AIM: Can manual or automated assessment of retinopathy predictfuture risk of cardiovascular disease in people with diabetes?
METHODS: This was a retrospective cohort study using data fromthe regional diabetes register linked to hospital morbidity andmortality data. Retinal photographs and routine clinical datarelated to cardiovascular risk and cardiovascular outcome datawere obtained for 13,124 people with diabetes who attended retinal screening between 2003 and 2004. The retinal photographswere also assessed using an automated system for detection ofdiabetic retinopathy that has been validated for use in screening.Patients with a history of cardiovascular disease were excludedleaving 9,232 subjects [male 52.6%, median age 61years(IQR 21),Type 1 diabetes 16.7%]; 1,038 cardiovascular events (123 deaths)were recorded over the following 5 years.
RESULTS: The adjusted risk of developing cardiovascular diseasewas higher in patients with any retinopathy by manual assessmentat baseline (HR 1.21, p<0.05) and the risk was higher in patientswith Type 1 diabetes (HR 2.24, p<0.01). The increased risk wasmaintained with the sub-categories of cardiovascular diseaseincluding coronary heart disease (HR 1.21, p<0.05) andperipheral vascular disease(HR 1.54, p<0.001) but not withcerebrovascular disease. Automated features associated with anincreased cardiovascular risk were high microaneurysm count (HR1.23, p<0.05) and high haemorrhage score (1.57, p<0.05), andthe risk was higher in people with Type 1 diabetes for highmicroaneurysm count (HR2.70, p<0.001).
CONCLUSION: In our cohort the presence of retinopathy or a highautomated microaneurysm count at retinal screening increased thefuture 5 year risk of cardiovascular disease.
