乳酸/白蛋白比值与重症冠心病短期结局——双数据库验证

Lactate-to-Albumin Ratio and Short-Term Outcomes in Critically Ill Patients with Coronary Heart Disease—A Two-Database Validation Study

摘要: 目的：在既往单一数据库研究基础上，进一步评价乳酸/白蛋白比值(Lactate-to-Albumin Ratio, LAR)与重症冠心病患者入住重症监护病房(ICU)后28 d死亡风险的关系，并观察该关联在独立重症数据库中的一致性。方法：基于MIMIC-IV v3.1和eICU Collaborative Research Database开展回顾性队列研究。纳入年龄 > 18岁、首次入住ICU且诊断为冠心病的患者。MIMIC-IV主队列纳入4324例，按LAR三分位分为<0.46组、0.46~0.81组和>0.81组；eICU队列纳入331例，作为支持性验证队列。采用Kaplan-Meier生存曲线、Cox回归模型、限制性立方样条、亚组分析及模型诊断评估LAR与短期死亡风险的关系。结果：在MIMIC-IV队列中，较高的LAR伴随白蛋白降低，白细胞计数、肌酐、尿素氮、SOFA评分及SAPSⅡ评分升高，提示炎症负荷加重和器官功能障碍更明显。三组28 d死亡率分别为16.3%、22.9%和33.7%。完全调整后，连续型LAR仍与28 d死亡风险增加相关(HR = 1.136, 95% CI: 1.071~1.204, P < 0.001)；与LAR < 0.46组相比，LAR > 0.81组死亡风险升高(HR = 1.471, 95% CI: 1.236~1.750, P < 0.001)。eICU队列中，连续型LAR与死亡风险的关联方向一致(HR = 2.276, 95% CI: 1.740~2.978, P < 0.001)，但分类分析置信区间较宽。限制性立方样条和亚组分析均支持总体上呈正向关联，但eICU队列的曲线和分类分析结果应谨慎解读。结论：入ICU早期LAR升高与重症冠心病患者短期死亡风险增加相关。该结果更适宜作为既有证据的扩大样本和跨数据库支持性验证，提示LAR可作为早期个体化风险评估的辅助指标，其阈值稳定性和临床增量价值仍需前瞻性研究确认。

Abstract: Objective: To further evaluate the association between the lactate-to-albumin ratio (LAR) measured early after intensive care unit (ICU) admission and short-term mortality in critically ill patients with coronary heart disease (CHD), with particular attention to whether a previously reported single-database association could be supported in an independent critical care database. Methods: This retrospective cohort study used two publicly available critical care databases, MIMIC-IV version 3.1 and the eICU Collaborative Research Database. Adult patients were eligible if they were older than 18 years, had CHD identified by International Classification of Diseases codes, and were admitted to the ICU for the first time. LAR was calculated as serum lactate divided by serum albumin, using the first available measurements after ICU admission. In the MIMIC-IV cohort, patients were categorized by LAR tertiles into <0.46, 0.46~0.81, and >0.81 groups. The primary outcome was 28-day mortality after ICU admission in the main cohort. The eICU cohort was used for supportive validation because of its smaller sample size and differences in outcome availability. Kaplan-Meier curves, Cox regression models, restricted cubic splines, subgroup analyses, proportional hazards assumption tests, and collinearity diagnostics were used to characterize the association. Results: The study included 4324 patients from MIMIC-IV and 331 patients from eICU. In MIMIC-IV, higher LAR was accompanied by lower albumin and higher white blood cell count, creatinine, blood urea nitrogen, SOFA score, and SAPS II, suggesting greater inflammatory burden and organ dysfunction. The 28-day mortality rates were 16.3%, 22.9%, and 33.7% across increasing LAR tertiles. After full adjustment, continuous LAR remained associated with increased mortality risk (HR = 1.136, 95% CI: 1.071~1.204, P < 0.001). Compared with the lowest tertile, the highest LAR tertile had a higher mortality risk (HR = 1.471, 95% CI: 1.236~1.750, P < 0.001), whereas the middle tertile did not reach statistical significance. In the eICU cohort, continuous LAR showed a directionally consistent association after full adjustment (HR = 2.276, 95% CI: 1.740~2.978, P < 0.001), although categorical estimates were imprecise. Restricted cubic spline and subgroup analyses supported a generally positive association, but the eICU curves and categorical results should be interpreted cautiously. Conclusion: Elevated early LAR is associated with higher short-term mortality among critically ill patients with CHD. The present study should be interpreted as an expanded and supportive validation of previous evidence rather than a claim of a first discovery. LAR may provide a simple auxiliary marker for individualized early risk assessment, but its optimal threshold, incremental predictive value, and clinical utility require further prospective validation.