In Reply: As noted by Dr Büchner, prospective biologic assignment trials based on donor availability are not randomized comparisons. Analyzed on a donor vs no-donor basis, they have nevertheless been widely accepted as providing good-quality evidence of treatment effect, as no evidence of major bias has been identified.
Büchner identifies potential bias in the inclusion of both “no-sibling” and “no HLA-matched sibling” patients in the no-donor group. One scenario is that no-sibling patients would be immediately assigned to the no-donor group, while patients with siblings may have indeterminate status and be excluded from the analysis if they experienced an early event (eg, death) before the completion of all HLA typing. The differential exclusion of early events in patients who may indeed have had an HLA-matched sibling donor represents a potential bias. An alternative scenario is that patients with siblings who were untyped at the time of an early event may incorrectly be included in the no-donor group, introducing potential bias. However no such bias has been documented, and Büchner offers no evidence that it occurs. There are additional hypothetical biases such as lag-time bias based on family size (more siblings require more time to complete HLA typing), which may lead to differential exclusion of patients with early events or misassignment based on unknown sibling status.