建立和评价包括经皮肾镜等微创手术治疗在内的重症急性胰腺炎(severe acute pancreatitis, SAP)患者住院期间死亡的预测模型。

回顾性分析2015年1月至2019年12月期间，解放军总医院第一医学中心重症急性胰腺炎专病中心收治的314例SAP患者的临床资料。男225例、女89例，中位年龄45.8岁(16，84)，中位体质量指数24.26 kg/m²(14.88，52.08)。针对胰腺和胰腺周围坏死感染病灶的局部治疗方法实行升阶梯微创治疗方案，首先行CT引导下经皮穿刺引流，然后根据病情缓解程度，选择经皮肾镜微创或开放手术治疗。收集分析患者住院死亡的潜在危险因素，将单变量逻辑回归中与结局相关的变量(P<0.1)纳入模型筛选。使用基于赤池信息量准则的逐步回归分析法进行模型筛选。

314例SAP患者中，44例患者住院期间死亡，病死率为14.0%(44/314)。逐步回归分析法筛选出变量进入最终模型：年龄>60岁(OR＝4.01，P＝0.004)、>2个器官功能支持(OR＝26.56，P<0.010)、ICU住院时间>2 d(OR＝2.38，P＝0.162)及住院后微创手术(OR＝0.17，P<0.010)。基于此模型制作的列线图可产生<5%(0个不良因素)和>90%(4个不良因素)的住院期间死亡可能性。

该模型可较准确地预测包括经皮肾镜等微创手术治疗在内的SAP患者在院期间的死亡概率。

To establish a model to predict in-hospital mortality of patients with severe acute pancreatitis undergoing minimally invasive procedure including percutaneous nephrolithotomy.

Patients diagnosed with acute severe pancreatitis during Jan. 2015 till Dec. 2019 at the Pancreatitis Specialized Center were retrospectively analyzed. A total of 314 patients were included, 225 male and 89 female, median age 45.8 years old (16, 84), and median body mass index 24.26 kg/m² (14.88, 52.08). A surgical step-up approach for pancreatic necrosis or infected necrosis was implemented. The minimally invasive intervention, such as percutaneous puncture and drainage, is preferred. The decision was made whether perform percutaneous nephroscope or open surgery according to the degree of remission. Logistic regression analysis was used to evaluate the association between in-hospital mortality and relevant factors. Significant predictors of in-hospital mortality on univariate analysis (P<0.1) were entered into a stepwise selection using the Akaike information criterion to define the final model.

Of the 314 patients with severe acute pancreatitis, 44 cases (14.0%) died. In stepwise selection, the following factors were identified as predictors of in-hospital mortality: age over 60 (OR=4.01, P=0.004), multiple organ support (OR= 26.56, P<0.010), ICU stay longer than 2 d(OR=2.38, P= 0.162) and minimal invasive surgery after admission (OR= 0.17, P<0.010). The prognostic nomogram based on this model yielded a probability of in-hospital mortality ranging from <5% (0 factors) to >90% (all 4 factors).

This prediction model can accurately predict in-hospital mortality in patients with severe acute pancreatitis including those who undergoing step-up minimally invasive approach including percutaneous nephrolithotomy.