Data and code for cross national reproducibility of atomoxetine pharmacovigilance signals across FAERS Canada Vigilance and JADER

Supporting data and executed analysis code for the revision of the manuscript "Cross National Reproducibility of Atomoxetine Pharmacovigilance Signals Across FAERS Canada Vigilance and JADER" (PONE-D-26-24370R1). The materials document report-level backgrounds, atomoxetine report selection, reaction terms, contingency counts, drug-role and calendar-period comparisons, numerical Gamma-Poisson Shrinker fitting and posterior validation, cross-database signal inventories, Table 1 profiles and figure source values. FAERS covers 87 selected quarterly sources from 2004Q1 through 2025Q3. The reconstructed FAERS background contains 19,682,792 reports; primary-suspect and expanded primary/secondary-suspect atomoxetine cohorts contain 18,228 and 21,189 reports. Canada Vigilance and JADER backgrounds contain 1,026,986 and 1,002,922 reports, with 767 and 290 exposed reports, respectively. The four-criterion screen includes GPS EB05>2. The Canada/JADER deposit contains the six complete processed DEMO, DRUG and REAC exports used for reconstruction. Four separately named FAERS parquet files provide the complete zero-inclusive weighted N/E likelihood inputs, preserving the audited source sizes and SHA-256 values. These are fitting inputs, not the 87 original FAERS ZIP archives. The supporting-data packages provide the corresponding code, selected rosters, counts, accepted parameter estimates, model checks and plotting inputs. File identities and reproduction paths are documented in the accompanying manifests and README. Source provenance limitations are retained. Canada selection uses latest received date from 2004-01-01 through 2025-09-30. JADER uses stored period codes from 200401 through 202503, with observed maximum 202502; an independent local calendar-conversion record is unavailable. Original official source-download dates and official MedDRA release provenance are not verified. No official MedDRA dictionary is supplied. The files are author-processed research extracts, not official regulator distributions. The source databases and their terms must be acknowledged. Results are hypothesis-generating reporting associations and do not establish incidence or causality. AI assistance: ChatGPT/Codex assisted with code development, debugging, numerical checks, plotting and revision. Executed code, input identities and validation outputs are retained; authors remain responsible for interpretation. The stage scripts retain their original path assumptions and are not a newly validated one-command installer. First author: Linhong Yang. Corresponding author: Jian Yang. FundingHunan Provincial Science and Technology Program (2018SK51813 to Jian Yang; https://kjt.hunan.gov.cn/).Shaoyang City Science and Technology Program (2019GZ66 to Jian Yang; https://kjj.shaoyang.gov.cn/).Shaoyang City Science and Technology Program (2024CX6159 to Zhengjie Xue; https://kjj.shaoyang.gov.cn/).National Health Commission of the People’s Republic of China—Institute of Hospital Administration Research Fund on Chronic Disease Prevention and Control Management Models in Medical Institutions (YM2024ZD032 to Pengcheng Zhu; https://www.niha.org.cn/).Natural Science Foundation Program (Medical and Health Industry Joint Fund) administered by the Hunan Provincial Natural Science Foundation Committee Office (2025JJ81114 to Jin Luo; https://kjt.hunan.gov.cn/). The funders had no role in the study design; data acquisition, curation, or analysis; interpretation of the findings; manuscript drafting; or the decision to submit the work for publication.