危重症患者中疾病特异性的最佳血糖控制策略：回顾性队列研究与网络荟萃分析。
完整摘要 · 中文译文
背景：重症患者的理想血糖目标可能因基础疾病而异，但当前仍沿用“一刀切”的方法。目的：确定疾病特异性的血糖阈值可能有助于实现最低死亡率。研究设计：本研究将重症监护医疗信息集市-IV（MIMIC-IV）数据库的回顾性队列分析与随机对照试验（RCT）的频率学派网络荟萃分析（NMA）相结合。血糖控制目标，以时间加权平均血糖（TWAg）衡量，分为四个层级：严格、中等严格、宽松和非常宽松控制。结果：回顾性分析纳入了21 119例脓毒症、16 957例心脏疾病和4708例神经系统疾病患者。在脓毒症和心脏疾病患者中，与中等严格对照组（110-150 mg/dL，参考）相比，严格对照组（80-110 mg/dL：HR 1.10/1.26，均p < 0.05）和宽松对照组（150-180 mg/dL：HR 1.38/1.78，均p < 0.05）的28天死亡率均显著更高。相比之下，在神经系统疾病患者中，严格控制带来了生存获益（HR 0.76，p = 0.013）。包含35项RCT的NMA证实了脓毒症和神经系统疾病的这些发现，其中中等严格控制在脓毒症中排名最高（P评分74.9%），严格控制在神经系统疾病中排名最高（P评分71.2%），而宽松控制在心脏疾病中排名最高（P评分83.5%）；但所有两两比较均无统计学显著差异。结论：本研究提供了证据表明理想的血糖目标 在重症监护病房（ICU）中的血糖控制目标是疾病特异性的。在回顾性分析中，中度严格控制在脓毒症和心脏疾病中与最低死亡率相关，而严格控制可能对神经科患者有益且不增加低血糖风险。然而，对于心脏疾病，网络荟萃分析（NMA）将宽松控制列为最优，使得最佳目标仍不确定。 与临床实践的相关性：这些发现可能支持在危重患者中采用基于病理生理学指导的个体化血糖管理策略。
完整摘要 · 英文原文
BACKGROUND: Optimal glucose targets for critically ill patients are likely to vary by underlying condition, yet a 'one-size-fits-all' approach persists. AIM: Determining disease-specific glucose thresholds may help achieve the lowest mortality rates. STUDY DESIGN: This study combined a retrospective cohort analysis of the Medical Information Mart for Intensive Care-IV (MIMIC-IV) database with a frequentist network meta-analysis (NMA) of randomized controlled trials (RCTs). The glucose control targets, measured by time-weighted average glucose (TWAg), were categorized into four strata: strict, intermediate strict, liberal and very liberal control. RESULTS: The retrospective analysis included 21 119 sepsis, 16 957 cardiac and 4708 neurological patients. In both sepsis and cardiac patients, compared with the intermediate strict control group (110-150 mg/dL, reference), both the strict (80-110 mg/dL: HR 1.10/1.26, both p < 0.05) and liberal (150-180 mg/dL: HR 1.38/1.78, both p < 0.05) control groups showed significantly higher 28-day mortality. In contrast, among neurological patients, strict control conferred a survival benefit (HR 0.76, p = 0.013). The NMA of 35 RCTs corroborated these findings for sepsis and neurological disorders, with intermediate strict control ranking highest for sepsis (P-score 74.9%) and strict control highest for neurological disorders (P-score 71.2%), whereas liberal control ranked highest for cardiac disease (P-score 83.5%); no pairwise differences were statistically significant. CONCLUSIONS: This study provides evidence that optimal glucose targets in the ICU are disease-specific. In the retrospective analysis, intermediate strict control was associated with the lowest mortality in sepsis and cardiac disease, while strict control may benefit neurological patients without excess hypoglycaemia. However, for cardiac disease, the NMA ranked liberal control highest, leaving the optimal target uncertain. RELEVANCE TO CLINICAL PRACTICE: These findings may support a pathophysiology-guided, individualized approach to glycaemic management in critically ill patients.