目的:探讨白细胞源性标志物对心脏瓣膜手术患者术后谵妄(POD)的预测价值。
方法:进行前瞻性队列研究。纳入2021年10月至2023年3月在首都医科大学附属北京安贞医院接受心脏瓣膜手术的患者。人口统计,收集基线和围手术期数据,术前和术后24小时内计算中性粒细胞与淋巴细胞比值(NLR)和血小板与白细胞比值(PWR)。术后1-5天内或5天内出院的患者每天进行两次谵妄评估。根据评价结果,将患者分为谵妄组和非谵妄组。比较两组患者的临床指标。采用多因素Logistic回归分析筛选POD的独立危险因素,并构建了POD预测模型。采用受试者操作特征曲线(ROC曲线)评价POD预测模型的预测价值。
结果:共有235名患者被纳入分析,其中83例患者有POD(35.32%),152例患者无POD(64.68%)。与非谵妄组相比,谵妄组患者的Charlson合并症指数(CCI)评分较高,简易精神状态检查(MMSE)评分较低.在围手术期数据方面,与非谵妄组相比,谵妄组患者手术时间较长,体外循环的持续时间,重症监护病房(ICU)住院时间,机械通气的持续时间,以及术后住院时间,围手术期房颤发生率较高,和较低的出院寿命评分。在白细胞来源的标志物方面,两组患者术后24小时内NLR均明显高于术前,PWR明显低于手术前。NLR在手术后24小时内,谵妄组的PWR差异和NLR差异显著高于非谵妄组。多因素Logistic回归分析显示,CCI评分[比值比(OR)=1.394,95%置信区间(95CI)为1.038~1.872,P=0.027]。围手术期房颤(OR=3.697,95CI为1.711~7.990,P<0.001),体外循环时间(OR=1.008,95CI为1.002-1.015,P=0.016),ICU住院时间(OR=1.006,95CI为1.002-1.010,P=0.002),NLR差异(OR=1.029,95CI为1.009-1.050,P=0.005)和PWR差异(OR=1.044,95CI为1.009-1.080,P=0.013)与POD独立相关。多因素Logistic回归分析结果:POD预测模型指数=-4.970+0.336×CCI评分+1.317×围术期房颤+0.009×体外循环时间+0.006×ICU住院时间+0.030×NLR差值+0.044×PWR差值。ROC曲线分析显示预测POD的NLR差值的ROC曲线下面积(AUC)为0.659(95CI为0.583~0.735),最佳临界值为16.62,灵敏度为60.2%,特异性为70.4%(P<0.05)。预测POD的PWR差异的AUC为0.608(95CI为0.528-0.688),最佳临界值为25.68,灵敏度为51.8%,特异性为75.7%(P<0.05)。POD预测模型预测POD的AUC为0.805(95CI为0.745-0.865),最佳临界值为0.39,灵敏度为74.7%,特异性为79.6%(P<0.05)。
结论:NLR和PWR的差异与POD独立相关,这对预测心脏瓣膜手术后的POD具有潜在价值。
OBJECTIVE: To explore the predictive value of leukocyte derived markers for postoperative delirium (POD) in patients undergoing cardiac valve surgery.
METHODS: A prospective cohort study was conducted. The patients who underwent cardiac valve surgery admitted to Beijing Anzhen Hospital of Capital Medical University from October 2021 to March 2023 were enrolled. The demographic, baseline and perioperative data were collected, and the neutrophil to lymphocyte ratio (NLR) and platelet to white blood cell ratio (PWR) were calculated before operation and within 24 hours after operation. Delirium assessment was conducted twice a day for patients within 1-5 days after surgery or discharged within 5 days. According to the evaluation results, the patients were divided into delirium group and non-delirium group. The clinical indexes between the two groups were compared. Multivariate Logistic regression analysis was used to screen the independent risk factors of POD, and the POD predictive model was constructed. The predictive value of POD predictive model was evaluated by receiver operator characteristic curve (ROC curve).
RESULTS: A total of 235 patients were enrolled in the analysis, of which 83 patients had POD (35.32%) and 152 patients did not have POD (64.68%). Compared with the non-delirious group, the patients in the delirious group had higher Charlson comorbidity index (CCI) score and lower mini-mental state examination (MMSE) score. In terms of perioperative data, compared with the non-delirium group, the patients in the delirium group had longer operative time, duration of cardiopulmonary bypass, length of intensive care unit (ICU) stay, duration of mechanical ventilation, and postoperative hospital stay, higher incidence of perioperative atrial fibrillation, and lower discharge life score. In terms of leukocyte derived markers, NLR within 24 hours after surgery in both groups were significantly higher than those before surgery, and PWR were significantly lower than those before surgery. The NLR within 24 hours after surgery, PWR difference and NLR difference in the delirium group were significantly higher than those in the non-delirium group. Multivariate Logistic regression analysis showed that CCI score [odds ratio (OR) = 1.394, 95% confidence interval (95%CI) was 1.038-1.872, P = 0.027], perioperative atrial fibrillation (OR = 3.697, 95%CI was 1.711-7.990, P < 0.001), duration of cardiopulmonary bypass (OR = 1.008, 95%CI was 1.002-1.015, P = 0.016), length of ICU stay (OR = 1.006, 95%CI was 1.002-1.010, P = 0.002), NLR difference (OR = 1.029, 95%CI was 1.009-1.050, P = 0.005) and PWR difference (OR = 1.044, 95%CI was 1.009-1.080, P = 0.013) were independently correlated with POD. POD predictive model was constructed by multivariate Logistic regression analysis result: POD predictive model index = -4.970+0.336×CCI score+1.317×perioperative atrial fibrillation+0.009×duration of cardiopulmonary bypass+0.006×length of ICU stay+0.030×NLR difference+0.044×PWR difference. ROC curve analysis showed that the area under the ROC curve (AUC) of NLR difference for predicting POD was 0.659 (95%CI was 0.583-0.735), the optimal critical value was 16.62, the sensitivity was 60.2%, and the specificity was 70.4% (P < 0.05). The AUC of PWR difference for predicting POD was 0.608 (95%CI was 0.528-0.688), the optimal critical value was 25.68, the sensitivity was 51.8%, and the specificity was 75.7% (P < 0.05). The AUC of POD predictive model for predicting POD was 0.805 (95%CI was 0.745-0.865), the optimal critical value was 0.39, the sensitivity was 74.7%, and the specificity was 79.6% (P < 0.05).
CONCLUSIONS: The differences of NLR and PWR are independently related to POD, which has potential value in predicting POD after cardiac valve surgery.