Erythrocytosis has been increasingly observed during sodium-glucose cotransporter 2 (SGLT2) inhibitor therapy, but its implications in patients with atrial fibrillation (AF) receiving oral anticoagulants (OAC) have remained uncertain. A population-based observational cohort study published in Cardiovascular Diabetology evaluated the incidence of SGLT2 inhibitor–associated erythrocytosis and its association with thromboembolic and bleeding outcomes in this population.

Using the Tianjin Health and Medical Data Platform, adults with AF receiving OAC between 2011 and 2024 were identified. In Part 1, a new-user design with a stratified pseudo-time strategy was applied to reduce immortal time bias. Of 47,401 eligible individuals, 39,190 were included in the analysis, including 6,697 SGLT2 inhibitor users. Incident erythrocytosis was assessed using Poisson regression to estimate incidence rate ratios.

Within 6 ± 2 months, erythrocytosis occurred in 11.3% of SGLT2 inhibitor users compared with 3.1% of non-users, corresponding to an adjusted incidence rate ratio of 3.85 (95% CI, 2.78-5.31). In Part 2, restricted to SGLT2 inhibitor users, erythrocytosis served as the exposure for thromboembolic and bleeding outcomes. During full follow-up, erythrocytosis was associated with thromboembolic events (Model 5 adjusted hazard ratio 2.34; 95% CI, 1.39-3.94; P = .001), whereas the association with bleeding was not statistically significant (adjusted hazard ratio 1.42; 95% CI, 0.81–2.49; P = .22). Subgroup and sensitivity analyses demonstrated consistent results across age, sex, kidney function, and anticoagulant type. Counterfactual mediation analysis indicated that the rise in hemoglobin levels functioned primarily as a risk marker rather than a causal mediator.

In this large real-world cohort of patients with AF receiving anticoagulation, SGLT2 inhibitor use was associated with a higher incidence of erythrocytosis, and erythrocytosis among users was associated with thromboembolic events but not bleeding. The hemoglobin rise appeared to represent a risk marker rather than a causal mediator based on mediation analyses.