Scholarlink

<p>Non-traumatic stroke (NTS) is associated with high mortality, and dynamic glucose patterns may provide prognostic information beyond static glucose measurements. This study aimed to identify glucose trajectory phenotypes in ICU patients with NTS and evaluate their association with in-hospital mortality. This multicenter retrospective cohort study included ICU patients with NTS from MIMIC-IV, eICU, and NSICU. Glucose trajectories during the first 168&#xa0;h after ICU admission were identified using latent class growth modeling. Associations with in-hospital mortality were assessed using Cox regression, Kaplan–Meier analysis, competing-risk analysis, and subgroup analyses. Incremental predictive value was evaluated using AUC, integrated discrimination improvement (IDI), net reclassification improvement (NRI), and median improvement in predicted risk. Among 11,177 patients, four glucose trajectory phenotypes were identified: normoglycemia (NG, <i>n</i> = 5,515), stable mild hyperglycemia (SMH, <i>n</i> = 3,425), persistent hyperglycemia (PH, <i>n</i> = 2,080), and highly variable hyperglycemia (HVH, <i>n</i> = 157). Compared with NG, SMH, PH, and HVH were progressively associated with higher in-hospital mortality in the fully adjusted model, with HRs of 1.668 (95% CI: 1.469–1.894), 2.406 (95% CI: 2.083–2.779), and 5.766 (95% CI: 4.319–7.699), respectively. HVH remained associated with increased mortality in both diabetic and non-diabetic patients. Adding glucose trajectories to baseline prediction models modestly improved AUC, IDI, NRI, and predicted risk classification across LASSO, Boruta, and BSS-BIC models. In ICU patients with NTS, glucose trajectory phenotypes were associated with different risks of in-hospital mortality, with HVH showing the strongest association. Glucose trajectories may provide additional prognostic information for risk stratification beyond static glucose measurements.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Blood glucose trajectory phenotypes are associated with in-hospital mortality in a multicenter ICU stroke cohort

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年6月3日

查看全文
  • Juan Wang,
  • Hai-Bo Li,
  • Wen-Juan Li,
  • Man-Man Xu,
  • Chun-Hua Hang,
  • Peng-Lai Zhao

摘要

Non-traumatic stroke (NTS) is associated with high mortality, and dynamic glucose patterns may provide prognostic information beyond static glucose measurements. This study aimed to identify glucose trajectory phenotypes in ICU patients with NTS and evaluate their association with in-hospital mortality. This multicenter retrospective cohort study included ICU patients with NTS from MIMIC-IV, eICU, and NSICU. Glucose trajectories during the first 168 h after ICU admission were identified using latent class growth modeling. Associations with in-hospital mortality were assessed using Cox regression, Kaplan–Meier analysis, competing-risk analysis, and subgroup analyses. Incremental predictive value was evaluated using AUC, integrated discrimination improvement (IDI), net reclassification improvement (NRI), and median improvement in predicted risk. Among 11,177 patients, four glucose trajectory phenotypes were identified: normoglycemia (NG, n = 5,515), stable mild hyperglycemia (SMH, n = 3,425), persistent hyperglycemia (PH, n = 2,080), and highly variable hyperglycemia (HVH, n = 157). Compared with NG, SMH, PH, and HVH were progressively associated with higher in-hospital mortality in the fully adjusted model, with HRs of 1.668 (95% CI: 1.469–1.894), 2.406 (95% CI: 2.083–2.779), and 5.766 (95% CI: 4.319–7.699), respectively. HVH remained associated with increased mortality in both diabetic and non-diabetic patients. Adding glucose trajectories to baseline prediction models modestly improved AUC, IDI, NRI, and predicted risk classification across LASSO, Boruta, and BSS-BIC models. In ICU patients with NTS, glucose trajectory phenotypes were associated with different risks of in-hospital mortality, with HVH showing the strongest association. Glucose trajectories may provide additional prognostic information for risk stratification beyond static glucose measurements.