Chunk #5 — Bias in a protocol for a meta-analysis of 5-HTTLPR, stress, and depression — Issue 1: “Primary Analysis Plan 2” to study lifetime depression does not allow for establishing temporal order between stress and depression
The BMC Psychiatry methods paper includes two plans. It includes Primary Analysis 1, a new separate plan to study only those data sets that can establish temporal order between stress and depression. We applaud the addition of Plan 1. The paper also retains Primary Analysis 2, the original plan to study lifetime depression including all studies available, which we argued above is inappropriate. We must query why Plan 2 is still retained. The abovementioned flaws in Plan 2 should come as no surprise, because our point about the importance of accurate measurement of temporal order is not new. We addressed it empirically in our original Science paper reporting the GxE in question [2]. In that paper we estimated the GxE effect using a measure of life events that occurred prior to depression and we estimated it again using a measure of life events that occurred after depression. Results showed empirically that unless the stress occurred before the depression, the GxE finding was not observed. Culverhouse et al. carefully and rightly emphasize the importance of matching the design features of a
