Statistics from Altmetric.com
We are grateful to Dr Bianchi and colleagues for their interest in our study.1 They argue that ‘knowledge of local and even personal bronchoscopic practice and performance’ is necessary to determine the level of risk to the patient from the procedure and hence the degree of information that must be provided.2 This is certainly true if there is reason to believe that the risks in an institution or for an individual differ significantly from the norm—in either direction.
A database, such as that used in the Sheffield Teaching Hospitals, for recording complications following bronchoscopy is a valuable resource for auditing outcomes and quality assurance. However, one must be cautious when interpreting the absence of a serious complication in any given series. Hanley and Lippman-Hand, in a now-classic paper, described the ‘rule of three’ for such series: if none of n patients showed the event of interest, we can be 95% confident that the chance of this event is at most 3/n.3 For example, the Sheffield data showing no death with 1261 fibreoptic bronchoscopies translate into a 95% confidence limit ranging from zero to an upper limit of 1 death in 420 procedures (Clinicians may find the other implication of using CI—that occurrence of an uncommon complication is not of itself an evidence of poor performance—more comforting). The absence of an uncommon complication in a personal or an institutional series will not of itself help the clinician strike the difficult balance between providing too much and too little risk information.