基于Cox模型的美沙酮维持治疗患者脱失预测研究  

Prediction of patients' dropping out in community-based methadone maintenance treatment using Cox Model

作　　者：牛姬飞[1] 杨长贵[2] 康春香[2] 严燚[3] 司徒潮满[1] 舒彬[1] 姜世强[3] 

机构地区：[1]深圳市福田区疾病预防控制中心,广东深圳518040 [2]深圳市福田区人民医院第三分院 [3]深圳市南山区疾病预防控制中心

出　　处：《实用预防医学》2016年第5期529-532,共4页Practical Preventive Medicine

基　　金：2013年度福田区卫生公益性科研项目(编号:FTWS201348)

摘　　要：目的构建Cox比例风险模型预测美沙酮维持治疗门诊患者脱失概率,及时识别患者脱失风险,实施个体化干预,提高患者维持治疗率。方法收集深圳市福田区和南山区2009-2014年所有美沙酮维持治疗患者相关信息,将样本分为训练样本和测试样本,训练样本用来拟合Cox比例风险预测模型,测试样本用来评估模型信度和效度。结果经拟合模型,筛选出8个变量对脱失预测有统计学意义(P<0.05),经测试样本评估,模型的灵敏度为82.24%,特异度为80.76%,一致率为81.76%,说明模型预测结果和实际脱失情况具有较高的一致性。结论 Cox比例风险模型可以用于美沙酮维持治疗门诊患者的脱失预测。Objective To construct the Cox proportional hazards model to predict the probability of patients＇ dropping out in methadone maintenance treatment （MMT）, and to identify dropout risk factors in time, implement individualized intervention measures and improve MMT rate. Methods The relevant information of patients with MMT in Futian and Nanshan District in Shenzhen City from 2009 to 2014 were collected. The data collected were divided into the training sample data and testing sample data. The training sample data were used to construct the Cox model, and the testing sample data were used to evaluate the reliability and validity of the model. Results From the constructed Cox model, 8 variables with significance in predicting patients＇ dropping - out were selected. Evaluation with the testing sample data indicated that the model＇ s sensitivity was 82.24 %, specificity was 80.76 % and the consistent rate was 81.76 %, indicating that predictions by the model had high consis- tency with the actual dropout situation. Conclusions The Cox proportional hazards mode Drop out can be used to predict the patient＇s dropout in methadone maintenance treatment.

关 键 词：美沙酮维持治疗 脱失 COX比例风险模型 

分 类 号：R512.91[医药卫生—内科学]