Predicting the risk of diabetic retinopathy in type 2 diabetic patients

J Diabetes Complications. 2011 Sep-Oct;25(5):292-7. doi: 10.1016/j.jdiacomp.2010.12.002. Epub 2011 Feb 21.


Aims: Diabetic retinopathy (DR) is often asymptomatic even in its more advanced stages. Timely and repeated screening for DR avoids a late diagnosis of DR, but the high number of diabetic patients precludes a frequent screening; thus, the need for a method to identify patients at higher risk for DR becomes crucial.

Methods: A prospective analysis of 5034 type 2 diabetic patients followed from 1996 to 2007 and not affected by retinopathy at the time of the recruitment was performed. Patients were randomly divided (ratio 2:1) into two groups: the train data set and the test set (3327 and 1707 patients, respectively). Factors associated with the occurrence of DR were assessed by the Cox's proportional hazard model.

Results: Duration of diabetes, glycosylated hemoglobin, systolic blood Pressure, male gender, albuminuria and diabetes therapy other than diet were all significantly associated with the occurrence of DR.

Conclusions: The nomogram could help in ranking the type 2 diabetic patients at higher risk to develop DR and thus with a need for more frequent ophthalmologic checks, without enhancing neither the time nor the costs.

Publication types

  • Validation Study

MeSH terms

  • Aged
  • Albuminuria / complications
  • Artificial Intelligence
  • Decision Trees
  • Diabetes Mellitus, Type 2 / blood
  • Diabetes Mellitus, Type 2 / complications*
  • Diabetes Mellitus, Type 2 / drug therapy
  • Diabetes Mellitus, Type 2 / urine
  • Diabetic Retinopathy / epidemiology*
  • Female
  • Follow-Up Studies
  • Glycated Hemoglobin / analysis
  • Humans
  • Hypertension / complications
  • Italy / epidemiology
  • Male
  • Middle Aged
  • Nomograms
  • Proportional Hazards Models
  • Prospective Studies
  • Retrospective Studies
  • Risk Factors
  • Sex Factors
  • Vision Screening


  • Glycated Hemoglobin A
  • hemoglobin A1c protein, human