职业性肺癌危险因素的多水平生存分析

The Investigation on the Selection of Time-Scale in Cox Proportional Hazard Model

目的 探讨多水平生存分析中的Cox比例风险模型中在职业性肺癌危险因素量化评价中的应用。方法 以云南锡业总公司劳动防护研究所提供的高危人群队列的随访资料为例 ,拟合 3个水平的截断数据的多重成员Cox模型 ,并分析各水平的随机效应及其与肺癌发生间的关系。结果 模型结果显示 ,云锡所属单位中“马拉格锡矿”与其他单位有不同的致癌随机效应 ,个体因素也与肺癌的发生有关。结论 多水平生存分析对于处理具有系统层次结构的数据 ,有着较为宽广的应用前景。多水平Cox模型还可结合多重成员模型 ,处理复杂层次结构的资料。

Objective To investigate on the application of multilevel Cox model to the qualificatively determinating of occupational lung cancer risk factors.Methods A 3-level multiple membership Cox model was fitted for the truncated cohort data among high risk population in Yunnan Tin Cooperation.Results “Malage Tin Mine” was found to be heterogeneous compared to others, meanwhile individual factor also presented magnitude random effect associated to lung cancer.Conclusion Multilevel Cox model is relatively broadly applicable to the analysis of data with hierarchical structure. Moreover, it could be flexible to combine with models for more complicated data structure, e.g. the multiple membership model.