高甘油三酯血症性急性胰腺炎并发急性呼吸衰竭危险因素及预测模型的建立

Risk factors and prediction model of acute respiratory failure in patients with hypertriglyceridemic acute pancreatitis

目的分析高甘油三酯血症性急性胰腺炎(HTG-AP)患者并发急性呼吸衰竭(ARF)的危险因素,并构建风险预测模型。方法222例HTG-AP患者根据ARF诊断指南分为非ARF组176例和ARF组46例,比较2组临床资料并筛选预测因子,将筛选的因子纳入多因素Logistic回归分析并建立回归模型,绘制受试者工作特征(ROC)曲线、校准图及决策曲线分析(DCA)分别评估模型的预测能力、准确性及临床实用性。结果与非ARF组相比,ARF组高密度脂蛋白胆固醇(HDL-C)、低密度脂蛋白胆固醇(LDL-C)、白蛋白(ALB)水平降低(P<0.05),而肌酐(Cr)、尿素氮(BUN)、天冬氨酸转氨酶(AST)、C-反应蛋白(CRP)水平及胸水、腹水发生率升高(P<0.05)。多因素Logistic回归分析结果显示,较高水平的Cr、AST,较低水平的ALB、HDL-C及有腹水为HTG-AP并发ARF的独立危险因素(P<0.05)。基于该结果建立HTG-AP并发ARF的列线图预测模型ROC曲线下面积为0.952(95%CI:0.923~0.981),约登指数0.808,敏感度、特异度分别为93.33%、87.43%。校准曲线显示,模型预测HTG-AP并发ARF概率与实际发生概率具有较好一致性。DCA曲线显示,模型具有一定的临床实用价值。结论构建的列线图预测模型可为临床预防HTG-AP并发ARF提供方案。

Objective To analyze risk factors of acute respiratory failure(ARF)in patients with hypertriglyceridemia acute pancreatitis(HTG-AP)and construct a risk prediction model.Methods A total of 222 HTG-AP patients were included in this study and divided into the non-ARF group(176 cases)and the ARF group(46 cases)according to diagnostic guidelines for ARF.Clinical data of the two groups were compared and the predictive factors were screened.These selected factors were then utilized in a multivariate Logistic regression analysis to construct a Logistic regression model.Subsequent evaluation of the model's predictive ability,accuracy and clinical utility was conducted through ROC,curve analysis,calibration plot examination and decision curve analysis(DCA),respectively.Results Compared with the non-ARF group,the levels of high density lipoprotein cholesterol(HDL-C),low density lipoprotein cholesterol(LDL)-C and albumin(ALB)were decreased in the ARF group(P<0.05),while the levels of creatinine(Cr),urea nitrogen(BUN),aspartate aminotransferase(AST)and C-reactive protein(CRP)were increased,and the incidence of pleural fluid and ascites was also increased(P<0.05).Multivariate Logistic regression analysis showed that higher levels of Cr and AST,lower levels of ALB,HDL-C and ascites were independent risk factors for HTG-AP complicated ARF(P<0.05).Based on these results,a column-line prediction model for HTG-AP complicated ARF was established.After internal verification,the area under curve(AUC)of receiver operating characteristic(ROC)curve of the nomogram model was 0.952(95%CI:0.923-0.981),the Youden index was 0.808 and the sensitivity and specificity were 93.33%and 87.43%,respectively.The calibration curve showed that the probability of HTG-AP concurrent ARF predicted by the model was in good agreement with the actual probability.The DCA curve showed that the model had certain clinical value.Conclusion The nomogram prediction model combined could provide a scheme for the clinical prevention of HTG-AP complicated with ARF.