Adults with major depression were more likely to stick with their antidepressants and reported greater symptom improvements when their medication was chosen with the help of a web-based decision-support tool, according to a study
issued today in
JAMA.
Why It’s Relevant
A
growing share of Americans take antidepressants (16%), but the majority of people with major depressive disorder (MDD) don’t stick with them long enough to benefit. Enter PETRUSHKA, a machine learning program that selects three preferred antidepressant choices based on patient demographics, symptom severity, past antidepressant use, childhood maltreatment, medical conditions, and patient preferences.
By the Numbers
- Among 540 adults with MDD across three countries (Brazil, Canada, and the United Kingdom), those who received their antidepressant with the aid of PETRUSHKA were 38% less likely to discontinue their medication after eight weeks compared with those who received usual care (17% discontinuation versus 27%, respectively).
- There were no significant differences in self-reported depression or anxiety (assessed with the PHQ-9 and GAD-7, respectively) between the groups at eight weeks. However, at 24 weeks, the PETRUSHKA group reported lower scores on both measures: 7.1 on the PHQ-9 and 4.6 on the GAD-7, versus 9.2 and 5.8, respectively, in the usual care group.
- The three most prescribed antidepressants in the PETRUSHKA group were mirtazapine, escitalopram, and vortioxetine (29%, 28%, and 24%, respectively), whereas the usual care group strongly favored sertraline, followed by citalopram and fluoxetine (52%, 15%, and 9%, respectively).
What’s More
The PETRUSHKA tool was more effective in Brazil and the UK—where it was tested in primary care—compared with Canada, where it was tested in a psychiatric setting. “A psychiatrist’s experience with a wider range of antidepressants and knowledge about specific mechanisms of action may allow them to better tailor antidepressant treatment,” the researchers wrote.
The Other Side
Patients’ self-assessments of depression and anxiety might be biased because they knew whether their antidepressant was selected by the tool. Psychiatric assessors blinded to treatment allocation didn’t uncover significant symptom differences between the groups at eight or 24 weeks. There was also a large amount of missing data on secondary outcomes.
Takeaway Message
“Our results suggest that the PETRUSHKA tool could be ideally placed in clinical settings where MDD is managed by [non-psychiatrist] clinicians at community hospitals and specialist outpatient clinics, in low-resource or rural areas, and in low- and middle-income countries,” the researchers wrote.
Related Information
Source
Andrea Cipriani, et al. A decision-support system to personalize antidepressant treatment in major depressive disorder: a randomized clinical trial. Journal of American Medical Association. Published March 4, 2026. doi:10.1001/jama.2026.1327
(Image: Getty Images/iStock/filo)