In Reply: Dr Elkin is correct that her trial, 1 of 6 we included, yielded an interaction between initial severity and the drug-placebo difference, and this should have been stated in our article. We conducted numerous tests of the robustness of our findings, including omitting each of the studies one by one. If the data from any single study are omitted, the test of the interaction between baseline depression severity and drug vs placebo on outcome remains significant.
DeRubeis RJ, Fournier JC, Fawcett J. Depression Severity and Effect of Antidepressant Medications—Reply. JAMA. 2010;303(16):1596-1599. doi:10.1001/jama.2010.510