To differentiate tuberculous enteritis and Crohn's disease with barium study using Bayes theory.

The study group consisted of 34 patients with tuberculous enteritis (age range 16-86 years, mean age 43.3 years, M:F=19:15) and 36 patients with Crohn's disease (age range 19-78 years, mean age 35.2 years, M:F=18:18). These diagnoses were confirmed by therapeutic tests (tuberculous enteritis: 15, Crohn's disease:16) or histopathological examinations (tuberculous enteritis: 19, Crohn's disease: 20) conducted from January 1993 to May 2003. Three radiologists (two abdominal specialists and one trainee) analyzed each radiological finding of tuberculous enteritis and Crohn's disease by means of a barium enema and/or small bowel series. We used Fisher's exact test to verify the statistical significance of each radiological finding and p-values less than 0.05 were considered to be significant. We calculated the likelihood ratio (LR) of tuberculous enteritis versus Crohn's disease for each finding by employing Bayes theory.

The radiological findings associated with a high likelihood ratio for tuberculous enteritis were the involvement of the cecum (LR=2.65) and ascending colon (LR=1.99), rigid narrowing (LR=1.94), shortening of the bowel (LR=1.99), haustral loss (LR=1.97) and sacculation (LR=3.88). The radiological findings associated with a high LR for Crohn's disease (low LR for tuberculous enteritis) were age between 20 and 29 years (LR=0.53), the involvement of the jejunum (LR=0.12), terminal ileum (LR=0.19), sigmoid colon (LR=0.30) or rectum (LR=0.17), and the presence of skip lesions (LR=0.19) or strictures (LR=0.21). With these LRs, the probability of the subject having tuberculous enteritis versus Crohn's disease could be calculated using Bayes theory.

The analysis of a barium study using Bayes theory could provide an objective, easy and fast method of differentiating tuberculous enteritis and Crohn's disease.