We examined the cost-effectiveness of three screening strategies for diabetic retinopathy (DR); using a personalized model, annual screening, and the most recent Dutch guideline. The Dutch guideline has variable intervals based on pre-existing retinopathy scores from immediate referring for treatment up to 3 years.

For each individual, DR screening intervals were determined for personalized screening using different STR risk margins. Observational data (1998-2017) from the Hoorn Diabetes Care System cohort of people with type 2 diabetes, were used (N=5,514). In order to evaluate the performance of the model, the actual time to develop STR in the cohort was used and for missing values of this variable, two alternative scenarios were assumed: slow and fast STR progression. Missed cases, the outcome of each strategy, were determined by comparing model based screening intervals to observed time to develop STR. Costs were calculated based on screening and travel costs. Finally, outcomes and costs were compared for the different screening strategies.

Comparing personalized screening with annual screening resulted in 11.0% and 11.6% more missed cases with €10.4 and €8.3 less cost per patient for slow and fast STR progression assumptions, respectively. The personalized screening strategy performed better in terms of diagnosing STR cases and it had 7.1% and 9.1% less missed cases compared to Dutch guideline screening strategy. While for a slow STR progression assumption, personalized screening strategy reduced costs with €0.2 per patient, assuming fast STR progression personalized screening was €1.9 per patient more expensive than current Dutch guideline strategy. Missing cases would be found at a later time, with a median delay of 19 months for personalized screeening, and 12 months for the Dutch guideline strategy.

Personalized retinopathy screening is more cost-effective than the Dutch guideline screening strategy. Although the personalized screening strategy was less effective than annual screening, the number of late diagnosed STR patients is low and the saving is considerable. With around 1,000,000 people with type 2 diabetes in the Netherlands, implementing this personalized model could save 8.5 to 10.6 million euros