Hospital Universitario Rey Juan Carlos, Universidad Rey Juan Carlos, 28933 Móstoles, Spain.
Hospital Universitario Fundación Alcorcón, Instituto de Investigación Hospital Universitario La Paz, 28046 Madrid, Spain.
Int J Environ Res Public Health. 2021 Apr 6;18(7):3809. doi: 10.3390/ijerph18073809.
The aim of this study was to develop a predictive model of gait recovery after hip fracture. Data was obtained from a sample of 25,607 patients included in the Spanish National Hip Fracture Registry from 2017 to 2019. The primary outcome was recovery of the baseline level of ambulatory capacity. A logistic regression model was developed using 40% of the sample and the model was validated in the remaining 60% of the sample. The predictors introduced in the model were: age, prefracture gait independence, cognitive impairment, anesthetic risk, fracture type, operative delay, early postoperative mobilization, weight bearing, presence of pressure ulcers and destination at discharge. Five groups of patients or clusters were identified by their predicted probability of recovery, including the most common features of each. A probability threshold of 0.706 in the training set led to an accuracy of the model of 0.64 in the validation set. We present an acceptably accurate predictive model of gait recovery after hip fracture based on the patients' individual characteristics. This model could aid clinicians to better target programs and interventions in this population.
本研究旨在建立一个预测髋部骨折后步态恢复的模型。数据来自 2017 年至 2019 年期间纳入西班牙国家髋部骨折登记处的 25607 名患者的样本。主要结局是恢复基线步行能力水平。使用样本的 40%建立逻辑回归模型，并在剩余 60%的样本中验证模型。模型中引入的预测因素包括：年龄、骨折前独立行走能力、认知障碍、麻醉风险、骨折类型、手术延迟、术后早期活动、负重、压疮的存在和出院去向。根据预测恢复概率，将患者分为五个组或聚类，包括每个聚类的最常见特征。在训练集中，概率阈值为 0.706，验证集的模型准确率为 0.64。我们提出了一种基于患者个体特征的髋部骨折后步态恢复的可接受的准确预测模型。该模型可以帮助临床医生更好地针对该人群的项目和干预措施。
