Endo Atsushi, Baer Heather J, Nagao Masashi, Weaver Michael J
Harvard T.H. Chan School of Public Health, Boston, MA.
Division of General Internal Medicine and Primary Care, Brigham and Women's Hospital, Boston, MA.
J Orthop Trauma. 2018 Jan;32(1):34-38. doi: 10.1097/BOT.0000000000001026.
Mortality in elderly patients after the surgical treatment of hip fractures remains high. Although individual clinical risk factors have been widely studied, there has been limited research on prediction models in this population. The purpose of this study was to develop a prediction model for in-hospital mortality after hip fracture surgery and to evaluate the performance of this model.
Using the National Inpatient Sample database from 2012 to 2013, we collected data on 535,475 patients older than 50 years who had hip fracture surgery. Patient characteristics, surgery-specific factors, and Elixhauser comorbidities were used as candidate variables. The patients were randomly divided into training and testing cohorts. The Lasso (least absolute shrinkage and selection operator) method was used to select predictor variables, and points were assigned to each variable based on its coefficient.
We identified 8 essential predictors (age, timing of surgery, male sex, congestive heart failure, pulmonary circulation disease, renal failure, weight loss, and fluid and electrolyte disorders) for mortality, with a maximum prediction score of 20. The model's area under the curve was 0.74, and the Hosmer-Lemeshow test P value was 0.59 on the testing set. With the application of cutoff values (scores 0-5, 6-9, and 10-20), the observed in-hospital postoperative mortality was 0.6%, 2.5%, and 7.5%, respectively.
We built a simple prediction model with 8 essential clinical factors that predict in-hospital mortality after hip fracture surgery. This model may assist in counseling patients and families and measuring hospital quality of care.
Prognostic Level III. See Instructions for Authors for a complete description of levels of evidence.
老年患者髋部骨折手术治疗后的死亡率仍然很高。尽管对个体临床危险因素已进行了广泛研究，但针对该人群预测模型的研究却很有限。本研究的目的是开发一种髋部骨折手术后院内死亡率的预测模型，并评估该模型的性能。
利用2012年至2013年的国家住院患者样本数据库，我们收集了535475例年龄大于50岁且接受髋部骨折手术患者的数据。患者特征、手术相关因素和埃利克斯豪泽共病情况被用作候选变量。患者被随机分为训练队列和测试队列。采用套索（最小绝对收缩和选择算子）法选择预测变量，并根据其系数为每个变量赋值。
我们确定了8个死亡率的重要预测因素（年龄、手术时机、男性、充血性心力衰竭、肺循环疾病、肾衰竭、体重减轻以及液体和电解质紊乱），最大预测得分为20分。该模型在测试集上的曲线下面积为0.74，Hosmer-Lemeshow检验P值为0.59。应用临界值（分数0 - 5、6 - 9和10 - 20）时，观察到的术后院内死亡率分别为0.6%、2.5%和7.5%。
我们构建了一个包含8个重要临床因素的简单预测模型，用于预测髋部骨折手术后的院内死亡率。该模型可能有助于为患者及其家属提供咨询，并衡量医院的护理质量。
预后III级。有关证据水平的完整描述，请参阅作者指南。
