Estimation using all available covariate information versus a fixed look-back window for dichotomous covariates.

Estimation using all available covariate information versus a fixed look-back window for dichotomous covariates.
复制标题

DOI:
10.1002/pds.3434
复制
发表时间:
2013-05
影响因子:
2.6
通讯作者:
Seeger, John D.
Seeger, John D.
中科院分区:
医学4区
文献类型:
--
作者:
Brunelli, Steven M.;Gagne, Joshua J.;Huybrechts, Krista F.;Wang, Shirley V.;Patrick, Amanda R.;Rothman, Kenneth J.;Seeger, John D.

文献摘要

参考文献

被引文献

相似文献

当使用声明数据时,通常假设不存在二分协变量(C),除非观察到该条件的声明。当受试者之间可用的历史数据不同时,研究者必须选择使用所有可用的历史数据还是使用固定窗口的数据来评估C。我们的目的是比较这两种方法下的估计。我们模拟了20000名受试者的队列,其中二分变量代表暴露(E)、结局(D)和单个时不变C,以及不同的历史数据可用性。在每个范例下操作性地定义C,并通过Mantel-Haenszel方法用于估计E对D的校正风险比。在基础情况下,与固定窗口相比,使用所有可用信息观察到的偏倚更小,MSE更低;在较高的建模混杂强度下,差异被放大。在引入未测量的协变量(F)后,所有可用方法在大多数情况下仍然具有较小的偏倚,并且在所有情况下都能更好地近似估计,并根据混杂因素的实际(建模)值进行调整。在考虑的大多数情况下,基于所有可用的历史数据,而不是基于在共同共享的固定历史窗口上观察到的数据,在操作上定义时不变的二分C,导致偏差较小的估计。
When using claims data, dichotomous covariates (C) are often assumed to be absent unless a claim for the condition is observed. When available historical data differs among subjects, investigators must chose between using all available historical data versus data from a fixed window to assess C. Our purpose was to compare estimation under these two approaches. We simulated cohorts of 20000 subjects with dichotomous variables representing exposure (E), outcome (D) and a single time-invariant C, as well as varying availability of historical data. C was operationally defined under each paradigm and used to estimate the adjusted risk ratio of E on D via Mantel-Haenszel methods. In the base case scenario, less bias and lower MSE were observed using all available information compared with a fixed window; differences were magnified at higher modeled confounder strength. Upon introduction of an unmeasured covariate (F), the all-available approach remained less biased in most circumstances and better approximated estimation that was adjusted for the actual (modeled) value of the confounder in all instances. In most instances considered, operationally defining time-invariant dichotomous C based on all available historical data, rather than on data observed over a commonly shared fixed historical window, results in less biased estimates.
DOI: 10.1002/pds.2061
发表时间: 2011-03-01
影响因子: 2.6
作者:
Tagalakis, Vicky;Kahn, Susan R.
通讯作者: Kahn, Susan R.
DOI: 10.1093/ije/21.3.537
发表时间: 1992-06-01
影响因子: 7.7
作者:
CHAVANCE, M;DELLATOLAS, G;LELLOUCH, J
通讯作者: LELLOUCH, J
DOI: 10.1111/j.0006-341x.2002.00878.x
发表时间: 2002-12-01
期刊: BIOMETRICS
影响因子: 1.9
作者:
Gustafson, P;Le, ND
通讯作者: Le, ND
DOI: 10.1002/sim.4780100803
发表时间: 1991-08-01
影响因子: 2
作者:
WALKER, AM;LANES, SF
通讯作者: LANES, SF
DOI: 10.1111/j.1525-1497.2005.0242.x
发表时间: 2005-12-01
影响因子: 5.7
作者:
Schauer, DP;Moomaw, CJ;Eckman, MH
通讯作者: Eckman, MH