Related Experiment Video
Updated: Jun 29, 2025

Implementation of a Real-Time Psychosis Risk Detection and Alerting System Based on Electronic Health Records using CogStack
Published on: May 15, 2020
Using a data-driven approach to define post-COVID conditions in US electronic health record data
Kathleen M Andersen1, Farid L Khan2, Peter W Park2
1Vaccines Real World Evidence, Pfizer Inc, New York, New York, United States of America.
A new data-driven definition identified post-COVID conditions (PCC) in 20% of individuals after COVID-19 infection. This definition, based on symptom changes, captured more cases than the U09.9 code, especially in older adults.
Area of Science:
- Medical Informatics
- Public Health
- Epidemiology
Background:
- Post-COVID conditions (PCC) present a significant public health challenge.
- Existing diagnostic criteria for PCC lack standardization and may not fully capture the patient experience.
- A data-driven approach is needed to accurately define and identify PCC.
Purpose of the Study:
- To develop a data-driven definition of post-COVID conditions (PCC) using electronic health record (EHR) data.
- To quantify the incidence of PCC based on symptom changes following a COVID-19 infection.
- To compare the performance of the data-driven definition against the ICD-10-CM code U09.9 for identifying PCC.
Main Methods:
- Retrospective cohort study utilizing a de-identified EHR dataset from April 2020 to September 2021.
- Developed a "COVID symptom score" based on new diagnoses weighted by incidence ratios compared to a control group.
- Compared the data-driven PCC definition with the U09.9 code for cases diagnosed in September 2021.
Main Results:
- The cohort included 588,611 individuals with COVID-19; 20% developed PCC.
- PCC incidence increased with age, affecting 7.8% (0-17), 17.3% (18-64), and 33.3% (65+).
- The data-driven definition identified 19.0% of September 2021 cases with PCC, compared to only 2.9% identified by the U09.9 code.
Conclusions:
- Symptom-based and U09.9 code-based definitions identify distinct patient populations.
- A combined approach may be necessary for maximal capture of PCC cases.
- The findings highlight the limitations of current coding systems and the need for robust, data-driven definitions, especially prior to widespread code utilization and consensus.
Related Concept Videos
Steps in Outbreak Investigation
Statistical Methods for Analyzing Epidemiological Data
Methods of Documentation VII: EMR
Statistical Software for Data Analysis and Clinical Trials
Introduction to Epidemiology
Types of Biopharmaceutical Studies: Controlled and Non-Controlled Approaches
Non-controlled studies, commonly employed for initial exploration, lack a control group, rendering them susceptible to biases and external influences. In contrast,...

