目的 本研究旨在评估白蛋白校正阴离子间隙（albumin-corrected anion gap，ACAG）与急性胰腺炎（acute pancreatitis，AP）患者短期至长期死亡结局之间的关联。 方法 这项回顾性研究基于美国重症监护医学信息数据库（Medical Information Mart for Intensive Care，MIMIC）-IV，将诊断为AP并入住重症监护病房的成人患者纳入了研究，通过COX回归风险分析、受试者工作特征（receiver operating characteristic，ROC）曲线分析、卡普兰-迈耶（Kaplan-Meier，K-M）生存曲线分析、限制性立方样条以及亚组分析判断ACAG对AP患者死亡结局的预测能力。 结果 本研究共入组444例患者，根据入院后28 d 的患者死亡情况将患者分为存活组（412例）及死亡组（32例），入院后28 d死亡率为7.2%。多因素COX回归显示ACAG是AP患者入院28 d 全因死亡率的独立预测因子（HR=1.18；95%CI=1.05~1.32），而在90 d（HR=1.05，95%CI=0.97~1.14），180 d（HR=1.01，95%CI=0.94~1.09）和1年（HR=1.02，95%CI=0.95~1.10）的预后判断能力上，ACAG并不能作为AP患者死亡结局的独立预测因子。ROC曲线分析显示ACAG预测28 d 死亡结局的曲线下面积（area under the curve，AUC）为0.732（95%CI=0.632~0.832），优于AG［AUC=0.665（95%CI=0.550~0.781）］和血清白蛋白（serum albumin，ALB）［AUC=0.655（95%CI=0.550~0.761）］，与序贯器官衰竭评分（sequential organ failure score，SOFA）相近［AUC=0.745（95% CI=0.651~0.838）］。根据ROC曲线得出ACAG的最佳界值为21.375。将ACAG以21.375 分为高值组与常值组，K-M分析显示ACAG高值患者死亡率高于ACAG常值患者（P<0.001）。亚组分析显示结果稳定。 结论 ACAG可作为AP患者入院后28 d全因死亡率的独立预测因子，其检验效能优于AG及ALB，与SOFA相近，但与AP患者90 d、180 d，1年死亡结局的相关性并不具有显著性。

Objective To investigate the association between albumin-corrected anion gap（ACAG） and short- to long-term death outcomes in patients with acute pancreatitis（AP）. Methods This retrospective study was based on the Medical Information Mart for Intensive Care-IV database，and the adult patients who were diagnosed with AP and were admitted to the intensive care unit were enrolled in this study. Cox regression risk analysis，receiver operating characteristic（ROC） curve analysis，Kaplan-Meier survival curve analysis，restricted cubic spline，and subgroup analysis were used to investigate the value of ACAG in predicting the death outcome of AP patients. Results A total of 444 patients were enrolled in this study，and according to the death status of the patients on day 28 after admission，the patients were divided into survival group with 412 patients and death group with 32 patients，with a mortality rate of 7.2% on day 28 after admission. The multivariate Cox regression analysis showed that ACAG was an independent predictive factor for all-cause mortality rate on day 28 after admission in AP patients（hazard ratio［HR］=1.18，95%CI=1.05-1.32），while it was not an independent predictive factor for death outcome on days 90（HR=1.05，95%CI=0.97-1.14） and 180（HR=1.01，95%CI=0.94-1.09） and at 1 year（HR=1.02，95%CI=0.95-1.10）. The ROC curve analysis showed that ACAG had an area under the ROC curve（AUC） of 0.732（95%CI=0.632-0.832） in predicting 28-day death outcome，which was better than that of AG（AUC=0.665，95%CI=0.550-0.781） and serum albumin（Alb）（AUC=0.655，95%CI=0.550-0.761） and was similar to that of Sequential Organ Failure Assessment（SOFA） score（AUC=0.745，95%CI=0.651-0.838）. The ROC curve showed that the optimal cut-off value of ACAG was 21.375. Based on the cut-off value of ACAG of 21.375，the patients were divided into high-value group and normal-value group，and the Kaplan-Meier curve analysis showed that the patients with a high level of ACAG had a significantly higher mortality rate than those with normal ACAG（P<0.001）. The subgroup analysis showed that the results were stable. Conclusion ACAG can be used as an independent predictive factor for all-cause mortality rate on day 28 after admission in AP patients，with a better efficacy than AG and Alb and a similar efficacy to SOFA. However，it is not significantly associated with 90-day，180-day，and 1-year death outcomes in AP patients.