Treatment expectancies and psilocybin vs escitalopram for depression
This interesting research letter reads: "The potential for treatment expectancies to affect clinical trial results is a major concern in modern psychiatric research, especially for studies involving interventions that are difficult to blind, such as psilocybin, 3,4-methylenedioxymethamphetamine (MDMA), and ketamine. Methods In the trial of psilocybin vs escitalopram for major depressive disorder conducted by Carhart-Harris et al, participants reported their expected symptomatic improvement for each treatment arm before randomization using visual analog scales ranging from 0% improvement to 100% improvement. Therefore, although the original report did not make use of these data, all participants had a baseline escitalopram expectancy measurement and a baseline psilocybin expectancy measurement. We sought to test the hypothesis that baseline escitalopram expectancy or psilocybin expectancy or both would predict relative treatment efficacy. We fit a linear model to the only (week 6) posttreatment Hamilton Depression Rating Scale 17-Item (HDRS-17) score with the following model terms: baseline score, group, escitalopram expectancy, and a group by escitalopram expectancy interaction. An analogous model was fit using psilocybin expectancy instead of escitalopram expectancy. HDRS-17 was selected for analysis because it exhibited the largest between-group standardized effect size of the depression outcomes,2 and therefore, the highest statistical power to detect effect modification. Analysis methods are available in the eMethods in Supplement 1. Results A statistically significant interaction between group and escitalopram expectancy was identified (coefficient [SE], −0.106 [0.033]; P = .002), indicating that the between-group effect size was dependent on baseline expectations of escitalopram. For individuals with low expectations of escitalopram, psilocybin was significantly better than escitalopram at reducing depressive symptoms, but for individuals with higher escitalopram expectancies, the treatments were not significantly different (Figure). The Pearson correlation between escitalopram expectancy and HDRS-17 score was r = −0.49 (P = .008) for escitalopram recipients (n = 28) and r = 0.30 (P = .14) for psilocybin recipients (n = 26). No statistically significant main or interaction effects were observed for psilocybin expectancy. Results were identical when the change score was used as the outcome (Table). In sensitivity analyses, the interaction between escitalopram expectancy and group was significant for the Montgomery-Åsberg Depression Rating Scale (coefficient [SE], −0.153 [0.057]; P = .009) and the Beck Depression Inventory (coefficient [SE], −0.154 [0.070]; P = .03) but not the Quick Inventory of Depressive Symptomatology (coefficient [SE], −0.064 [0.037]; P = .09). Median (IQR) expectancies were 23% (11%-50%) improvement for escitalopram and 60% (40%-71%) improvement for psilocybin. The relationship between escitalopram expectancy and psilocybin expectancy was linear and had a Pearson correlation of r = 0.25 (P = .06). Discussion Limitations of this work include the small sample size, the post hoc nature of the analyses, and the unclear generalizability of the findings to individuals with low psilocybin expectancies. Nevertheless, our findings have several important implications. First, the large between-group effect size that has been reported for psilocybin in depression may be specific to individuals with low or relatively low expectations of other treatment options. Second, given the magnitude of the apparent expectancy effect observed within escitalopram recipients, in clinical trials in depression and perhaps in psychiatry more broadly, it may be appropriate to routinely measure treatment expectancies at baseline. At a minimum, this information could be reported descriptively for use in evaluating the generalizability of findings. Further, where a trial sample does not have similar expectations of all treatment arms, analyses could test for treatment-expectancy interactions and present effect size estimates stratified by expectancy level, although these analyses may be limited by low between-participant variability in expectancies. Additionally, even where no effect modification is present, the inclusion of expectancies in statistical models may increase statistical power and predictive accuracy. Finally, further work is needed to better understand the roles in the management of depression of assessing and modifying treatment expectancies."
For more psychedelic news and research, visit the psychedelic health professional network homepage.