Abstract 17455: The Illusion of Big Data: Interim Manual Adjudication for Validation of Administrative Claims in Acute Illness Associated Atrial Fibrillation (AIAAF)
Circulation · 02.11.2026 · 154 (Suppl_1)
DOI: 10.1161/circ.154.suppl_1.17455 ↗Kurz gefasst
Fachartikel in der Zeitschrift Circulation, erschienen am 02.11.2026. Verfasst von Hari Chandana Kalangi und 9 weiteren Autor:innen.
Abstract (Originalsprache)
Background
Large-scale observational studies of AIAAF heavily rely on administrative billing claims (ICD-10 codes) to determine stroke and bleeding event rates. However, the validity of these endpoints in critically ill patients is unverified. Research and Hypothesis: We sought to quantify the accuracy of administrative event reporting compared to a clinical gold standard.
Methods
Within the SAFE AF (Stroke Prevention, Anticoagulation, Fibrillation Control, Evaluation for Reversible Causes in Atrial Fibrillation) multi-center registry, a rigorous manual adjudication protocol is underway to validate administrative claims against comprehensive electronic health record review. We report interim validation findings from 104 ICD-10 flagged randomly selected events (52 ischemic strokes and 52 major bleeding events). Events were independently evaluated by two clinical reviewers, with a third adjudicating discordant case. Inter-rater reliability was assessed via Cohen’s Kappa, and administrative positive predictive value (PPV) was calculated.
Results
Interim analysis revealed profound inaccuracy in administrative claims data. For ischemic stroke, reviewers demonstrated high agreement (Cohen’s Kappa = 0.95; 95% CI: 0.85-1.00), yet only 14 of 52 events were confirmed true, yielding a dismal administrative PPV of 26.9% (95% CI: 15.7%-41.0%). For major bleeding, reviewers had moderate agreement (Cohen’s Kappa = 0.40; 95% CI: 0.17-0.64). Upon adjudication, 33 of 52 events were confirmed true (PPV 63.5%; 95% CI: 49.0%-76.4%). Given this moderate inter-rater reliability, a novel, advanced adjudication methodology is currently being deployed to process the remaining 1,600 pending hemorrhagic events.
Conclusion
Claims-based cardiovascular literature suffers from profoundly high false-positive rates for stroke and bleeding in AIAAF populations. Unadjudicated administrative data lacks the precision required for clinical decision-making. Relying on it obscures true event rates, severely complicates risk-benefit assessments for anticoagulation, and underscores the absolute necessity of rigorous manual adjudication methodologies in acute care registries.
Zitieren
APA
Kalangi, H. C., Elashry, M., Mehta, S. M., Choulet, S., Vahora, I., Zhang, N., Butterfield, R., Schultz, A., Nelson, C., & Hozayen, S. S. (2026). Abstract 17455: The Illusion of Big Data: Interim Manual Adjudication for Validation of Administrative Claims in Acute Illness Associated Atrial Fibrillation (AIAAF). Circulation, 154(Suppl_1). https://doi.org/10.1161/circ.154.suppl_1.17455
Vancouver
Kalangi HC, Elashry M, Mehta SM, Choulet S, Vahora I, Zhang N, et al. Abstract 17455: The Illusion of Big Data: Interim Manual Adjudication for Validation of Administrative Claims in Acute Illness Associated Atrial Fibrillation (AIAAF). Circulation. 2026;154(Suppl_1). doi:10.1161/circ.154.suppl_1.17455
BibTeX
@article{Kalangi2026,
author = {Kalangi, Hari Chandana and Elashry, Marwa and Mehta, Sachin M and Choulet, Samuel and Vahora, Ilyas and Zhang, Nan and Butterfield, Richard and Schultz, Alison and Nelson, Chad and Hozayen, Sameh S},
title = {Abstract 17455: The Illusion of Big Data: Interim Manual Adjudication for Validation of Administrative Claims in Acute Illness Associated Atrial Fibrillation (AIAAF)},
journal = {Circulation},
year = {2026},
volume = {154},
number = {Suppl_1},
doi = {10.1161/circ.154.suppl_1.17455}
}
EUVIMED