S. Emamipour, A.A.W.A. van der Heijden, M. Postma, J.F.M. van Boven, T. Feenstra

Research output: Contribution to journalMeeting AbstractAcademic


Objectives: Diabetic retinopathy (DR) is one of the leading causes of visual impairment and blindness. However, only a minority of the type 2 diabetes population is at risk of sight-threatening retinopathy (STR). Most current DR screening programs recommend that these patients should be screened annually or biennially. Recently, a prediction model for DR was developed. Combined with the appropriately chosen STR risk margin this resulted in personalized screening intervals. We examined the cost-effectiveness of using this personalized model compared with annual screening and the most recent Dutch guideline algorithm. Methods: DR screening intervals were determined for STR risk margins ranging from 0.0% to 4.0%, for each individual. Observational data (1998-2017) of the Diabetes Care System, a cohort of people with type 2 diabetes, were used (N=5,514). Missed cases were determined by comparing model based screening to observed grades of DR, then, costs were calculated based on screening and travel costs. Real time to develop STR was missing for 22 percent of STR cases, for these an optimistic and a pessimistic scenario were assumed. Finally, savings per missed case were determined, as compared with annual screening for finding the best risk margin and then this model was compared with the Dutch guideline algorithm. Results: The risk margins that maximized the savings per missed case compared to annual screening were 2.7% (€12,500) and 3.0% (€15,900) for the pessimistic and optimistic scenarios, respectively. Screening patients according to the Dutch guideline led to more missed cases (N=10) as well as higher costs compared to personalized screening (C=€2.1 per patient). Conclusions: A personalized DR screening model with a risk margin of around 3.0% is cheaper and more effective than the Dutch guideline. Assuming 800,000 diabetes patients in the Netherlands, implementing this personalized model could save 8.3 to 8.8 million euros annually.
Original languageEnglish
Pages (from-to)578
Number of pages1
JournalValue in Health
Publication statusPublished - 1-Nov-2019


  • adult
  • algorithm
  • animal experiment
  • animal model
  • blindness
  • cohort analysis
  • conference abstract
  • controlled study
  • cost effectiveness analysis
  • diabetic patient
  • diabetic retinopathy
  • female
  • male
  • Netherlands
  • non insulin dependent diabetes mellitus
  • nonhuman
  • practice guideline
  • prediction
  • travel

Cite this