Wang Wei, Yao Wei, Tang Wanyun, Li Yuhao, Liu Yazhou, Lv Qiaomei, Ding Wenbo
Department of Orthopedics, Dandong Central Hospital, China Medical University, Dandong, China.
School of Clinical Medicine, Dalian Medical University, Dalian, China.
J Arthroplasty. 2025 Jun;40(6):1573-1581.e4. doi: 10.1016/j.arth.2024.11.037. Epub 2024 Nov 27.
Predicting postoperative delirium (POD) in patients who have hip fractures is challenging due to its complex mechanism. Therefore, there is a critical need to explore and evaluate a novel predictive indicator.
There were four hematological markers independently associated with POD that were utilized to construct and evaluate a more reliable predictive indicator for POD. The study employed random sampling, dividing the data into training and validation cohorts in a 7:3 ratio. The strength of association between each predictive indicator and POD was assessed using multivariable logistic analysis and propensity score matching analysis. Predictive indicators with significant correlations underwent receiver operating characteristic curve and characteristic parameter comparisons to identify the optimal predictive indicator. Subsequent validation included the assessment of discriminative ability, correlation, and predictive performance. Furthermore, subgroup analysis was conducted to explore potential interactions. A total of 1,807 patients were included in this study, with a POD incidence rate of 16.5%.
Multivariable logistic analysis and propensity score matching analysis demonstrated that the glucose-to-albumin ratio (GAR) was independently positively associated with POD. Specifically, for every 0.1 unit increase in preoperative GAR levels in hip fracture patients, the risk of POD increased by 1.6 times. The receiver operating characteristic curve curve indicated that the optimal cutoff value for the GAR was 0.2, with an area under the curve of 0.8, sensitivity of 81.2%, and specificity of 59.0%.
Preoperative GAR has a certain predictive value for the occurrence of POD and can function as a novel indicator for predicting POD in geriatric hip fracture patients.
由于机制复杂，预测髋部骨折患者术后谵妄（POD）具有挑战性。因此，迫切需要探索和评估一种新的预测指标。
利用与POD独立相关的四种血液学标志物构建并评估一种更可靠的POD预测指标。该研究采用随机抽样，将数据按7:3的比例分为训练队列和验证队列。使用多变量逻辑分析和倾向评分匹配分析评估每个预测指标与POD之间的关联强度。对具有显著相关性的预测指标进行受试者工作特征曲线和特征参数比较，以确定最佳预测指标。随后的验证包括评估判别能力、相关性和预测性能。此外，进行亚组分析以探索潜在的相互作用。本研究共纳入1807例患者，POD发生率为16.5%。
多变量逻辑分析和倾向评分匹配分析表明，血糖与白蛋白比值（GAR）与POD独立正相关。具体而言，髋部骨折患者术前GAR水平每升高0.1个单位，POD风险增加1.6倍。受试者工作特征曲线表明，GAR的最佳截断值为0.2，曲线下面积为0.8，灵敏度为81.2%，特异性为59.0%。
术前GAR对POD的发生具有一定的预测价值，可作为老年髋部骨折患者预测POD的新指标。
