Weight loss and fewer eating concerns did not necessarily travel alongside better mood in a small Japanese study of people starting or switching incretin medicines. The findings raise a useful question about follow-up, but they cannot establish that the medicines caused anyone’s mood to worsen.
The study published online on 31 October 2026 followed 15 adults for 85 days. Fourteen had type 2 diabetes; only one had obesity without diabetes. This was an exploratory observation of routine treatment, not a randomized obesity trial. Its most useful contribution is showing why weight, eating experiences and depressive symptoms should be read as separate outcomes.
Three measures, different trajectories
Participants recorded weight and eating experiences daily and completed the nine-item Patient Health Questionnaire, or PHQ-9, weekly. Researchers used repeated-measures models to examine changes over time. Weight and the daily eating score declined overall. Depressive-symptom scores moved in different directions depending on the initial score: downward in the lower-score group and upward in the higher-score group.
The published supplement puts those groups in perspective: nine people started below a PHQ-9 score of 10, and six were at or above it. Those are small groups, even though repeated questionnaires supplied many observations. A hundred assessments from a few people do not provide the same breadth of evidence as a hundred independently enrolled participants.
PHQ-9 is an established symptom measure, with its severity assessment supported by an original validation study. Here, a score threshold divided participants for analysis. The observed trajectories do not by themselves establish new clinical diagnoses or explain what caused the changes.
What the eating score actually measured
The daily questionnaire asked about two experiences: eating an unusually large amount and feeling a loss of control while eating. Researchers adapted these from an existing eating-disorder questionnaire, changing them into daily yes-or-no items and adding the responses into a score from zero to two.
A lower combined score could reflect a change in either experience. It is not a count of diagnosed binge-eating disorder cases or proof that an eating disorder went into remission. That distinction matters when an appetite-related result becomes a headline about treating binge eating.
NIDDK’s guidance on diagnosis and treatment explains that assessment involves symptoms and eating patterns with a professional trained in eating disorders. Our binge-eating assessment guide explores why those care goals can differ from a weight target.
Why this cannot identify a drug effect
The UMIN registry describes an observational study. There was no randomized untreated comparison group against which to judge the mood changes. Life events, the course of existing symptoms and other care could matter; the available evidence does not isolate their contributions.
Treatment was also varied. Nine participants were beginning therapy and six were switching. The supplement lists semaglutide, tirzepatide and dulaglutide as study therapies, with some participants previously using another incretin medicine. These experiences should not be pooled into a claim that one drug improves or worsens mood more than another. Our semaglutide profile provides separate medicine context.
The accessible abstract and supplement do not provide a complete account of doses or concurrent psychiatric treatment. The subscription-only main text was not available for this review. Those gaps limit interpretation; they are not evidence that other treatment was absent. Our guide to randomized and real-world evidence explains what an observational design can contribute.
Missing questionnaires also shape the result
Researchers obtained 166 of 195 possible PHQ-9 assessments and analyzed the observed scores without filling in missing assessments. For two people who missed the initial questionnaire, the score seven days later determined their group. Their classification therefore was not based on a questionnaire completed at treatment initiation.
Missing responses might be unrelated to mood, or they might reflect circumstances that also affect symptoms. This small dataset cannot settle that question. Nor can it establish whether the pattern persists beyond the 85-day observation period or applies broadly to people receiving treatment for obesity without diabetes.
A more useful follow-up conversation
The practical lesson is a question to bring to care: “How are we following mood and eating concerns alongside weight?” A favorable scale trend does not answer that question. If mood changes or distressing loss-of-control eating occur, describing their timing and impact can make the conversation more specific. They deserve attention without first proving a medication caused them.
This study does not justify choosing, stopping or switching a medicine on the basis of its subgroup findings. It also does not measure suicide risk; that is a separate outcome from a total depressive-symptom score. The site’s guide to interpreting psychiatric safety updates addresses that different evidence question.
Larger studies with clearer treatment comparisons and longer follow-up are needed. For now, the paper offers a reason to keep asking about several dimensions of health, rather than treating weight loss as a complete picture of how someone is doing.



