Li C Y, Thériault G, Lin R S
Department of Epidemiology and Biostatistics, Faculty of Medicine, McGill University Montreal, Quebec, Canada.
J Expo Anal Environ Epidemiol. 1997 Oct-Dec;7(4):493-504.
Between November 1994 and May 1995, indoor power frequency magnetic fields of 407 residences in northern Taiwan were assessed by short-term on site measurements and by modeling computerized power lines information from high-voltage transmission lines. The study residences were selected according to the distance from the transmission lines with cutoff points of 50 meters (m), 100 m, and 150 m, which gave rise to four categories of residences. The analysis showed that the measured magnetic fields were higher than the calculated magnetic fields for most residences, especially for those with lower measured magnetic fields. The measured and calculated magnetic fields showed an agreement of 0.93 (intra-class correlation coefficient, [ICC] = 0.93) for the residences within 50 m of the transmission lines. The ICC declined with the distance from the transmission lines with a lowest ICC of 0.42 for the residences located more than 149 m away from power lines. When both measured and calculated magnetic fields were categorized into three levels with cutoff points of 1 milligauss (mG) and 2 mG, the indices of agreement were very similar for the three residential categories within 149 m of the lines with Kappa (K) between 0.51 and 0.55. The K for the residences more than 149 m from the lines was low at 0.29. The ICC and K observed from a reduced sample of 114 residences presumably representative of all residences in northern Taiwan with respect to the distribution of household magnetic fields was 0.90 and 0.64, respectively. Our data show that using exclusively computerized power lines information to assess residential magnetic fields exposure entails a certain degree of exposure misclassification, but in the absence of direct measurement, this way of doing provides reasonable estimates of magnetic fields exposure within 100 m of transmission lines.
1994年11月至1995年5月期间，通过短期现场测量以及对高压输电线路的计算机化电力线信息进行建模，对台湾北部407户住宅的室内工频磁场进行了评估。根据与输电线路的距离选择研究住宅，截止点分别为50米（m）、100米和150米，由此产生了四类住宅。分析表明，大多数住宅的实测磁场高于计算磁场，尤其是那些实测磁场较低的住宅。对于距离输电线路50米以内的住宅，实测磁场与计算磁场的一致性为0.93（组内相关系数，[ICC]=0.93）。ICC随着与输电线路距离的增加而下降，对于距离电力线超过149米的住宅，ICC最低为0.42。当实测和计算磁场都分为三个级别，截止点分别为1毫高斯（mG）和2 mG时，对于线路149米以内的三类住宅，一致性指数非常相似，卡帕（K）在0.51至0.55之间。对于距离线路超过149米的住宅，K值较低，为0.29。从114户住宅的缩减样本中观察到的ICC和K，就家庭磁场分布而言，大概代表了台湾北部所有住宅，分别为0.90和0.64。我们的数据表明，仅使用计算机化电力线信息来评估住宅磁场暴露会导致一定程度的暴露误分类，但在没有直接测量的情况下，这种方法可以对输电线路100米以内的磁场暴露提供合理估计。
