Xuxi Zhang

2.09) 186 fewer to 182 more) DR Worsening: Follow ‐ up 2 ‐ 5 years (follow ‐ up 4.3 to 5 years) 3 randomised trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 87/414 (21%) 115/407 (28.3%) OR 0.68 (0.49 to 0.94) 71 fewer per 1000 (from 12 fewer to 121 fewer)  HIGH IMPORTANT DR Worsening: Follow ‐ up>5 years (follow ‐ up 6 to 7.5 years) 2 randomised trials serious 6 no serious inconsistency no serious indirectness serious 3 none 44/155 (28.4%) 49/101 (48.5%) OR 0.41 (0.24 to 0.69) 206 fewer per 1000 (from 91 fewer to 301 fewer)  LOW IMPORTANT 1 Three studies had high risk of performance bias [Kumamoto, 1995; Tovi, 1998; VA CSDM, 1996] 2 7 RCTs 3 Only two studies 4 Only one study 5 Two studies had high risk of performance bias [Tovi, 1998; VA CSDM, 1996] 6 One study had high risk of performance bias [Kumamoto, 1995] DR, diabetic retinopathy 198 Chapter 7