We read with interest the article by Larocheet al on the cost effectiveness of performing computed tomographic (CT) scanning before fibreoptic bronchoscopy (FOB) in patients with suspected endobronchial carcinoma.1 Compared with the group in which FOB was performed blind to the result of the CT scan, in the group in which the CT result was available before FOB was performed the FOB was obviated in 24% of patients, a higher proportion of the FOBs were diagnostic, and the initial invasive investigation was more frequently diagnostic. The authors do not specify which CT appearances determined that FOB should not be undertaken.

We have recently performed a complementary study2evaluating the ability of CT scanning to predict in which patients with an abnormal chest radiograph and high clinical suspicion of bronchial carcinoma FOB is likely to provide a positive histological diagnosis. In distinction to Laroche et al, we excluded patients with pulmonary collapse but included patients with peripheral lesions.

We identified four CT features which predict a positive FOB. In descending order of accuracy these were: (1) a segmental or larger airway leading to the mass; (2) an endobronchial component; (3) situation of the mass within 4 cm of the origin of the nearest lobar bronchus; and (4) an ill defined edge to the mass. A “gestalt” assessment of the likelihood of positive FOB was slightly more accurate than any individual factor. When two or more predictive factors were present, FOB was positive in >70% of our selected series. Of interest, proximity of the lesion to the hilum per se was a poor predictor of positive FOB.

We agree with Laroche et al that CT scanning should be performed before FOB in most patients with an abnormal chest radiograph and a high clinical suspicion of bronchial carcinoma. Our study clarifies how

the CT appearances can be used to select those patients in whom FOB is unlikely to be diagnostic and should probably not be performed.