• diabetes mellitus;
  • diabetic retinopathy;
  • digital photographs;
  • maculopathy;
  • screening

Diabet. Med. 28, 741–746 (2011)


Aims  To compare agreement level and identify reasons for disagreement between grading of mydriatic digital photographs in a diabetic retinopathy screening service and hospital eye service biomicroscopy grading.

Methods  Structured examination findings leading to automatically calculated National Screening Committee grades recorded on an electronic medical record system in the hospital eye service at the first clinic visit after diabetic retinopathy screening service referral between April 2006 and November 2007 were retrospectively compared with the grade at the screening visit that prompted referral. In cases of disagreement, screening images were reviewed.

Results  Data on 452 eyes (226 patients) were analysed. For retinopathy, hospital eye service slit-lamp biomicroscopy grades were: R0 (no diabetic retinopathy) in 63 eyes; R1 (background retinopathy) in 251eyes; R2 (pre-proliferative) in 129 eyes and R3 (proliferative) in nine eyes. Diabetic retinopathy screening service grades were in agreement in 350 eyes (77.4%), showed a lower grade in 59 eyes and a higher grade in 43. Agreement was moderate (κ = 0.60). The most common reason for disagreement was overgrading of R1 by clinicians. Hospital eye service biomicroscopy maculopathy grades were: M0 (no maculopathy) in 366 eyes and M1 (maculopathy) in 86 eyes. Diabetic retinopathy screening service grades were in agreement in 327 eyes (72.3%), showed a lower grading in five eyes and a higher grade in 120 eyes. Agreement was moderate (κ = 0.41). The commonest cause for disagreement was clinicians failing to identify fine macular exudates.

Conclusions  This study of routine clinical services demonstrates moderate agreement between non-medical grading of mydriatic digital retinal photography images and hospital slit-lamp biomicroscopy grading of patients referred with diabetic retinopathy. The majority of errors in grading were attributable to errors by hospital doctors, usually in the direction of under-grading which could be a potential source of clinical risk if treatment is delayed.