test was used to compare the incidence of adverse perinatal outcomes between the transfusion group and the non-transfusion group.Univariate and multivariate Firth-penalized logistic regression analyses were performed to identify independent predictors of adverse outcomes.Based on the identified inde-pendent risk factors,a combined prediction model was developed.Internal validation was performed using the Bootstrap method(1 000 resamplings)to assess the model's generalizability.Model performance was evaluated using receiver operat-ing characteristic(ROC)curves and calibration curves.Finally,a nomogram for predicting adverse perinatal outcomes in AFLP was constructed.Results Overall,57.1%(32/56)of the patients received blood product transfusion during the ter-mination of pregnancy.Multivariate Firth-penalized logistic regression analysis showed that,given the limited sample size,intraoperative blood transfusion had no independent statistical effect on adverse outcomes(OR=0.812,95%CI:0.133-3.698,P=0.797).Mediation analysis also revealed no significant indirect effect.The independent risk factors were de-creased preoperative fibrinogen(OR=14.062,95%CI:2.389-126.656,P=0.003),pregnancy with infection(OR=4.536,95%CI:1.143-22.107,P=0.031),and cesarean section(OR=8.691,95%CI:1.321-90.081,P=0.023).The combined prediction model achieved an AUC of 0.881(95%CI:0.793-0.969,P<0.001),indicating good discrimi-nation.At the optimal cut-off value,the sensitivity was 65.6%,specificity 95.8%,positive predictive value 95.5%,and negative predictive value 67.6%.Internal validation by the Bootstrap method showed that the predictive model had good dis-crimination and no obvious overfitting.The calibration curve demonstrated that the model's predicted risk was consistent with the actual observed risk.Based on this,an AFLP perinatal adverse outcome nomogram model was constructed.Con-clusion Under the limited sample size of this study,no independent statistical effect of intraoperative blood product infu-sion on adverse perinatal outcomes in AFLP patients was detected.A multivariate Firth-penalized logistic regression model incorporating decreased preoperative fibrinogen,pregnancy with infection,and cesarean delivery demonstrated good discrimi-nation and calibration.The derived nomogram may serve as an exploratory tool for early risk stratification and proactive inter-vention.