Improving the value of screening for diabetic macular oedema using surrogate photographic markers

Diabetes is a major problem for the National Health Service. In 2004 its prevalence was estimated to be 3.2%, however this is expected to double in the next ten to fifteen years. Diabetic retinopathy affects one in three of every person diagnosed with diabetes. Experience in the UK would suggest that at least 4% of those screened are subsequently referred to ophthalmology. The main indication for referral is to confirm or refute the presence of macular oedema. The macula is the central part of the retina that gives us our central vision enabling us, for example, to read and to recognise people. Oedema is the term for accumulated fluid which, in diabetic eye disease, results from leaking blood vessels. In the United Kingdom it is the most common cause of visual loss in working age people. Screening programmes are intended to reduce sight loss from this condition by detecting it early. However, oedema is transparent and cannot be seen directly on a retinal photograph. Guidelines based on other features seen on photographs have therefore been devised. Unfortunately the existing guidelines result in many people having to attend hospital unnecessarily, creating anxiety and reducing the time available for those who actually have oedema. Optical coherence tomography is the best method doctors have for seeing this fluid. Although a rapid and safe test, it is expensive compared with normal photography. Our aims are (1) to determine how effective existing photographic referral criteria are compared with optical coherence tomography, and (2) to see whether computer analysis of photographic features improves the effectiveness of photography.