青少年术后长期使用阿片类药物的预测

阿片类药物是围术期最常使用的镇痛药，但存在成瘾性等风险，术后长期使用阿片类药物会对健康带来负面影响，如果有办法早期识别或预测高危人群，可以给予相应的预防措施，降低风险。各种机器算法在很多领域都发挥了重要的作用，那么其能否预测青少年术后长期使用阿片类药物的风险呢？分享一篇最新发表在Anesth Analg杂志上的文章，看看其结果如何。

从机器学习视角：

青少年术后长期使用阿片类药物的预测

背景：

长期使用阿片类药物会对健康带来负面影响。

手术患者术后长期使用阿片类药物的风险增加。

虽然已经识别出危险因素，但是目前仍没有办法来识别高危人群！

目的：

评估各种机器学习算法对青少年术后长期使用阿片类药物（prolonged opioid use after surgery ，POUS）的预测能力并鉴别出相关危险因素。

方法

采用回顾性队列研究；

研究人群：接受1297种全麻手术之一的12-21岁青少年；

来源：国家保险理赔数据库；

时间：2011年1月1日至2017年12月30日。

采用L2正则化Logistic回归和L1(Lasso)正则化Logistic回归、随机森林、梯度推进机和极限梯度提升模型进行训练，利用患者和提供者的特征预测POUS(术后90-180天内阿片类药物处方使用量≥1)风险。

采用受试者工作特征曲线下面积(AUC)/ c统计量(C-statistic)、算术平均数精度(MAP)评估预测能力。

使用敏感性、特异性、约登指数、F1评分和需要评估的数量来比较个体决策阈值。

利用排列重要性确定与POUS风险最密切相关的变量。

结果：

在符合条件的186493例手术患者中，8410例（4.51%）有POUS。

最佳算法的总体AUC为0.711（95%置信区间[CI]，0.699-0.723），

某些手术的AUC明显更高（例如，脊柱融合手术为0.823，牙科手术为0.812）。

与POUS相关性最强的变量是术前一年阿片类药物的供应天数和口服吗啡毫克当量。

结论：

预测青少年POUS风险的机器学习模型在不同的手术中显示出中等到强的结果，并揭示了与高风险相关的变量。

这些结果可能会为卫生保健系统识别POU高危患者提供信息，并推动预防措施的制定。

原文摘要

Prediction of Prolonged Opioid Use After Surgery inAdolescents: Insights From Machine Learning

Background:

Long-term opioid use has negative health care consequences. Patients whoundergo surgery are at risk for prolonged opioid use after surgery (POUS).While risk factors have been previously identified, no methods currently existto determine higher-risk patients. We assessed the ability of a variety ofmachine-learning algorithms to predict adolescents at risk of POUS and toidentify factors associated with this risk.

Methods:

A retrospective cohort study was conducted using a national insuranceclaims database of adolescents aged 12-21 years who underwent 1 of 1297surgeries, with general anesthesia, from January 1, 2011 to December 30, 2017.Logistic regression with an L2 penalty and with a logistic regression with anL1 lasso (Lasso) penalty, random forests, gradient boosting machines, andextreme gradient boosted models were trained using patient and providercharacteristics to predict POUS (≥1 opioid prescription fill within 90-180 daysafter surgery) risk. Predictive capabilities were assessed using the area underthe receiver-operating characteristic curve (AUC)/C-statistic, mean averageprecision (MAP); individual decision thresholds were compared usingsensitivity, specificity, Youden Index, F1 score, and number needed toevaluate. The variables most strongly associated with POUS risk were identifiedusing permutation importance.

Results:

Of 186,493 eligible patient surgical visits, 8410 (4.51%) had POUS. Thetop-performing algorithm achieved an overall AUC of 0.711 (95% confidenceinterval [CI], 0.699-0.723) and significantly higher AUCs for certain surgeries(eg, 0.823 for spinal fusion surgery and 0.812 for dental surgery). Thevariables with the strongest association with POUS were the days' supply ofopioids and oral morphine milligram equivalents of opioids in the year beforesurgery.

Conclusions:

Machine-learning models to predict POUS risk among adolescents show modestto strong results for different surgeries and reveal variables associated withhigher risk. These results may inform health care system-specific identificationof patients at higher risk for POUS and drive development of preventativemeasures.

原文链接

Ward A, Jani T, De Souza E, Scheinker D, Bambos N, Anderson TA. Predictionof Prolonged Opioid Use After Surgery in Adolescents: Insights From MachineLearning. Anesth Analg. 2021 May 3. doi: 10.1213/ANE.0000000000005527. Epubahead of print. PMID: 33939656.