Introduction Dysplasia in ulcerative colitis (UC) is often multifocal and flat, making it easy for significant lesions to be overlooked. Dye spraying the mucosal surface is believed to enhance visualisation of subtle mucosal abnormalities. Despite studies suggesting that chromoendoscopy could improve surveillance yields in patients with UC, the technique continues to be underutilised both in Europe and North America.

Aims To compare the diagnostic yield of dysplastic lesions in patients with ulcerative colitis undergoing surveillance colonoscopy between chromoendoscopy (CE) and standard white light (WLE).

Design Meta-analysis of randomised trials or prospective cohort studies.

Methods We searched electronic databases for full journal articles published after 1990 reporting on chromoendoscopy in patients with IBD compared to standard white light endoscopy. We hand searched the reference lists of all retrieved articles. We carried out random effects meta-analysis, checked for bias using Funnel plots and calculated Cochran's Q and the I² statistic to assess for heterogeneity.

Results A total of 5 studies involving 1220 patients provided data on a total number of lesions detected by each method. The difference in yield between CE and WLE was 14% (95% CI 8% to 20%) with an NNT of 7.1. A total of five studies involving 1124 patients provided data on the number of targeted lesions detected by each method. The difference in yield of CE over WLE was 13% (7–19%) with an NNT of 7.7. A total of four studies involving 1128 patients provided data on number of flat lesions detected by either method. The difference in yield of CE over WLE was 13% (4–22%) with an NNT of 7.7. There was significant heterogeneity between the studies (I²>50%) mainly due to the different populations of UC enrolled in the studies. Sensitivity analysis done by leaving each study out in turn did not affect the significance of the findings.

Conclusion CE is significantly better than WLE in detecting dysplasia in patients with UC. This holds true for all lesions, number of targeted lesions and number of flat lesions detected. An average of 7–8 patients needs to have CE in order to detect an additional dysplastic lesion.