Multimodal Pharmacovigilance Needs a Hierarchy for Claims Derived from Reports, Records, and Digital Traces

Modern pharmacovigilance increasingly combines spontaneous reports, electronic health records, administrative claims, registries, regulatory documents, published literature, and patient-generated digital traces. Their coexistence creates an apparent opportunity for stronger safety inference, but multimodality alone does not determine what can legitimately be claimed. Each source observes a different part of the exposure–event process, imposes different selection and measurement mechanisms, and supports different transitions from observation to signal, corroboration, comparative estimation, causal interpretation, and regulatory action. This Evidence-Integration Perspective argues that multimodal pharmacovigilance therefore requires an explicit hierarchy of safety claims rather than an implicit hierarchy of data sources. The proposed approach separates evidence acquisition from claim strength and treats cross-source concordance as informative only after considering phenotype compatibility, exposure definition, risk-window alignment, population comparability, analytical design, source independence, and source-specific bias. Spontaneous reporting remains essential for hypothesis generation; longitudinal healthcare data can contextualize and test selected concerns; registries can provide structured clinical depth; regulatory and literature sources document assessment and action; and digital traces may supply timely patient-generated context while often lacking the conditions necessary for population-risk inference. The resulting hierarchy is intended as a disciplined framework for deciding what additional evidence is needed before a safety claim should be escalated.