(C-066) Channeling Bias in First-Line Prescribing of SGLT2 Inhibitors and GLP-1 Receptor Agonists for Individuals with Severe Mental Illness and Diabetes

Background: Sodium-glucose cotransporter-2 inhibitors (SGLT2i) and glucagon-like peptide-1 receptor agonists (GLP-1) offer cardiovascular (CV) benefits in the general population. Adults with severe mental illness (SMI) face increased CV and metabolic risks, making these agents promising first-line treatment options. As evidence of their CV benefits has grown, selective prescribing (i.e., channeling) to high-risk individuals with SMI may bias comparative effectiveness studies of these first-line treatments in this population.

Objectives: To evaluate channeling bias among new users of SGLT2i, GLP-1, and metformin as first-line treatment in individuals with SMI and diabetes.

Methods: Using a 20% random sample of US Medicare fee-for-service claims data, we conducted a 3-arm new-user retrospective cohort study of beneficiaries aged ≥18. We identified individuals with SMI (schizophrenia, bipolar disorder, depressive disorder) and diabetes who initiated SGLT2i, GLP-1, or metformin 1/1/2018 – 12/31/2021. Eligible individuals had continuous medical and pharmacy enrolment and no oral antihyperglycemic agent in the 12-month baseline period prior to the index date (first prescription claim for study drug of interest). We excluded individuals with type 1 diabetes, gestational diabetes, polycystic ovary syndrome, organ transplant, end-stage renal disease, or HIV/AIDS at baseline. We assessed imbalance in baseline characteristics using standardized differences (SD); metformin as reference. A positive SD indicates a higher prevalence for SGLT2i or GLP-1 than metformin (and a negative SD a lower prevalence), with |SD|>0.1 reflecting imbalance. Data analyses identified temporal changes in channeling bias.

Results: Among individuals with SMI and diabetes, we identified new users of first-line treatment with SGLT2i (n=1,322), GLP-1 (n=2,306), and metformin (n=21,144). In 2018, SGLT2i users were younger (SD -0.11) and had higher prevalence of atrial fibrillation (AF; SD 0.01), heart failure (HF; SD 0.07), ischemic heart disease (IHD; SD 0.29), and chronic kidney disease (CKD; SD 0.18) than metformin users. By 2021, SGLT2i users were older (SD 0.19); with notably larger SDs for AF (SD 0.47), HF (SD 0.72), IHD (SD 0.56), and CKD (SD 0.62). GLP-1 users consistently showed a higher prevalence of CKD (SD 0.42 in 2018; SD 0.28 in 2021) and obesity (SD 0.33 in 2018; SD 0.41 in 2021) than metformin users; other comorbidities were stable over time.

Conclusions: From 2018–2021, profiles shifted, with SGLT2is increasingly prescribed to those at high CV risk. GLP-1s were consistently prescribed to those with CKD or obesity. This channeling should be accounted for in comparative effectiveness research using real-world data.