南方某县居民尿镉浓度和相关影响因素研究

本文采用整群随机抽样的方法,抽取污染区、对照区10个村的1274个居民开展内暴露及问卷调查,通过采集这些居民的晨尿和10个村的土壤、大气颗粒物、饮用水、大米样品,对采集的所有样品进行了镉含量的测定.同时,将这些检测调查资料通过MLwiN 2.02软件进行多水平模型拟合,以评估南方某县污染区和对照区居民尿镉含量在不同水平之间的差异性,并找出影响这种差异的社会人口和环境方面的因素.结果表明,居民的尿镉含量在村和个体不同水平上均存在显著差异,其中,QL-3、FJ-4污染区人体尿镉含量明显高于其它研究区,远远超过其临界值.多水平模型的结果显示,土壤镉含量、大米镉含量、居民性别、居民的外出工作时长和居住地离污染企业的距离对居民尿镉水平有影响,而居民的年龄、涉镉工作时长、吸烟时长和在当地生活时长这4个因素对居民尿镉水平的影响差异没有统计学意义.与一般线性模型相比,多水平模型拟合的效果更佳.研究表明,多水平模型能够较合理地揭示居民尿镉水平的地区差异及社会人口和环境因素的影响.

Urinary cadmium concentration and their related factors in a southern county, China

This study evaluated the differences in urinary cadmium concentrations of residents in a polluted area and a control area. A series of socio-demographic and environmental factors were identified to explain the differences. General and occupational characteristics were gathered from 1274 participants using a structured questionnaire, and their urina sanguinis samples along with samples of soil, atmospheric particulates, drinking water and rice were collected from both groups. A multilevel statistical model was used to fit the survey data. The results demonstrated that the urinary cadmium concentrations of residents were significantly different at different levels. In the QL-3 and FJ-4 polluted areas, the urinary cadmium concentrations were significantly higher than those from other study areas and greatly exceeded the critical value. The results of multi-level model showed that the soil cadmium concentration, rice cadmium concentration, resident gender, time working outside and distance from pollution enterprises had impacts on the urinary cadmium concentration of residents, while age, smoking duration, cadmium-exposing working time, and local life span had no significant difference. Compared with the general linear model, the multi-level model had better fitting effects. The individual and regional differences of the urinary cadmium concentrations of residents and the impact of socio-demographic and environmental factors were effectively revealed through the application of multi-level models.