Prediction of Postoperative Hypokalemia in Patients with Severe Carotid Artery Stenosis undergoing Standard Carotid Endarterectomy: A Retrospective Cohort Study.
{"title":"Prediction of Postoperative Hypokalemia in Patients with Severe Carotid Artery Stenosis undergoing Standard Carotid Endarterectomy: A Retrospective Cohort Study.","authors":"Suiyuan Hu, Xuan Lai, Yunfeng Han","doi":"10.4235/agmr.25.0164","DOIUrl":null,"url":null,"abstract":"<p><strong>Background: </strong>Postoperative hypokalemia is a common electrolyte disturbance associated with adverse outcomes, particularly in older adults. This study aimed to identify risk factors and develop predictive models for hypokalemia within 24 hours after carotid endarterectomy (CEA) for severe carotid artery stenosis, a condition that primarily affects older patient populations.</p><p><strong>Methods: </strong>A retrospective cohort of 1,076 CEA patients (October 2021 to May 2023) was analyzed. Risk factors were identified using univariate and multivariate logistic regression. A predictive nomogram was developed and internally validated via bootstrapping. Machine learning models (Random Forest and XGBoost) were developed and interpreted using SHAP (SHapley Additive exPlanations) analysis. Subgroup analyses were performed in patients aged ≥70 years and by comparing postoperative potassium levels >4.0 mmol/L versus 3.5-4.0 mmol/L.</p><p><strong>Results: </strong>The cohort had a median age of 65 years. Multivariate analysis identified preoperative potassium (odds ratio [OR]=0.60, 95% confidence interval [CI] 0.50-0.72), hemoglobin (OR=0.74, 95% CI 0.63-0.88), BMI (OR=0.74, 95% CI 0.63-0.88), and postoperative visual analogue scale score (OR=1.28, 95% CI 1.09-1.51) as independent predictors. Frailty showed borderline significance (OR=1.56, 95% CI 1.00-2.44, p=0.05). The nomogram achieved an area under the curve (AUC) of 0.710, demonstrating good discrimination and calibration. Machine learning models similarly performed well (AUC 0.707-0.709).</p><p><strong>Conclusion: </strong>We developed a validated tool to predict postoperative hypokalemia after CEA. The model highlights that in addition to biochemical and surgical factors, geriatric syndromes like frailty and nutritional status are pivotal risk determinants. This facilitates early, individualized management, including tailored potassium supplementation, nutritional support, and pain control, especially for vulnerable older adults, to mitigate complications and promote recovery.</p>","PeriodicalId":44729,"journal":{"name":"Annals of Geriatric Medicine and Research","volume":"30 1","pages":"18-27"},"PeriodicalIF":3.2000,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13054564/pdf/","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Annals of Geriatric Medicine and Research","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.4235/agmr.25.0164","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"2025/12/29 0:00:00","PubModel":"Epub","JCR":"Q3","JCRName":"GERIATRICS & GERONTOLOGY","Score":null,"Total":0}
引用次数: 0
Abstract
Background: Postoperative hypokalemia is a common electrolyte disturbance associated with adverse outcomes, particularly in older adults. This study aimed to identify risk factors and develop predictive models for hypokalemia within 24 hours after carotid endarterectomy (CEA) for severe carotid artery stenosis, a condition that primarily affects older patient populations.
Methods: A retrospective cohort of 1,076 CEA patients (October 2021 to May 2023) was analyzed. Risk factors were identified using univariate and multivariate logistic regression. A predictive nomogram was developed and internally validated via bootstrapping. Machine learning models (Random Forest and XGBoost) were developed and interpreted using SHAP (SHapley Additive exPlanations) analysis. Subgroup analyses were performed in patients aged ≥70 years and by comparing postoperative potassium levels >4.0 mmol/L versus 3.5-4.0 mmol/L.
Results: The cohort had a median age of 65 years. Multivariate analysis identified preoperative potassium (odds ratio [OR]=0.60, 95% confidence interval [CI] 0.50-0.72), hemoglobin (OR=0.74, 95% CI 0.63-0.88), BMI (OR=0.74, 95% CI 0.63-0.88), and postoperative visual analogue scale score (OR=1.28, 95% CI 1.09-1.51) as independent predictors. Frailty showed borderline significance (OR=1.56, 95% CI 1.00-2.44, p=0.05). The nomogram achieved an area under the curve (AUC) of 0.710, demonstrating good discrimination and calibration. Machine learning models similarly performed well (AUC 0.707-0.709).
Conclusion: We developed a validated tool to predict postoperative hypokalemia after CEA. The model highlights that in addition to biochemical and surgical factors, geriatric syndromes like frailty and nutritional status are pivotal risk determinants. This facilitates early, individualized management, including tailored potassium supplementation, nutritional support, and pain control, especially for vulnerable older adults, to mitigate complications and promote recovery.
背景:术后低钾血症是一种常见的与不良后果相关的电解质紊乱,特别是在老年人中。本研究旨在确定颈动脉内膜切除术(CEA)后24小时内低钾血症的危险因素并建立预测模型,颈动脉内膜切除术主要影响老年患者人群。方法:对1076例CEA患者(2021年10月至2023年5月)进行回顾性队列分析。使用单因素和多因素logistic回归确定危险因素。我们开发了一个预测nomogram,并通过bootstrap进行了内部验证。机器学习模型(随机森林和XGBoost)的开发和解释使用SHAP (SHapley加性解释)分析。对年龄≥70岁的患者进行亚组分析,比较术后钾水平(4.0 mmol/L和3.5-4.0 mmol/L)。结果:该队列的中位年龄为65岁。多因素分析确定术前钾(比值比[OR]=0.60, 95%可信区间[CI] 0.50-0.72)、血红蛋白(OR=0.74, 95% CI 0.63-0.88)、BMI (OR=0.74, 95% CI 0.63-0.88)和术后视觉模拟量表评分(OR=1.28, 95% CI 1.09-1.51)为独立预测因子。虚弱有临界意义(OR=1.56, 95% CI 1.00-2.44, p=0.05)。该图的曲线下面积(AUC)为0.710,具有良好的判别性和定标性。机器学习模型同样表现良好(AUC为0.707-0.709)。结论:我们开发了一种有效的预测CEA术后低钾血症的工具。该模型强调,除了生化和外科因素外,虚弱和营养状况等老年综合症也是关键的风险决定因素。这有利于早期、个性化的管理,包括量身定制的钾补充、营养支持和疼痛控制,特别是对脆弱的老年人,以减轻并发症和促进康复。