Recognizing the problem of delayed entry in time-to-event studies: Better late than never for clinical neuroscientists.

Abstract

One of the greatest differences I encountered in moving from being Editor-in-Chief of a basic science journal to the same position at Annals of Neurology was the much greater importance of meticulous review of the statistical treatment of data. In basic science, the conditions of an experiment can be set up by the investigator so that relatively simple statistical treatments can often be used with clear-cut results. Comparing the treatment and outcomes of human studies is much messier. Humans have a history before the study started; they have lives that often cause them to deviate from the protocol; and it is much harder to measure the outcomes because our methods have to be so much less invasive than they can be in animal studies. In addition, unlike mice or rats, which are deliberately inbred in a laboratory to minimize variation between animals, we are a wild species with enormous genetic and environmental variability. We solved the problem of the lack of statistical sophistication of the scientific editors by bringing in an expert in study design and statistical analysis, Dr Rebecca Betensky, a Professor of Biostatistics at the Harvard T. H. Chan School of Public Health, who serves as our Statistical Editor. We have weekly editorial conferences where the Associate Editors, Rebecca, and I consider which of the papers that were reviewed in the previous week should be published. Each week, it seems, we are treated to a private tutorial (sometimes more than one) on the complexities of study design and statistical analysis in the papers we are considering. I have learned a great deal at these meetings and thought that this education should be extended to our readers, and so I have prevailed upon Dr Betensky to address some recurring topics in our editorial conferences. We begin this month with the concept of delayed entry, a common problem in human studies that many investigators and reviewers fail to take into account. We have decided to publish this series under the NeuroGenesis section of the Annals, because this section is devoted to the career development of neurologists, and it seems critical to the professional judgment of every academic neurologist to assimilate the concepts in this series, both to improve our own work and to evaluate the work of other neurologists more critically. -C.B.S.