Joseph Anny-Claude, Fuentes Montserrat, Wheeler David C
Department of Mathematical Sciences, United States Military Academy, West Point, NY.
Office of the Executive Vice President and Provost, University of Iowa, Iowa City, IA.
Stat Med. 2020 May 20;39(11):1610-1622. doi: 10.1002/sim.8501. Epub 2020 Feb 14.
In many studies of environmental risk factors for disease, researchers use the location at diagnosis as a geographic reference for environmental exposures. However, many environmental pollutants change continuously over space and time. The dynamic characteristics of these pollutants coupled with population mobility in the United States suggest that for diseases with long latencies like cancer, historic exposures may be more relevant than exposure at the time of diagnosis. In this article, we evaluated to what extent the commonly used assumption of no population mobility results in increased bias in the estimates of the relationship between environmental exposures and long-latency health outcomes disease in a case-control study. We conducted a simulation study using the residential histories of a random sample from the National Institutes of Health-AARP (formerly American Association of Retired Persons) Diet and Health Study. We simulated case-control status based on subject exposure and true exposure effects that varied temporally. We compared estimates from models using only subject location at diagnosis to estimates where subjects were assumed to be mobile. Ignoring population mobility resulted in underestimates of subject exposure, with largest deviations observed at time points further away from study enrollment. In general, the effect of population mobility on the bias of the estimates of the relationship between the exposure and the outcome was more prominent with exposures that showed substantial spatial and temporal variability. Based on our results, we recommend using residential histories when environmental exposures and disease latencies span a long enough time period that mobility is important.
在许多关于疾病环境风险因素的研究中，研究人员将诊断时的位置作为环境暴露的地理参考。然而，许多环境污染物会随空间和时间不断变化。这些污染物的动态特性以及美国的人口流动性表明，对于像癌症这样潜伏期较长的疾病，历史暴露可能比诊断时的暴露更具相关性。在本文中，我们评估了在病例对照研究中，常用的无人口流动假设在多大程度上会导致环境暴露与长期健康结果疾病之间关系估计的偏差增加。我们使用来自美国国立卫生研究院-美国退休人员协会（原美国退休人员协会）饮食与健康研究的随机样本的居住史进行了一项模拟研究。我们根据受试者暴露情况和随时间变化的真实暴露效应模拟病例对照状态。我们将仅使用诊断时受试者位置的模型估计与假设受试者可流动的估计进行了比较。忽略人口流动会导致对受试者暴露的低估，在离研究入组时间点更远的时间点观察到最大偏差。一般来说，对于显示出显著空间和时间变异性的暴露，人口流动对暴露与结果之间关系估计偏差的影响更为突出。根据我们的结果，我们建议当环境暴露和疾病潜伏期跨越足够长的时间段以至于流动性很重要时，使用居住史。
