老年急性脑出血患者并发低钾血症的相关危险因素及风险预测模型构建

Related risk factors and construction of risk prediction model for hypokalemia in elderly patients with acute cerebral hemorrhage

目的 探讨老年急性脑出血患者并发低钾血症的相关危险因素,并且基于logistic回归构建风险预测模型。方法 选取2022年6月至2024年5月佛山市中医院诊治的老年急性脑出血患者190例,根据有无并发低钾血症分为低钾血症组(血钾<3.5 mmol/L)51例和正常组(血钾3.5~5.5 mmol/L)139例。采用logistic回归分析影响急性脑出血患者并发低钾血症的危险因素,并根据危险因素构建综合指数模型;ROC曲线分析综合指数预测急性脑出血患者并发低钾血症危险因素的诊断价值。结果 低钾血症组女性、入院时美国国立卫生研究院卒中量表(National Institutes of Health Stroke Scale, NIHSS)评分、入院时尿素、入院时血清肌酐、肾小球滤过率≤60 ml/min、呋塞米剂量>20 mg/d比例高于正常组(P<0.01)。单因素logistic回归分析显示,女性、入院时NIHSS评分、入院时尿素、入院时血清肌酐、肾小球滤过率≤60 ml/min、呋塞米剂量>20 mg/d为老年急性脑出血患者并发低钾血症的危险因素(P<0.05,P<0.01);多因素logistic回归分析显示,女性、入院时NIHSS评分、肾小球滤过率≤60 ml/min、呋塞米剂量>20 mg/d为老年急性脑出血患者并发低钾血症的危险因素(OR=6.393,95%CI:2.138~19.112,P=0.001;OR=3.123,95%CI:2.161~4.513,P=0.000;OR=3.327,95%CI:1.137~9.736,P=0.028;OR=3.111,95%CI:1.083~8.933,P=0.035)。ROC曲线分析显示,女性、入院时NIHSS评分、肾小球滤过率≤60 ml/min、呋塞米剂量>20 mg/d及综合指数预测急性脑出血患者并发低钾血症的曲线下面积分别为0.621、0.897、0.601、0.613、0.857,敏感性分别为52.90%、76.50%、49.00%、54.90%、72.50%,特异性分别为71.20%、88.50%、71.20%、67.60%、87.80%。结论 女性、入院时NIHSS评分、肾小球滤过率≤60 ml/min、呋塞米剂量>20 mg/d可能会对急性脑出血患者并发低钾血症产生影响,并依据这些危险因素构建综合指数模型对预测急性脑出血患者并发低钾血症具有较高的效能。

Objective To explore the related risk factors of hypokalemia in elderly patients with acute cerebral hemorrhage(ACH),and construct a risk prediction model based on logistic regression.Methods A total of 190 elderly ACH patients treated in Foshan Hospital of Traditional Chinese Medicine from June 2022 to May 2024 were enrolled as study objects, and were divided into hypokalemic group(potassium<3.5 mmol/L,n=51) and normal group(potassium 3.5-5.5 mmol/L,n=139) according to whether hypokalemia occurred.Logistic regression model was used to analyze the risk factors of hypokalemia in the elderly ACH patients.Based on the identified risk factors, a comprehensive index model was constructed.ROC curve was drawn to analyze the diagnostic value of the index for occurrence of hypokalemia in the patients.Results Larger female ratio, higher NIHSS score at admission, elevated urea nitrogen and blood creatinine at admission, and higher glomerular filtration rate(GFR)≤60 ml/min, and ratio of using hydrochlorothiazide >20 mg/d were observed in the hypokalemic group than the normal group(P<0.01).Univariate logistic regression analysis showed that female, NIHSS score at admission, urea nitrogen at admission, serum creatinine at admission, GFR ≤60 ml/min, and hydrochlorothiazide dose >20 mg/d were risk factors for hypokalemia in the ACH patients(P<0.05,P<0.01).Multivariate logistic regression analysis indicated that female, NIHSS score at admission, GFR ≤60 ml/min, and hydrochlorothiazide dose >20 mg/d were independent risk factors for hypokalemia in the elderly ACH patients(OR=6.393,95%CI:2.138-19.112,P=0.001;OR=3.123,95%CI:2.161-4.513,P=0.000;OR=3.327,95%CI:1.137-9.736,P=0.028;OR=3.111,95%CI:1.083-8.933,P=0.035).ROC curve analysis revealed that the AUC values of female, NIHSS score at admission, GFR,hydrochlorothiazide dose and comprehensive index in predicting hypokalemia in elderly ACH patients were 0.621,0.897,0.601,0.613 and 0.857,with a sensitivity of 52.90%,76.50%,49.00%,54.90% and 72.50%,and a specificity of 71.20%,88.50%,71.20%,67.60% and 87.80%,respectively.Conclusion Female, NIHSS score at admission, GFR ≤60 ml/min, and hydrochlorothiazide dose >20 mg/d may affect the occurrence of hypokalemia in elderly ACH patients, and our comprehensive index model based on these risk factors has high performance in predicting the occurrence of hypokalemia in elderly ACH patients.