Objective:

To identify patients meeting criteria for postacute sequelae of SARS-CoV-2 infection (PASC) using an AI-enabled electronic health record phenotyping approach.

Approach:

• Study Design: Retrospective cohort study analyzing electronic health record data from 457,950 adults with COVID-19 across 58 hospitals in 4 US regions.
• Algorithm Used: The Precision Phenotyping for Research Cohorts (P2RC) algorithm operationalized the WHO case definition for PASC, identifying symptom patterns occurring at least 3 months post-infection.
• Data Analysis: Temporal trends were assessed from 2020 to 2024, with a focus on identifying chronic conditions and excluding preexisting conditions.

Key Findings:

• The algorithm identified PASC in 16% of patients with COVID-19, significantly higher than existing diagnostic codes.
• Regional prevalence of PASC was 19% in New England, 20% in Southeast Texas, 23% in Southern California, and 14% in Western Pennsylvania.
• 89% of patients with PASC had at least one chronic condition requiring ongoing clinical management.

Interpretation:

Limitations:

• The algorithm may have underestimated PASC among patients with limited health care engagement.
• Electronic health record review validation was not conducted in all regions.
• Lack of a COVID-19–negative comparator group limits quantification of excess incidence.

Conclusion:

Sources:

• JAMA Network Open