Review of: Dowd SM et al, World J Psychiatry 2020;10(3):21–28

Study TYPE: Randomized, double-blind, placebo-controlled trial

The idea that good sleep promotes mental health has led researchers to test whether hypnotics have psychiatric benefits beyond their effects on sleep. Eszopiclone (Lunesta) fulfilled that promise in depression and generalized anxiety disorder, where it brought about more rapid and pronounced responses when added to SSRIs in two large, randomized, placebo-controlled trials. Although those trials enrolled patients with comorbid insomnia, their primary goal was to test whether eszo­piclone could improve core symptoms of depression and anxiety, which it did (Pollack M et al, Arch Gen Psychiatry 2008;65(5):551–562; Fava M et al, Biol Psychiatry 2006;59(11):1052–1060).

The current study was a small, randomized, double-blind, placebo-controlled trial that tested whether eszopiclone improved core symptoms of PTSD. After 12 weeks, the medication worked no better than placebo on measures of sleep or PTSD. In contrast, an earlier randomized crossover trial found significant effects for eszopiclone in both sleep and non-sleep symptoms of PTSD (Pollack MH et al, J Clin Psychiatry 2011;72(7):892–897), so which results should we believe?

The new study was hindered by a high dropout rate (36%). The Cochrane group considers dropout rates problematic when they rise above 20% in short-term trials and 30% in long-term ones. But the bigger lesson here is that the results of small trials—ie, fewer than 50 subjects—should never be taken at face value.

Randomization attempts to cancel out the effects of confounding variables, which is why most papers list a study’s baseline demographics before describing the results. If the randomization is successful, there should be a balanced distribution of women and men, older and younger subjects, mild and severe cases, etc, in both groups. However, no list can account for all of the confounding variables that might skew the results, and small samples make it more likely that the results were skewed by events that were unaccounted for, such as marital discord or changes in physical activity.

TCPR’s Take
While this study does little to clarify eszopiclone’s role in PTSD, it reminds us to give extra scrutiny to results from small (n < 50) trials, even when they are randomized and controlled.