Diabetic blindness is a preventable condition. Unfortunately diabetic screening and early intervention are not yet available for most of the diabetic population in the state sector in South Africa (SA). A recent review of 248 diabetic patients attending a day hospital in Cape Town revealed that only 5.2% had regular annual fundus examinations and only 10.4% were aware that annual fundoscopy was required. [1] Although screening programmes have been piloted or partially implemented in a few localised centres, [2,3] the severe deficiencies in current screening and laser management mean that large numbers of patients present with advanced ocular complications of diabetes. Management of these patients usually requires surgical intervention (vitrectomy), and visual outcomes remain unpredictable. Fifty per cent of vitrectomies performed at Groote Schuur Hospital (GSH), Cape Town, are for the complications of diabetic retinopathy, and we anticipate that the proportion may be similar at other state hospitals. In resource-limited settings it is appropriate to examine the outcomes of retinal surgery and to assess context-specific risk factors that may influence surgical outcomes, as well as decisions affecting resource allocation. We are not aware of any publications of this type from SA state facilities. Methods We performed a retrospective cohort study of consecutive patients undergoing diabetic vitrectomy at GSH from January 2012 to December 2012 with up to 6 months' follow-up. All patients were assessed and listed for surgery by four vitreoretinal consultants (one full-time and three sessional consultants) and one trainee vitreoretinal consultant in the department. Preoperative visual acuities were recorded using Snellen charts. The time from listing for surgery to the date of the operation was recorded, as well as any change in visual acuity during this period. Surgical cancellations or postponements were recorded. Routine preoperative tests included an assay for urea and electrolytes. Intraoperative details were recorded by each surgeon. Early (within 1 month) and late postoperative complications were noted. If silicone oil was inserted during the procedure, time until removal of oil was recorded. Postoperative visual acuity was assessed at 3 and 6 months, and cases were considered anatomical successes if the retina was completely attached with no features of proliferation in the absence of a tamponading agent (e.g. silicone oil). All data were collected in Epidata and transferred to STATA 12 (Stata Statistical Software: Release 12, StataCorp, USA) for analysis. Visual acuity was converted to the logarithm of the minimum angle of resolution (LogMAR) for statistical analysis. The χ 2 test or Fisher's exact test was used for categorical variables and Student's t-test and the rank sum test for continuous variables. Logistic regression was used to examine prognostic factors. The study was approved by the University of Cape Town Human Research Ethics Committee (approval number: HREC 851/2014). Results During the study peri...