广义线性混合模型在煤工尘肺发病影响因素研究中的应用

摘要

Objective Explore the risk factors of coal worker’s pneumoconiosis (CWP) by using generalized linear mixed models. Methods The study was performed by using longitudinal retrospective data in Tiefa colliery in northern China. The generalized linear model was implemented in the NLMIXED procedure in SAS and compared to chi square and logistic regression. Result Chi square test showed the duration of exposure to dust, period of first exposure to dust, and job type were the risk factors of CWP. Logistic regression showed the duration of exposure to dust, period of first exposure to dust, and job type were the risk factors of CWP. The generalized linear mixed models: duration of exposure to dust, period of first exposure to dust, and job type were the risk factors of CWP, whereas the working place is not a random effect in statistics. Conclusion The generalized linear mixed model considers the random effects and the errors, therefore it is a superior approach in assessing the risk factors of CWP.%目的研究广义线性混合模型在煤工尘肺发病影响因素的应用，为煤工尘肺及类似疾病的发病影响因素研究提供新的方法。方法收集煤矿集团所属8个煤矿所有接尘矿工的资料，用SAS进行卡方检验、Logistic 回归进行尘肺发病影响因素研究，并与SAS nlmixed模块实现的广义混合现行模型结果进行比较。结果单因素分析、Logistic回归和广义线性混合模型显示：工龄、接尘时间和工种均为尘肺发病的影响因素，工作地点作为随机效应结果差异无统计学意义。结论广义线性混合模型因考虑了不同工人工作地点的不同和随机效应，结果虽然和其他方法一致，但是更有说服力，与其他方法相比是研究尘肺发病影响因素较好的方法。