Abstract:

Objective　To explore risk factors of childhood attention deficit hyperactivity disorder (ADHD) and construct a risk prediction model. Methods　From July 2023 to May 2024, 78 children with ADHD who were admitted to Hunan Children's Hospital were selected as the ADHD group. Meanwhile, 312 children who were admitted for treatment during the same period but did not have ADHD were selected as the non-ADHD group at a ratio of 1:4. In the ADHD group, there were 41 boys and 37 girls, aged (8.42±1.35) years.In the non-ADHD group, there were 169 boys and 143 girls, aged (8.57±1.29) years. A uniformly designed data collection form combined with electronic medical records was used to collect the relevant information of the children including age, gender, educational level of parents, mode of delivery, etc. Independent sample t test and χ² test were used for statistical analysis. Multivariate logistic regression analysis was used to analyze the influencing factors of the diagnosis of children with ADHD,and a risk prediction model was constructed. The diagnostic efficacy of the model on children with ADHD was analyzed by using receiver operating characteristic curve (ROC). The calibration curve and Hosmer-Lemeshow test were used to test the calibration degree and goodness of fit of the model. Results　From July 2023 to May 2024, Hunan Provincial Children's Hospital admitted a total of 1 391 children, among whom 78 cases were ADHD patients, with an incidence rate of 5.61%. The proportions of fetal distress in utero, prolonged labor, broken parent-child relationship in the family, and educational types of crude and excessive indulgence in the ADHD group were higher than those in the non-ADHD group [69.23% (54/78) vs. 12.82% (40/312), 70.51% (55/78) vs. 16.35% (51/312), 76.92% (60/78) vs. 16.67% (52/312), 73.08% (57/78) vs. 10.58% (33/312)] (all P<0.05). The results of logistic regression analysis indicated that fetal distress in utero, prolonged labor, broken parent-child relationship in the family, and educational types of crude and excessive indulgence were independent risk factors influencing the diagnosis of children with ADHD (all P<0.05). The area under the ROC of the risk prediction model for ADHD in children was 0.803 (95%CI: 0.706-0.951), the Youden index was 0.877, the critical value was 0.301, the sensitivity was 81.42%, and the specificity was 87.51%. The Hosmer-Lemeshow test results showed that χ²=16.889,P=0.302. Conclusion　The risk prediction model for childhood ADHD constructed in this study has considerable clinical value. Medical staff can take targeted preventive measures based on the risk factors of childhood ADHD.

Key words:  , Children's attention deficit hyperactivity disorder,  , Risk factors,  , Prediction model

摘要：

目的　探讨儿童注意缺陷多动障碍（ADHD）的危险因素，并构建风险预测模型。方法　选取2023年7月至2024年5月湖南省儿童医院收治的78例ADHD患儿作为ADHD组，按1∶4的比例选取同期入院治疗的312例未发生ADHD患儿作为未发生ADHD组。ADHD组男41例，女37例；年龄（8.42±1.35）岁；未发生ADHD组男169例，女143例；年龄（8.57±1.29）岁。采用统一设计的资料收集表结合电子病历收集患儿相关信息，包括：年龄、性别、父母文化程度、分娩方式等。采用独立样本t检验、χ2检验进行统计学分析；采用logistic回归分析ADHD患儿诊断的影响因素，并构建风险预测模型；采用受试者操作曲线（ROC）分析该模型对ADHD患儿的诊断效能；采用校准曲线和Hosmer-Lemeshow检验模型校准度和拟合优度。结果　2023年7月至2024年5月湖南省儿童医院共收治1 391例患儿，其中78例为ADHD患儿，发生率为5.61%。ADHD组胎儿宫内窘迫、产程延长、家庭父母关系破裂、教育类型为粗暴/过度溺爱的占比均高于未发生ADHD组［69.23%（54/78）比12.82%（40/312）、70.51%（55/78）比16.35%（51/312）、76.92%（60/78）比16.67%（52/312）、73.08（57/78）比10.58（33/312）］（均P<0.05）。logistic回归分析结果显示，胎儿宫内窘迫、产程延长、家庭父母关系破裂、教育类型为粗暴/过度溺爱均是影响ADHD患儿诊断的独立危险因素（均P<0.05）。儿童ADHD风险预测模型ROC曲线下面积为0.803（95%CI：0.706～0.951），约登指数为0.877，临界值为0.301，灵敏度为81.42%，特异度为87.51%。Hosmer-Lemeshow检验结果显示，χ2=16.889，P=0.302。结论　本研究构建的儿童ADHD风险预测模型具有较好的临床价值，医护人员可针对儿童ADHD风险因素，采取针对性预防措施。

关键词: 儿童注意缺陷多动障碍,  ,  , 危险因素,  ,  , 预测模型