Network Analysis of Headache Diagnoses Using International Classification of Headache Disorders, 3rd Edition

BACKGROUND AND OBJECTIVE: The International Classification of Headache Disorders, Third Edition (ICHD-3), significantly influences clinicians' understanding of headache disorders. In this study, we aim to elucidate how the hierarchical structure of ICHD-3 shapes the understanding of interconnectivity among headache disorders.METHODS: A network comprises elements known as "nodes," with the connections between them referred to as "edges." In our study, a node represents a headache diagnosis that meets at least one ICHD-3 diagnostic criterion of the ICHD-3. We developed two network models for ICHD-3: a non-hierarchical model, where edges are only formed by cross-references found within the text of diagnoses, and a hierarchical model that incorporates the ICHD-3's structural organization by adding extra edges between sections and their subsections. We identified the top 10 disorders in terms of their centrality, which assesses their popularity, their role as bridges in the network, and their proximity to other disorders. These measurements are calculated using the network's degree, betweenness, and closeness centrality.RESULTS: Both our models contain 387 nodes. The choice between a non-hierarchical or hierarchical model affects which diagnoses occupy the top 10 centrality nodes. In both models, migraine and medication-overuse headaches consistently rank among the top 10 diagnoses according to all three centrality metrics. The hierarchical model includes a greater number of secondary headache diagnoses among its top 10 compared to the non-hierarchical model.CONCLUSION: Migraine and medication overuse headaches are the most interconnected nodes in ICHD-3. The addition of a diagnostic hierarchy facilitates the unification of secondary headaches, which would otherwise be considered isolated, miscellaneous diagnoses. When interconnected hierarchically, these secondary headache diagnoses become the majority of the most well-connected nodes in our field.