People seeking treatment for depression often experience a reduction in symptoms, whether they are given an active drug or an inactive placebo. By integrating data from different symptom studies, the researchers found that although the pattern of mood improvement was very similar in both scenarios, active drug treatment caused a stronger recovery, which was uniquely related to patients’ baseline brain connectivity. The results of these studies were published in the journal Psychological Medicine.
Mood improvement is notoriously difficult to measure. Clinicians typically rely on standard questionnaires that condense a wide range of symptoms into a single score. This approach can blur the lines between different aspects of mental health, such as sadness, anxiety, and suicidal thoughts. It also becomes difficult to distinguish between drug and placebo effects.
A placebo effect occurs when a patient’s condition improves simply because the patient expected the treatment to work. Previous studies comparing antidepressants and placebos often show little statistical difference when using traditional broad rating scales. When a patient takes a drug, the expectation of feeling better often triggers actual neurobiological changes. To understand a drug’s true effects, researchers need tools that can distinguish a drug’s unique benefits from the baseline responses generated by the mind.
Lucy Berkovich, a researcher in the Department of Psychiatry at Yale University School of Medicine, led a team investigating this measurement question. The researchers suspected that standard clinical evaluations masked subtle differences between pharmacological and placebo responses. They wanted to know whether the underlying pattern of symptom relief was the same in both groups. They also sought to determine whether an individual’s brain wiring before treatment could predict the likelihood of recovery.
To answer these questions, the research team analyzed data from a previous clinical trial of 192 patients with major depressive disorder. In the first phase of the trial, patients were randomly assigned to receive either a common antidepressant called sertraline or a placebo pill for eight weeks. Researchers in the first trial collected detailed information about patients’ depression, anxiety, suicidal thoughts, and manic symptoms. They also performed magnetic resonance imaging scans of the patients’ brains before starting treatment.
During clinical trials, clinicians used a simple seven-point rating system called the Clinical Global Impression Scale to determine whether patients were improving. Based on this extensive evaluation, the original results showed no statistical difference between the sertraline and placebo groups. The proportion of people considered to have responded to treatment was about the same for active drugs and sugar pills.
Berkovic and his team took a different approach to the data. They used statistical methods to evaluate responses to all individual questions from four separate psychological surveys. Rather than just looking at the final score calculated by the doctor, the researchers had a computer algorithm find the most dominant patterns of change across 73 individual symptom questions. This data-driven approach compressed a wide variety of patient responses into a single mathematical dimension of clinical improvement.
The results revealed that patients in both the medication and placebo groups improved along exactly the same path. Symptom relief followed a common geometry, whether people were given active drugs or sugar pills. The mathematical type of symptoms that changed over time was consistent regardless of the medication taken.
However, patients taking sertraline went further down this path. Mathematical models showed that antidepressants promote stronger overall recovery than placebo. This increased effect was primarily driven by a significant reduction in anxiety and a lower risk of suicidal ideation.
This finding highlighted the limitations of classic clinician rating scales. The basic 7-point rating did not detect this difference in response strength. Standard research often focuses on physical symptoms, which can obscure the specific psychological improvements tracked by mathematical models.
The research team also looked at patients’ symptoms at the start of the study to see if their initial level of illness could predict recovery. They found that severe anxiety and suicide risk at baseline predicted greater improvement in the mathematical model for both groups. Conversely, high depression-specific baseline scores only predicted recovery in patients taking sertraline.
After the first eight weeks, the trial included a second phase in which patients who did not see improvement were switched to the new treatment. Patients who did not respond to placebo were given sertraline, and those who did not respond to sertraline were given bupropion, a different type of antidepressant. The researchers ran the mathematical model during this second stage and found the same common pattern of improvement. This result suggests that the shape of symptoms is consistent across drugs.
The researchers achieved the most revealing insight when they analyzed baseline brain scans. In a resting-state scan, a machine measures how different areas of the brain communicate with each other when the patient is awake but not performing a specific task. Researchers mapped the brain’s global connectivity. They determined how strongly connected each small region was to the rest of the neural network.
They found that in symptom models for patients taking antidepressants, higher overall brain connectivity before treatment predicted stronger recovery. This means that the biological settings of a patient’s brain can predict their response to the actual drug. This predictive effect was not statistically significant in patients who received placebo.
Specific networks in the brain also showed different predictive patterns. Connectivity in the amygdala, an almond-shaped collection of neurons involved in processing fear and emotion, predicted improvement in symptoms in both groups. Wider, more comprehensive brain networks were only correlated with drug success. This pharmacological treatment appears to target specific, reproducible brain circuits. The biological causes of the placebo effect turned out to be noisier and more difficult to predict than the drug response.
This study relied on secondary analysis of previously completed trials, so the data were not collected specifically for this new mathematical approach. The sample size was relatively small for the type of statistical modeling used. Additionally, the original study design did not include brain scans taken at the end of the 8-week treatment period. Without follow-up imaging, researchers could only observe what predicted recovery, rather than observing how the brain physically changed in response to drugs or placebos.
Future studies with larger patient groups could help confirm whether this single pathway of mood improvement holds true across different demographics and depression subtypes. New trials involving multiple scans over time will allow scientists to map how these neural networks actually reorganize as symptoms fade. Comparing different types of antidepressants side by side using the same computer modeling could reveal how different chemical mechanisms influence recovery. Improving how we measure the mind may eventually allow doctors to use brain scans to match patients with the most effective individualized treatments.
The study, “General symptom geometry of mood improvement with sertraline and placebo is associated with different neural patterns,” was authored by Lucie Berkovitch, Kangjoo Lee, Jie Ji, Markus Helmer, Masih Rahmati, Jure Demsar, Aleksij Kraljic, Andraz Matkovic, Zailyn Tamayo, John Murray, Grega Repovs, John Krystal, William Martin, and Clara. Fontenot and Alain Antisevich.
