In the 1970s Engerman et al showed that strict control of diabetes could prevent the development of retinopathy in diabetic dogs. 1 When close monitoring of glycaemia and methods of giving frequent insulin injections became possible in routine practice clinicians had great expectations that similarly strict control in humans could prevent the development of diabetic retinopathy and arrest or even reverse early lesions. These expectations were dashed, however, when it became clear that tightening control in patients with existing retinopathy could make the lesions worse. 2 This normoglycaemic re-entry phenomenon has puzzled clinicians, 3 and, although there is still much to learn about its mechanism, it is becoming clearer how to manage it.

Early findings that patients with varying degrees of retinopathy suffered a paradoxical worsening of their lesions as their diabetic control improved 2 3 4 5 6 have been confirmed in a recent large study of non-insulin dependent diabetics started on insulin. 7 But even before then the Diabetes Control and Complications Trial had shown beyond doubt that this early deterioriation of retinopathy at the time of normoglycaemic re-entry was a true finding. 8 In this study 1441 patients with insulin dependent diabetes were randomly allocated either to continue with their conventional treatment (control group) or to change to intensive treatment aimed at maintaining near normal glycated haemoglobin concentrations. During the first year of the study retinopathy deteriorated by three or more levels in only 2% of controls but 5% of intensively treated patients. Moreover, the deterioration was most marked in those with more advanced retinopathy. 9 10