CHEST 2026 asthma abstract

Expansion of the INSIGHTS Asthma Pragmatic Registry to a Nationally Representative Population: Delivering Rich, Reliable Real-World Data at Scale

Download the poster

Purpose
Prior publications have described high-accuracy phenotyping in asthma and its application within the INSIGHTS Asthma Pragmatic Registry, conducted across multiple US-based health systems. This program was expanded to include a broader and more representative population to support a more comprehensive understanding of the asthma patient journey. The expansion will enable robust evaluation of real-world outcomes and comparative effectiveness, informing opportunities to improve patient care. The objective was to expand the scale of the INSIGHTS Asthma Pragmatic Registry while maintaining data richness and reliability through application of the Verantos Evidence Platform to the Optum Market Clarity data, including unstructured clinical notes.

Methods
Eligible patients included individuals aged 12 or older who were diagnosed with or were being treated for moderate-to-severe asthma and were drawn from a source population of more than 23 million asthma (of any severity) patients. Applying the Verantos Evidence Platform, curation and linkage across multiple data source types (electronic health records [EHR], administrative healthcare claims, and death registry data) were used in combination with data reliability measurement and optimization. Data reliability was measured as accuracy (correct capture of data), completeness (availability of data), and traceability (ability to link data back to the source of truth). The F1-score, which is the harmonic mean of recall and precision, was used to measure accuracy of curated concepts against a manual reference standard derived from chart abstraction by a team of clinical annotators. Completeness was estimated as a weighted mean of available data sources during each calendar year of data. Traceability was estimated as the proportion of data elements identified in clinical source documentation. Accuracy, completeness, and traceability were expressed as percentages. All data were de-identified and the final dataset was conformed to the Observational Medical Outcomes Partnership (OMOP) Common Data Model.

Results
N=349,600 US patients with moderate-to-severe asthma were identified, with a median follow-up time of 83 months (interquartile range [IQR], 45 to 118 months). Patient characteristics were aligned with expected distributions. Patients were predominantly female (64%), white (71%), non-Hispanic (82%) and had a median age of 55 years (IQR, 40 to 66 years). Nearly a third (31%) had documentation of smoking. High data reliability was maintained with 91% accuracy (F1-score), 72% completeness, and 64% traceability. Richness was maintained through curation of unstructured electronic health record (EHR) data and linkage to traditional claims and structured EHR data for all patients. There was an average of 25,560 data elements per patient year. Richness was further expanded during this period through curation of pulmonary function tests (PFT) including result and unit, with 713,855 PFT results in the data set.

Conclusion
The INSIGHTS Asthma Pragmatic Registry was successfully expanded through collaborative efforts across organizations with complementary resources. The Pragmatic Registry population was expanded 31-fold while maintaining richness and reliability.

Clinical implications
INSIGHTS now represents the largest and one of the richest severe asthma data sets reported to date, with substantial potential to support asthma research and positively impact patient care.