Introduction Meta-analysis of published literature suggests that up to 14% of upper gastrointestinal cancer (UGIC) subjects had a negative oesophagogastroduodenoscopy (OGD) up to 3 years prior to diagnosis. We have examined how often UGIC is missed at OGD in our organisation and associated risk factors.

Methods Computerised OGD records from Sandwell General and City Hospitals between 1999 and 2007 were retrieved and submitted to the West Midlands Cancer Intelligence Unit (WMCIU) for UGIC registrations linkage. Subjects undergoing OGD 3 months to 5 years before diagnosis were identified as potentially missed UGIC cases and those with no OGD 3 months to 5 years before diagnosis served as controls. The influence of age, gender, indication, endoscopist specialty, trainee involvement, sedation use, number of biopsies taken, site and histology of UGIC on missed UGIC were examined by logistic regression analysis.

Results 36577 OGD records were submitted to WMCIU for cancer linkage and 524 UGIC were linked. 419 control UGIC: 388 (93%) diagnosed at initial OGD; 31 (7%) diagnosed at repeat OGD appropriately (within 90 days, with previous benign histology). 105 (20%) potentially missed UGIC cases: 39 (37%) oesophageal cancer (OC) and 66 (63%) gastric cancer (GC); 42 (40%) had OGD 3 months to 1 year prior to diagnosis; 33 (31%) had OGD 1–3 years prior and 30 (29%) had OGD 3–5 years prior. Furthermore, 30% of missed UGIC cases had OGD where abnormality at site of UGIC was seen but not biopsied or benign histology from an inadequate number of biopsies ( < 4) within 3 months to 1 year prior to diagnosis.

Lack of alarm symptoms (2.51, 95%CI 1.58–4.00, p = 0.0001) and female gender (1.79, 1.16–2.79, p = 0.009) were associated with missed UGIC. The number of biopsies taken was significantly lower in the missed UGIC group than in the controls (2.1 ± 0.2 vs 5.4 ± 0.1, p < 0.05).

In subjects with OC, mid-oesophageal OC appeared more likely to be missed than lower-oesophageal OC (2.04, 0.99–4.23, p = 0.05). Oesophageal squamous cell carcinoma was much more likely to be missed than oesophageal adenocarcinoma (4.47, 1.88–10.65, p < 0.001). In GC subjects, there was no association between missed UGIC and tumour site (1.10, 0.61–1.97, p = 0.3) or histology subtype (1.10, 0.46–2.67, p = 0.8).

Age (1.5, –0.8–3.8, p = 0.2), endoscopist specialty (1.39, 0.70–2.76, p = 0.34), trainee involvement (1.2, 0.78 – 1.86, p = 0.39) and sedation use (0.98, 0.64–1.51, p = 0.9) were not associated with increased risk of missing UGIC.

Conclusion Missing UGIC at OGD was seen in 14.3% of subjects within 3 years of diagnosis. It was associated with lack of alarm symptoms, female gender, oesophageal squamous cell carcinoma and an insufficient number of biopsies from recognised abnormalities.

Disclosure of Interest None Declared