文档下载 文档下载导航
设为首页 | 加入收藏
搜索 请输入内容:  
 导航当前位置: 文档下载 > 所有分类 > 医药卫生 > 预防医学 > Accounting for Selection Bias in Association Studies
侵权投诉

Accounting for Selection Bias in Association Studies

Original article

Accounting for Selection Bias in Association Studies with

Complex Survey Data

Kathleen E. Wirtha and Eric J. Tchetgen Tchetgena,b

Abstract: Obtaining representative information from hidden and btaining representative information from hidden and O

hard-to-reach populations is fundamental to describe the epidemiol-ogy of many sexually transmitted diseases, including HiV. Unfortu-nately, simple random sampling is impractical in these settings, as no registry of names exists from which to sample the population at ran-dom. However, complex sampling designs can be used, as members of these populations tend to congregate at known locations, which can be enumerated and sampled at random. For example, female sex workers may be found at brothels and street corners, whereas injec-tion drug users often come together at shooting galleries. Despite the logistical appeal, complex sampling schemes lead to unequal probabilities of selection, and failure to account for this differential selection can result in biased estimates of population averages and relative risks. However, standard techniques to account for selection can lead to substantial losses in efficiency. consequently, researchers implement a variety of strategies in an effort to balance validity and efficiency. Some researchers fully or partially account for the survey design, whereas others do nothing and treat the sample as a realization of the population of interest. We use directed acyclic graphs to show how certain survey sampling designs, combined with subject-matter considerations unique to individual exposure-outcome associations, can induce selection bias. Finally, we present a novel yet simple maximum likelihood approach for analyzing complex survey data; this approach optimizes statistical efficiency at no cost to validity. We use simulated data to illustrate this method and compare it with other analytic techniques.(Epidemiology 2014;25: 444–453)

Submitted 16 December 2012; accepted 24 September 2013; posted 4 March 2014.

From the aDepartment of epidemiology, Harvard School of Public Health,

Boston, Ma; and bDepartment of Biostatistics, Harvard School of Public Health, Boston, Ma.

the authors report no conflicts of interest.

this project was supported by the national institutes of general Medi-cal Sciences (U54 gM088558), environmental Health Sciences (r21 eS019712), allergy and infectious Diseases (r37 ai51164), and Heart, lung, and Blood (r01 Hl080644). Supplemental digital content is available through direct Url cita-tions in the HtMl and PDF versions of this article (doc.guandang.net). this content is not peer-reviewed or copy-edited; it is the sole responsibility of the author.

correspondence: Kathleen e. Wirth, Department of epidemiology, Harvard School of Public Health, 677 Huntington avenue, Suite 501, Boston, Ma 02115. e-mail: kwirth@hsph.harvard.edu.copyright © 2014 by lippincott Williams & WilkinsiSSn: 1044-3983/14/2503-0444

DOi: 10.1097/eDe.0000000000000037

444 | doc.guandang.net

hard-to-reach populations is fundamental to epidemio-logic description of sexually transmitted diseases. Members of these populations often contribute disproportionately to transmission. For example, female sex workers often act as reservoirs of infection from which HiV spreads to the general population through their male clients.1–4 Unfortunately, sim-ple random sampling of such persons is difficult and imprac-tical. there are no lists of names from which to sample the population at random. even if such a registry existed, these groups are small proportions of the population, making it unlikely that random sampling would capture enough people for subgroup analyses. Finally, these populations often con-ceal themselves because their activities are illegal or highly stigmatized. Persons who can be reliably identified by tradi-tional sampling approaches may differ in important ways from those who cannot.

complex sampling has been used successfully to obtain representative samples of hidden and hard-to-reach popula-tions.5–8 this type of sampling differs from simple random sampling, in that the primary sampling unit is a group of persons rather than a single person. groups are first enumer-ated and then the specific groups to be sampled are selected at random. then, within the selected groups, individuals are sampled into the survey. complex sampling can be effective at capturing hard-to-reach populations because members tend to congregate at known locations. For example, female sex work-ers may be found at brothels and street corners, whereas drug users often come together at shooting galleries. the gathering locations serve as “clusters” from which individuals can be selected for participation.

although complex sampling has logistical appeal, it presents analytic challenges. the sampling framework leads to an unequal probability of selection across individuals and thus to potentially biased estimates of population averages. although it is a standard practice to weight each person by the inverse of the probability of selection to estimate population averages, it is unclear what adjustment, if any, should be made in regression analyses.9,10 Some complex surveys recommend incorporating the sampling weights in every analysis,11–13 whereas others explicitly instruct users to disregard sampling weights when estimating regression coefficients.14 in practice, researchers accessing the same data may implement a variety of adjustment strategies; some fully or partially account for

Epidemiology Volume 25, Number 3, May 2014

第1页

你可能喜欢

热门文档

相关文档

站点地图 | 文档上传 | 侵权投诉 | 手机版
本站所有资源均来自互联网,本站只负责收集和整理,均不承担任何法律责任,如有侵权等其它行为请联系我们.
文档下载 Copyright 2013 doc.guandang.net All Rights Reserved.  -  email