Depression scores fell and suicide screening did not worsen in a 44‑teenager GLP‑1 chart review

Researchers at the Medical College of Wisconsin looked back through the records of 44 young people with obesity who were treated with a GLP-1 medicine at their pediatric hormone clinic, and found no sign that the drugs made depression or suicidal thinking worse in the months that followed. The review was published online on September 14 in the Journal of Pediatric Endocrinology and Metabolism.

The finding lands on a question that has followed these medicines since 2023, when reports of suicidal thoughts in people taking them prompted a formal FDA investigation.

Only 15 of the 44 teenagers had a depression score at both visits

This was a chart review, not a trial. Nobody was assigned to a treatment and there was no comparison group. The researchers pulled records for patients aged 12 to 21 who were prescribed liraglutide or semaglutide at a single academic center between June 2022 and December 2024, then looked for mental health screening scores recorded before treatment started and again at a later visit.

Two screening tools were in use. The PHQ-9 is a nine-question survey that adds up to a depression symptom score, where a higher number means more symptoms. The ASQ is a four-question suicide risk screen short enough to run at a routine clinic visit.

Because the analysis counted only patients screened twice with the same tool, most of the group dropped out of the numbers. Fifteen patients had a PHQ-9 at both points. Twenty-three had an ASQ at both points. Those are the samples the conclusions actually rest on.

Median depression scores went from 6 to 1, and the drop did not track weight change

Among the 15 patients with paired PHQ-9 scores, the median score fell from 6 to 1. The spread was wide. The middle half of patients scored between 2 and 14 at the start and between 0 and 5 at follow-up, so the top quarter began at 14 or above.

The authors report the change as statistically significant, with a p-value of 0.003. In plain terms, a drop that large would be unlikely to turn up by chance alone in a sample this size.

They also report that the improvement held independently of how much each patient's BMI changed, meaning the better scores were not simply following weight loss.

Among the 23 patients with paired ASQ screens, no one screened positive for new suicidal thinking and no screen got worse.

FDA asked drugmakers in January to drop the suicide warning from three obesity labels

The study covers a stretch of time when this question was formally open. FDA began investigating reports of suicidal thoughts and behavior in people taking GLP-1 medicines in July 2023.

On January 13, 2026, the agency said a comprehensive review had found no increased risk, and asked the companies behind Saxenda (liraglutide), Wegovy (semaglutide) and Zepbound (tirzepatide) to remove the suicidal ideation and behavior warning from their prescribing information. FDA noted the warning had been carried over from reports seen with a range of older weight loss medicines, and that GLP-1 products approved for type 2 diabetes never carried it. The agency told patients to keep taking their medication as prescribed and to raise any concerns with their clinician.

That request is the current regulatory position, and it rests on far more evidence than this chart review does. The Wisconsin paper is a small addition pointing the same direction, not the basis for the change.

A chart review with no comparison group cannot separate the drug from time

The clearest limits are the ones the authors name. With 15 and 23 patients in the two analyses, the study is too small to pick up an uncommon harm. With no untreated comparison group, there is no way to know how these scores would have moved on their own.

There is also a selection problem the design cannot fix. Patients got a second screen because a clinician ordered one, and the patients who were screened twice may differ from the ones who were not. A starting median of 6 with the top quarter at 14 or above leaves room for scores to drift downward without any treatment effect, a pattern statisticians call regression to the mean.

The authors call the work exploratory and say larger prospective studies are needed. Read alongside the regulatory review, it is one small reassuring data point rather than a safety guarantee. New or worsening depression, or any unusual change in mood, is still worth reporting to the prescribing clinician.