Data Sharing Failures Limit Accuracy of Neuroimaging Meta-Analyses

Modern clinical practice relies on the synthesis of vast amounts of data to guide interventions, such as the management of atrial fibrillation or the use of intraarterial treatment within 6 hours for acute ischemic stroke [1, 2]. Systematic reviews and meta-analyses (statistical methods that combine results from multiple independent studies to reach a single conclusion) serve as the primary tools for aggregating disparate findings, including the identification of neuroimaging biomarkers for social cognitive impairments or the prediction of outcomes in neurodegenerative diseases [3, 4, 5]. However, the utility of these syntheses depends entirely on data transparency, as evidenced by meta-analyses of resting-state functional magnetic resonance imaging where findings in anxiety disorders were not robust to file-drawer effects (the statistical bias where negative or non-significant results remain unpublished, leading to an overestimation of treatment efficacy) [6]. When critical information remains inaccessible or reporting is non-standardized, the resulting evidence base may be incomplete, potentially skewing the clinical guidelines that physicians use for daily decision-making [7, 8]. A new analysis now quantifies the extent of this information gap and its impact on the reliability of medical literature.