Introduction Upper gastrointestinal bleeding (UGIB) is a common reason for admission and “coffee ground vomiting” (CGV) has classically been considered a sign but clinical experience hasn’t always borne this out. There is a paucity of data concerning endoscopic findings and outcomes in patients presenting with CGV. The aim of this study was to analyse endoscopic yield and outcomes at 30 days in patients presenting with CGV alone compared to those with haematemesis or melaena.

Methods The Aberdeen bleeding unit opened in 1991 and has recorded demographics, presenting symptoms, endoscopic diagnosis and outcomes on all admissions. Endoscopic diagnostic groups were varices, gastric ulcers (GU) or duodenal ulcers (DU), cancer, trivial (e.g. Mallory-Weiss tear (MWT)), no source found, no bleed and colonic. Analysis was performed over the period 1991 to 2005 and three groups identified; CGV alone (group A), haematemesis (group B) and melaena (group C). Endoscopic diagnosis, rebleeding and mortality rate were calculated and using group A as reference, odds ratios calculated (shown in brackets).

Results 6105 patients were admitted over the study period with suspected UGIB (A, n = 1708 (923 males), B, n = 1663 (968 males) and C, n = 2734 (1640 males)). Trivial, no source found and no bleed diagnoses were found in 1390 (75%) group A, 1291 (50%) group B and 1130 (40%) group C. Group B was younger than groups A and C (mean age 50.4 vs. 64.6 and 64.8 respectively, p < 0.001) therefore groups were stratified into <50 or ≥ 50. In < 50, group B had significantly more varices (OR 2.3) and MWT (OR 2.8) whereas group C had significantly more GU (OR 2.6), DU (OR 9.9) and colonic bleeds (OR 6.9). Only 4 cases had upper GI cancer all presented with melaena. 30 day mortality (<50) for groups A, B and C was 2.7%, 1.6% and 1.9% but rebleeding was significantly higher in group C (OR 3.5). In ≥ 50, group B had significantly more varices (OR 6.9), GU (OR 1.8), gastric cancer (OR 2.4), oesophageal cancer (OR 3.2) and MWT (OR 3.1) whereas group C had significantly more varices (OR 2.9), GU (OR 2.5), DU (OR 3.7), gastric cancer (OR 2.0), and colonic bleeds (OR 6.1). 30 day mortality (≥50) for groups A, B and C was 11.8%, 12.5% and 11.1% but rebleeding was significantly higher in groups B and C (OR 5.3 and OR 5.0 respectively).

Conclusion Presenting with CGV is associated with the same mortality as haematemesis or malaena but has significantly lower endoscopic yield and rebleeding suggesting a non-gastrointestinal cause. CGV should not be synonymous with UGIB and needs to be considered for investigations other than endoscopy.

Disclosure of Interest None Declared.