Authors' Response

Abstract

To the Editor: We thank Hansen et al for their comments and agree that the arbitrary categorization of Kasai procedure caseload (<5 vs >5 cases per year) by McKiernan et al (1) does not allow for complete statistical analysis, nor can it determine the precise volume cutpoint for optimal outcomes. Categorization of continuous variables can result in loss of information, reduced statistical power, and increased risk of statistical error (2). In addition, categorization establishes arbitrary cutpoints, which can be problematic when the volume–outcome relation is not linear (as demonstrated by Hansen et al in their reevaluation of cutpoints). We suspect that surgical expertise and specialized medical therapy play a role in improving outcomes; however, to this point the biliary atresia literature has not adjusted for the strong effect of clustering. The similarities of patients and care provision in a single centre or from the same region results in statistical clustering of outcomes, and correction of this effect is crucial to reduce the likelihood of error in volume–outcome research (3). In addition, raw volume of procedures at a centre does not tell the full story; it does not account for individual surgeon volume, surgeon and other health care provider education and experience, the resources in that centre, or the associated health system capability to address postoperative complications. All of these could affect Kasai success and should be addressed in an analysis of the volume–outcome relation. We would encourage further large, well-designed epidemiologic research to determine the role of surgeon and centre volume on the outcomes of patients with biliary atresia.