Choosing an antidepressant can involve more than matching a diagnosis to a medicine. People vary in how they respond to treatment and in which side effects they most want to avoid. Those differences can affect whether they continue taking the medicine.
PETRUSHKA is a digital tool designed to help clinicians make that choice more individualised. It combines prediction models based on personal characteristics with a patient’s preferences about adverse effects, producing an overall recommendation score for each antidepressant.
What the trial tested
The tool was compared with usual care in a randomised controlled trial conducted across 47 sites in Brazil, the United Kingdom and Canada.
Of 540 people initially randomised, 20 were later found to be ineligible. The primary analysis included 493 eligible participants diagnosed with major depressive disorder: 241 were assigned to PETRUSHKA and 252 to usual care. Participants had a median age of 35; 58% were female and 84% were white.
The prediction models included characteristics such as age, sex, childhood maltreatment and other medical conditions. The trial considered antidepressant treatment alone and did not include other effective interventions, such as psychotherapy.
Fewer early treatment stops
After eight weeks, 41 people in the PETRUSHKA group had stopped taking their antidepressant, compared with 69 people receiving usual care. This represented an almost 40% lower risk of discontinuation with PETRUSHKA, with a risk ratio of 0.62.
Stopping because of an adverse event was also less common with PETRUSHKA at eight weeks. The risk ratio was 0.59. However, neither difference was clearly present at 24 weeks.
The medicines prescribed also differed between the groups. In the PETRUSHKA group, the most common choices were mirtazapine, escitalopram and vortioxetine. In usual care, sertraline, citalopram and fluoxetine were most common.
Symptoms and quality of life
There were no clear differences between the groups in depression or anxiety symptoms at eight weeks.
At 24 weeks, people in the PETRUSHKA group reported lower depression and anxiety symptoms and higher health-related quality of life than those receiving usual care. The symptom differences were relatively small compared with reported minimal clinically important changes, and the 24-week findings were based on fewer participants than were initially randomised.
Observer-rated depression and anxiety followed a similar pattern, but those estimates were more uncertain and were not statistically significant.
What the findings do—and do not—show
The results suggest that bringing patient preferences into antidepressant selection, alongside prediction based on personal characteristics, may help some people remain on treatment during the first eight weeks.
It is not yet clear whether the benefit came from more accurate predictions about treatment fit, greater patient involvement and motivation, or both. Missing data at 24 weeks also make the longer-term results harder to interpret. Selective dropout may have influenced the findings for symptoms and quality of life.
The apparent effect may have been larger in primary care than in secondary care, but relatively few people were recruited from secondary care, so that comparison needs caution. Patients and clinicians also knew which treatment approach had been assigned, which may have influenced discussions or symptom ratings.
For now, PETRUSHKA is best understood as a decision-support approach intended to support conversations between clinicians and patients. Further research is needed to confirm the findings and assess whether the tool works differently across care settings.
AI tools were used to assist with the preparation of this article.
Leave a Reply