ST段抬高型心肌梗死直接经皮冠状动脉介入治疗后慢血流/无复流：一项Meta分析。
完整摘要 · 中文译文
直接经皮冠状动脉介入治疗（PCI）后冠状动脉慢血流/无复流会限制ST段抬高型心肌梗死（STEMI）患者的心肌再灌注。本研究综合了临床、实验室、血管造影和手术相关关联，同时将粗未经调整的主要分析与仅调整后的敏感性分析分开。检索了七个数据库，检索时间从建库至2026年5月31日。随机效应模型采用限制性最大似然估计和Hartung-Knapp置信区间；当至少三项研究提供数据时报告95%预测区间。共纳入43项观察性研究，包括27,405例患者，其中5,631例存在慢血流/无复流。粗合并平均值支持糖尿病、Killip II级或更高/急性心力衰竭、晚期Killip分级、基线心肌梗死溶栓试验血流受损、高血栓负荷、较长支架长度、血栓抽吸和直接支架置入的关联，尽管若干预测区间包括无效值。仅调整后分析对年龄、高血压、糖尿病、Killip II级或更高/急性心力衰竭和高血栓负荷的结果一致。结局定义和纳入标准敏感性分析未能消除重要的不确定性。基于领域的偏倚风险判断主要为存在一些担忧或高担忧，推荐分级的评估、制定与评价（GRADE）证据确定性为低或极低。因此，本综述提供的是跨领域证据图谱，而非因果目录或床旁预测评分。
完整摘要 · 英文原文
Coronary slow flow/no-reflow after primary percutaneous coronary intervention (PCI) limits myocardial reperfusion in ST-segment elevation myocardial infarction (STEMI). Clinical, laboratory, angiographic, and procedural associations were synthesized while separating crude primary analyses from adjusted-only sensitivity analyses. Seven databases were searched from inception through May 31, 2026. Random-effects models used restricted maximum likelihood estimation and Hartung-Knapp confidence intervals; 95% prediction intervals were reported when at least three studies contributed. Forty-three observational studies included 27,405 patients, including 5,631 with slow flow/no-reflow. Crude pooled averages supported associations for diabetes mellitus, Killip class II or higher/acute heart failure, advanced Killip class, impaired baseline Thrombolysis in Myocardial Infarction flow, high thrombus burden, longer stent length, thrombus aspiration, and direct stenting, although several prediction intervals included the null. Adjusted-only analyses were consistent for age, hypertension, diabetes, Killip class II or higher/acute heart failure, and high thrombus burden. Outcome-definition and eligibility sensitivity analyses did not eliminate important uncertainty. Domain-based risk-of-bias judgments were predominantly some concerns or high concern, and Grading of Recommendations Assessment, Development and Evaluation certainty was low or very low. The review, therefore, provides a cross-domain evidence map rather than a causal catalog or bedside prediction score.