基于CT Balthazar评分联合CRP的急性胰腺炎复合重症结局预测模型研究

Study on a Predictive Model for Composite Severe Outcomes of Acute Pancreatitis Based on CT Balthazar Score and CRP

摘要: 目的：构建基于Balthazar评分、C反应蛋白(C-reactive protein, CRP)和年龄的急性胰腺炎(acute pancreatitis, AP)入院早期预测复合重症结局风险的列线图模型，并评估其在早期风险分层中的临床应用价值。资料与方法：回顾性纳入2024年1月1日至12月31日安徽医科大学附属六安医院住院并确诊为AP的患者。所有患者入院72小时内完成腹部CT扫描，24小时内完成相关实验室检查。研究主要终点为复合重症结局，包括重症监护病房(ICU)收治、多器官功能衰竭及院内死亡。基于Balthazar评分、年龄和CRP构建仅含Balthazar评分的模型M0、加入年龄的模型M1及三指标联合的模型M2，采用受试者工作特征曲线下面积(area under the curve, AUC)、Bootstrap校准曲线及决策曲线分析评价模型性能，并基于最佳模型构建列线图。结果：共纳入193例AP患者，其中非重症组175例、重症组18例。重症组的Balthazar评分和CRP水平明显高于非重症组(P < 0.001)。多变量Logistic回归分析显示，Balthazar评分(OR = 3.108, P < 0.001)和CRP (OR = 5.841, P = 0.009)与复合重症结局发生相关，而年龄未达统计学显著。M2的AUC为0.888 (95% CI: 0.814~0.962)，优于M0 (ΔAUC = 0.107, P = 0.002)和M1 (ΔAUC = 0.080, P = 0.017)，且通过Bootstrap校准曲线验证了列线图预测模型的预测值与实际观察值吻合较好。决策曲线分析显示，M2模型在阈值概率为0.10~0.30时具有较高的净获益。结论：基于Balthazar评分、年龄及CRP构成的联合模型能够在入院早期较为有效地预测AP复合重症结局，简便易用，适用于临床推广，有望为监护等级安排及治疗决策提供参考。

Abstract: Objective: To develop a nomogram model for early prediction of composite severe outcomes in acute pancreatitis (AP) patients based on Balthazar score, C-reactive protein (CRP), and age, and to evaluate its clinical application value in early risk stratification. Methods: This retrospective study included patients diagnosed with AP who were admitted to Lu’an Hospital of Anhui Medical University between January 1 and December 31, 2024. All patients underwent abdominal CT scanning within 72 hours of admission and relevant laboratory tests within 24 hours. The primary endpoint was the composite severe outcome, including admission to the intensive care unit (ICU), multiple organ failure, and in-hospital mortality. A Logistic regression model was constructed based on Balthazar score, age, and CRP, including M0 (Balthazar score only), M1 (Balthazar score and age), and M2 (Balthazar score, age, and CRP). The model performance was evaluated using the area under the receiver operating characteristic curve (AUC), Bootstrap calibration curve, and decision curve analysis (DCA), and a nomogram was constructed based on the best model. Results: A total of 193 AP patients were included, with 175 in the non-severe group and 18 in the severe group. The Balthazar score and CRP levels were significantly higher in the severe group compared to the non-severe group (P < 0.001). Multivariable Logistic regression analysis showed that Balthazar score (OR = 3.108, P < 0.001) and CRP (OR = 5.841, P = 0.009) were associated with the occurrence of composite severe outcomes, while age did not reach statistical significance. The AUC of M2 was 0.888 (95% CI: 0.814~0.962), which was superior to M0 (ΔAUC = 0.107, P = 0.002) and M1 (ΔAUC = 0.080, P = 0.017), and the Bootstrap calibration curve validated a good agreement between the predicted values of the nomogram model and the actual observed values. Decision curve analysis showed that the M2 model provided higher net benefit within the threshold probability range of 0.10~0.30. Conclusion: The combined model based on Balthazar score, age, and CRP can effectively predict the composite severe outcomes in AP patients early in the admission process. It is simple, easy to use, and suitable for clinical application, providing a reference for monitoring intensity and treatment decision-making.