OBJECTIVE: The aim of this study was to validate the endoscopic findings against histologic features of the distal esophageal mucosa for the diagnosis of reflux esophagitis in infants. METHODS: The data records of 167 patients (88 M; 79F) aged 38-364 days, referred for investigation of reflux esophagitis, between January 1995 and December 2000 were retrospectively reviewed. The association between nominal (presence or absence of esophagitis) and ordinal (grades of esophagitis) variables was analyzed through a correlation between the results of endoscopic findings and histology. RESULTS: Endoscopy when compared with histologic analysis had a sensitivity of 45%; specificity of 71%; positive and negative predictive value of 89% and 21% respectively; and accuracy of 50%. Additionally, this study demonstrated that there was a poor correlation between endoscopic and histologic findings when endoscopy was normal or when endoscopic grade I esophagitis was observed (p = 0.10). Normal esophageal appearance failed to identify 79.2% of patients with histologic esophagitis. Conversely, amongst patients with endoscopic grade I esophagitis, 12.1% had normal histology. CONCLUSIONS: We concluded that whilst endoscopy had a specificity of 71%, it did not attain an acceptable range of sensitivity (45%) to justify performing an endoscopy without biopsy, as many true cases of esophagitis would not be detected; and that the presence of grade I (non-erosive) esophagitis at endoscopy did not increase the value of the test in predicting histologic abnormality.
Reflux esophagitis; endoscopy; histology; diagnosis
