Objective To assess UK gastroenterology registrars' and consultants' competence in in vivo characterisation of polyps, plus the impact of a dedicated colonic lesion characterisation training intervention.

Design Prospective evaluation of a computer-based colonic lesion training module.

Setting Three UK general hospitals.

Patients High-quality endoscopic images obtained during colonoscopy in a previous study.

Interventions 30 min computer-based training module covering in vivo characterisation of colonic lesions viewed under white light, chromoendoscopy and magnification chromoendoscopy.

Main outcome measures Accuracy of characterisation of colonic lesions (hyperplastic vs adenoma vs cancer) before and after training and differences between groups (bowel cancer screening (BCS) nurses vs gastroenterology trainees vs consultant gastroenterologists).

Results Mean accuracy pretraining was 61.1%, 67.6% and 60.0% for the trainee, consultant and nurse groups respectively with no significant difference between the groups. Mean accuracy post-training improved significantly to 71.2%, 72.6% and 67.2% for the trainee, consultant and nurse groups (p<0.001 vs pretraining) with no significant difference between the three groups. Mean sensitivity and specificity improved significantly for the 15 participants overall.

Conclusions The baseline level of colonic lesion characterisation skills is limited for gastroenterology consultants and trainees and does not differ from that of non-endoscopist BCS nurses. Accuracy of lesion characterisation can be modestly improved with a brief computer-based training intervention. Lesion characterisation should become a standard part of training in colonoscopy, and should be learnt alongside technical skills for endoscope handing and therapeutic procedures.