Different incidences of diabetic retinopathy requiring treatment since diagnosis according to the course of diabetes diagnosis: a retrospective cohort study.

We aimed to estimate the cumulative incidence of treatment-requiring diabetic retinopathy since clinical diagnosis of diabetes based on the course of diagnosis in a retrospective cohort study using Japan's medical claims and health checkup data (JMDC Claims Database; 2009-2020). We included patients whose diabetes was first diagnosed at medical facilities (hospitals/clinics). We grouped them by health checkup participation before diagnosis, health checkup results, and antidiabetic medication promptly after the diagnosis. The incidence of treatment-requiring diabetic retinopathy (laser photocoagulation, intraocular injection, or vitrectomy) was compared among the groups. Of 126,696 patients, those who started an antidiabetic medication promptly after diabetes diagnosis without a recent health checkup faced the highest risk of treatment-requiring diabetic retinopathy (1-/5-year cumulative incidence: 3.1%/6.0%). This increased risk was consistently observed across various analyses, including the Cox proportional hazard model, sensitivity analysis restricting to those with an eye examination, and sensitivity analysis using vitrectomy as the outcome. Among patients with HbA1c ≥ 6.5% at recent health checkups, those who promptly started an antidiabetic medication had a higher risk (1.4%/3.8%) than those who did not (0.7%/2.7%). Taking the information about the course of diabetes diagnosis is important to manage risk stratification for diabetic retinopathy appropriately.