The Impact of Unmeasured Within- and Between-Cluster Confounding on the Bias of Effect Estimators from Fixed Effect, Mixed effect and Instrumental Variable Models

TL;DR

This paper analyzes how unmeasured confounding within and between clusters affects the bias of effect estimators across different statistical methods, providing guidance on method selection based on confounding structure.

Contribution

It derives bias formulas for IV, FE, and LMM estimators considering unmeasured confounding, clarifying when each method yields consistent estimates.

Findings

IV methods are consistent with within-cluster confounding but not with between-cluster confounding.

Fixed effects and linear mixed models are consistent with between-cluster confounding but not with within-cluster confounding.

The impact of between-cluster confounding on IV estimates is larger than within-cluster confounding on FE and LMM estimates.

Abstract

Instrumental variable methods are popular choices in combating unmeasured confounding to obtain less biased effect estimates. However, we demonstrate that alternative methods may give less biased estimates depending on the nature of unmeasured confounding. Treatment preferences of clusters (e.g., physician practices) are the most f6requently used instruments in instrumental variable analyses (IVA). These preference-based IVAs are usually conducted on data clustered by region, hospital/facility, or physician, where unmeasured confounding often occurs within or between clusters. We aim to quantify the impact of unmeasured confounding on the bias of effect estimators in IVA, as well as alternative methods including ordinary least squares regression, linear mixed models (LMM) and fixed effect models (FE) to study the effect of a continuous exposure (e.g., treatment dose). We derive bias formulae of estimators from these four methods in the presence of unmeasured within- and/or between-cluster confounders. We show that IVAs can provide consistent estimates when unmeasured within-cluster confounding exists, but not when between-cluster confounding exists. On the other hand, FEs and LMMs can provide consistent estimates when unmeasured between-cluster confounding exits, but not for within-cluster confounding. Whether IVAs are advantageous in reducing bias over FEs and LMMs depends on the extent of unmeasured within-cluster confounding relative to between-cluster confounding. Furthermore, the impact of unmeasured between-cluster confounding on IVA estimates is larger than the impact of unmeasured within-cluster confounding on FE and LMM estimates. We illustrate these methods through data applications. Our findings provide guidance for choosing appropriate methods to combat the dominant types of unmeasured confounders and help interpret statistical results in the context of unmeasured confounding