The Directions of Selection Bias

TL;DR

This paper demonstrates that under certain conditions, the observed odds ratio in a selected population provides a lower bound for the true odds ratio, highlighting implications for bias in epidemiological studies.

Contribution

It establishes a theoretical relationship between selection bias and the observed versus true odds ratios when exposure and outcome influence selection similarly.

Findings

Observed odds ratio is a lower bound under specified conditions

Selection bias affects the interpretation of epidemiological measures

Conditions involve same direction effects and non-positive interaction

Abstract

We show that if the exposure and the outcome affect the selection indicator in the same direction and have non-positive interaction on the risk difference, risk ratio or odds ratio scale, the exposure-outcome odds ratio in the selected population is a lower bound for true odds ratio.