Adjusting for Data Observation Delay in near Real-time Safety Surveillance of COVID-19 Vaccines Using Administrative Claims Databases

Friday, August 26, 2022

6:15 PM – 6:30 PM CEST

Location: Congress Hall D4

Publication Number: 82

Background: The U.S. Food and Drug Administration's (FDA) Biologics Effectiveness and Safety (BEST) Initiative performed near real-time safety surveillance of COVID-19 vaccines by comparing observed to expected (historical comparator) rates of pre-specified adverse events of special interest (AESI) following COVID-19 vaccination in a network of large administrative claims databases. There is an inherent delay between the service date when an event is recorded and the date when the event is accessible in claims databases. Ignoring observation delay will underestimate the rate ratios (RRs) and potentially miss detection of a safety signal.

Objectives: To describe the assessment and adjustment for observation delay in near real-time safety surveillance of COVID-19 vaccines in claims databases.

Methods: We conducted sequential hypothesis testing for pre-specified AESI in pre-adjudicated Medicare Shared Systems Data (SSD) and three commercial claims databases (Optum pre-adjudicated, HealthCore adjudicated, and CVS Health adjudicated). Using fully accrued data with all claims observable from 2019 in each database, we estimated a delay distribution — the probability that an event is observed within any specified time interval after the service date — for each AESI. Delay distributions were used to adjust the expected rates for near real-time surveillance.

Results: Amount of observation delay varied by AESI and database. Overall, claims accrual was faster in pre-adjudicated claims databases: In the pre-adjudicated claims databases, a median of 58—77% of AESIs claims accrued within 4 weeks and 92—93% within 12 weeks. In the adjudicated claims databases, a median of 11% of AESI claims accrued within 4 weeks and 78—83% within 12 weeks. We observed safety signals for four AESI (acute myocardial infarction [AMI], pulmonary embolism, disseminated intravascular coagulation, and immune thrombocytopenia) following COVID-19 vaccination in the Medicare population that would have remained unidentified without adjustment for the observation delay until near complete data accrual. For instance, AMI would not have signaled early in the surveillance period without adjustment for observation delays in Medicare (delay-adjusted RR = 1.42 versus unadjusted RR= 0.67).

Conclusions: Adjustment for data observation delays facilitates timely signal detection in near real-time surveillance activities which must use partially accrued data. The variability in observation delay amount by each AESI and database requires specific adjustment for each element, and our proposed analytical framework is flexibly adaptable to accommodate various adjustments.

Presenting Author(s)

YC

Yoganand Chillarige, MPA

Research Manager
Acumen LLC
Burlingame, California, United States

Slides

Co-Author(s)

MH

Mao Hu

Acumen, LLC
Burlingame, United States
YF

Yuhui Feng

Acumen LLc, United States
XS

Xiangyu Shi

Acumen LLC
Cupertino, CA, United States
CZ

Cindy Zhou, PhD

Food and Drug Administration, United States
HW

Hui Lee Wong, PhD

Associate Director for Innovation and Development
US Food and Drug Administration
Silver Spring, Maryland, United States
YJ

Yixin Jiao

Acumen, LLC
Washington, District of Columbia, United States
Patricia C. Lloyd

US Food and Drug Administration
Silver Spring, Maryland, United States
Kandace Amend

Epidemiologist
Optum
Bloomfield Hills, MI, United States
Daniel C. Beachler, PhD, MHS

Director, Safety & Epidemiology
HealthCore Inc.
Wilmington, Delaware, United States
Alex Secora, PhD

IQVIA Government Solutions Inc.
Charlottesville, Virginia, United States
AK

Anne Marie Kline

CVS Health, United States
Tainya C. Clarke, PhD, MPH, MSc

Epidemiologist/Health Statistician
US Food and Drug Administration
Silver Spring, Maryland, United States
RO

Rachel Ogilvie

Optum Epidemiology
Boston, Massachusetts, United States
CR

Christian Reich

IQVIA, Cambridge, MA, USA
Cambridge, MA, United States
CM

Cheryl N McMahill-Walraven, PhD

Director, Safety & Surveillance
CVS Health Clinical Trial Services
Blue Bell, PA, United States
John D. Seeger, PharmD, FISPE

Optum Epidemiology
Boston, Massachusetts, United States
SA

Steven Anderson, PhD, MPP

Food and Drug Administration, United States
AS

Azadeh Shoaibi, PhD

Associate Director Post-Market Surveillance
US Food and Drug Administration
Silver Spring, Maryland, United States