GLP-1 Receptor Agonists and Suicide Risk in a Nationwide Study

Key Takeaways

• In these Danish cohorts, GLP-1 receptor agonist initiation was not linked to a higher adjusted hazard versus SGLT-2 inhibitors after 1 year of follow-up and was associated with a lower adjusted hazard versus DPP-4 inhibitors.
• Within-person analyses showed lower incidence during exposed periods than during the pretreatment reference period.
• Overall findings did not show increased incidence, although observational limits and low event counts remain important when interpreting the results.

In a nationwide Danish registry study conducted when GLP-1 receptor agonists were approved only for type 2 diabetes in Denmark, initiation of a GLP-1 receptor agonist was associated with a lower adjusted hazard of suicide or suicide attempts versus DPP-4 inhibitors after 1 year of follow-up, with an adjusted HR of 0.58 (95% CI, 0.37-0.91). Investigators also found no higher adjusted hazard versus SGLT-2 inhibitors after 1 year of follow-up in parallel cohort analyses. A complementary within-person analysis used a self-controlled case series to compare periods before and after treatment initiation. Overall, the pattern did not indicate increased incidence of suicide or suicide attempts.

The analysis used two new-user, active-comparator cohorts and a self-controlled case series to study suicide or suicide attempts. One cohort compared GLP-1 receptor agonists in 83,464 new users with SGLT-2 inhibitors in 78,366 users, and the other compared GLP-1 receptor agonists in 108,322 users with DPP-4 inhibitors in 55,411 users. In intention-to-treat analyses, cohort follow-up continued through 31 December 2022; in as-treated analyses, a gap longer than 90 days between prescription fills marked treatment end. The analyses adjusted for measured sociodemographic and clinical factors, accounted for competing risk of death, and used the within-person design as a complementary check.

In the active-comparator cohorts, the adjusted HR for suicide or suicide attempts after 1 year of follow-up was 0.93 (95% CI, 0.57-1.52) versus SGLT-2 inhibitors and 0.58 (95% CI, 0.37-0.91) versus DPP-4 inhibitors. In intention-to-treat analyses, median follow-up was 1 year [1-3] for GLP-1 users and 2 years [1-3] for SGLT-2 users. Median follow-up was 2 years [1-4] for GLP-1 users and 4 years [2-7] for DPP-4 users in the second cohort. Overall, the active-comparator results did not suggest a higher hazard.

The self-controlled case series used the 12 months before treatment initiation as the reference period and compared it with 0-12 months and 13-24 months after initiation. During the first exposed window, the incidence rate ratio for suicide or suicide attempts was 0.45 (95% CI, 0.10-0.50). The later exposed period was also lower than pretreatment, although the exact estimate was not provided in the available text. Sensitivity analyses by recent psychiatric history and interaction testing by sex, age, and prior suicide attempts did not change the overall no-increase pattern, aside from one interaction involving prior attempts and sex in the SGLT-2 comparison.

The authors noted several limitations, including possible exposure misclassification because adherence was not directly measured and residual confounding despite adjustment. Low numbers of suicide and suicide attempt events limited precision, and competing risk of death could have led to underestimation. Time-varying factors could still affect the self-controlled analysis, and obesity subgroup assessment was limited by underreporting in Danish registries. Across both analytic approaches, GLP-1 receptor agonist use was not associated with increased incidence of suicide or suicide attempts in this Danish registry setting.