Editor,—We read with interest the paper by Gascheet al (Gut1999;44:112–117) on the accuracy of transabdominal ultrasound in the detection of complications in Crohn's disease. The authors evaluated 33 patients with Crohn's disease who had resective bowel surgery. The results were impressive: 87% sensitivity with 90% specificity in the diagnosis of entero-enteric fistulas; 100% sensitivity with 92% specificity in the diagnosis of intra-abdominal abscesses; and 100% sensitivity with 91% specificity in the diagnosis of strictures. However, these data are in contrast with those reported by Maconi and colleagues1 who found very low sensitivity (50%) with 95.5% specificity in ultrasound detection of entero-enteric fistulas.

The difference in levels of sensitivity in these studies could be explained by the use of different standards and also, in our opinion, by varying definitions of fistulas. Gasche and colleagues considered fistulas to be any hypoechoic peri-intestinal lesion measuring less than 2 cm. However, although this arbitrary cut off point may be useful to differentiate between fistulas and abscesses, it does not allow for precise differentiation between fistulas and strictures, for which we usually adopt a cut off criterion of a diameter of less than 1 cm.

Strictures and abscesses are often considered to be more easily detectable by ultrasound than fistulas, but contrasting data exist even on this point. Gasche et al found 100% sensitivity for intra-abdominal abscesses, whereas in contrast Maconiet al found an overall sensitivity of 83.3%, with only 66.6% for intra-abdominal abscesses. Schwerk and colleagues2 found levels of sensitivity for parietal and intra-abdominal abscesses that were similar to those of Gasche et al, although they emphasised a lower sensitivity for ultrasound in detecting retroperitoneal and perianal lesions. We agree that abscesses located in the small pelvis or in the pararectal space are the most difficult to detect, although intra-abdominal and parietal abscesses are easily recognised.

Finally, we agree with Gasche et al on the accuracy of ultrasound in detecting strictures; they reported 100% sensitivity and 91% specificity, with bowel wall thickening of at least 3 mm. Different values of bowel wall thickening have been considered to be pathological, which is probably due to the use of different type of probes and to operator experience. Di Candio and Sheridan3 4 defined bowel wall thickness of greater than 5 mm as pathological, whereas Maconi and Schwerk1 2considered wall thickening of 4 mm or more to be abnormal. Hata and colleagues5 reported that the mean overall wall thickness of normal bowel specimens was 2.8 mm and that no normal specimens exceeded 4 mm in thickness. More recently other studies by Solvig, Van Oostayen, and even Gasche defined bowel wall thickening of 3 mm or more as pathological.6 7

Previously,8 we considered 4 mm to be the pathological value of bowel wall thickness in patients with inflammatory bowel disease, but we have now reduced this value to 3 mm or more,9 having excluded patients with ipoalbuminaemia or portal hypertension, in which bowel wall thickness is due to an oedematous imbibition. Recently, we conducted a prospective study (unpublished data) in which bowel wall thickness was shown to have a prognostic value. We found that patients with Crohn's disease with a bowel wall thickness greater than 6 mm, who are in clinical remission, showed a significantly higher relapse rate (90%) in the subsequent 18 months, when compared with patients with bowel wall thickness of less than 6 mm (40%).

In conclusion, the diagnostic accuracy of transabdominal ultrasound has improved progressively and the differences found in the literature are due principally to the introduction of new technologies, the level of experience of the operators, and the growing interest in the application of ultrasound to the study of the digestive tract.