Aims/IntroductionMean blood glucose (MBG) level is associated with mortality among critically ill patients. We undertook a cohort study to investigate the relationship between MBG and mortality in critically ill patients.Materials and MethodsCritically ill patients were enrolled from the Medical Information Mart for Intensive Care IV (MIMIC‐IV) database. MBG was calculated to represent the overall glycemic status during intensive care unit (ICU) hospitalization, and a multivariate logistic regression determined the relationship between MBG and ICU mortality in different subgroups of critically ill patients.ResultsA total of 8,973 patients were included in the study, 1,244 of whom died within 28 days, including 5,402 men and 3,571 women. Multivariate adjusted restricted cubic spline analyses suggested that the relationship between MBG and ICU mortality was a “J” shape. Logistic regression showed 28 day mortality in group 3 (glucose ≥10 mmol/L): the adjusted odds ratio was 2.06 (95% confidence interval 1.65–2.57). The results of subgroup analysis showed that hyperglycemia had a more significant impact on ICU mortality in patients without diabetes, hypoglycemia and liver disease, and the ICU mortality risk of non‐diabetes patients was always higher than that of diabetes patients with the same hyperglycemia level.ConclusionsCurrent evidence suggested a J‐shaped relationship between MBG and mortality in critically ill patients.

中文翻译：

---

重症患者血糖与死亡率之间的关系：MIMIC-IV数据库的回顾性分析

目的/简介平均血糖 (MBG) 水平与危重患者的死亡率相关。我们进行了一项队列研究来调查重症患者的 MBG 与死亡率之间的关系。材料和方法重症患者从重症监护 IV 医疗信息集市 (MIMIC-IV) 数据库中入组。计算MBG来代表重症监护病房（ICU）住院期间的总体血糖状态，并通过多变量逻辑回归确定MBG与ICU危重患者不同亚组死亡率之间的关系。结果共有8,973名患者纳入研究，其中 1,244 人在 28 天内死亡，其中男性 5,402 人，女性 3,571 人。多变量调整限制三次样条分析表明MBG与ICU死亡率之间呈“J”形关系。Logistic 回归显示第 3 组（血糖≥10 mmol/L）的 28 天死亡率：调整后的优势比为 2.06（95% 置信区间 1.65–2.57）。亚组分析结果显示，高血糖对无糖尿病、低血糖和肝病患者的ICU死亡率影响更为显着，且非糖尿病患者的ICU死亡风险始终高于相同高血糖水平的糖尿病患者。结论 目前的证据表明，重症患者的 MBG 与死亡率之间存在 J 形关系。