Prediction of visual acuity after early vitrectomy in diabetics.

In proliferative diabetic retinopathy the indication of early vitrectomy remains controversial. At present, no decision rule exists for the assessment of the various factors predicting the postoperative visual outcome. We reviewed 75 vitrectomies in 68 diabetics from our clinic. All vitrectomies were done by one surgeon. In all cases, vitrectomy was indicated because of nonclearing vitreous hemorrhage and/or fibrovascular proliferation. A linear regression model was used to identify factors correlating with the visual outcome. By means of univariate analysis, six of nine clinical variables were found to be associated with the final visual outcome. Dividing the patients into two groups according to their preoperative visual acuity (group 1 hand movement, group 2 better than hand movement), we identified two predictors that were independently associated with the postoperative visual acuity: group 1-the visual acuity of the fellow eye (P < 0.05) and rubeosis iridis (P < 0.05); group 2-the visual acuity of the fellow eye (P < 0.001) and preexisting systemic diabetic vascular disorders (P < 0.01). Based on this model, a formula was derived to predict the visual acuity at 6 months postsurgery. For easier handling the prognostic factors of a poor visual outcome (less than 0.1) were summarized in a flow chart. The test is a useful tool for the preoperative evaluation of various risk factors and, hence, for more reliable prediction of a poor visual outcome. Thus, it may be especially useful to objectify the risk-benefit ratio for early vitrectomy in diabetics.