Lundberg M, Hallqvist J, Diderichsen F
Department of Public Health Sciences, Karolinska Institutet, Stockholm, Sweden.
Epidemiology. 1999 Sep;10(5):545-9.
The objectives of this paper are to analyze the consequences of exposure misclassification on effect estimates in interaction analysis, and to develop a mathematical equation for the potentially biased estimate. The main point is to identify situations in which misclassification of the first exposure, dependent on the second exposure but independent on outcome status, leads to overestimation or underestimation of the interaction effect. We show that misclassification theoretically can cause overestimation of the interaction effect. Nevertheless, because the categories that yield overestimation due to misclassification are fewer than the categories that yield underestimation, and misclassification in reality mostly is multidimensional (more than one category are biased simultaneously), it is more likely that the effect of misclassification is underestimation rather than overestimation. Misclassification in the categories that lead to overestimation is compensated by misclassification in the categories that lead to underestimation. The magnitude of the biased estimate depends on the prevalences of the misclassified exposure, stratified for the second exposure and the outcome-the lower the prevalence, the smaller the bias.
本文的目的是分析交互作用分析中暴露错误分类对效应估计的影响，并推导一个关于潜在偏差估计的数学方程。重点是确定在哪些情况下，首次暴露的错误分类（取决于第二次暴露但独立于结局状态）会导致交互作用效应的高估或低估。我们表明，理论上错误分类会导致交互作用效应的高估。然而，由于因错误分类导致高估的类别少于导致低估的类别，且现实中的错误分类大多是多维度的（多个类别同时存在偏差），所以错误分类的影响更有可能是低估而非高估。导致高估的类别中的错误分类会被导致低估的类别中的错误分类所抵消。偏差估计的大小取决于错误分类暴露的患病率，按第二次暴露和结局分层——患病率越低，偏差越小。
