GLP-1 Drugs Improve Diabetes Outcomes but Show No Measurable Gains in Mental Health or Employment, Study Finds

The story patients and payers have told about GLP-1 drugs goes something like this: lose the weight, lower the blood sugar, and the rest of life follows. A new working paper released through the National Bureau of Economic Research puts some pressure on that narrative.

Economist Robert Kaestner of the University of Chicago Harris School of Public Policy and coauthor Cuiping Schiman analyzed more than a decade of nationally representative data, 2012 through 2023, drawn from the Medical Expenditure Panel Survey, a federal household survey that tracks adults with diabetes. According to the University of Chicago News, the researchers found little evidence that using GLP-1 medications produced measurable changes in mental health, self-rated health, employment, or marital status among that population.

The design matters here. The team didn't simply compare GLP-1 users against non-users, which would pick up all kinds of differences that exist before anyone opens a prescription bottle. Instead, they used a within-person, longitudinal approach, tracking how each individual's circumstances shifted after they started the medications. When they did that, the apparent advantages that GLP-1 users seemed to have over non-users largely disappeared, suggesting those gaps reflected pre-existing differences between the two groups rather than anything the drugs were doing.

Results held whether the researchers looked at roughly one year of use or followed patients through a two-year window, according to reporting on the NBER paper by SciTechDaily.

Kaestner was careful about what the findings do and don't say. "We measured outcomes like whether someone became employed or unemployed, whether they got married or divorced, and standard indicators of mental health," he said in a statement reported by the University of Chicago News. "Those are important measures, but they don't necessarily capture changes in self-esteem, relationship quality, or other day-to-day experiences."

That's a clinically honest caveat worth sitting with. Survey-based instruments like the Medical Expenditure Panel Survey aren't built to catch shifts in how someone feels getting dressed in the morning, or whether they're showing up differently in a conversation. Those are real things. They're just hard to measure at population scale.

The study doesn't disturb the established physical case for GLP-1 treatment. Better glycemic control, reduced cardiovascular risk, and weight loss are documented across large randomized trials in people with Type 2 diabetes and obesity. The question the Chicago paper asks is a different one: whether those physical gains ripple outward into the social and psychological fabric of a person's life in ways we can actually detect.

For now, the answer appears to be not reliably, at least not in adults with diabetes and not over a one-to-two-year horizon.

The timing of the paper lands in a crowded moment for GLP-1 research. Separate observational work published earlier this year, including a large Swedish registry study covered by ScienceDaily, reported associations between semaglutide use and reduced psychiatric hospitalizations, lower rates of depression and anxiety diagnoses, and fewer substance-use-related absences from work. A Griffith University team found similar patterns specifically in people with bipolar disorder. Those studies relied on administrative health records, not survey-reported wellbeing, and used associational designs that can't rule out confounding.

What's happening across these papers is a methods conversation as much as a clinical one. Observational data that compares users to non-users will almost always find some advantage for the treated group, because people who get and stay on an expensive medication aren't a random slice of any population. The Chicago team's within-person approach is a meaningful step toward controlling for that.

The practical stakes are not abstract. As ScienceDaily noted in its coverage of the paper, insurers and government programs are actively weighing whether to expand coverage for GLP-1 medications. If the drugs' value is understood to extend beyond glycemic control and cardiovascular risk reduction into employment productivity and mental health burden, that calculus shifts. If it doesn't, the cost-benefit math looks different.

For clinicians writing these prescriptions, the core guidance hasn't moved: GLP-1 medications remain well-supported for their approved indications. What this paper adds is a reminder to be straightforward with patients about what the evidence actually shows, and what it's still working out.

Sources cited:
- University of Chicago News (https://news.uchicago.edu/story/glp-1-drugs-may-hold-few-benefits-beyond-physical-health-study-finds)
- SciTechDaily (https://scitechdaily.com/ozempic-changes-your-body-but-does-it-change-your-life-a-new-study-suggests-not/)
- ScienceDaily (NBER paper coverage) (https://www.sciencedaily.com/releases/2026/09/260917003728.htm)
- NBER Working Paper No. 35198 (https://www.nber.org/system/files/working_papers/w35198/w35198.pdf)
- ScienceDaily (Swedish registry / bipolar study) (https://www.sciencedaily.com/releases/2026/09/260907201601.htm)