The cerebral palsy directed acyclic graph: A structural causal model of aetiology

AIM: To describe the development of the initial version of the cerebral palsy (CP) directed acyclic graph (DAG). DAGs are visual representations of causal assumptions and powerful tools for representing and communicating complex causation. METHOD: An international working group of clinicians, researchers, and individuals with lived experience developed the CP-DAG, using methods adapted from the US National Aeronautics and Space Administration's risk modelling approach. The team developed a network of nodes and causal links spanning prenatal, perinatal, neonatal, and postneonatal pathways, along with a node dictionary ('DAGtionary'). Iterative drafts were created with feedback incorporated from presentations at international meetings and workshops. RESULTS: As of August 2025, the CP-DAG includes 106 nodes and 378 links, yielding a network density of 6.8%. It spans known and hypothesized causal pathways across biological, clinical, and social domains. The accompanying DAGtionary contains concise definitions and references for each node, totalling 210 citations. INTERPRETATION: The CP-DAG provides a visual, evidence-linked representation of causal reasoning in CP. It offers a foundation for identifying knowledge gaps, directing future research to under-researched areas, and, eventually, for probabilistic modelling. This initial version is intended as a starting point. We call on the global CP disability community to contribute to its ongoing refinement.