深圳市重症手足口病危险因素分析及其概率模型的建立

目的：探讨深圳市儿童手足口病重症病例发生的危险因素，构建其风险预测模型，为重症患儿的早期识别提供依据。方法回顾性分析2011年1月至2011年12月于本院住院的171例手足口病患儿的临床资料，采用性别、年龄、入院时间、按照1︰2匹配的方法（病例组例数与对照组例数比值为1︰2）将患儿分为重症组和轻症组。以儿童职业、居住地、性别、年龄（月）、体重（kg）、发热天数、EV71、手部皮疹、足部皮疹、口腔疱疹、呼吸系统症状、咳嗽、发病至初诊时间、初诊至入院时间、外周血白细胞计数（×10^9/L）、随机血糖、中性粒细胞及淋巴细胞比例等指标作为发生重症手足口病的可能影响因素，利用二分类多因素Logistic回归分析，在此基础上建立风险模型并进行预测，并评价模型的预测效果。结果经二分类多因素分析，手部皮疹分布、白细胞计数、年龄和EV71阳性4个因素为重症手足口病的独立危险因素，根据多因素Logistic回归分析的结果建立Logistic回归预测模型。对模型的预测概率进行ROC曲线分析，曲线下面积为0.870。根据该模型，对现有的数据进行预测，模型的敏感度为87.7%，特异度为93.8%，一致率为91.0%。结论重症手足口病模型可定量评估重症手足口病发生的概率。

Risk factors for severe cases in children with hand,foot and mouth disease and establishment of probability model in Shenzhen

Objective To explore the risk factors for severe cases in children with hand, foot and mouth disease （HFMD） in Shenzhen, and to establish a risk model for the early diagnosis of the severe patients. Methods A retrospective analysis was carried out about 171 hospitalized cases with HFMD. A case-control study was conducted on two groups of children matched by sex, age and hospital of time. The patients were divided into severe and mild groups. The clinical measures were studied as the possible risk factors, including gender, age, inhabitant environment, the time interval between onset date and clinic date, the days from the onset date to the admission date, rashes on the hands, EV71/CoxA16 infection, peripheral blood leukocyte, percentage of neutrophil leukocyte and lymphocyte. Binary Logistic regression was used to examine the relationship between risk factors and severe cases. A risk model was built according to the above factors. The predictive effect of the model was evaluated. Results In the multivariable analysis, 4 variables in the risk model （rashes on the hands, age, the higher number of leukocyte and EV71） were independent predictors for the outcome. The risk model highly predicted severe cases. The area under the ROC curve for the evaluation model was 0.870. When the prediction was performed based on the existing date using the present model, the sensitivity, speciifcity and consistency of the model were 87.7%, 93.8%and 91.0%, respectively. Conclusion The risk model could quantitatively predict severe cases in children with HFMD in Shenzhen.