Searching for Optimal Treatment Schedules for Longitudinal ECT
To the Editor: Kellner et al. (1), in the November 2016 issue of the Journal, randomly assigned elderly patients newly remitted after ECT for depression to 6 months of either aggressive pharmacotherapy alone or to a pharmacotherapy regimen combined with continuation ECT, which consisted initially of four weekly treatments thereafter followed by as-needed treatments for clinical worsening (a method termed “Symptom-Titrated, Algorithm-Based Longitudinal ECT,” or STABLE). STABLE contrasts with traditional continuation ECT whereby treatments are administered on a scheduled basis at increasingly spaced intervals over an entire 6-month interval. The primary outcome measure was serial Hamilton Depression Rating Scale (HAM-D) scores. The investigators concluded that patients treated with STABLE had HAM-D scores at the end of the 6-month interval that were approximately 4 points lower than patients treated with pharmacotherapy alone and that they also had lower relapse rates. However, inspection of the pattern of HAM-D ratings casts doubt on this conclusion.
As entry into phase 2 was limited to acute ECT remitters, both groups had phase 2 baseline HAM-D scores that were quite low (approximately 6 for each group). Thus, the purpose of phase 2 interventions would be to prevent return of depression as opposed to providing active treatment of current symptoms. If STABLE adds relapse prevention strength to pharmacotherapy alone, one would expect a difference between groups in HAM-D ratings, especially in the first month, during which all patients in the STABLE group were receiving ECT treatments. However, the ratings presented in the article’s Figure 1 and the survival curve data presented in Figure 2 both reveal no evidence of a protective effect of continuation ECT during that time period. At week 4, after which no ECT treatments were routinely administered, the groups had similar HAM-D scores (about a 1-point difference, which was not clinically significant). However, during weeks 4–8 and 22–24, the scores in the STABLE group inexplicably trend downward, and this accounts for the separation in scores from the pharmacotherapy-only group. Most STABLE patients during these time periods were not even receiving any ECT treatments. One might hypothesize that the initial phase 2 series of treatments might protect against clinical worsening over time, but there is no a priori reason to believe that four weekly treatments will cause a lowering of depression scores in the weeks thereafter. The difference between the end of phase 2 HAM-D scores comes from improvement at two unpredicted, fortuitous time points in the STABLE group. This points to the likelihood of a type I error (a false positive) as the explanation of the findings. Indeed, if the trial had been stopped earlier than week 24, the difference in HAM-D scores at all other time points would have been no greater than about 3 points. At a minimum, the authors should acknowledge that STABLE should not be considered the standard of care for continuation ECT.
1 : A novel strategy for continuation ECT in geriatric depression: phase 2 of the PRIDE study. Am J Psychiatry 2016; 173:1110–1118Link, Google Scholar