BACKGROUND: People with diabetes have an elevated risk of developing depression, anxiety, and suicide. GLP-1 receptor agonists are licensed to treat diabetes and obesity, but data on whether these medications alleviate or exacerbate anxiety, depression, and self-harm are mixed. We studied the risk of worsening mental illness in people already diagnosed with depression, anxiety, or both who were prescribed antidiabetic medications including GLP-1 receptor agonists.
METHODS: The study cohort, identified from national Swedish electronic health registers, included people with a diagnosis of depression or anxiety disorder who used any antidiabetic medication between the years 2009 and 2022. GLP-1 receptor agonists, individually and as a group, were compared with non-use of GLP-1 receptor agonists and directly with other second-line antidiabetic medications. A within-individual design was used for all comparisons to reduce confounding, comparing periods of use versus periods of non-use of a medication in the same individual. The primary outcome was worsening of mental illness, defined as a composite of psychiatric hospitalisation; sick leave from work for more than 14 days for psychiatric reasons; hospitalisation due to self-harm; or death by suicide. Secondary outcomes were worsening of depression or anxiety, analysed separately, worsening of substance use disorder, and self-harm. Within-individual stratified Cox models with adjusted hazard ratios (aHRs) and 95% CIs were used. A person with related lived experience was involved in the design and write-up of this study.
FINDINGS: The cohort included 95 490 people (56 976 [59·7%] female and 38 514 [40·3%] male) with a mean age of 50·6 years (SD 12·3). Ethnicity data were not available. GLP-1 receptor agonists were used by 22 480 individuals during the follow-up period. Compared with non-use of GLP-1 receptor agonists, semaglutide (aHR 0·58 [95% CI 0·51-0·65]) and liraglutide (0·82 [0·76-0·89]) were associated with lower risk of worsening mental illness, whereas exenatide (1·01 [0·69-1·46]) and dulaglutide (1·01 [0·85-1·20]) were not. Semaglutide was associated with a decreased risk of worsening depression (0·56 [0·44-0·71]), of worsening anxiety (0·62 [0·52-0·73]), and of worsening substance use disorder (0·53 [0·35-0·80]). Liraglutide was associated only with lower risk of worsening depression (0·74 [0·64-0·87]). GLP-1 receptor agonists as a group were associated with a reduced risk of self-harm (0·56 [0·34-0·92]).
INTERPRETATION: For anxiety and depression that co-occur with diabetes and obesity, semaglutide and, to a lesser extent, liraglutide might be useful dually effective therapeutic options. Randomised controlled trials evaluating these findings are warranted.
FUNDING: Sigrid Jusélius Foundation, Jane and Aatos Erkko Foundation, and Finnish Ministry of Social Affairs and Health.
