The appearance of retinopathy and progression to proliferative retinopathy: the WHO multinational study of vascular disease in diabetes

Aims/hypothesis. We aimed to estimate incidences of any retinopathy and proliferative diabetic retinopathy (PDR) by direct ophthalmoscopy and relate them to baseline risk factors in re-examined diabetic survivors from 10 centres of the WHO Multinational Study of Vascular Disease in Diabetes. Methods. After a mean follow-up of 8.4 years (11.7 years in Oklahoma), 2877 (71.6%) survivors were resubmitted to standardised direct ophthalmoscopy as at baseline. The presence of any retinopathy and PDR were recorded at each centre and their incidence estimated in those without retinopathy and PDR at baseline. The independent associations of these incidences with baseline risk factors are expressed as odds ratios derived from multiple logistic regression analyses, within individual centres (which included fasting plasma glucose in 8 and triglyceride in 5) and in pooled data. Results. Of the 4662 original patients, 465 (10.4%) of those without and 77 (43.0%) of those with baseline PDR had died (p < 0.001). Any retinopathy was newly reported at follow-up in 47.7% and PDR in 9.7% of those free of them at baseline, with reported incidences varying substantially among centres. Incident retinopathy appeared earlier in the known course of diabetes but incidence rates rose more slowly with duration in patients with Type IT (non-insulin-dependent) diabetes mellitus than in those with Type I (insulin-dependent) diabetes mellitus. In pooled data and in some individual centres, any retinopathy incidence gave significantly positive odds ratios with age, diabetes duration, systolic pressure, plasma cholesterol, BMI, insulin treatment and proteinuria, and with fasting plasma glucose in the centres where it was measured. Positive odds ratios for PDR were similarly obtained for age, duration, insulin treatment, cholesterol, proteinuria and fasting glycaemia. Smoking status odds ratios were negative for both outcomes. Conclusion/interpretation. Incidence of ophthalmoscopically ascertained any retinopathy varied about twofold and of PDR about threefold among centres. Although, in part attributable to differences between observers, variation in incidence in all centres and in some cases within centres was associated with a number of baseline risk factors. Such associations are not likely due to observer variation or selection biases and emerged despite the imprecision of clinical ophthalmoscopy. Improved detection and control of these risk factors should reduce the impact of diabetic retinopathy and its consequences.