应用CTM及LR分析儿童烧烫伤危险因素的结果比较

Application and comparison between the classification tree model and LR in the influencing-factor analyses of unintentional pediatric burns

目的探讨决策树模型(classification tree model,CTM)在分析黔北农村学龄儿童非故意烧烫伤影响因素中的应用及与Logistic回归(Logistic Regression,LR)分析结果的差异。方法采用多阶段整群随机抽样的方法,在黔北地区抽取地6所小学和3所中学共1842名在校学龄儿童,采用问卷调查其过去1年内的烧烫伤发生情况,同时运用CTM与LR分析其影响因素并比较二者的结果差异。结果 (1)LR分析发现学龄儿童烧烫伤的危险因素有低年级;女童(OR=1.70,[95% CI]=1.29～2.25);留守儿童(OR=1.35,[95% CI]=1.02～1.80);母亲文化程度为小学及以下的(OR=2.43,[95% CI]=1.61～3.69)。(2)CTM结果显示,性别是儿童非故意烧烫伤发生的主要影响因素,影响男童烧烫伤发生最为显著的因素是其母亲高中文化程度以下者,影响女童烧烫伤发生的主要影响因素是其家庭经济状况。结论CTM与LR两法认为黔北地区农村学龄儿童非故意烧烫伤的共有影响因素有性别、年级和母亲文化;其余结果存在差异,互补使用效果可能更好。

Objective To explore the application and difference between classification tree model（CTM） and logistic regression（LR） to analyze the influencing factors of unintentional burns among school-aged children in rural northern Guizhou Province. Methods A stratified sampling method was used to recruit 1842 rural children from 9 schools（6 primary schools and 3 middle schools）,self-reported burns during the past 12 months and relevant concerns of burns were collected by questionnaires. CTM and LR analysis were used to find out the influencing factors. Results（1） The main risk factors for burns included grade,female（OR = 1. 70,[95% CI]= 1. 29-2. 25）,left-behind children（OR = 1. 35,[95% CI]= 1. 02-1. 80）,and poor mother education（OR = 2. 43,[95% CI]= 1. 61-3. 69）.（2） CTM analysis showed that the main influencing factor of unintentional burns was gender,the most significant factor affecting boys＇ burns was the level of their mothers＇ education level,and the main factors influencing girls＇ burns are their family economic situation. Conclusion Gender,grade,and mother＇s education degree were the joint influencing factors of burns between CTM and LR,while other results were some different,and the complementary use of CTM and LR was recommended.