SGLT2 inhibitor use is associated with more polycythemia vera diagnosis codes but not with JAK2-positive polycythemia vera: a propensity score–matched cohort study

Background Treatment with sodium-glucose cotransporter 2 inhibitors (SGLT2is) is associated with modest increases in hematocrit, which may prompt hematologic evaluation. We examined whether SGLT2i use is associated with increased detection of previously unrecognized myeloproliferative neoplasms (MPNs) and compared diagnosis code–based with JAK2 -positive ascertainment of polycythemia vera (PV). Methods We conducted a retrospective cohort study using electronic health records from Clalit Health Services, including adults aged ≥40 years with diabetes who initiated SGLT2is or dipeptidyl peptidase-4 inhibitors (DPP4is) between January 2015 and July 2025, with follow-up through November 2025. Baseline clinical, laboratory, and medication data were collected, and participants were followed for incident MPNs. SGLT2i users were matched 1:1 to DPP4i users using propensity scores. Hazard ratios (HRs) were estimated using Cox and Fine–Gray competing-risk models. Results A total of 73,147 SGLT2i and 48,639 DPP4i users were eligible. After propensity score matching, 40,550 patients remained in each group, with balance across most baseline characteristics; residual creatinine imbalance was addressed in an extended model. Median follow-up was 3.8 years with SGLT2is and 7.0 years with DPP4is. A new PV diagnosis code (ICD-10 D45) was recorded in 204 SGLT2i users and 205 DPP4i users [1.22 vs. 0.76 per 1,000 person-years; adjusted HR (aHR) 1.69, 95% confidence interval (CI) 1.39–2.05], with the excess concentrated in the first year after initiation. Among the 204 patients who underwent JAK2 testing, 64 (31.4%) had a positive result, representing 15.6% of all 409 patients with a D45 code. JAK2 positivity among tested patients was lower with SGLT2is than DPP4is (24.5% vs. 37.7%). JAK2 -positive coded PV was identified in 24 and 40 patients, respectively (0.14 vs. 0.15 per 1,000 person-years; aHR 1.00, 95% CI 0.60–1.65). No significant differences were observed for essential thrombocythemia or primary myelofibrosis. Results were consistent in competing-risk and sensitivity analyses. Conclusions SGLT2i use was associated with more PV diagnosis codes but not with JAK2 -positive PV or other MPNs. The findings are consistent with increased diagnostic evaluation of SGLT2i-related erythrocytosis rather than increased occurrence of clonal PV. Diagnosis codes alone may substantially overestimate confirmed disease.