Large Study Links GLP-1 Drugs to Lower Heart and Kidney Risk, and Higher Risk of Some Other Conditions

What happened / Why it matters

What happened. A study published online Oct. 28, 2026 in the Journal of Diabetes Science and Technology compared adults with type 2 diabetes who started a GLP-1 receptor agonist with similar adults who started other diabetes drugs. Among 70,982 matched adults, it found 118 “robust” associations: 65 with lower risk and 53 with higher risk of a range of conditions.

Why it matters. The findings point in both directions, which headlines can flatten into “protective” or “dangerous.” The study was observational, so it can show associations, not cause and effect. This article covers what the abstract reports and what it does not. We read only the abstract: the full text is under embargo in PubMed Central until October 2027 and we could not open it on the journal’s site, so we could not check the paper’s methods, tables or its authors’ own limitations section.

What did the researchers do?

According to the abstract:

• Design. A “target trial emulation,” which uses existing health records to mimic a randomized trial. People were not randomly assigned. The records came from five University of California academic medical centers, 2012 to 2025.
• Who. Adults with type 2 diabetes who newly started a GLP-1 drug, a sulfonylurea, an SGLT2 inhibitor or a DPP-4 inhibitor. Patients were matched one-to-one, with a statistical model that drew on more than 11,000 covariates, and followed for up to five years.
• What they looked at. 181 prespecified outcomes across 15 body systems.
• What counted as “robust.” The authors required results to hold when any one medical center was left out, an E-value of at least 1.5 (a measure of how strongly an unmeasured factor would have to be tied to both the drug and the outcome to explain the result away) and a false discovery rate-adjusted P value below .10.

What lower-risk associations did it find?

The abstract says the risk reductions were “most consistent” for cardiovascular outcomes. Examples it gives, as hazard ratios (below 1.0 means lower risk), include:

• Myocarditis or cardiomyopathy: 0.38 (95% CI 0.32 to 0.44), compared with SGLT2 inhibitors.
• Heart attack (acute myocardial infarction): 0.53 (0.42 to 0.66).
• Heart failure: 0.71 (0.58 to 0.88), compared with DPP-4 inhibitors.
• Nephritis or nephrosis (kidney conditions): 0.73 (0.67 to 0.79).
• Also reported: thromboembolism, nutritional anemia, coagulation disorders and tobacco-related disorders.

The abstract does not say which comparison drug applies to every figure. Where it names one, we have named it.

What higher-risk associations did it find?

The abstract reports 53 associations with higher risk, including 22 for musculoskeletal conditions. Examples:

• Osteoarthritis: 1.48 (1.29 to 1.68).
• Stress fractures: 1.68 (1.15 to 2.47).
• Diverticular disease: 1.18 to 1.33.
• Sleep-wake disorders: 1.27 to 1.54.
• Menopausal disorders: 1.32 to 1.52.

The authors write that increased risks for musculoskeletal, gastrointestinal and sleep-wake disorders “warrant monitoring as the clinical use of GLP-1RAs continues to expand.”

What can this study not tell us?

• Cause. Because patients were not randomly assigned, the study cannot show that the drugs caused any of these differences. As in any non-randomized comparison, people prescribed one drug rather than another can differ in ways that matching does not capture.
• How big the risks are in practice. The abstract reports relative measures, not absolute risk. A hazard ratio of 1.48 does not say how many people in a hundred are affected.
• Which drug. The abstract does not break results down by individual drug or dose. The abstract does not say which GLP-1 drugs people took, and the study period, 2012 to 2025, spans many years of prescribing. Its results cannot be assigned to any single medicine, including semaglutide or tirzepatide. Both are FDA-approved for type 2 diabetes in specific products (Ozempic, Mounjaro). Compounded versions are not FDA-approved.
• Obesity without diabetes. Everyone in the study had type 2 diabetes.
• Some false positives. The authors accepted a false discovery rate below 10%. As we read it, that allows some chance findings among the 118 results.
• The authors’ own caveats. We have not read their limitations section.

Is this an FDA finding?

No. The PubMed record lists grants from FDA and from the National Center for Advancing Translational Sciences at NIH. Funding a study is not the same as FDA reviewing or endorsing its results, and nothing in the abstract is an FDA position or a change to any drug’s label. The authors report no conflicts of interest.

What does this mean for a reader?

A single observational study does not settle whether any diabetes drug is better or riskier. Decisions about starting, stopping or switching a medicine depend on a person’s health and goals, and belong with a prescribing clinician. Anyone with questions about their own treatment can raise this study with their clinician. Do not stop or change a medicine because of a news article.

U.S. status of semaglutide and tirzepatide checked Oct. 31, 2026. Based on the study abstract only, as of that date.