病情严重程度改变ICU成人脓毒症死亡负担:64,703例入院中的严重度依赖性休克异质性证据
这项MIMIC-IV回顾性队列纳入64,703例ICU入院,脓毒症占42.7%,30天死亡率12.9%。脓毒症与死亡的关系随SAPS II评分显著变化(交互p<0.001):SAPS II为30时脓毒症平均边际效应为+3.8%(95%CI 3.2-4.3%),增加死亡风险;SAPS II为50-55时不再显著;SAPS II为80时脓毒症患者校正后死亡率反而更低(AME -7.3%,95%CI -9.6至-5.0%)。该交叉在未校正数据中也存在,且在8项敏感性分析中一致。
要点
- SAPS II为30时,脓毒症与更高30天死亡率相关,校正后AME +3.8%(95%CI 3.2-4.3%),死亡风险增加。
- SAPS II为50-55时,脓毒症与死亡的关系减弱且不再显著,提示死亡负担随严重度增加而递减。
- SAPS II为80时,脓毒症患者校正后30天死亡率反而更低,AME -7.3%(95%CI -9.6至-5.0%)。
- 未校正数据同样显示交叉:SAPS II≥70时脓毒症与非脓毒症30天死亡率分别为61.5%与69.2%。
- 该交叉模式在8项敏感性分析及不同ICU类型中一致,提示脓毒症死亡负担并非随严重度恒定不变。
适用人群与场景
- MIMIC-IV数据库2008-2019年64,703例ICU入院;脓毒症按Sepsis-3标准定义,占42.7%;基础严重度用SAPS II评分。
诊断标准与关键阈值
- 脓毒症采用Sepsis-3标准;基础严重度采用SAPS II评分并结合分数多项式建模;30天死亡率为主要结局。
处置流程与药物剂量
- 原文未提及
红旗征与不能漏的情况
- 低至中度严重度(SAPS II约30)的脓毒症患者死亡风险增加,需按脓毒症路径积极处理。
- 高严重度(SAPS II约80)脓毒症患者校正后死亡风险反而较低,解读预后时不应简单将脓毒症视为额外死亡负担。
- SAPS II≥70时未校正死亡率仍呈交叉(61.5% vs 69.2%),高严重度层仍需警惕但避免单因脓毒症过度悲观。
研究设计与关键结果
- 回顾性队列研究,64,703例ICU入院,脓毒症占42.7%,30天死亡率12.9%。多变量logistic回归含交互项评估效应修饰,SAPS II为30时脓毒症AME +3.8%(95%CI 3.2-4.3%),升高死亡率;SAPS II为50-55时变为不显著;SAPS II为80时AME -7.3%(95%CI -9.6至-5.0%),脓毒症患者校正后死亡率更低。未校正数据SAPS II≥70时脓毒症与非脓毒症死亡率分别为61.5
证据类型、等级与局限
- 回顾性队列研究,基于单中心MIMIC-IV数据库,存在残余混杂、回顾性设计、数据库编码与缺失数据等局限;原文未给出证据等级或推荐强度。
对床旁实践的改变
- 在抢救室和ICU床旁,不应把脓毒症在所有严重度水平都视为同等死亡负担;低中度严重度脓毒症需积极识别和按脓毒症处理,高严重度时脓毒症本身不必然意味着额外死亡风险。该发现提示脓毒症相关预后判断、临床试验设计和质量对标应考虑严重度依赖的效应修饰。
指南或共识出处
- 原文未提及
以上提炼依据论文摘要(全文未开放获取,或许可不允许翻译转载);全文请点「阅读原文」。
英文原文(摘要)· 点击展开
BACKGROUND: Illness severity scores such as SAPS II implicitly assume that condition-specific mortality contributions - including that of sepsis - remain constant across the severity spectrum. From a shock pathophysiology perspective, distributive shock from sepsis may carry different reversibility profiles than cardiogenic or neurogenic shock at equivalent severity. Whether illness severity modifies the mortality burden of sepsis remains unevaluated.
METHODS: We conducted a retrospective cohort study of 64,703 ICU admissions from MIMIC-IV (2008-2019). Sepsis was defined by Sepsis-3 criteria; baseline severity by SAPS II with fractional polynomial modeling. Multivariable logistic regression with interaction terms assessed effect modification; average marginal effects (AMEs) were estimated across the SAPS II spectrum. ICU-type interaction heterogeneity was tested.
RESULTS: Sepsis occurred in 42.7% of admissions; 30-day mortality was 12.9%. The sepsis-mortality association varied significantly by severity (interaction p<0.001). At SAPS II 30, sepsis was associated with higher mortality (AME +3.8%, 95%CI 3.2-4.3%); this attenuated and became non-significant at SAPS II 50-55. At SAPS II 80, septic patients had lower adjusted mortality (AME -7.3%, 95%CI -9.6 to -5.0%). This crossover was present in unadjusted data (SAPS II ≥70: 61.5% vs 69.2%) and consistent across eight sensitivity analyses.
CONCLUSIONS: The mortality burden of sepsis is not constant across the severity spectrum: it confers additional risk at low-to-moderate severity, but a crossover pattern emerges with lower mortality at high severity - robust in crude and adjusted analyses and consistent across ICU types. These findings implicate severity-dependent shock reversibility and have implications for prognostication, clinical trial design, and quality benchmarking.
参考来源
- Derakhshan, Ghomshei, Ansarian, et al. Illness Severity Modifies the Mortality Burden of Sepsis in Critically Ill Adults: Evidence of Severity-Dependent Shock Heterogeneity in 64,703 ICU Admissions. Shock. 2026. PMID: 42928527. doi:10.1097/SHK.0000000000002900
供执业医师抢救与诊疗参考,不替代临床判断;剂量请结合体重、肝肾功能和本院药品规格复核,以指南原文为准。