The effect of joint exposures: examining the presence of interaction
The effect of joint exposures: examining the presence of interaction
复制标题
DOI：
10.1038/ki.2008.645
复制
发表时间：
2009-04-01
影响因子：
19.6
通讯作者：
Dekker, Friedo W.
中科院分区：
文献类型：
--
作者：
de Mutsert, Renee;Jager, Kitty J.;Dekker, Friedo W.
临床流行病学研究调查暴露或危险因素是否与疾病或死亡的发展或进展有因果关系。研究这种关系在不同类型的患者中是否有所不同可能是有意义的。为了解决这样的研究问题，可以检查风险因素之间是否存在相互作用。在流行病学中，两个危险因素之间的因果相互作用被认为是最具临床相关性的。当两个风险因素共同作用导致疾病时，因果交互作用就会发生，并被明确定义为偏离风险差异表上的可加性。根据所使用的模型，统计相互作用可以在加性(绝对风险)和乘性(相对风险)两个尺度上进行评估。在使用Logistic回归模型(即乘性模型)时，本文提出了几个相加交互作用的量度来评估关联的大小是否因子组而异：交互作用所致的相对超额风险(RERI)、交互作用所致的归因比例(AP)或协同作用指数(S)。为了透明地呈现相互作用的影响，最近的加强流行病学观察研究报告(STROBE)声明建议报告每个暴露的单独影响以及与未暴露组相比的联合影响作为联合参考类别，以便能够评估相加和相乘的相互作用。
Clinical epidemiological studies investigate whether an exposure, or risk factor, is causally related to the development or progression of a disease or mortality. It might be of interest to study whether this relation is different in different types of patients. To address such research questions, the presence of interaction among risk factors can be examined. Causal interaction between two risk factors is considered most clinically relevant in epidemiology. Causal interaction occurs when two risk factors act together in causing disease and is explicitly defined as a deviation from additivity on a risk difference scale. Statistical interaction can be evaluated on both an additive (absolute risk) and multiplicative (relative risk) scale, depending on the model that is used. When using logistic regression models, which are multiplicative models, several measures of additive interaction are presented to evaluate whether the magnitude of an association differs across subgroups: the relative excess risk due to interaction (RERI), the attributable proportion due to interaction (AP), or the synergy index (S). For a transparent presentation of interaction effects the recent Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement advises reporting the separate effect of each exposure as well as the joint effect compared with the unexposed group as a joint reference category to permit evaluation of both additive and multiplicative interaction.