Guo Junfei, Di Jun, Gao Xian, Zha Junpu, Wang Xiuli, Wang Zhiqian, Wang Qingxian, Hou Zhiyong, Zhang Yingze
Department of Orthopaedic Surgery, Third Hospital of Hebei Medical University, Shijiazhuang, China.
Orthopaedic Research Institute of Hebei Province, Shijiazhuang, China.
Gerontology. 2022;68(1):62-74. doi: 10.1159/000515526. Epub 2021 Apr 23.
Preoperative risk assessment can predict adverse outcomes following hip fracture surgery, helping with decision-making and management strategies. Several risk adjustment models based on coded comorbidities such as Charlson Comorbidity Index (CCI), modified Elixhauser's Comorbidity Measure (mECM), and modified frailty index (mFI-5) are currently prevalent for orthopedic patients, but there is no consensus regarding which is optimal. The primary purpose was to identify the risk factors of CCI, mECM, and mFI-5, as well as patient characteristics for predicting (1) 1-month, 3-month, 1-year, and 2-year mortality, (2) perioperative complications, and (3) extended length of stay (LOS) following hip fractured surgery. The secondary aim was to compare the best-performing comorbidity index combined with characteristics identified in terms of their discriminative ability for adverse outcomes.
We retrospectively reviewed 3,379 consecutive patients presenting with intertrochanteric fractures at our Level I trauma center from 2013 to 2018. After eliminated by exclusion criteria, 2,241 patients undergoing hip fracture surgery by PFNA, with age ≥65 years, were included. Three main multivariate logistic regression models were constructed. Cox proportional hazards models were used to calculate hazard ratios for mortality. A base model included age, BMI, surgical delay, anesthesia type, hemoglobin record at admission, and American Society of Anesthesiologists grade (ASA) also was constructed and assessed.
Base model + mECM outperformed other models for the occurrence of major complications including severe complications, cardiac complications, and pulmonary complications [the area under the receiver operating characteristic curve (AUC), 0.647; 95% CI, 0.616-0.677; AUC, 0.637; 95% CI, 0.610-0.664; AUC, 0.679; 95% CI, 0.642-0.715, respectively], while base model + CCI provided better prediction of minor complications of neurological complications and hematological complications (AUC, 0.659; 95% CI, 0.609, 0.709; AUC, 0.658; 95% CI, 0.635, 0.680). In addition, BMI, surgical delay, anesthesia type, and ASA were found highly relevant to extended LOS. Age-group (with a 10-year interval) was indicated to be mostly associated with all-cause mortality with fully adjusted hazard ratio of 1.35 and 95% CI range 1.20-1.51.
In comparison with mFI-5 and CCI, mECM so far may be the best comorbidity index combined with the base model for predicting major complications following hip fracture. The base model already achieved good discrimination for all-cause mortality and extended LOS, further addition of risk adjustment indices led to only 1% increase in the amount of variation explained.
术前风险评估可预测髋部骨折手术后的不良结局，有助于制定决策和管理策略。目前，基于编码合并症的几种风险调整模型，如Charlson合并症指数（CCI）、改良的Elixhauser合并症测量法（mECM）和改良衰弱指数（mFI-5）在骨科患者中普遍使用，但对于哪种模型最佳尚无共识。主要目的是确定CCI、mECM和mFI-5的风险因素，以及用于预测（1）1个月、3个月、1年和2年死亡率，（2）围手术期并发症，以及（3）髋部骨折手术后延长住院时间（LOS）的患者特征。次要目的是比较表现最佳的合并症指数与所确定特征在不良结局判别能力方面的差异。
我们回顾性分析了2013年至2018年在我院一级创伤中心连续就诊的3379例转子间骨折患者。经排除标准筛选后，纳入2241例年龄≥65岁、接受PFNA髋部骨折手术的患者。构建了三个主要的多因素逻辑回归模型。采用Cox比例风险模型计算死亡率的风险比。还构建并评估了一个基础模型，该模型包括年龄、体重指数（BMI）、手术延迟、麻醉类型、入院时血红蛋白记录以及美国麻醉医师协会分级（ASA）。
基础模型+mECM在预测包括严重并发症、心脏并发症和肺部并发症在内的主要并发症发生率方面优于其他模型[受试者操作特征曲线（AUC）下面积分别为0.647；95%置信区间（CI），0.616 - 0.677；AUC为0.637；95% CI，0.610 - 0.664；AUC为0.679；95% CI，0.642 - 0.715]，而基础模型+CCI对神经并发症和血液学并发症等次要并发症的预测效果更好（AUC分别为0.659；95% CI，0.609，0.709；AUC为0.658；95% CI，0.635，0.680）。此外，发现BMI、手术延迟、麻醉类型和ASA与延长住院时间高度相关。年龄组（以10年为间隔）被表明与全因死亡率最相关，完全调整后的风险比为1.35，95% CI范围为1.20 - 1.51。
与mFI-5和CCI相比，mECM目前可能是与基础模型相结合用于预测髋部骨折后主要并发症的最佳合并症指数。基础模型在全因死亡率和延长住院时间的判别方面已经表现良好，进一步添加风险调整指数仅使可解释变异量增加了1%。
