Author: Dewi Fathin Romdhoniyyah (United Kingdom)
Co-authors: Nicholas Beare, Simon Harding, Christopher Cheyne, Marta Garcia-Finana, Ayesh Alshukri, David Parry
Purpose
Age-related macular degeneration (AMD) is the leading cause of certified visual loss in England. Diabetes mellitus and diabetic retinopathy (DR) have been inconsistently related to the risk of AMD. The UK has universal photographic screening programmes for DR enabling the retrospective study of AMD in people with diabetes. In this study, we would aim to study the prevalence and the type of AMD in people with diabetes attending a DR screening program, with a view to identifying any associations between AMD and DR.
Setting/Venue
Retrospective, observational study in Liverpool, United Kingdom, utilising fundus photographs routinely collected by the Liverpool Diabetic Eye Screening Programme (LDESP). Patients are enrolled into this programme when their general practitioner makes the diagnosis of diabetes. We utilised patient data collected by the Individualised Screening for Diabetic Retinopathy (ISDR) Cohort Study, which included patients from LDESP unless they opted out of ISDR.
Methods
We randomly selected 1,500 subjects of 10,336 people aged ≥50 years who attended the LDESP in 2011 and were enrolled in the ISDR Cohort Study. One researcher (DFR) graded the retinal images taken from the LDESP database for AMD based on the Age-Related Eye Disease Study grading system. We modified the classification by including the non-central geographic atrophy into level 4 (late AMD) in our study. In the analysis, we used the eye with the worst AMD grading result. We randomly selected 10% of the images to be graded for AMD by a reading centre senior grader to assess the quality control. DR grading was performed for routine screening by trained accredited graders within LDESP following published national standards. For the purpose of this analysis, DR was categorised as: (1) no DR (R0M0 in both eyes or only assessable eye); (2) background/mild non-proliferative DR (NPDR) includes R1M0 in both eyes/R1M0_R0M0/R1M0 in one assessable eye; (3) sight-threatening DR (STDR) includes R2 (pre-proliferative/moderate/severe NPDR), R3 (proliferative), M1 (maculopathy), and P (photocoagulation). Multiple logistic regression with multiple imputation was used for data analysis, and some variables were discounted due to high proportion of missing values (≥20%).
Results
Gradable photographs were obtained in 1,492 subjects including 374 with background DR and 40 with STDR. Early, intermediate, and late AMD was seen in as follows: 221 (14.8%), 54 (3.6%), and 13 (0.9%) of all subjects; 166 (15.6%), 33 (3.1%), and 11 (1.0%) of those with no DR; 50 (13.4%), 17 (4.5%), and 1 (0.3%) of those with background DR; 4 (10.0%), 2 (5.0%), and 1 (2.5%) of those with STDR. There was no statistically significant association between the binary presence of DR and presence of AMD (p=0.5, chi-squared test). In multiple logistic regression analysis, and after adjusting for possible confounding factors, prevalence of AMD increased with increasing age (OR = 1.06; 95%CI: 1.04–1.08, p<0.001) as expected. No significant association was identified between any AMD and the presence of any DR (OR = 0.87; 95%CI: 0.63–1.19, p=0.4) or diabetes control (OR = 0.99; 95%CI: 0.98 - 1.01, p=0.4). The interobserver agreement for AMD severity level of the right eye (kappa=0.86) was “almost perfect agreement” and “substantial” for the left eye (kappa=0.76).
Conlusions
We found that detecting and grading AMD during routine DR screening is feasible and reliable, indicated by good interobserver agreement in our study. This could provide a model for adding AMD to DR screening programmes should a treatment for dry AMD be identified. We did not find any evidence that DR is a protective, or risk factor for AMD. This suggests the mechanisms of these prevalent retinal diseases are not shared.
Financial Disclosure
None
Comments
-