Response to letter, ‘Immortal time bias in retrospective analysis’

We appreciate the response from Hsieh et al. 1 to our publication, 'Efficacy and safety of long-term treatment with lenalidomide and dexamethasone in patients with relapsed/refractory multiple myeloma'. As a part of our original study, we assessed the effect of lenalidomide and dexamethasone treatment on 'humoral improvement' in multiple myeloma patients, which was defined as an increase in uninvolved immunoglobulin A levels from baseline. We found that, compared with humoral nonresponders whose mean immunoglobulin A levels remained unchanged, humoral responders to lenalidomide and dexamethasone treatment showed improvements in long-term outcomes such as progression-free survival (PFS) and overall survival (OS), as determined by log-rank tests (P o 0.0001 for PFS and OS; Figures In the letter by Hsieh et al., 'Immortal time bias in retrospective analysis', 1 the group questions the use of log-rank tests (or Cox regression models using response as a timeindependent covariate), and remarks that this could have resulted in 'immortal time bias '. 3 In this context, this means that a bias towards improved outcomes may have occurred in the humoral responder patient group, dependent on the longer observation time in that group. In other words, patients who remained alive and progression-free had a higher chance of achieving a humoral response.