探讨甘油三酯葡萄糖（TyG）指数对急性胰腺炎（AP）严重程度的预测价值。

检索PubMed、CNKI等数据库，纳入比较重症急性胰腺炎（SAP）与非SAP患者TyG指数的原始研究。采用随机效应模型合并标准化均数差（SMD），进行异质性、亚组、敏感性和发表偏倚分析，并依据纽卡斯尔-渥太华量表（NOS）评价文献质量。发表偏倚通过Egger、Begg检验及漏斗图进行评估。

最终纳入12项研究，共3 407例患者。Meta分析显示，SAP患者TyG指数明显高于非SAP患者（SMD=1.09，95%CI：0.99~1.18，P<0.05）；异质性分析显示研究间存在较高异质性（I²=93.4%，P<0.001）。按研究质量分组，高质量研究合并SMD为1.13（95%CI：1.00~1.26，I²=82.0%）；中等质量研究合并SMD为1.04（95% CI：0.90~1.17，I²=95.8%），两组间差异无统计学意义（P=0.334）。按SAP组平均年龄分组，年龄<50岁的研究SMD为1.41（95% CI：1.26~1.57，I²=96.9%）；年龄≥50岁组SMD为0.88（95% CI：0.76~1.00，I²=0.0%），TyG指数在年轻人群中的预测性能更高（P<0.001）。逐项剔除单个研究重复合并效应值，结果显示合并SMD变动不大，所有分析均保持差异有统计学意义，提示本研究结果具有良好稳健性，未受到个别研究的明显影响。漏斗图整体分布基本对称，仅个别研究点偏离。Egger和Begg回归检验显示无发表偏倚（t=0.64，Z=0.89；P>0.05）。

现有证据提示，TyG指数在预测AP严重程度方面具有一定潜在价值，尤其在年轻患者人群中表现出更高的预测效能。

To explore the predictive value of triglyceride glucose (TyG) index for the severity of acute pancreatitis (AP).

Original studies comparing TyG index between severe acute pancreatitis (SAP) and non-SAP patients were searched from PubMed, CNKI and other databases. Random effect model combined with standardized mean difference (SMD) was used to perform heterogeneity, subgroup, sensitivity and publication bias analyses. The quality of studies was evaluated according to Newcastle-Ottawa Scale (NOS). Publication bias was evaluated by Egger, Begg test and funnel plot.

12 studies comprising 3 407 patients were included. Meta-analysis showed that the TyG index in SAP patients was significantly higher than that in non-SAP counterparts (SMD=1.09, 95%CI: 0.99-1.18, P<0.05). Heterogeneity analysis revealed that high degree of heterogeneity was observed among different studies (I²=93.4%, P<0.001). According to the research quality, the SMD of high-quality studies was 1.13 (95%CI: 1.00-1.26, I²=82.0%),1.04 for medium-quality studies (95%CI: 0.90-1.17, I²=95.8%), and no statistical difference was noted between two groups (P=0.334). According to the average age in the SAP group, the SMD for SAP patients aged<50 years was 1.41 (95%CI: 1.26-1.57, I²=96.9%), and 0.88 for those aged≥50 years (95%CI: 0.76-1.00, I²=0.0%), indicating the prediction performance of TyG index was higher in the young SAP patients (P<0.001). A single study was excluded and the effect value was repeated combined. The results showed that combined SMD caused slight changes, and statistical significance was maintained in all analyses, indicating that the results of this study were robust and not significantly affected by individual studies. The overall distribution of funnel plot was basically symmetrical, with deviations of only a few studies. Egger and Begg regression tests detected no publication bias (t=0.64, Z=0.89; both P>0.05).

The existing evidence suggests that TyG index has certain potential value in predicting the severity of AP, especially in young patients.