Description
Gene expression patterns in the brain are strongly influenced by the severity of physiological stress at death. This agonal effect, if not well controlled, can lead to spurious findings in case-control comparisons. While many recent studies match samples by tissue pH and clinically recorded agonal conditions, we found that these commonly used indicators were sometimes at odds with observed stress-related patterns of gene expression, and that matching by these criteria still sometimes results in identifying differences between cases and controls that are primarily driven by residual agonal effects. This problem is analogous to the one in genetic studies, where race and ethnicity are often imprecise proxies for complex environmental and genetic factors.