Observer variation in the detection of colorectal neoplasia on double-contrast barium enema: implications for colorectal cancer screening and training.

AIM

To assess inter-observer error for the diagnosis of neoplasia on double contrast barium enema (DCBE) in the light of claims that no additional interpretative training would be needed for implementation in a national screening programme.

MATERIALS AND METHODS

10 experts, 10 consultants, and 10 experienced trainees each reported 20 DCBE studies, of which two showed cancer, three showed large polyps, four showed small polyps, and 12 were normal. Inter-observer variation was compared using odds ratios with the trainee group as reference (baseline group).

RESULTS

Experts were significantly more likely to correctly identify neoplasia on DCBE than trainees. The odds of a correct diagnosis for experts were 2.79 (95% CI 2.04, 3.81) for cancer, 2.36 (1.88, 2.97) for large polyps, and 3.50 (1.98, 6.18) for small polyps. While consultants were more likely to correctly diagnose a large polyp than trainees, 1.45 (1.15, 1.84), there was no significant difference between these two groups for the correct diagnosis of either cancer, 1.24 (0.52, 2.96), or small polyps, 1.26 (0.83, 1.90). A cancer was missed by 6 (60%) experts, 9 (90%) consultants, and 8 (80%) trainees. Large polyps were missed by 4 (40%) experts, 5 (50%) consultants, and 6 (60%) trainees. There was no significant difference between any group when false positive diagnoses were considered.

CONCLUSIONS

There is considerable inter-observer perceptive error for the diagnosis of neoplasia on DCBE. Experts performed significantly better than other observers but the overall standard of performance was poor, even amongst experts.