Higher semaglutide doses linked to fewer new psychiatric diagnoses in large records study

A study published September 1 in the journal npj Metabolic Health and Disease reports that among 63,215 people taking semaglutide who already had a neurological or psychiatric diagnosis, those who reached higher doses were less likely to receive new diagnoses of mood, anxiety or substance-use disorders in the following two years. The pattern did not track with how much weight they lost ^[1].

The research team, led by Karthik Murugadoss and Venky Soundararajan of nference, used electronic health records from a network of 489,785 semaglutide-treated patients. From that group they built an analysis cohort of 63,215 patients with baseline neurological or psychiatric conditions. Parts of the cohort were matched one-to-one with patients taking metformin (56,859), DPP-4 inhibitors (4,757) or SGLT2 inhibitors (30,680) ^[1].

Compared with metformin, semaglutide was linked to lower two-year rates of mood disorders (10.5 percent versus 13.2 percent), substance-related disorders (3.8 versus 6.1 percent), dementia or degenerative central nervous system disease (0.95 versus 1.96 percent), cognitive symptoms (3.1 versus 4.3 percent), psychotic disorders (0.31 versus 0.93 percent) and cerebrovascular disease (2.6 versus 3.7 percent), among other outcomes. Those differences were reported as statistically significant ^[1]. Metformin was linked to a lower rate of one outcome, incident neuromuscular disease (0.7 versus 1.1 percent) ^[1]. The study also checked negative control outcomes — eczema, kidney stones, rotator cuff disorders, hernia and tinea corporis — and found no significant differences between the two drugs ^[1].

For the dose analysis, patients were grouped by the highest dose they reached in their first two years of treatment and by the most weight they lost in the same period, with outcomes measured over the next two years. A higher dose was defined as 1.7 milligrams or more; the lower group ranged from 0.25 to 1.0 milligrams ^[2]. Patients on the higher dose had a 29 percent lower relative risk of a substance-related disorder and an 18 percent lower relative risk of a mood disorder, Soundararajan told PsyPost. He described the absolute differences as about 23 and 14 fewer diagnoses per 1,000 people at risk ^[2].

Higher dose was also associated with lower incidence of anxiety- and stress-related disorders, CNS atrophies and personality or impulse-control disorders, with P values below 0.05 ^[1]. One outcome went the other way: dementia or CNS degenerative disease occurred at similar rates in the high-dose and low-dose groups, a difference the authors reported as not statistically significant (P = 0.15) ^[1].

Weight loss told a different story. When patients were grouped by how much weight they lost — from less than 5 percent to more than 20 percent of body weight — researchers did not see the same pattern for substance, mood or anxiety disorders ^[2]. For cognitive and speech or language symptoms, weight loss was the closer correlate (P < 0.001 and P = 0.003) ^[1]. Recorded cognitive symptoms rose from 2.1 percent in the group that lost the least weight to 4.7 percent in the group that lost the most, which the researchers say does not necessarily mean the drug harms cognition; in real-world records, large weight loss can be a sign of underlying decline or frailty ^[2].

The authors also examined public gene-expression datasets and reported low-level, regionally restricted signals for the GLP-1 receptor in central and peripheral nervous system tissue, including the hypothalamus and caudate nucleus — context they describe as hypothesis-generating rather than proof of how the drug acts ^[1][2].

Why it matters for patients

The findings are observational. They cannot show that semaglutide protects against psychiatric diagnoses, or that raising a dose would prevent one. "Whether changing someone's dose would produce those differences requires a randomized control trial," Soundararajan said ^[2]. He also cautioned that "independent of weight loss" should be read carefully, because comparing separate dose groups and weight-loss groups does not establish a direct drug effect independent of weight. People who reach higher doses may differ in treatment tolerance, continuity of care and affordability, and those who have side effects early often stay on lower doses or stop, which can make the high-dose group look healthier from the start ^[2].

The authors say the study should not be used to justify prescribing semaglutide for psychiatric conditions, and it was not designed to evaluate psychiatric side effects right after starting the drug ^[2]. The data come from diagnostic codes entered by doctors, which capture formal diagnoses and lag behind when a condition begins ^[2]. No difference in dementia risk appeared between dose groups, so the paper offers no support for taking semaglutide to protect memory or thinking ^[1].

What happens next

The paper's authors say the results motivate prospective mechanistic and clinical studies ^[1]. Soundararajan said the group's long-term aim is to understand which patients benefit, at what dose, and with what trade-offs across the body and brain, and that review of clinical records with artificial intelligence is a priority ^[2]. No timelines for such trials have been reported ^[2].