Diabetic retinopathy is a leading cause of blindness in the US, and current guidelines recommend annual ophthalmologic screening for all patients with diabetes. Compliance with screening among diabetic patients remains poor despite the availability of vision-preserving treatments, possibly because patients remain asymptomatic until the late stages of the disease and because current screening programs require a visit to a specialist. The inconvenience and costs of screening in the specialist setting have motivated interest in alternative screening programs that can be implemented in primary care settings. A discrete-time Markov model was developed to estimate the health benefits and costs of hypothetical new diabetic retinopathy screening programs. The model tracks the underlying epidemiology of diabetic retinopathy and accounts for the influence of screening and subsequent treatment on progression and vision loss. Screening attributes considered by the model include setting, frequency, level of retinopathy targeted, accuracy, failure rate, and compliance. Data to parameterize the model were obtained from publicly available sources, and the model was validated with a clinician and against published clinical and economic studies. The health benefits, costs, and cost-effectiveness of hypothetical new screening programs were estimated in comparison with current screening programs.
