Adjusting for confounding by neighborhood using complex survey data

Abstract

Recently, we examined methods of adjusting for confounding by neighborhood of an individual exposure effect on a binary outcome, using complex survey data; the methods were found to fail when the neighborhood sample sizes are small and the selection bias is strongly informative. More recently, other authors have adapted an older method from the genetics literature for application to complex survey data; their adaptation achieves a consistent estimator under a broad range of circumstances. The method is based on weighted pseudolikelihoods, in which the contribution from each neighborhood involves all pairs of cases and controls in the neighborhood. The pairs are treated as if they were independent, a pairwise pseudo-conditional likelihood is thus derived, and then the corresponding score equation is weighted with inverse-probabilities of sampling each case-control pair. We have greatly simplified the implementation by translating the pairwise pseudo-conditional likelihood into an equivalent ordinary weighted log-likelihood formulation. We show how to program the method using standard software for ordinary logistic regression with complex survey data (e.g. SAS PROC SURVEYLOGISTIC). We also show that the methodology applies to a broader set of sampling scenarios than the ones considered by the previous authors. We demonstrate the validity of our simplified implementation by applying it to a simulation for which previous methods failed; the new method performs beautifully. We also apply the new method to an analysis of 2009 National Health Interview Survey (NHIS) public-use data, to estimate the effect of education on health insurance coverage, adjusting for confounding by neighborhood.

Area-Level Socioeconomic Position and Repeat Mammography Screening Use: Results from the 2005 National Health Interview Survey

“…However, as shown in previous work (19,21), extensions of generalized linear mixed models for complex survey data are prone to bias when the neighborhood sample sizes are small and the sampling is informative. We are currently developing more suitable extensions, based on ideas analogous to those in (24). Nevertheless, for inference concerning effects of measured neighborhood characteristics, such as area-level SEP, the use of ordinary logistic regression with survey data is completely adequate.…”

Accounting for Interactions and Complex Inter-Subject Dependency in Estimating Treatment Effect in Cluster-Randomized Trials with Missing Outcomes

“…Similarly, the characteristics of subgroups to which the index case belongs (household, neighborhoods, etc. ), whether known or not, may be interfering covariates (Brumback and He, ). In this article, we consider the phenomenon of covariate interference where there exists at least one individual such that .…”

Adjusting for confounding by neighborhood using generalized linear mixed models and complex survey data

“…Neuhaus and Kalbfleisch [20] promoted the use of GLMM regression with model (4), and Neuhaus and McCulloch [21] dubbed the 'poor man's' alternative to conditional likelihood methods. When the inverse-link function h in model (1) is the identity function and var.Y ij jX i ; b i / is constant in i and j , it so happens [22] that consistent estimation ofˇis achieved by setting .X i ; / equal to model (4), regardless of whether model (4) is the correct model for .X i ; /. However, misspecifying .X i ; / can lead to an inconsistent estimator ofˇwhen h is the exponential function [22], or the expit function [23], and therefore also for models (2) and (3).…”

“…Second, we have needed to generalize these approaches for use with complex survey data. So far, we have generalized conditional likelihood methods for use with binary, ordinal, or multinomial outcomes and complex survey data [1,2,4], via a weighted composite conditional likelihood [5], based on ideas presented in [6][7][8][9]. We refer to these methods as conditional pseudolikelihood methods.…”