How reliable is determination of ulcer size by endoscopy?

The suface areas of 23 artificial ulcers in a rubber manikin and of 35 ulcers in 35 consecutive patients admitted for endoscopy of the upper gastrointestinal tract were estimated by six endoscopists. Of the 138 estimations made in the manikin 80% underestimated the true size of the ulcer: the mean (+/- SD) was -29 +/- 40%. The largest and the smallest estimate of the same ulcer by different endoscopists varied on average by a factor of 4.5 +/- 3.8, and the estimates by the same endoscopists of ulcers with the same size varied by a factor of 2.3 +/- 0.6. In the patients the scatter of the estimates was even larger, the mean factor being 7.8 +/- 6.3. Changes in ulcer size are therefore an unsuitable criterion for assessing ulcer healing. Even if consecutive examinations are performed by the same endoscopist, changes in ulcer area smaller than by a factor of 3 are not discernible.