Inter-observer variability in the detection of coffee ground emesis

Abstract

Listen

Translate article

Background: Hematemesis often presents as dark altered blood. In the abscence of fresh blood, it is often difficult to distinguish between the coffee ground emesis and bile or altered food. The importance of accurate judgement is important, as misdiagnosis may result in unnecessary diagnostic testing or missed diagnosis. Objectives: To Compare the ability of different groups of observers to distinguish different substances in emesis including bile, blood, and food material and measure the inter-observer variability within groups with different levels of training and experience. Methods: Double blind model was used. Eight groups of 10 volunteers each, including medical and paramedical personnel at different levels of experience and training were enrolled. Eleven test tubes were used containing coffee ground vomitus, bile, bile with old blood, gastric aspirate with and without blood, and vegetable material mixed with gastric aspirate. The observers were asked to identify presence or abscence of blood in each test tube. Percent correct score, sensitivity, specificity and kappa statistics were computed for each group. Results: Nearly 30% of the observers failed to recognize coffee grounds. Further, nearly 50% of the observers failed to recognize blood when it was mixed with bile. There was moderate agreement to presence or abscence of blood among all groups (k=0.40-0.75) except among surgical and internal medicine staff physicians where the agreement was poor (k :5 .40). Gastroenterologists had highest percent correct score of 81% and highest specificityof 100%.This was statistically significant. Hemeoccult testing was 100% accurate. Conclusion: Presence of bile and food material makes assesment less reliable and makes it difficult to identify blood if it is infact present. The ability to detect blood increases with speciality training and not with general experience. Hemeoccult testing may also help to identify blood in the vomitus but its clinical application in eliminating gross blood is limited. Awareness of observerreliability is essential and the complete clinical picture needs to be kept in mind before making any decisions.