Independent replication of classic trials in neurosurgery: A missing validation practice

摘要

A concerning phenomenon has been identified within the scientific literature. In multiple fields, systematic efforts to replicate peer reviewed studies have been successful in fewer than half of the originals attempted. Currently it is unclear how many influential neurosurgical trials have undergone direct replication. In this study we conducted a publication search to estimate the proportion of classic trials (>400 citations) to have undergone independent direct replication. Published replications would then be assessed for agreement with the original study findings. Our search included all journal articles in the Web of Science written in English and published between 1900 and 2019. This yielded 110 highly cited trials published in 28 medical journals. Screening of the citation index for these classic studies (113,387 articles) identified 4632 articles to be scrutinised for evidence of replication. Review of these articles did not find any self-identified direct replication studies. This apparent absence of direct replication of classic trials in neurosurgery raises questions about the strength of the evidence base for widespread neurosurgical practices. Direct replication studies must be attempted and published in order to confirm the robustness of influential findings from neurosurgical research. (C) 2020 Elsevier Ltd. All rights reserved.