(253) Combining COVID-19 Vaccine Data from Claims with Data from Immunization Information Systems (IIS) to Address Exposure Misclassification: An Approach Within the FDA Biologics Effectiveness and Safety (BEST) Initiative

Sunday, August 28, 2022

8:00 AM – 1:45 PM CEST

Location: Hall C

Publication Number: 1298

Presenting Author(s)

Elizabeth Bell
– Optum, Minneapolis, Minnesota, United States

Co-Author(s)

Cindy Zhou, PhD
– Food and Drug Administration, United States
Karen L. Schneider, PhD
– Managing Consultant, OptumServe, Falls Church, VA, United States
Patricia C. Lloyd
– US Food and Drug Administration, Silver Spring, Maryland, United States
Tainya C. Clarke, PhD, MPH, MSc
– Health Statistician, DHHS/FDA, Silver Spring, Maryland, United States
Grace Yang
– OptumServe, United States
Michael Wilkinson
– Optum, United States
Kandace Amend
– Epidemiologist, Optum, Bloomfield Hills, MI, United States
Emily Myers
– OptumServe, United States
Steven Anderson, PhD, MPP
– Food and Drug Administration, United States
Azadeh Shoaibi, PhD
– Associate Director Post-Market Surveillance, US Food and Drug Administration, Silver Spring, Maryland, United States
John D. Seeger, PharmD, FISPE
– Optum Epidemiology, Boston, Massachusetts, United States
Hui Lee Wong, PhD
– Associate Director for Innovation and Development, US Food and Drug Administration, Silver Spring, Maryland, United States

Background: While claims data are readily available to support COVID-19 vaccine effectiveness and safety research, it does not capture vaccination status if the vaccine administration does not generate a claim. Immunization Information Systems (IIS) collecting immunization data in each jurisdiction are an additional source of COVID-19 vaccination data that could enhance the completeness of vaccine capture. This study was conducted to assess the feasibility of linking claims data to IIS data.

Objectives: To enhance claims-based COVID-19 vaccine capture with IIS data and to estimate the residual misclassification of vaccinated individuals as unvaccinated.

Methods: U.S. individuals who were < 65 years, residing in the catchment areas of available IIS, and commercially insured by Optum’s affiliated health plans from 12/2020 through 12/2021 were included. Pharmacy and pre-adjudicated medical claims from these individuals were linked to IIS COVID-19 vaccine records. We estimated the percentage of individuals with at least one dose and with a complete COVID-19 vaccine series. The latter was defined based on manufacturer and observed sequence of vaccine records at least three days apart. To assess remaining misclassification, we used two approaches: external references and a data-driven estimate. Specifically, we compared our estimates to 1) age-standardized CDC vaccine administration estimates, 2) age-standardized state department of health (DOH) vaccine administration estimates, and 3) capture-recapture-adjusted estimates.

Results: Of the 5,295,385 people residing in the jurisdictions, at least one COVID-19 vaccine dose was observed in 1,726,512 people (33%). This increased to 48% in the linked claims-IIS data and varied widely by jurisdiction (38% to 58%). The percentage with a complete COVID-19 series increased from 24% to 42% after addition of IIS vaccine records and also varied widely by jurisdiction. IIS/claims combined vaccine capture is in general lower, as compared with the three references (12–47% lower than CDC estimates, 3–47% lower than DOH estimates, and 9–51% lower than capture-recapture estimates), and the extent of under-recording varies by jurisdiction.

Conclusions: Enhancing claims-based COVID-19 vaccine data with IIS records substantially increased the number of individuals identified as vaccinated, but misclassification remains; the extent of remaining misclassification varies by jurisdiction. These findings provide guidance for addressing misclassification of COVID-19 vaccine exposure in vaccine safety and effectiveness studies.