Background Vitamin D levels have been shown to be associated with diabetic retinopathy, however to date, no review has examined the relationship between vitamin D and sight threatening diabetic retinopathy (STDR) and non-sight threatening diabetic retinopathy (NSTDR). The aim of this review, therefore, was to pool associations between vitamin D deficiency (25(OH)D < 20 ng/mL) and STDR/NSTDR. A further aim was to examine associations between circulating 25(OH)D levels and STDR/NSTDR. Methods A systematic review of major databases was undertaken for studies published from inception to 22/04/2022, using a pre-published protocol. Studies reporting prevalence of STDR or NSTDR versus a control group with diabetes and no DR or DME and either (a) vitamin D deficiency prevalence, or (b) circulating 25(OH)D levels, were included. A random effects meta-analysis was undertaken. Results Following screening, 12 studies (n = 9057) were included in the meta-analysis. STDR was significantly associated with vitamin D deficiency (OR = 1.80 95%CI 1.40–2.30; p = <0.001), whereas NSTDR was not (OR = 1.07 95%CI 0.90–1.27; p = 0.48). Both conclusions were graded as low credibility of evidence. Furthermore, circulating 25(OH)D levels were significantly associated with both NSTDR (SMD = -0.27 95%CI -0.50; −0.04; p = 0.02) and STDR (SMD = −0.49 95%CI -0.90; −0.07; p = 0.02), although these were graded as low credibility of evidence. Conclusion Vitamin D deficiency is significantly associated with STDR (including DME), but not with NSTDR. Given the well-reported associations between vitamin D deficiency and other unfavourable outcomes, it is important that vitamin D deficiency is managed appropriately and in a timely manner to reduce the risk of blindness in people with diabetes.

Background: Diabetes mellitus can cause several long-term macrovascular and microvascular complications including nephropathy, neuropathy, and retinopathy (DR). Several studies have reported positive associations between eating pathologies and DR; however, these studies have not been aggregated and subgrouped into type of pathological eating behaviour, and the differences in risk according to type of eating behaviour is unknown. The aim of this review, therefore, was to aggregate risks of DR in populations with and without pathological eating behaviours, stratified according to eating behaviour. Methods:A systematic review and meta-analysis was conducted. Major databases and grey literature were search from inception until 1/6/2021. Studies reporting the prevalence of pathological eating behaviours (against a control group with no pathological eating behaviours) in diabetic people with and without DR were included. Odds ratios were calculated from primary data.Results: Seven studies with eight independent outcomes with a total of 1162 participants were included. The odds ratio of DR in the total pooled analysis was 2.94 (95%CI 1.86-4.64; p=<0.001; I 2 =29.59). Two types of eating behaviour yielded enough data for sub-group analysis. Eating disorder not otherwise specified yielded an odds ratio of 2.73 (95%CI 1.81-4.10; p=<0.001; I 2 =0.00), and binge eating disorder yielded an non-significant odds ratio of 0.92 (95%CI 0.31-2.77; p=0.887;I 2 =0.00). Discussion:The likelihood of DR increases almost three times in the presence of pathological eating behaviours. More studies are required to confirm this in clinical populations stratified by eating disorder. Practitioners working with people with diabetes should closely monitor eating behaviours to preclude this risk.