Introduction Upper gastrointestinal (UGI) cancer continues to have a very poor prognosis; it tends to present late and at an advanced stage. The best hope for long term survival therefore remains early diagnosis with radical treatment. There has been increasing interest recently in measuring the accuracy of UGI endoscopy in diagnosing cancer. Depending on the population studied published missed rates vary between 3% and 20%. We hypothesised that concentrating the practice of UGI endoscopy into specialist hands would reduce the rate of missed diagnosis.

Methods This is a historical cohort study. In 2001 our institution employed an UGI nurse endoscopist and concentrated the practise of UGI endoscopy into her hands and those of the only UGI surgeon in the hospital. Rates of missed cancer diagnosis were calculated for the 7 years up to and subsequent to 2001 by cross-referencing the regional electronic endoscopy reporting system and the regional UGI cancer registry. As in other similar studies, we defined a definitely missed cancer as one diagnosed within 1 year of previous endoscopy and a possibly missed cancer as one diagnosed between 1 and 3 years of previous endoscopy. Missed diagnoses were sub-classified as being due to endoscopist error, pathologist error or follow-up error.

Results From 1994 to 2001 a total of 13 589 UGI endoscopies were performed—of a total of 305 UGI cancers diagnosed in this time 30 (10%) were missed (22 (7%) definitely missed and 8 (3%) possibly missed). From 2002 to 2009 a total of 16 503 UGI endoscopies were performed—of a total of 344 UGI cancers diagnosed in this time 20 (6%) were missed (11 (3.2%) definitely missed and 9 (2.6%) possibly missed). There was a statistically significant (p<0.05) difference between definite miss rates but between total miss rates. The difference observed between groups was mostly explained by reductions in pathology errors and follow-up errors and not by improvements in endoscopist performance.

Conclusion Missed diagnosis rates at our institution are within the ranges reported in other studies of Western populations. Performance was not significantly improved by concentrating the practice of UGI endoscopy into specialist hands.

Competing interests None declared.