Diabetes Mellitus Prevalence by Uveitis Etiology at a Japanese Tertiary Center

Diabetes mellitus (DM) may influence susceptibility to ocular infection and inflammatory phenotypes; however, disease-specific DM prevalence across uveitis etiologies has not been well characterized. We evaluated DM prevalence across major uveitis entities in a large single-center cohort and compared observed prevalence estimates with age-specific national estimates from Japan. A total of 3163 adult patients out of 4751 patients newly diagnosed with uveitis at the Tokyo Medical University Hospital between 2004 and 2023 were included in the final analysis after excluding patients without DM-related laboratory parameters, those who declined to participate through the institutional opt-out procedure, pediatric patients, and patients with diabetic iritis. DM was defined as an HbA1c level of 6.5% or higher, current treatment for DM, or a random blood glucose level of 200 mg/dL or higher. For reference comparisons with national data, we analyzed seven adult entities with sufficient sample size. For each entity, expected DM prevalence was based on the corresponding national age-stratum estimate according to the mean age category of that entity. Observed and expected DM prevalence estimates were compared using Fisher's exact test. Multivariable modified Poisson regression with robust variance estimation was used to estimate adjusted prevalence ratios (aPRs) for DM prevalence, including age, sex, and individual uveitis entities as covariates. Overall, 166 of 3163 patients (5.3%) had DM. DM prevalence differed substantially by etiology, ranging from 18.0% in endophthalmitis to 1.8% in Behçet disease. Selected infectious uveitis showed a higher crude prevalence of DM than selected noninfectious uveitis (10.4% vs. 3.8%; p < 0.01). In comparisons with corresponding national age-stratum estimates, the observed prevalence estimates of DM in endophthalmitis and herpetic iritis did not differ significantly from the expected national prevalence estimates, whereas acute retinal necrosis, intraocular lymphoma, sarcoidosis, and Behçet disease showed a significantly lower prevalence than expected. In the multivariable analysis, older age (aPR per 10-year increase, 1.49; 95% CI, 1.33-1.67; p < 0.001) and male sex (aPR, 2.11; 95% CI, 1.40-3.19; p < 0.001) were independently associated with DM prevalence, whereas individual uveitis entities were not significantly associated after adjustment for covariates. DM prevalence in uveitis is heterogeneous and disease-specific. Although selected infectious uveitis showed a higher crude prevalence of DM, comparisons with corresponding national age-stratum estimates suggested that some of the differences reflected the underlying age structure. Older age and male sex were the strongest independent correlates of DM prevalence.