Despite being theorized as a genetic, heritable trait, studies of depression with large sample sizes and statistical power in this review and meta-analysis from a large list of collaborators around the world did not reach genome-wide satistical significance for any DNA loci. An overall meta-analysis combining discovery and replication studies, however, did reach significance for an association between depression and one single nucleotide polymorphism. The authors conclude that “only a large sample comprising more than 50,000 subjects may be sufficiently powered to detect genes for depressive symptoms.” Results were published online January 3, 2013 by Biological Psychiatry.
© 2024 Mad in America Foundation
I’m not by any stretch expert in genetics or statistics. But as I understand it from Jay Joseph’s writing, “heritability” isn’t what it sounds like. It does NOT mean that something is genetically caused – the fact that it runs from one generation to the next could be due to environmental factors. So it’s interesting that the abstract of this article makes the claim about heritability and appears to proceed on the assumption that depression IS genetic.
Moreover, David Healy’s writing about clinical trials indicates that when gigantic sample sizes are required to produce “statistically significant” findings, it is a sign that the actual clinical difference is miniscule. “Real” and “significant” effects don’t require 50,000 subjects to “tease out” associations.
This study’s conclusion seems odd to me. They could as easily have noted the billions and decades spent on this research, and wondered if they’re barking up the wrong tree – six of seven candidates didn’t prove out, and the seventh showed an extremely weak association.
I’d be interested in comments from people who know a lot more about genes and statistics.
Report comment