全髋关节置换术后持续阿片类药物处方的机器学习预测算法开发
本研究基于2家学术医疗中心和3家社区医院2000年1月1日至2018年8月1日因骨关节炎行全髋关节置换术的成人患者电子健康记录进行回顾性分析,开发5种机器学习算法术前预测术后持续阿片类药物处方。共5507例患者中345例(6.3%)出现术后持续阿片类药物处方,弹性网惩罚逻辑回归模型区分度最佳(c统计量=0.77),已整合为可提供预测与解释的数字应用。
要点
- 全髋关节置换术后持续阿片类药物处方发生率为6.3%(345/5507)。
- 预测因素包括年龄、阿片类药物暴露时长、术前血红蛋白及术前用药。
- 弹性网惩罚逻辑回归模型区分度最佳,c统计量=0.77。
核心结论
- 若在独立人群中获得外部验证,本研究开发的算法可改善对全髋关节置换术后持续阿片类药物处方高风险患者的术前筛查与支持。
- 早期识别和干预高风险病例可能减轻阿片类药物依赖的长期不良后果。
研究设计与样本
- 回顾性审查2家学术医疗中心和3家社区医院的电子健康记录,纳入2000年1月1日至2018年8月1日因骨关节炎行全髋关节置换术的成人患者。
- 术后持续阿片类药物处方定义为术后连续阿片类药物处方持续至术后至少90天。
- 开发5种机器学习算法预测该结局,并通过区分度、校准和决策曲线分析进行评估。
- 证据等级为III级。
主要发现
- 共5507例患者行全髋关节置换术,其中345例(6.3%)出现术后持续阿片类药物处方。
- 用于预测术后持续阿片类药物处方的因素为年龄、阿片类药物暴露时长、术前血红蛋白和术前用药(抗抑郁药、苯二氮䓬类药物、非甾体抗炎药和β2受体激动剂)。
- 弹性网惩罚逻辑回归模型在区分度(c统计量=0.77)、校准和决策曲线分析中表现最佳。
- 该模型已整合至可提供预测和解释的数字应用(https://sorg-apps.shinyapps.io/thaopioid/)。
关键数值与推荐
- 术后持续阿片类药物处方定义为术后连续阿片类药物处方持续至术后至少90天。
- 弹性网惩罚逻辑回归模型的c统计量为0.77。
- 共5507例患者,345例(6.3%)出现术后持续阿片类药物处方。
临床怎么用
- 原文结论指出,若在独立人群中获得外部验证,该算法可改善对全髋关节置换术后持续阿片类药物处方高风险患者的术前筛查与支持。
- 原文结论指出,早期识别和干预高风险病例可能减轻阿片类药物依赖的长期不良后果。
局限性
- 原文结论指出该算法需在独立人群中获得外部验证。
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
原文
BACKGROUND: Postoperative recovery after total hip arthroplasty (THA) can lead to the development of prolonged opioid use but there are few tools for predicting this adverse outcome. The purpose of this study is to develop machine learning algorithms for preoperative prediction of prolonged opioid prescriptions after THA. METHODS: A retrospective review of electronic health records was conducted at 2 academic medical centers and 3 community hospitals to identify adult patients who underwent THA for osteoarthritis between January 1, 2000 and August 1, 2018. Prolonged postoperative opioid prescriptions were defined as continuous opioid prescriptions after surgery to at least 90 days after surgery. Five machine learning algorithms were developed to predict this outcome and were assessed by discrimination, calibration, and decision curve analysis. RESULTS: Overall, 5507 patients underwent THA, of which 345 (6.3%) had prolonged postoperative opioid prescriptions. The factors determined for prediction of prolonged postoperative opioid prescriptions were age, duration of opioid exposure, preoperative hemoglobin, and preoperative medications (antidepressants, benzodiazepines, nonsteroidal anti-inflammatory drugs, and beta-2-agonists). The elastic-net penalized logistic regression model achieved the best performance across discrimination (c-statistic = 0.77), calibration, and decision curve analysis. This model was incorporated into a digital application able to provide both predictions and explanations (available at https://sorg-apps.shinyapps.io/thaopioid/). CONCLUSION: If externally validated in independent populations, the algorithms developed in this study could improve preoperative screening and support for THA patients at high risk for prolonged postoperative opioid prescriptions. Early identification and intervention in high-risk cases may mitigate the long-term adverse consequence of opioid dependence. LEVEL OF EVIDENCE: III.
引用
- Karhade AV, Schwab JH, Bedair HS, et al. Development of Machine Learning Algorithms for Prediction of Sustained Postoperative Opioid Prescriptions After Total Hip Arthroplasty. J Arthroplasty. 2019. PMID: 31327647. doi:10.1016/j.arth.2019.06.013
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。