Introduction The frequency of upper gastrointestinal cancer being diagnosed following a negative oesophagogastroduodenoscopy (OGD) up to 3 years previously in selected single unit series ranges from 6.7–14%. We have examined how often oesophageal cancer (OC) is missed at OGD in a large unselected primary care cohort and associated risk factors.

Methods All subjects with OC from a UK primary care database (The Health Improvement Network (THIN)) were studied. THIN covers over 6 million patients and is regionally and demographically representative of the UK population. A nested case-control study was performed with cases of ‘missed OC’ subjects who underwent OGD 1–5 years prior to their OC diagnosis and controls subjects who did not have OGD 1–5 years prior to their OC diagnosis. Logistic regression analysis was used to study associations with having an OGD that potentially missed OC.

Results 5354 subjects with OC were identified (3561 (67%) male, mean age 71 ± 11years) and 9187 OGD were analysed. 380 subjects (237 (62%) male, mean age 72 ± 11years) had OGD 1–5 years prior to OC diagnosis (715 OGD, mean 1.9 per subject). 266 (5.0%) subjects had OGD 1–3 years and 144 (2.7%) had OGD 3–5 years prior to OC diagnosis. 180 (47.4%) of OGD which did not diagnose OC 1–5 years previously had an abnormality detected (Oesophagus – Barrett’s 59 (15.5%), hiatus hernia 45, oesophagitis 16, gastro-oesophageal reflux disease 17, stricture 8, ulcer 7; Stomach – gastritis 25, ulcer 7, polyp 4; Duodenum – duodenitis 17, ulcer 8, coeliac disease 1). The ‘missed OC’ subjects had a total 329 primary care consultations 1 year prior to their OGD that missed OC. ‘Alarm symptoms’ were present at 139 consultations (38 anaemia, 1 abdominal mass, 78 dysphagia, 20 haematemesis/melaena and 6 weight loss). Subjects presenting with anaemia (OR 2.05, 95%CI 1.4–3.1, p = 0.0006), haematemesis/melaena (1.82, 1.1–3.2, p = 0.03) or dyspepsia (2.20, 1.7–2.8, p < 0.05) were more likely to have had an OGD that missed OC. In contrast, subjects with dysphagia (0.74, 0.5–0.9, p = 0.04) or weight loss (0.39, 0.2–0.9, p = 0.02) were less likely to have an OGD that missed OC. Logistic regression did not find any association between age (0.7, –0.5–1.8, p = 0.3) or female gender (1.2, 0.9–1.5, p = 0.07) with OGD that failed to diagnose OC 1–5 years prior to OC diagnosis.

Conclusion Missing OC at OGD was relatively uncommon in this unselected cohort with 5% of OC subjects having had an OGD 1 – 3 years prior to diagnosis that did not diagnose OC. Presenting with anaemia, haematemesis/melaena or dyspepsia were positively associated with missing OC at OGD. In contrast, subjects with dysphagia or weight loss were less likely to have OC missed at OGD. Many of the OGD which missed OC had an abnormality recorded, with Barrett’s oesophagus reported in 15.5%.

Disclosure of Interest None Declared