潜在类别分析在出生缺陷高危人群识别中的应用

Application of Latent Class Analysis in the Identification of High-risk Population in Birth Defects

目的 应用潜在类别分析（latent class analysis,LCA）对具有不同出生缺陷相关因素分布特征的人群进行分类,识别出出生缺陷高危人群,为采取有重点、有针对性的干预措施提供依据。方法 选取近亲结婚、亲属缺陷儿、自然流产史、既往缺陷儿、孕期发热、孕期感冒、居住地存在污染源、孕期用药、育龄9个出生缺陷相关因素作为显变量,进行潜在类别分析,识别出潜在类别,进而对人群进行分类。结果 当潜在类别数目为4时模型拟合最佳,依据各因素在4个潜在类别中的条件概率的分布特征,将类别1~类别4分别命名为：一般人群组、家族缺陷史组、孕期感冒发热组、单纯用药组。对个体进行聚类后,4类人群的出生缺陷率比较,χ~2值为3099.254,P〈0.001。不同潜在类别人群缺陷率有差别。家族缺陷史组（73.7%）,感冒发热组（3.8%）,单纯用药组（2.5%）,一般人群组（1.0%）缺陷率依次降低,其中家族缺陷史组缺陷率最高,提示家族缺陷史组人群为出生缺陷高危人群,应该有重点有针对地进行干预。结论 潜在类别分析可以应用于出生缺陷高危人群识别中。依据各因素在潜在类别中的条件概率的分布特征进行人群的分型,通过比较不同类别人群的缺陷率的差异,识别出生缺陷高危人群,揭示出生缺陷预防的重点人群和内容,为制定出生缺陷预防措施提供依据。

Objective This topic expect to identify the high-risk population in birth defects, by classifing the people with different distribution characteristics of risk factors of birth defects, applicating the latent class analysis （ LCA）, and to provide decision support for birth defects prevention and treatment. Methods Latent class analysis was used to identify the latent class and classify the population with different distribution characteristics of risk factors for birth defect on the basis of nine key indicator variables ： intermarriage, abnormal child-bearing of relatives, history of natural abortion, abnormai child-bearing history, catching a cold during pregnancy, fever during pregnancy, environment pollution resources near residence, drug use during pregnancy, mother＇s child-bearing age. Then chi-square test was used to compare the rate of birth defects of different types of mothers and identify the high-risk groups of birth defects. Results Population of risk factors for birth defects could be divided into four types which were named：family heredity history group（ the mothers were mainly characterized by abnormal child-bearing history and relatives having a baby with birth defect） ,cold and fevers group（ the mothers were mainly characterized by catching a cold and having a fever during pregnancy）, medication group （the mothers were mainly characterized by taking medicine during pregnancy）, and general group （the mothers did not have distinct distribution characteristics of risk factors ）. The rate of birth defects of different types of mothers were as follows ： family heredity history group（ 73.7% ） ;cold and fevers group （3.8 % ） ;medication group （2.5 % ） ;general group （1.0%）, and the difference was statistically significant （X2 = 3099. 254, P 〈 0. 001 ）. We could find that the family heredity history group had a highest rate of birth defects, so the family heredity history group was the high-risk population in birth defects. Conclusion Latent class model can be used to classify the population basis on the different distribution characteristics of risk factors and identify the high-risk population in birth defects, which indicate the main contents and the most important tasks in birth defects prevention and provide decision support for birth defects prevention and treatment.