What the study measured
Murugadoss and colleagues studied 63,215 people with pre-existing neuropsychiatric conditions and evaluated 24 groups of new diagnoses in electronic health records [1]. Their separate dose analysis compared 39,795 semaglutide users whose highest dose in the first two years was 0.25–1.0 mg with 15,515 whose highest dose was 1.7, 2.0 or 2.4 mg. Researchers then counted diagnoses recorded during the following two years.
The investigators found lower recorded rates for substance-related disorders (rate ratio 0.71, 95% confidence interval 0.62–0.81), mood or affective disorders (0.82, 0.75–0.89), and anxiety and related disorders, including trauma-related disorders and OCD (0.88, 0.81–0.94). Other categories are listed in the full paper. The analysis did not detect a statistically clear difference for dementia or degenerative central nervous system diseases (P = 0.15).
Why reaching a higher dose matters
People who reach a higher dose may differ in treatment tolerance, access and how long they continue therapy. In a separate Vanderbilt obesity clinic program that covered prescribed medication costs, 81% of semaglutide users reached at least 1 mg and 23% reached 2.4 mg [3]. A Danish primary-care study on dose titration likewise illustrates how prescribing varied within one practice. That study tracked dose and weight, and the Vanderbilt figures describe how often patients reached certain doses.
Using separate windows for dose and outcomes helps establish the order of recorded events. Early symptoms could still affect who was able to increase their dose, and the analysis could miss symptoms that occurred during the first two years. The authors also reported that most outcomes did not differ clearly across weight-loss groups. The retrospective design limits what those findings can tell us about the causes of the observed differences. The authors worked for nference, which collaborates with drugmakers whose products were included, and reported that those collaborators had no role in or funding for the study [1].
How this fits with other evidence
The Swedish register study of four GLP-1 medicines examined worsening illness in people with existing depression or anxiety and recorded antidiabetic medication use. Its semaglutide estimate was an adjusted hazard ratio of 0.58 (95% CI 0.51–0.65) for a composite that included psychiatric hospitalization, extended psychiatric sick leave, hospitalization for self-harm and death by suicide [2]. That comparison was based on periods of medication use and non-use. A separate South Korean hospital cohort found more newly recorded depression and anxiety diagnoses among Wegovy initiators than among matched non-initiators. The populations, comparisons and outcomes differed, and each study was observational [4].
In the separate evidence on suicidal thoughts and behavior, the FDA found no increased risk of those outcomes and requested removal of the corresponding warning from Wegovy, Saxenda and Zepbound labeling [5]. The dose study answers a different question about recorded neuropsychiatric diagnoses.