Validation of ICD Code-Based Identification of Myocarditis Among mRNA COVID Vaccine Immunized Patients and COVID-Infected Patients
Author(s)
Chenan Zhang1, Ana-Claudia IANOS, MSc, MD2, Jill Dreyfus, MPH, PhD2, Stephen E. Schachterle, PhD2, Jonathan Johnson, MS, MBA3, Irene Margolin-Katz, BS3, Ankita Gupta, MS3, Cynthia Senerchia, MS3, Scott Kelly, PhD2.
1South San Francisco, CA, USA, 2Pfizer, New York City, NY, USA, 3Optum, Eden Prairie, MN, USA.
1South San Francisco, CA, USA, 2Pfizer, New York City, NY, USA, 3Optum, Eden Prairie, MN, USA.
Presentation Documents
OBJECTIVES: Myocarditis, a cardiovascular disease with heterogeneous clinical presentations, is often studied in real-world studies using International Classification of Diseases (ICD) codes. We examined the validity of this case identification approach of myocarditis in COVID-vaccinated and COVID-infected cohorts.
METHODS: Optum’s de-identified Market Clarity Data (2020-2023) was used to find patients with >=1 inpatient or >=2 outpatient ICD-10-CM myocarditis diagnostic codes. The two cohorts were: 1) patients with a documented mRNA COVID vaccination <=21 days before the myocarditis code and no record of COVID infection in the +/-8 weeks around the myocarditis code; and 2) patients with a record of a positive SARS-CoV-2 test and no record of COVID-19 vaccination <=8 weeks before the myocarditis code. Two clinicians reviewed medical records for myocarditis based on the Brighton Collaboration criteria, with levels 1-3 defined as validated myocarditis cases.
RESULTS: In the vaccinated cohort, 49% of patients with myocarditis ICD codes were validated by clinician review compared to 77% validated in the COVID-infected cohort. Accuracy was higher in patients <25 years old (90% vs 41%), in inpatient/emergency department (IP/ED) settings (70% vs 30%), and with <=1 Charlson Comorbidity Index (CCI) (63% vs 37%). In the COVID-infected cohort, accuracy was similar by age and CCI, but higher in IP/ED settings (80% vs 62%). No accuracy differences were noted by sex, race or ethnicity, geographical region, insurance type, or prior healthcare utilization for either cohort. Among the 24 patients in the vaccinated cohort with clinician confirmed Brighton level 5 (invalid case), 71% had ICD code D86.85 for sarcoid myocarditis.
CONCLUSIONS: The accuracy of ICD-code based identification of myocarditis cases varies by age, care setting, comorbidities, and cohort. Future directions include the development of ICD-code based algorithms using correlates of accuracy for improved myocarditis case identification in both COVID and non-COVID contexts.
METHODS: Optum’s de-identified Market Clarity Data (2020-2023) was used to find patients with >=1 inpatient or >=2 outpatient ICD-10-CM myocarditis diagnostic codes. The two cohorts were: 1) patients with a documented mRNA COVID vaccination <=21 days before the myocarditis code and no record of COVID infection in the +/-8 weeks around the myocarditis code; and 2) patients with a record of a positive SARS-CoV-2 test and no record of COVID-19 vaccination <=8 weeks before the myocarditis code. Two clinicians reviewed medical records for myocarditis based on the Brighton Collaboration criteria, with levels 1-3 defined as validated myocarditis cases.
RESULTS: In the vaccinated cohort, 49% of patients with myocarditis ICD codes were validated by clinician review compared to 77% validated in the COVID-infected cohort. Accuracy was higher in patients <25 years old (90% vs 41%), in inpatient/emergency department (IP/ED) settings (70% vs 30%), and with <=1 Charlson Comorbidity Index (CCI) (63% vs 37%). In the COVID-infected cohort, accuracy was similar by age and CCI, but higher in IP/ED settings (80% vs 62%). No accuracy differences were noted by sex, race or ethnicity, geographical region, insurance type, or prior healthcare utilization for either cohort. Among the 24 patients in the vaccinated cohort with clinician confirmed Brighton level 5 (invalid case), 71% had ICD code D86.85 for sarcoid myocarditis.
CONCLUSIONS: The accuracy of ICD-code based identification of myocarditis cases varies by age, care setting, comorbidities, and cohort. Future directions include the development of ICD-code based algorithms using correlates of accuracy for improved myocarditis case identification in both COVID and non-COVID contexts.
Conference/Value in Health Info
2025-05, ISPOR 2025, Montréal, Quebec, CA
Value in Health, Volume 28, Issue S1
Code
EPH125
Topic
Epidemiology & Public Health
Topic Subcategory
Disease Classification & Coding, Safety & Pharmacoepidemiology
Disease
SDC: Cardiovascular Disorders (including MI, Stroke, Circulatory), SDC: Infectious Disease (non-vaccine), STA: Vaccines