Document text
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 1of 194NON-INTERVENTIONAL ( NI) STUDY CONCEPT PR OTOCOL
Title Post-Emergency Use Authorization Active
Safety Surveillance Study among Individuals
in the Veteran’s Affairs Health Sy stem
Receiving Pfizer -BioNTech Coronavirus
Disease 2019 (COVID -19) Vaccine
Protocol number C4591012
Protocol version identifier Version 2.0
Date of last version of protocol 27January2021
EUPost Authori zation Study (PAS)
register numberEUPAS39779
Activesubstance COVID-19 mRNA Vaccine is single -stranded,
5’-capped messenger RNA (mRNA) produced
using a cell -free in vitro transcription from the
corresponding DNA templates, encoding the
viral spike (S) protein of SARS -CoV-2.
Medicinal product Pfizer-BioNTech COVID- 19 Vaccine
Research question and objectives Research question: what are the incidence rates
of safety events of interest (based on adverse
events of special interest[AESI]) among
individuals vaccinated with the Pfizer-
BioNTech COVID-19 vaccine within the US
Veterans Health Administration (VHA) s ystem
overall and in sub- cohorts of interest, as
compared to expected rates of those events?
Primary study objectives:
To assess whether individuals in the
VHA system experience increased risk
of safety events of interest following
receipt of the Pfizer- BioNTech
COVID-19 vaccine;
To assess whether sub- cohorts of
interest (i.e., immunocompromised,
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077341
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 2of 194elderly, individuals with specific
comorbidities, individuals receiving
only one dose of the Pfizer -BioNTech
COVID-19 vaccine, and individuals
with prior SARS -CoV-2 infection) in
the VHA s ystem experience increased
risk of safet y events of interest
following receipt of the Pfizer-
BioNTech COVID -19 vaccine.
Secondary study objective :
To characterize utili zation patterns of
the Pfizer -BioNTech COVID-19
vaccine among individuals within the
VHA, including estimating the
proportion of individuals receiving
vaccine, 2 -dose vaccine completion
rate, and distribution of time gaps
between the first and second dose,
demographics and health histories of
recipients, overall and among the sub-
cohorts of interest.
Authors Yinong Young- Xu, ScD, MA, MS
Director, Clinical Epidemiology Program
Veterans Affairs Medical Center
White River Junction, VT
Cynthia de Luise, PhD, MPH
Senior Epidemiologist/ Safety Surveillance
Research Scientist ; Risk Management and
Safety Surveillance Research
Pfizer, Inc.
New York, NY
Mei Sheng Duh, ScD, MPH
Managing Principal and Chief Epidemiologist
Analysis Group, Inc.
Boston, MA
Thisdocument contains confidential information belonging to Pfizer. Except as otherw ise agreed to in w riting,
by accepting or reviewing this document, you agree to hold this information in confidence and not copy or
disclose it to others (except where required by applicable law) or use it for unauthorized purposes. In the event
of any actual or suspected breach of this obligation, Pfizer must be promptly notified.
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077342
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 3of 1941.TABLE OF CONTENTS
1.TABLE OF CONTENTS.......................................................................................................3
2. LIST OF ABBREVIAT IONS................................ ................................ ................................ 5
3. RESPONSIBLE PARTI ES................................ ................................ ................................ ....8
4.ABSTRACT ................................ ................................ ................................ ........................... 9
5.AMENDMENTS AND UP DATES................................ ................................ ..................... 23
6.MILESTONES ................................ ................................ ................................ ..................... 29
7. RATIONALE AND BAC KGROUND ................................ ................................ ................ 30
8. RESEARCH QUESTION AND OBJECTI VES................................ ................................ .31
9. RESEARCH METHODS................................ ................................ ................................ ....32
9.1. Study Design................................ ................................ ................................ ...........32
9.1.1. Self -Controlled Risk I nterval (SCRI) Design with Post -Vaccination
Control Interval................................ ................................ ................................ 32
9.1.2. Active Comparator Design ................................ ................................ .........35
9.1.3. Additional Study Designs in the Signal Evaluation Phase ......................... 36
9.1.4. Study Period................................ ................................ ................................ 36
9.2. Setting ................................ ................................ ................................ ...................... 36
9.2.1. Inclusion Criteria ................................ ................................ ........................ 36
9.2.2. Exclusion criteria ................................ ................................ ........................ 36
9.2.3. Subgroups ................................ ................................ ................................ ...36
9.3. Variables ................................ ................................ ................................ .................. 38
9.3.1. Exposure of I nterest................................ ................................ .................... 38
9.3.1.1. Pfizer -BioNTech COVID- 19 Vaccine Groups of Interest ........38
9.3.2. Baseline Characteristics ................................ ................................ ..............39
9.3.3. Outcomes ................................ ................................ ................................ ....40
9.4. Data Source................................ ................................ ................................ .............47
9.5. Study Size................................ ................................ ................................ ................ 48
9.5.1. Power ................................ ................................ ................................ ..........48
9.6. Data Management ................................ ................................ ................................ ...51
9.6.1. Case report forms (CRFs)/Electronic data record ................................ ......51
9.6.2. Recor d retention ................................ ................................ .......................... 51
9.7. Data Anal ysis................................ ................................ ................................ ..........52
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077343
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 4of 1949.7.1. Baseline Characteristics ................................ ................................ ..............52
9.7.2. Vaccine Utilization Patterns ................................ ................................ .......53
9.7.3. Safet y Signal Analyses ................................ ................................ ...............53
9.7.3.1. Signal Detection ................................ ................................ ........54
9.7.3.2. Signal Evaluation ................................ ................................ ......58
9.7.3.3. Signal Verification ................................ ................................ ....61
9.7.4. Seasonality -Adjusted Cases -Centered Method ................................ ...........62
9.7.5. End-of- Season and End -of-Surveillance Analy ses................................ .....62
9.7.6. Subgroup Analy sis................................ ................................ ...................... 63
9.7.7. Incidence Rates and Time to Safety Event of Interest Anal ysis................. 64
9.7.8. Prioritized Safety Analysis of Myocarditis/Pericarditis ............................. 64
9.8. Quality Control................................ ................................ ................................ ........65
9.9. Strengths and Limitations of the Research Methods ................................ ...............66
9.10. Other Aspects ................................ ................................ ................................ ........67
10. PROTECTI ON OF HU MAN SUBJECTS ................................ ................................ ........68
10.1. Patient I nformation ................................ ................................ ................................ 68
10.2. Patient Consent ................................ ................................ ................................ ......68
10.3. Institutional Review board (I RB)/Independent Ethics Committee (I EC).............68
10.4. Ethical Conduct of the Study ................................ ................................ ................ 68
11. MANAGEMENT AND R EPORTING OF ADVERSE EVENTS/ADVERSE
REACTIONS................................ ................................ ................................ ...................... 69
12. PLANS FOR DI SSEMINATING AND COMMUNI CATING STUDY RESUL TS........70
13. REFERENCES ................................ ................................ ................................ .................. 71
14. LIST OF TABLES ................................ ................................ ................................ .............76
15. LIST OF FIGURES ................................ ................................ ................................ ...........76
16. ANNEX 1. LIST OF STAND ALONE DOCUMEN TS................................ ................... 76
17. ANNEX 2. ENCEPP CHECKLIST FOR STUDY PROTOCOL S................................ ...76
18. ANNEX 3. ADDITIO NAL INFORMATION ................................ ................................ ...77
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077344
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 5of 1942. LIST OF ABBREVIATIONS
Abbreviation Definition
ACIP Advisory Committee on Immunization Practices
ACOS Associate Chief of Staff
AE Adverse event
AEM Adverse event monitoring
AESI Adverse event of special interest
AIDS Acquired immunodeficiency syndrome
AMI Acute myocardial infarction
BMI Body mass index
CAD Coronary artery disease
CBER Center for Biologics Evaluation and Research
CI Confidence Interval
CCI Charlson comorbidity index
CDC Centers for Disease Control and Prevention
CDW Corporate Data Warehouse
CEP Clinical Epidemiology Program
CMA Conditional Marketing Authorization
CMS Centers for Medicare & Medicaid Services
COPD Chronic obstructive pulmonary disease
COVID-19 Coronavirus Disease 2019
CPT Current Procedural Terminology
CRADA Cooperative Research and Data Agreement
CRFs Case report forms
DIC Disseminated intravascular coagulation
DVT Deep vein thrombosis
Tdap Diphtheria, tetanusand (acellular) pertussis
Td Diphtheria and tetanus
ED Emergency department
EMA European Medicines Agency
EMR Electronic medical records
EU European Union
EUA Emergency Use Authorization
EU PAS European Union Post -Authorization Safety
FDA Food and Drug Administration
GBS Guillain-Barré syndrome
GEP Good Epidemiological Practice
GPP Good Pharmacoepidemiology Practices
H0 Null hypothesis
Ha Alternative h ypothesis
HBV Hepatitis B virus
HCPCS Healthcare Common Procedure Coding S ystem
HCV Hepatitis C virus
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077345
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 6of 194Abbreviation Definition
HIV Human immunodeficiency virus
HPV Human papillomavirus
ICD-10-CM International Classification of Diseases, Tenth Revision, Clinical
Modification
ICD-10-PCS International Classification of Diseases, Tenth Revision, Procedure
Coding Sy stem
IEA International Epidemiological Association
IEC Independent Ethics Committee
IPTW Inverse probability of treatment weighting
IQR Interquartile range
IRB Institutional Review Board
KD Kawasaki disease
LLR Log-likelihood ratio
MaxSPRT Maximized sequential probability ratio test
MenACWY Meningococcal conjugate
MenB Serogroup B meningococcal
MIS-A Multisystem inflammatory syndromein adults
mRNA Messenger RiboNucleic Acid
MS Multiple sclerosis
NDC National Drug Codes
NIS Non-interventional study
NNERC VAMC Northern New England Research Consortium VA Medical Centers
NSAID Non-steroidal anti -inflammatory drug
ON Optic neuritis
PASS Post-Authorization Safety Study
PE Pulmonary embolism
PRISM Post-Licensure Rapid Immunization Safety Monitoring
PS Propensity score
R&D Research and Development
RCA Rapid cycle analysis
RR Relative risk
SAP Statistical analy sis plan
SARS-CoV-2 Severe acute respiratory syndrome coronavirus 2
SAS SAS Institute
SCCS Self-controlled case series
SCRI Self-controlled risk interval
SD Standard deviation
SJS Stevens-Johnson sy ndrome
SPEAC Safety Platform for Emergency vACcines
SRSS Subcommittee on Research Safet y and Securit y
TEN Toxic epidermal necrol ysis
TM Transverse m yelitis
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077346
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 7of 194Abbreviation Definition
TTS Thrombosis with thrombocy topenia sy ndrome
UK United Kingdom
US United States
VA Department of Veterans Affairs
VAIRRS VA Innovation and Research Review Sy stem
VAERS Vaccine Adverse Event Reporting S ystem
VHA Veterans Health Administration
VINCI VA Informatics and Computing Infrastructure
VINNE Veteran’s IRB of Northern New England
VISN Veterans Integrated Service Networks
VSD Vaccine Safet y Datalink
VTE Venous thromboembolism
WHO World Health Organization
WOC Without compensation
YRR Your Reporting Responsibilities
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077347
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 8of 1943.RESPONSIBLE PARTIES
Name, degree(s) Job Title Affiliation Address
Principal Investigator:
Yinong Young -Xu,
ScD, MA, MSDirector, Clinical Epidemiology
ProgramVeterans Affairs
(VA) Medical
Center163 Veterans Drive,
White River Junction,
VT 05009
Cynthia de Luise,
PhD, MPHSenior Epidemiologist /Safety
Surveillance Research Scientist ;
Risk Management and Safety
Surveillance ResearchPfizer, Inc. 235 East 42ndStreet,
New York, NY 10017
Mei Sheng Duh,
ScD, MPHManaging Principal and Chief
Epidemiologist
Visiting Scientist, Department of
BiostatisticsAnalysis Group,
Inc.
Harvard T. H.
Chan School of
Public Health111 Huntington Ave
14thFloor
Boston, MA 02199
677 Huntington Ave
Boston, MA 02115
Maral DerSarkissian,
PhD Vice President and Senior
Epidemiologist
Adjunct Assistant ProfessorAnalysis Group,
Inc.
Fielding School of
Public Health,
University of
California, Los
Angeles333 South Hope Street
27thFloor
Los Angeles, CA
90071
650 Charles E Young
DriveSouth
Los Angeles, CA
90095
Rachel Bhak, MS Manager and Senior Biostatistician Analysis Group,
Inc.111 Huntington Ave
14thFloor
Boston, MA 02199
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077348
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 9of 1944.ABSTRACT
Title:Post-Emergency Use Authorization Active Safety Surveillance Study among
Individuals in the Veteran’s Affairs Health S ystem Receiving Pfizer -BioNTech Coronavirus
Disease 2019 (COVID -19) Vaccine
Protocol Version: 2.0; Dateof Protocol : 31 Aug2021
Authors: Yinong Young Xu, ScD, MA, MS , Veterans Affairs Medical Center; Cy nthia de
Luise, PhD, MPH, Pfizer, I nc.; Mei Sheng Duh, ScD, MPH, Anal ysis Group, I nc.
Rationale and b ackground :
In March 2020, the World Health Organization (WHO) declared a global pandemic for the
coronavirus disease 2019 (COVID -19) due to the severe acute respiratory syndrome
coronavirus 2 (SARS -CoV-2), which w as first identified by public health officials in China
in December 2019.1The COVID -19 pandemic presents an unprecedented public health
crisis. As of January 7, 2021, over 21.4million COVID -19 cases and 364,000 deaths have
been reported in the United States (US) alone.2
Pfizer and BioNTech have partnered to develop a novel messenger RiboNucleic Acid
(mRNA)vaccine against SARS -CoV-2 for the prevention of COVID -19 (Candidate
BNT162b2). Pfizer is conducting a Phase 1/2/3, randomized, placebo -controlled, observer -
blind, dose -finding, vaccine candidate- selection, and efficacy study among healthy
individua ls (NCT04368728). The Food and Drug Administration (FDA) reviewed the
available safet y data from 37,586 participants 16 years of age and older and did not identify
any specific safet y concerns. In addition, the anal ysis of available efficacy data from 36,523
participants 12 years of age and older without evidence of prior SARS- CoV-2 infection at
least 7 day s after receiving the second dose demonstrated 95% efficacy of the vaccine in the
prevention of COVID -19 (as confirmed by 8 vs. 162 COVID -19 cases in the vaccine and
placebo groups, respectively ).3,4Based on these safety and efficacy data, as well as a review
of manufacturing information regarding product quality and consistency , the FDA
determined that the known and potential benefits of the vaccine outweighed the known and
potential risks for the prevention of COVID -19 in individuals 16 y ears of age and older.4
Therefore on December 11, 2020, the Pfizer -BioNTech COVID-19 vaccine was granted an
Emergency Use Authorization (EUA) by the FDA to prevent COVID -19 in individuals 16
years of age and older.5
With respect to geographic regions other than the US, on December 2, 2020, the United
Kingdom (UK) was the first country in the world to grant temporary authorization for
emergency use of the Pfizer -BioNTech COVID -19 vaccine.6On December 21, 2020, the
European Medicines Agency (EMA) grant ed the Pfizer -BioNTech COVID- 19 vaccine a
conditional marketing authorization (CMA) for use among individuals 16 years of age and
older throughout all of the European Union’s (EU) 27 member states.7
As required b y the EUA, post-authorization observational studies using real -world data are
needed in order to assess the association between Pfizer -BioNTech COVID- 19 vaccine and
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077349
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 10of 194pre-determined safet y events of interest (including deaths , hospitalizations, and severe
COVID-19)among individuals administered the vaccine in both the population at large and
in populations of interest ( e.g., immunocompromised individuals, elderl y, and those with
specific comorbidities) .4Pfizer in collaboration with the US Veterans Health Administration
(VHA) and Anal ysis Groupherein propose post -EUA active safet y surveillance of safety
events of interest based primarily on the Priorit y List of Adverse Events of Special Interest
from the Brighton Collaboration’s Saf ety Platform for Emergency vACcines (SPEAC)
Project and from the preliminary list of safet y events of interest presented at the September
22, 2020, meeting of Centers for Disease Control and Prevention’s (CDC ’s) Advisory
Committee on I mmunization Practices (ACIP)on the enhanced safet y monitoring of
COVID-19 vaccines.8,9This safety surveillance study willidentify and evaluate rapid ,near
real-time potential safet y signals associated with the Pfizer -BioNTech COVID-19 vaccine in
the large-scale VHA electronic medical record (EMR) database. The observed safet y event of
interest rates will be compared to expected rates derived from self -controls and active
comparators receiving seasonal influenza vaccination. Part of the methodologies used in this
study are constructed based on approaches previously used by the Post -Licensure Rapid
Immunization Safet y Monitoring (PRI SM) program for the H1N1 vaccine .10This non-
interventi onal study is designated as a Post -Authorization Safety Study (PASS) commitment
to the US FDA and is a Category 3 commitment in the EU Risk Management Plan .
Research question and objectives :
Research question: what arethe incidence rates of safet y events of interest (based on adverse
events of special interest [AESI ])among individuals vaccinated with the Pfizer-BioNTech
COVID-19 vaccine within the US VHA system overall and in sub -cohorts of interest, as
compared to ex pected rates of those events?
Primary study objectives:
To assess whether individuals in the VHA s ystem experience increased risk of safet y
events of interest following receipt of the Pfizer- BioNTech COVID -19 vaccine;
To assess whether sub- cohorts of inter est (i.e., immunocompromised, elderly ,
individuals with specific comorbidities, individuals receiving onl y one dose of the
Pfizer-BioNTech COVID
-19 vaccine, and individuals with prior SARS -CoV-2
infection) in the VHA s ystem experience increased risk of saf ety events of interest
following receipt of the Pfizer-BioNTech COVID- 19 vaccine.
Secondary study objective:
To characterize utilization patterns of thePfizer -BioNTech COVID -19 vaccine
among individuals within the VHA ,including estimating the proportion of
individuals receiving vaccine, 2-dose vaccine completion rate,anddistribution of
time gaps between the first and second dose , demographics and health histories of
recipients, overall and among the sub-cohorts of interest .
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077350
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 11of 194Study design: This post-EUAactive safet y surveillance program will employ a rapid-cycle,
longitudinal, observational cohort study designto provide early real-world safet y
information .
The self-controlled risk interval (SCRI) design will be used to sequentially monitor
occurrence of safety events of interest while controlling for time- invariant
confounders. The SCRI design uses data from cases ( i.e., individuals who experience
safety events of interest following vaccination) to compare the risk interval following
vaccination t o post-vaccination non -risk intervals (“post-vaccination control
interval”) in the same individual.
An active comparator design will be used to sequentially monitor occurrence of safety
eventsof interest with Pfizer -BioNTech COVID -19 vaccinations as compared to
recipients of influenza vaccinein the VHA during 2014/2015 through 2018/2019 flu
seasons. Data in peri -COVID time periods from January 2020 to present are excluded
because of pandemic- associated under -utilization of health resources and under -
reporting of medical events.
There will be additional study designs conducted during the signal evaluation phase if a
signal is detected from the above anal yses. These include self -controlled case series (SCCS)
andcomparison to unvaccinated contemporary controls. Additionally , signal evaluation
analyses may also be conducted based on signa ls detected in external sources or based on
regulatory request.
Population : The exposed population will be kept as broad as possible in order to capture
safety events of inte rest that occur among all individuals receiving the Pfizer -BioNTech
COVID-19 vaccine in the period from December 11, 2020 to present . Individuals will be
included if they have a record of at least one dose of Pfizer -BioNTech COVID-19 vaccine.
Individuals w ho receive at least one dose of COVID -19 vaccine from a manufacturer other
than Pfizer -BioNTech will be identified and summarized , but they will be excluded from
further anal ysis. All individuals will be required to be enrolled in and not disenrolled from
VHA benefits during the 1 y ear prior to vaccination date ( i.e., baseline period). Depending on
the attrition rate, the length of the baseline period may be modified to 6 months.
The influenza vaccine comparator cohort will be identified based on a record o f at least one
dose of seasonal influenza vaccine during prior flu seasons, from 2014/2015 through
2018/2019.
Variables :
Exposure s: Administration of Pfizer-BioNTech COVID- 19 vaccine post -EUA
approval will be identified based on the following (seeAppendix Table3for
additional details) :
oCurrent Procedural Terminology (CPT) and associated vaccine administration
Healthcare Common Procedure Coding S ystem (HCPCS)codes; OR
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077351
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 12of 194o10 and 11- digit National Drug Codes (NDCs); OR
oImmunization records that contain data on vaccine code descriptor, vaccine
manufacturer ( i.e., Pfizer), lot number, injection site, and date(s) of
immunization;11
Relevant codes will be continuously reviewed and amended if new codes are added.
Administration of the seasonal influenza vaccine during 2014/2015 through
2018/2019 flu seasons will be identified based on the following (see
Appendix Table3for additional details) :
oCPT codes and associated vaccine administration HCPCS codes ; OR
o10 and 11- digit NDCs; OR
oImmunization records that contain data on vaccine code descriptor, vaccine
manufacturer, lot number, injection site, and date(s) of immunization.
Outcomes : Safety events of interest for active surveillance (see Appendix Table2) are
based on the Priority List of Adverse Events of Special Interest from the Brighton
Collaboration’s SPEAC Project, the FDA and the CDC’sACIPenhanced safet y
monitoring recommendations .
The list of safet y events of interest may be re vised over the course of the study , and if
unanticipated potential safety eventsof interest are identified during the course of
surveillance, they will be added to the list and included in the anal yses. The risk and
control intervals for each safet y event of interest are based on biological plausibility
and precedents in the literature (see Table 1). Outpatient , emergency department
(ED),and/or inpatient settings will be used to identify safety events of interest
depending on the t ype of event. The specific encounter setting to be considered for
each safet y event of interest is summarized in Table 1and can be assigned to 1) the
risk interval following vaccination Pfizer-BioNTech COVID- 19 vaccin ation, 2) the
post-vaccination self -control interval, or 3 ) risk interval for the active comparators of
receiving seasonal influenza vaccine . Events outside the intervals will not be counted.
Only the individual’s first instance of asafety eventof interest following a specified
clean window ( i.e., the occurrence -free baseline period usedto define incide nt
outcomes during which individuals enter the study cohort only if the safety event of
interest did not occur during this period) will be captured; this means that if a safety
event of interest is identified but diagnosis codes corresponding to the safety event of
interest are also observed during the clean window, it will not be counted . The
duration of the pre-specified clean window will differ by safety event of interest (see
Appendix Table2) in order to rule out pre -existing events.
Key Covariates: Baseline demographic ( i.e., age, sex, race/ethnicit y, service region )
and clinical characteristics ( i.e., smoking, body mass index [ BMI],history of
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077352
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 13of 194anaphylaxis/allergic reactions, previous anaphy laxis to vaccine component, history of
hospitalizations, frailty index, Charlson Comorbidity Index [CCI ], selected
comorbidities, and concurrent immunizations )12will be assessed based on available
data (i.e., during 1- year baseline ) prior to the date of vaccination with Pfizer -
BioNTech COVID -19 vaccine and date of seasonal influenza vaccination for active
comparators .
Subgroups : Immunocompromised individuals (i.e., individuals diagnosed with
symptomatic human immunodeficiency virus(HIV)/acquired immunodeficiency
syndrome (AIDS), hematologic malignancy , or other immune conditions; individuals
diagnosed with solid malignancy , organ transplant, or rheumatologic/inflammatory
conditions, all of whom were administered chemotherap y or immune modulators;
individuals diagnosed with rheumatologic/inflammatory conditions and administered
systemic corticosteroids; individuals who were administered chemotherap y, immune
modulators, or systematic steroids for at least 14 day s),13elderly, individuals with
specific comorbidities,12those receiving onl y one dose of Pfizer -BioNTech COVID-
19 vaccine, those with prior SARS -CoV-2 infection, those with regular use of VHA
medical care, and VA priority group 1 veterans will be identified .Analyseswillalso
be performed among individuals enrolled in the VHA with dual coverage who are
also identified in linkedCenters for Medicare & Medicaid Services (CMS) Medicare
administrative claims data.
Data source :The VHA is the largest integrated health care s ystem in the US, providing both
inpatient and outpatient clinical care to over 9 million Veterans enrolled at mor e than 170
medical centers and 1,074 community -based outpatient clinics.14This studywill use data
from VHA’s Corporate Data Warehouse (CDW), which is an integrated EMR system with a
centralized data warehouse that is updat ed on a daily basis.The CDW does not include
information on an y care received outside of a VHA facility .The VA offers eligible Veterans
long-term care services ranging from nursing homes and assisted- living centers to caregiver
support in the Veterans’ o wn homes.15In a subgroup anal ysis of individuals with both VHA
and Medicare coverage, CDW data will be supplemented and linked with Medicare
administrative claims data at the patient level to ensure a more comprehensive eval uation of
the care an individual receives.
Study size: The sample size achieved will depend on the number of recipients of Pfizer -
BioNTech COVID -19 vaccine within the VHA database, which will increase over time with
subsequent anal yses. As of January 21, 2021, 112,201 doses of Pfizer -BioNTech COVID -19
vaccine have been administered within the VHA (based on CPT code 91300) to a total of
107,458 patients.
Data analy sis: A stepwise approach, illustrated in the diagram , will be performed for signal
detection, evaluation, and verification.
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077353
BNT162b2 (COVID- 19 vaccine)
C4591012 NON -INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 14of 194Notes:
[1] List of safety events of interest and corresponding definitions may be refined as the study progresses based on additiona l
available information.
[2] The risk and cont rol intervals selected for the SCRI analysis for each safety event of interest are based on biological
plausibility and precedents in the literature. Only the individual’s first instance during the specified clean window ( i.e., the
interval used to define incident outcomes) will be included. Note that only the first inpatient or outpatient occurrence of a
safety event of interest following the clean window will be used to identify incident events ( e.g., if an inpatient safety event
of interest occurs in th e clean window, a repeat occurrence will not be counted in the risk interval). However, event
worsening will be counted as a safety event of interest. For example, if an outpatient safety event of interest occurs in the
clean window and an inpatient occurr ence for the same type of safety event of interest occurs in the risk interval, the
inpatient occurrence will be counted as a safety event of interest.
1) Signal detection: The goal is to provide rapid -cycle, near real -time safety surveillance. In
the signal detection phase, the SCRI analysis will include post -vaccination control intervals
for certain safety events of interest that require a COVID -19 diagnosis (i.e., severe COVID -
19, multisystem inflammatory syndrome in adults [MIS-A]). To account for mult iple testing
and bi-weekly review of the data, the maximized sequential probability ratio test (MaxSPRT )
using a binomial probability model will be applied. For comparison with individuals who
received seasonal influenza vaccination, the Poisson -based MaxS PRT will be applied for all
other safet y events of interest .
Sequential anal yses for each safet y event of interest will commence once at least 3 events
occur. This approach is consistent with the FDA’s COVID -19 Vaccine Safety Surveillance
Projectto avoid spurious signals from a few early events .16Signals will be detected if the
critical values are reached via the SCRI or active comparator anal ysis.Critical values will be
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077354
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 15of 194determined for each safety event of inter est based on historical incidence rate, expected upper
limit of the number of events under the null hy pothesis, and pre -specified significance level
and power.
2) Signal evaluation: If signals are detected for safety events of interest based on the anal ysis
described above, further evaluation will be conducted to refine and confirm such detections.
This will include comprehensive quality assurance ( for example, check for possible
duplications of claims or medical records, checking for unusual clustering i n claim or
medical record accrual by service date for potential coding issues, check for geographical
distribution of cases that may be related to lot numbers or diagnostic practice ) and
multivariate adjustment using Poisson regression to account for basel ine differences between
Pfizer-BioNTech COVID-19 vaccinated and active comparator cohorts. SCRI analyses using
the post-vaccination control intervals and SCCS using post -vaccination control time periods
will be conducted as an additional inferential analys is once enough post -vaccination time has
accumulated. To address potential period effects, a comparison to contemporary
unvaccinated controls will also be performed , with adjustment using inverse probability of
treatment weighting (IPTW). The assessment of temporal clustering will also be conducted.
Incidence rates will also be calculated and Kaplan -Meier methods wi llbe used to anal yze
time to safety event of interest. Signal evaluation anal yses will be conducted every six
months.
3) Signal verification: diagnostic validation of the detected safety eventsof interest via
adjudication of medical records by VHA clinicians for outcome verification will be
conducted in a representative sample of cases. For rare events, potentiall y all cases may be
adjudicated.
End-of-season anal yses (over the course of the 30 -month period) and an end-of- surveillance
analysis (i.e., at 30 months) will be conducted. Various subgroup anal yses will also be
conducted, examining different age groups, immunocompromised individuals,13individuals
with specific comorbidities,12those who only received one dose of the Pfizer -BioNTech
COVID-19 vaccine, those with prior SARS -CoV-2 infection based on medical history or pre-
vaccination serology , those receiving care regularly at VA facilities ,those with VA Priority
group 1 status, which determines these individuals are of highest priority for VHA care and
likely receive all of their care within the VHA s ystem, and lastly, those with a dditional
Medicare coverage whose Medicare data can be linked to the CDW.
Notably, CDC recentl y investigated my ocarditis/pericarditis following mRNA COVID -19
vaccinations.17To provide additional context to the investiga tion conducted by CDC,
separate safety analyses will be prioritized and performed to assess the risk of
myocarditis/pericarditis following Pfizer -BioNTech COVID-19 vaccination. These analy ses
will be conducted to align with the rapid- cycle analysis perform ed by the Vaccine Safet y
Datalink (VSD).18The number of m yocarditis/pericarditis events in the risk interval will be
identified, and incidence rates per million doses will be summarized . Subgroup anal yses will
also be perform ed, stratified by age (e.g.,12-39 years, 40-49 years, 50-64 years, 65+ years),
gender, and race/ethnicity , respectivel y.Incidence rate ratios will be summarized to compare
the rate of m yocarditis/pericarditis events between vaccinated individuals who se event occurs
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077355
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 16of 194in a pre-specified risk interval versus vaccinated individuals who se event occurs in a
comparison interval on the same calendar day . Myocarditis/pericarditis events will also be
adjudicated via chart review and validated using the Brighton Collaboration’s case
definitions.19Risk factor anal ysis may also be conducted among confirmed cases. Lastly,
additional data surrounding risk factors, clinical course, and sequelae of identified
myocarditis/pericardi tis event up to 365 day s following the event will be collected and
summarized .
Milestones:
VHA CRADA execution: 8 January 2021;
Determination of Institutional Review Board ( IRB)exemption: 10 February 2021 ;
Determination of Research Safet y and Securit y exemption: 17 February 2021;
Approval b y Designated Member Review: 26 February 2021 ;
Registration in the EU PAS register: 5 March 2021 ;
Start of data collection: 11 May2021;
Interim reports: 30 June 2021; 31 December 2021; 30June 2022, 31 December 2022;
End of data collection : 30 June2023;
Final study report: 31 December 2023
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077356
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 17of 194SUMMARY
ObjectivePrimary 1 Primary 2 Secondary
Aim To assess whether individuals in the
Veterans Health Administration
(VHA) system experience increased
risk of safety events of interest
following receipt of the Pfizer -
BioNTech COVID-19 vaccine. To assess whether sub -cohorts of
interest (i.e.immunocompromised,
elderly, with specific comorbidities,
individuals receiving only one dose
of the Pfizer -BioNTech COVID -19
vaccine, and individuals with prior
SARS-CoV-2 infection ) in the VHA
system experience increased risk of
safety events of interest following
receipt of the Pfizer -BioNTech
COVID-19 vaccine. To characterize utilization patterns of
thePfizer-BioNTech COVID -19
vaccine among individuals within the
VHA including estimati ng the
proportion of individuals receiving
vaccine, 2 -dose vaccine completion
rate, and distribution of time gaps
between the first and second dose,
demographics and health histories of
recipients, overall and among the
sub-cohorts of interest.
Study design This post-EUA active safety surveillance program will employ a rapid -cycle, longitudinal, observational cohort study
design to provide early real -world safety information.
The self-controlled risk interval (SCRI) design to sequentially monitor occur rence of safety events of interest
while controlling for time -invariant confounders. This design allows inclusion of a post -vaccination control
interval.
An active comparator design will be used to sequentially monitor occurrence of safety events of inter est with
Pfizer-BioNTech COVID -19 vaccinations as compared to recipients of influenza vaccine in the VHA during
2014/2015 through 2018/2019 flu seasons. Data in peri- COVID time periods from January 2020 to present are
excluded because of pandemic -associated under -utilization of health resources and under -reporting of medical
events.
There will be additional study designs conducted during the signal evaluation phase if a signal is detected from the
above analyses. These include self -controlled case series (S CCS) and comparison of vaccinated to unvaccinated
contemporary control s. Additionally, signal evaluation analyses may also be conducted based on signals detected in
external sources or based on regulatory request (e.g., myocarditis/pericarditis) .
Studypopulation The study will be kept as broad as possible in order to capture safety events of interest that occur among vaccinated
individuals.
Inclusion criteria:
Record of at least one dose of Pfizer -BioNTech COVID -19 vaccine in the period of December 11, 2020 to
present, or
Record of at least one dose of seasonal influenza vaccine during prior flu seasons, from 2014/2015 through
2018/2019 ( applies to active comparator s only); and
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077357
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 18of 194ObjectivePrimary 1 Primary 2 Secondary
At least 1 year of enrollment in and no disenrollment from VHA benefits ( i.e., the baseline period) prior to
Pfizer-BioNTech COVID -19 or seasonal influenza vaccination date.
Exclusion criteria:
Individuals who receive at least one dose of Pfizer -BioNTech COVID -19 vaccine in addition to a COVID -19
vaccine from a manufacturer o ther than Pfizer -BioNTech will be identified and summarized , but they will be
excluded from further analysis. .
Study Period The study will be conducted for a period of 30 months, starting on December 11, 2020 onward, with data collection
concluding on June 10, 2023.
Exposure Administration of Pfizer -BioNTech COVID -19 vaccine post -EUA approval will be identified based on records of the
following (seeAppendix Table3for additional details) :
Current Procedural Terminology (CPT) and associated vaccine administration HCPCS codes; OR
10 and 11 -digit National Drug Codes (NDCs); OR
Immunization records that contain data on vaccine code descriptor, vaccine manufacturer ( i.e., Pfizer), lot
number, injection site, and date(s) of immunization;
Administration of the seasonal influenza vaccine during 2014/2015 through 2018/2019 flu seasons will be identified
based on records of the following (seeAppendix Table3for additional details) :
CPT codes and associated vaccine administration HCPCS codes ; OR
10 and 11 -digit NDCs; OR
Immunization records that contain data on vaccine code descriptor, vaccine manufacturer, lot number,
injection site, and date(s) of immunization.
Safety Events of Interest Safety events of interest for active surveillance were identified based on the Priority List of Adverse Events of Special
Interest from the Brighton Collaboration’s Safety Platform for Emergency vACcines (SPEAC) Project, the FDA and
the Centers for Disease Control and Prevention’s (CDC) Advisory Committee on Immunization Practices (ACIP)
enhanced safety monitoring recommendations. The list of safety events may be revised over the course of the study,
and if unanticipated potential safety events of interest are identified during the course of surveillance, they will be
added to the list and included in the analyses. The risk and control intervals for each safety event of interest are based
on biological plausibility and precedents in the literature. Outpatient , emergency department ,and/or inpatient settings
will be used to identify safety events of interest depending on the type of event. Safety events of interest can be
assigned to 1) the risk interval following vaccination Pfizer -BioNTech COVID -19 vaccination, 2) the post -vaccination
self-control interval, or 3) risk interval for the active comparators of receiving seasonal influenza vaccine. Events
outside the intervals will not be counted. Only the individual’s first instance of a safety event of interest following a
specified clean window ( i.e., the occu rrence-free baseline period used to define incident outcomes during which
individuals enter the study cohort only if the safety event of interest did not occur during this period) w ill be included;
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077358
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 19of 194ObjectivePrimary 1 Primary 2 Secondary
this means that if a safety event is identified but diagno sis codes corresponding to the safety event are also observed
during the clean window , it will not be counted . The duration of the pre -specified clean window will differ by type of
safety event of interest in order to rule out pre -existing events.
Neurologic:
Aseptic meningitis
Bell’s palsy
Cerebrovascular non -hemorrhagic stroke
Convulsions/seizures in individuals with controlled epilepsy
Encephalitis/encephalomyelitis
Guillain-Barré Syndrome (GBS)
Generalized convulsion/seizures
Multiple sclerosis (MS)
Optic neuritis (ON)
Other acute demyelinating diseases
Transverse myelitis (TM)
Immunologic :
Anaphylaxis
Arthritis and arthralgia/joint pain
Autoimmune thyroiditis
Fibromyalgia
Kawasaki disease (KD)
Multisystem inflammatory syndrome in adults (MIS -A)
Vasculitides
Cardiac:
Acute myocardial infarction (AMI)
Arrhythmia
Coronary artery disease (CAD)
Heart failure and cardiogenic shock
Microangiopathy
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077359
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 20of 194ObjectivePrimary 1 Primary 2 Secondary
Myocarditis
Pericarditis
Stress cardiomyopathy
Hematologic:
Cerebrovascular hemorrhagic stroke
Chilblain-like lesions
Disseminated intravascular coagulation (DIC)
Deep vein thrombosis (DVT)
Hemolytic anemia
Hemorrhagic disease
Limb ischemia
Pulmonary embol ism(PE)
Single organ cutaneous vasculitis
Thrombocytopenia
Thrombosis with thrombocytopenia syndrome (TTS)
Other:
Acute kidney injury
Appendicitis
Death
Erythema multiforme
Liver injury
Narcolepsy and cataplexy
Non-anaphylactic allergic reactions
Severe COVID- 19 disease
Stevens-Johnson syndrome (SJS)/Toxic epidermal necrolysis (TEN)
Data source The VHA Corporate Data Warehouse (CDW) database w ill be used and maybe supplemented with Medicare
administrative claims data from the Centers for Medicare & Medicaid Services (CMS) .
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077360
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 21of 194ObjectivePrimary 1 Primary 2 Secondary
Data analysis A stepwise approach w ill be performed for signal detection, evaluation, and verification.
1) Signal detection: The goal is to provide rapid -cycle, near real -time safety surveillance. SCRIanalyses using the
post-vaccination control intervals will be conducted for certain safety events of interest that require a COVID -19
diagnosis (i.e., severe COVID -19 illness, MIS -A). To account for multiple testing and bi -weekly review of the data, the
maximized sequential probability ratio test (MaxSPRT) using a binomial probability model w ill be applied. For
comparison with individuals who received seasonal influenza vaccination, the Poisson -based MaxSPRT w ill be applied
for all other safety events of i nterest.
Sequential analyses for each safety event of interest will commence once at least 3 events occur. This approach is
consistent with the FDA’s COVID -19 Vaccine Safety Surveillance Project to avoid spurious signals from a few early
events. Signals will be detected if the critical values are reached via the SCRI or active comparator analysis. Critical
values will be determined for each safety event of interest based on historical incidence rate, expected upper limit of
the number of events under the null hypothesis, and pre -specified significance level and pow er. Incidence rates will
also be calculated and Kaplan -Meier methods will be used to analyze time to safety event of interest.
2) Signal evaluation: If signals are detected for safety events of i nterest based on the analysis described above, further
evaluation will be conducted to refine and confirm such detections. This will include comprehensive quality assurance
(for example, check for possible duplications of claims or medical records, checkin g for unusual clustering in claim or
medical record accrual by service date for potential coding issues, check for geographical distribution of cases that may
be related to lot numbers or diagnostic practice) and multivariate adjustment using Poisson regre ssion to account for
baseline differences between Pfizer -BioNTech COVID -19 vaccinated and active comparator cohorts. SCRI analyses
using the post -vaccination control intervalsand the SCCS design with post -vaccination control time period will be
conducted as an additional inferential analysis once enough post -vaccination time has accumulated. To address
potential period effects, a comparison to contemporary unvaccinated controls will also be performed with adjustment
using inverse probability of treatment w eighting (IPTW). The assessment of temporal clustering will also be
conducted. Signal evaluation analyses will be conducted very six months.
3) Signal verification: diagnostic validation of the detected safety events of interest via adjudication of medical records
by VHA clinicians for outcome validation will be conducted in a representative sample of cases. For rare events,
potentially all cases may be adjudicated.
End-of-season analyses (over the course of the 30 -month period) and an end -of-surveillance analysis ( i.e., at 30
months) will be conducted. Various subgroup analyses will also be conducted, examining different age groups,
immunocompromised individuals, individuals with specific comorbidities, those who only received one dose of the
Pfizer-BioNTech COVID -19 vaccine, those with prior SARS -CoV-2 infection based on medical history or pre -
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077361
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 22of 194ObjectivePrimary 1 Primary 2 Secondary
vaccination serology, those receiving care regularly at VA facilities, those with VA Priority group 1 status, which
determines these individuals are of highest priori ty for VHA care and likely receive all of their care within the VHA
system, and lastly, those with additional Medicare coverage whose Medicare data can be linked to the CDW .
Notably, CDC recently investigated myocarditis/pericarditis following mRNA COVID -19 vaccinations. To provide
additional context to the investigation conducted by CDC, separate safety analyses will be prioritized and performed to
assess the risk of myocarditis/p ericarditis follow ing Pfizer -BioNTech COVID -19 vaccination .These analyses will be
conducted to align with the rapid -cycle analysis performed by the Vaccine Safety Datalink (VSD). The number of
myocarditis/pericarditis events in the risk interval will be identified, and incidence rates per million doses will be
summarized . Subgroup analyses will also be performed, stratified by age ( e.g., 12-39 years, 40 -49 years, 50 -64 years,
65+ years ), gender, and race/ethnicity, respectively. Incidence rate ratios w illbe summarized to compare the rate of
myocarditis/pericarditis events betw een vaccinated individuals who se events occur in a pre-specified risk interval
versus vaccinated individuals who se events occur in a comparison interval on the same calendar day.
Myocarditis/pericarditis events will also be adjudicated via chart review and validated using the Brighton
Collaboration’s case definitions. Risk factor analysis may also be conducted among confirmed cases. Lastly, additional
data surrounding the risk factors , clinical course, and sequelae of the identified myocarditis/pericarditis event up to 365
days following the event will be collected and summarized .
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077362
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 23of 1945.AMENDMENTS AND UPDAT ES
Amendment number Date Protocol
section(s)
changedSummary of amendment(s) Reason
1 31 August
20216 UpdatedtheMilestones section. To add additional information that
became available after the initial
protocol was submitted to FDA
regarding IRB review, EU PAS
registration, and data collection
dates.
1 31 August
20219.1.1 Added clarification on the self-
controlled risk interval ( SCRI)design,
including a description of the
measurements when there is a gap
between risk intervals for the first and
second dose and an illustration (new
Figure 2B) .To respond to a request from C enter
for Biologics Evaluation and
Research (CBER) to demonstrate
how the period after the risk interval
for dose 1 and prior dose 2 will be
handled in the anal ysis if there is no
overlap between the risk intervals
for the two doses.
1 31 August
20219.1.1 Added that a dditional doses of the
Pfizer-BioNTech COVID- 19 vaccine
may be included in the analy sis.To address the potentia l approval of
additional doses. Details for this
analysis will be further described in
the statistical analy sis plan.
1 31 August
20219.1.1, 9.3.3,
9.7.3, 9.7.5, 9.9Removed SCRI design with pre -
vaccination control interval and added
SCRI design with post -vaccination
control interval for 2 safety events of
interest (severe COVID -19,
multisystem inflammatory syndrome To address CBER request to
remove the pre -vaccination control
interval as its comparison to the risk
interval may introduce bias and
reduce the probabilit y of subsequent
vaccination. Note additional and
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077363
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 24of 194Amendment number Date Protocol
section(s)
changedSummary of amendment(s) Reason
in adults[MIS-A]) that could not be
evaluated with the seasonal influenza
vaccinated comparators.
Revised Figures 1 -5 to remove pre-
vaccination control interval and
provide examples for post -vaccination
interval. more robust anal yses were added to
signal evaluation phase (see new
sections under 9.7.3.2.5 and
9.7.3.2.6). SCRI with post -
vaccination control intervals was
included in the signal detection
phase to evaluate severe COVID -19
and MIS-A as they require COVID -
19 diagnosis, which would not be
observed in a seasonal influenza
comparator.
1 31 August
20219.2.3, 9.4 Added clarification for the
identification of subgroups who are
immunocompromised sand
individuals with specific
comorbidities .
Added one additional subgroup of
interest (individuals with Medicare
coverage for whom Veterans Health
Administration [VH A] records can be
linked to their Medicare claims).To provide additional detail
regarding how subgroups who are
immunocompromised and
individuals with specific
comorbidities will be defined and
operationalized.
To respond to a query from CBER
regarding the potential for
incomplete data for healthcare
encounters not received at VHA, an
additional subgroup of individuals
with linked Medicare data has been
added.
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077364
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 25of 194Amendment number Date Protocol
section(s)
changedSummary of amendment(s) Reason
1 31 August
20219.3.1 All measurement details concerning
how Pfizer -BioNTech COVID-19
vaccine and seasonal influenza
vaccine will be identified in the data
to an Appendix Table 3 in Section 18.
Appendix Table 3 includes all specific
CPT/HCPCS/NDC codes previously
listed in Section 9.3.1 as well as
additional codes identified at the time
of the data anal ysis. To update the protocol with all
relevant CPT/HCPCS/NDC codes,
while maintaining concise language
in the main text.
1 31 August
20219.3.1, 18 Added Appendix Table 4 in Section
18 regarding the LOINC codes used to
identify COVID-19 RT-PCR Test
among the stud y population and
corresponding reference.To provide additional details on
how individuals with prior SARS -
CoV-2 infection will be identified
in the data.
1 31 August
20219.3.2, 18 Added fra ilty index as a baseline
characteristic of interest.To describe the identification of
frailty in the Pfizer -BioNTech
COVID-19 and seasonal influenza
cohorts during the 1- year baseline
period prior to vaccination as frailt y
may be a prognostic factor for
safety events of interest.
1 31 August
20219.3.3, 18 Added fouradditional safety events of
interest: thrombosis with
thrombocy topenia sy ndrome,
convulsions/seizures in individuals To consider new safety events
based on emerging research and
align with codes from the FDA
CBER COVID-19 Vaccine Safet y
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077365
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 26of 194Amendment number Date Protocol
section(s)
changedSummary of amendment(s) Reason
with controlled epileps y, Steven-
Johnson sy ndrome/Toxic epidermal
necrolysis, and hemol ytic anemia
(increasing the number of safet y
events of interest from 42 to 46).
Reclassified COVID -19-related safet y
events of interest to be measured
independentl y of the patient ’s
COVID-19 infection status ; this
change had no impact on the number
of safety events of interest (reflected
both in the revised text and revised
Table 1).
Added that the clean window may be
extended (e.g., 2 years).Surveillance : Active Monitoring
Master Protocol.20,21,22
The COVID -19-related safet y
events were reclassified to more
closely align with the FDA CBER
COVID-19 Vaccine Safety
Surveillance : Active Monitoring
Master Protocol. COVID-19-related
safety events that were previously
listed may not necessarily be related
to COVID-19 infection (e.g.,
coronary artery disease), a nd
therefore are defined independent of
a COVID -19 diagnosis, with the
exception of “severe COVID -19
disease” and “MIS -A” which
requires a concurrent COVID -19
diagnosis.
Extending the clean window will
address the reduction in healthcare
resource utilizat ion during the
pandemic to more accurately
identify incident events.
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077366
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 27of 194Amendment number Date Protocol
section(s)
changedSummary of amendment(s) Reason
1 31 August
20219.7.3.2 Clarified in the Signal Evaluation
section that thesignal evaluation
analyses will beconducted every six
months.To provide additional detail on the
timing of the signal evaluation
analyses.
1 31 August
20219.1.3, 9.7.3.2.5 Added self-controlled case series
(SCCS)design with fullpost-
vaccination period as an additional
analysis in the Si gnal Evaluation
analysis. Added that Signal
Evaluation anal yses may also be
conducted based on signa ls detected
in external sources or based on
regulatory request(e.g.,
myocarditis/pericarditis) . To further align with the CBER
Master Protocol: Assessment of
Risk of Safety Outcomes Following
COVID-19 Vaccination ( March 23,
2021).23SCCS analy sis has
increased power compared to SCRI
design using post -vaccination
control interval and has been added
to complement the SCRI design. In
addition, clarified that Signal
Evaluation anal yses may al so be
conducted based on signals detected
in external sources or based on
regulatory request even if such
analyses were not first identified in
the Signal Detection phase of this
study.
1 31 August
20219.7.3.2.6 Added a comparison group of
contemporary unvaccinated controls
in the Signal Evaluation analysis.To address the recommendation
from CBER to include a
contemporary control group of
unvaccinated individuals due to
potential period effects of an active
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077367
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 28of 194Amendment number Date Protocol
section(s)
changedSummary of amendment(s) Reason
comparator design that uses
historical control s of influenza
vaccinated individuals.
1 31 August
20219.7.8 Added new section on
myocarditis/pericarditis safet y
analysis and risk factor analysis. To include a separate analy sis
focused on m yocarditis/pericarditis
based on emerging evidence
regarding this event in association
with mRNA COVI D-19 vaccines.17
1 31 August
20219.9 Added strengths and limitations
associated with the addition of the
SCCS design, contemporaneous
unvaccinated controls, and subgroup
analysis of individuals with linkage to
Medicare claims data.To further describe the rationale for
these additional anal yses.
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077368
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 29of 1946.MILESTONES
Milestone Planned date
VHA CRADA execution, Determination of IRB &
Research Safet y and Security exemptions,
Approval by Designated Member Review[1-3]January -February 2021
Registration in the EU PAS register 5 March 2021
Start of data collection 11 May2021[4]
Interim reports 30June 2021
31December 2021
30June 2022
31 December 2022
End of data collection 30June 2023[5]
Final study report 31December 2023
Abbreviations: ACOS, Associate Chief of Staff; COVID -19, Coronavirus disease 2019; CRADA ,
Cooperative Research and Data Agreement; IRB, Institutional Revie w Board; EUA, Emergency Use
Authorization; FDA, Food and Drug Administration; NNERC VAMC, Northern New England Research
Consortium VA Medical Centers; R&D, Research and Development; SRSS, S ubcommittee on Research
Safety and Security; VA, Veterans Affairs; VAIRRS, VA Innovation and Research Review System; VINNE,
Veteran’s IRB of Northern New England; VHA, Veterans Health Administration; US, United States.
Notes:
[1] IRB exemption determinati on was granted in accordance w ith 38 CFR 16 by the Veteran’s IRB of
Northern New England (VINNE), White River Junction VA Medical Center, White River Junction, VT for
the signal detection and signal evaluation phases. Prior to progressing to the signal ver ification phase for
chart review, a second IRB review application will be submitted for an expedited or full review. The tw o-
stage IRB application process is to expedite the initiation of the project.
[2] Research Safety and Security exemption determinatio n was granted by the Subcommittee on Research
Safety and Security (SRSS), VA Innovation and Research Revie w System (VAIRRS).
[3] Approved by Associate Chief of Staff for Research and Development (ACOS/R&D) and R&D
Committee of the Northern New England Rese arch Consortium VA Medical Centers (NNERC VAMC).
[4] Start of data collection is the date for starting data extraction for the purposes of the study analysis. The
initial data analysis includes Pfizer -BioNTech COVID -19 vaccine exposures from December 11, 2020 (the
EUA approval date by the US FDA) to March 12, 2021 (the data cutoff date).
[5] End of data collection is after the Pfizer -BioNTech COVID -19 vaccine exposure data reached 30 months
post-EUA approval and the last day of the month that the study wil l be completed.
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077369
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 30of 1947.RATIONALE AND BACKGR OUND
In March 2020, the World Health Organization (WHO) declared a global pandemic for the
coronavirus disease 2019 (COVID -19) due to the severe acute respiratory syndrome
coronavirus 2 (SARS -CoV-2), which was first i dentified by public health officials in China
in December 2019.1The COVID -19 pandemic presents an unprecedented public health
crisis. As of January 7, 2021, over 21.4million COVID -19 cases and 364,000 deaths have
been reported in the United States (US) alone.2To date, the incidence of COVID -19 has
continued to rise, large ly affecting the elderl y and middle -aged individuals, with worsening
clinical sequelae linked to increasing age and comorbid conditions ( e.g., cardiovascular
disease, active cancer, obesity , diabetes and chronic lung disease).24,25SARS-CoV-2 is a
well-adapted highl y infectious human pathogen with a case fatality rate that ranges between
0.5% and 20%, based on the individual’s age, gender, race, and comorbidites.26
Pfizer and BioNTech have partnered to develop a novel messenger RiboNucleic Acid
(mRNA)vaccine against SARS -CoV-2for the prevention of COVID -19 (Candidate
BNT162b2). To this end, Pfizer is conducting a Phase 1/2/3, randomized, placebo -controlled,
observer-blind, dose -finding, vaccine candidate- selection, and efficacy study among healthy
individuals (NCT04368728). In their Phase 1 trial evaluating safety and immunogenicit y of
two mRNA vaccine candidates ( i.e., BNT162b1, BNT162b2) at various dose leve ls,
candidate BNT162b2 was selected for advancement to a pivotal Phase 2/3 safet y and efficacy
evaluation due to its milder systemic reactogenicity profile, especially in older adults.27The
study was initiated in July 2020 with a target enrollment of 43,998 individuals.28
The US Food and Drug Administration (FDA) announced that regulatory emergency use
authorization (EUA) as well as full approval of any COVID-19 vaccine will require
demonstrating prevention of the disease or decrease in its severity in at least 50% of the
individuals who receive it. I n addition, data from Phase 3 studies are required to include a
median follow -up duration of at least 2 months after completion of the full vaccination
regimen to assess the vaccine’s benefit -risk profile, especiall y adverse events and cases of
severe COVID -19 in vaccinated study subjects.29The FDA reviewed the available safet y
data of the Phase 1/2/3 trial from 37,586 participants 16 y ears of age and older and did not
identify any specific safety concerns. In addition, the anal ysis of available efficacy data from
36,523 participants 12 years of age and older without evidence o f prior SARS -CoV-2
infection at least 7 day s after receiving the second dose demonstrated 95% efficacy of the
vaccine in the prevention of COVID -19 (as confirmed by 8 vs. 162 COVID -19 cases in the
vaccine and placebo groups, respectivel y).3,4Based on these safety and efficacy data, as well
as a review of manufacturing information regarding product quality and consistency , the
FDA determined that the known and potential benefits of the vacci ne outweighed the known
and potential risks for the prevention of COVID -19 in individuals 16 y ears of age and older.4
Therefore on December 11, 2020, the Pfizer -BioNTech COVID-19 vaccine was granted an
EUA by the FDA to preve nt COVID -19 in individuals 16 y ears of age and older.5
With respect to geographic regions other than the US, on December 2, 2020, the United
Kingdom (UK) was the first country in the world to grant temporary authorization for
emergency use of the Pfizer -BioNTech COVID -19 vaccine.6On December 21, 2020, the
European Medicines Agency (EMA) granted the Pfizer -BioNTech COVID- 19 vaccine a
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077370
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 31of 194conditional marketing authorization (CMA) for use among individuals 16 years of age and
older throughout all of the European Union’s (EU) 27 member states.7
As required b y the EUA, post-authorization observational studies using real -world data are
needed in order to assess the association between Pfizer -BioNTech COVID- 19 vaccine and
pre-determined safet y events of interest (including deaths , hospitalizations, and severe
COVID-19) among individuals administered the v accine in both the population at large and
in populations of interest ( e.g., immunocompromised individuals, elderl y, and those with
specific comorbidities) .4Post-authorization safety evaluations are important for identify ing
rare, serious safet y events of interest in larger populations that may not have been detected
during clinical trials (either due to sample size or selected study populations), and ensure a
favorable benefit -risk ratio post -trial.Pfizer in collaboration with the US Veterans Health
Administration ( VHA)of the Department of Veterans Affairs (VA) and Analy sis Group
herein propose post -EUA active safet y surveillance of safety events of interest based
primarily on the Priorit y List of Adverse Events of Specia l Interest from the Brighton
Collaboration’s Safet y Platform for Emergency vACcines (SPEAC) Project and from the
preliminary list of safet y events of interest presented at the September 22, 2020, meeting of
Centers for Disease Control and Prevention’s (CDC ’s) Advisory Committee on
Immunization Practices (ACI P)on the enhanced safet y monitoring of COVID -19 vaccines.8,9
This safet y surveillance study willidentify and evaluate rapid, near real-time potential safet y
signals associated with the Pfizer- BioNTech COVID -19 vaccine in the large -scale VHA
electronic medical record (EMR) database. The observed rates of safet y event of interest will
be compared to expected rates derived from self -controls and active comparators. Part of the
methodologies used in this study are constructed based on approachespreviously used by the
Post-Licensure Rapid Immunization Safety Monitoring (PRISM) program for the H1N1
vaccine.10This non-interventional study is designated as a Post -Authorization Safety Study
(PASS) commitment to the US FDA andisa Category 3 commitment in the EU Risk
Management Plan.
8.RESEARCH QUESTION AND OBJECTIVES
Research question: what are the incidence rates of safety events of interest (based on adverse
events of special interest [AESI ]) among individuals vaccinated with the Pfizer-BioNTech
COVID-19 vaccine within the US VHA system overall and in sub -cohorts of interest a s
compared to expected rates of those events?
Primary study objectives:
To assess whether individuals in the VHA s ystem experience increased risk of safet y
events of interest following receipt of the Pfizer- BioNTech COVID -19 vaccine;
To assess whether sub-cohorts of interest ( i.e., immunocompromised, elderly , with
specific comorbidities, individuals receiving onl y one dose of the Pfizer -BioNTech
COVID-19 vaccine, and individuals with prior SARS -CoV-2 infection) in the VHA
system experience increased risk of safety events of interest following receipt of the
Pfizer-BioNTech COVID- 19 vaccine.
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077371
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 32of 194Secondary study objective s:
To characterize utilization patterns of the Pfizer -BioNTech COVID -19 vaccine
among individuals within the VHA including estimating the proportion of individuals
receiving vaccine, 2-dose vaccine completion rate,and distribution of time gaps
between the first and second dose , demographics and health histories of recipients,
overall and among the sub-cohorts of interest.
9.RESEARCH METHODS
9.1.Study Design
This post-EUAactive safety surveillance program will employ a rapid-cycle, longitudinal,
observational cohort study design to provide earl y real-world safet y information . The self-
controlled risk interval (SCRI) design will be used to sequentially monitor occurrence of
safety events of interest while controlling for time- invariant confounders (such as sex, race,
chronic illness, and state). I n addition, safety events of interest associated with Pfizer-
BioNTech COVID -19 vaccinations will be sequentially monitored and compar ed to
recipients of influenza vaccine in the VHA between 2014/2015 to 2018/2019 .10,30
9.1.1.Self-Controlled Risk Interval (SCRI) Design with Post -Vaccination Control
Interval
The SCRI design uses data from cases ( i.e., individuals who experience safety events of
interest following vaccination) to compare the risk interval following vaccination to
post-vaccination non -risk intervals ( “post-vaccination control interval”) in the same
individual.31A length of 42 day s has been used to define the risk interval in SCRI design
studies for signal detection to ascertain the safet y profile of the H1N1 vaccine.10,30The same
length of risk interval is proposed here, subject to further modification based on clinical
input, clinical trial data, biologic plausibility , and published literature. The day of vaccination
will only be included in the risk period for thos e safety eventsof interest for which a same -
day occurrence is biologically plausible ( e.g.,anaphylaxis).
Apost-vaccination control interval will be used for certain safet y events of interest for the
following reasons: (1) a recent prior safet y event of interest might preclude vaccination
(i.e.,anaphylaxis), (2) individuals might have an underly ing condition that is also a
contraindication for vaccination ( i.e., seizure disorder), or (3) safet y event of interest and
vaccination may be seasonal in nature.32Thetime between the risk and control intervals will
be determined based on the biological mechanism of action for each safety eventsof interest
assessed, and may be subject to change based on further clinical input . Examples of the SCRI
design with a post -vaccination control interval (in an individual who only receives the first
dose of vaccine) is presented in Figure 1below.
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077372
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 33of 194Figure 1.Example of SCRI Design for Assessment of a Safety Event of Interest with a
42-day Risk Interval in an Individual who Receives Only One Vaccine Dose,
with Post -vaccination Control Intervals*
*The risk interv al may include day 0, date of Pfizer -BioNTech COVID -19 vaccination, for some of the safety
eventsof interest assessed ( e.g., anaphylaxis). The length of the risk interval will vary across each safety event
of interest and may be subject to change based on clinical input. Note that some individuals may not receive the
complete course of vaccination, and thus may only receive the first dose of vaccine. This is represented in
Figure 1while Figure 2represents an example where the complete course with 2 doses are receive d.
Two doses of the Pfizer -BioNTech COVID -19 vaccine are recommended 3 weeks apart.
This study program will monitor safety events of interest that occur after dose 1 before dose
2 (i.e., during risk interval 1), after dose 2 ( i.e., during risk interval 2), and aggregate for
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077373
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 34of 194doses 1 and 2 ( i.e., risk interval 1 + risk interval 2), respectively , for individuals receiving
both doses. Additional doses of the Pfizer -BioNTech COVID -19 vaccine may be included in
the analysis should they be approved , and those details will be described in the statistical
analysis plan(SAP).
Given the risk intervals for specific safety events of interest range from 1 day to 90 day s
(please see Table 1in Section 9.3.3), the time between the first and second dose may be
longer or shorter than the recommended risk interval for a given safety eventafter the first
dose. See Figure 2below for SCRI design exampleswhere asafety event with a 42 risk
interval window (e.g., Bell’s palsy ; Table 1 in Section 9.3.3 ) is assessed in hypothetical
individual swho receive two doses of P fizer-BioNTech COVID -19 vaccine: Figure 2Ashows
the SCRI design with the second dose received 21 day s after the first (i.e., the risk interval
for dose 1 overlaps with the risk interval for dose 2) , while Figure 2B shows the SCRI design
with the second dose received 60days after the first(i.e., there is a gap between the end of
the risk interval for dose 1 and dose 2 initiation) . For the first scenario ( Figure 2A), t he risk
interval for dose 1 will be censored at the time of dose 2; further, s afety events of interest that
occur during the overlapping period of risk interval 1 and risk interval 2 (shown in gray
shading in Figure 2A) may be flagged for separate anal yses to discern the additive effect of
Pfizer-BioNTech COVID- 19 vaccine dose 1 and dose 2. For the second scenario ( Figure 2B),
eventswill only be measured during the risk intervals, ignoring the gap between the end of
the risk interval for dose 1 and dose 2 initiation.
For each anal ysis, control intervals corresponding to the risk intervals will be defined either
at end of the risk interval for dose 1 (for individuals with only one dose observed) or after the
risk interval for dose 2 (for individuals with two doses observed), regardless of whether of
the analyses focus on safety events after dose 1, after dose 2, or aggregated for doses 1 and 2
(Figure 2A and Figure 2B).
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077374
BNT162b2 (COVID- 19 vaccine)
C4591012 NON -INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 35of 194Figure 2.Example of SCRI Design for Assessment of a Safety Event of Interest with a
42-day Risk Interval in an Individual who Receives TwoVaccine Dose
s, with
Post-vaccination Control Intervals
9.1.2.Active Comparator Design
In the active comparator design, the frequency of safety events of interest among individuals
who received Pfizer -BioNTech COVID- 19 vaccine from December 11, 2020 onward will be
compared wit h the event frequency among recipients of the seasonal influenza vaccination in
five prior seasons, between 2014/2015 t hrough 2018/2019. Data in peri -COVID time periods
from January 2020 to present are excluded because of pandemic -associated under -utilization
of health resources and under -reporting of medical events. The same risk interval length
(e.g.,42 days)will be used to evaluate safety eventsof interest following vaccination with
Pfizer-BioNTech COVID -19 vaccine and to assess safet y eventsof interest occurring after
vaccination for seasonal influenza in prior seasons. The observed number of safety eventsof
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077375
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 36of 194interest for Pfizer -BioNTech COVID -19 vaccine will be compared to the expected number
calculated for the influenza vaccine in past seasons.10
9.1.3.Additional Study Designs in the Signal Evaluation Phase
There will be additional study designs conducted during the signal evaluation phase if a
signal is detected from the above anal ysesin the signal detection phase . These in clude
analyses using self-controlled case series (SCCS) and comparison of vaccinated unvaccinated
contempora rycontrols.Additionally , signal evaluation anal yses may also be conducted for
signals detected in external sources or based regulatory request(e.g.,
myocarditis/pericarditis) . These analy ses are further detailed in Section 9.7.3.2 .
9.1.4.Study Period
The studywill be conducted for a period of 30 months , starting on December 11, 2020
onward, with data collection concluding on Ju ne 10, 2023.
9.2. Setting
The study population will be kept as broad as possible in order to capture safety eventsof
interest that occur among all vaccinated individuals.
9.2.1. Inclusion Criteria
Record of at least one dose of Pfizer-BioNTech COVID- 19 vaccine in the period of
December 11, 2020 to present, or
Record of at least one dose of seasonal influenza vaccine during prior flu seasons,
from 2014/2015 to 2018/2019 (applies to active comparator s only); and
At least 1 y ear of enrollment in and no disenrollment from VHA benefits ( i.e., the
baseline period) prior to Pfizer -BioNTech COVID- 19 or seasonal influenza
vaccination date.
9.2.2.Exclusion criteria
Individuals who receive at least one dose of Pfizer- BioNTech COVID-19 vaccine in
addition to a COVID -19 vaccine from a manufacturer other than Pfizer -BioNTech
will be identified and summarized , but they will be excluded from further anal ysis.
9.2.3.Subgroups
Safety surveillance may be conducted for subgroups of interest , including, but not limited to:
Immunocompr omised individuals , defined as individuals diagnosed with
symptomatic human immunodeficiency virus (HIV)/acquired immunodeficiency
syndrome (AIDS), hematologic malignancy , or other immune conditions; individuals
diagnosed with solid malignancy , organ transplant, or rheumatologic/inflammatory
conditions, all of whom were administered chemotherap y or immune modulators;
individuals diagnosed with rheumatologic/inflammatory conditions and administered
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077376
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 37of 194systemic corticosteroids; orindividuals who were administered chemotherapy ,
immune modulators, or systematic steroids for at least 14 day s;13
Different age groups, with a focus on the elderl y(e.g., <35, 35 - <45, 45 - <55, 55 -
<65, 65 -<75, > 75);
Individuals with specific comorbidities identified as high risk for COVID -19 by the
CDC (i.e., cancer, chronic kidney disease, chronic obstruction pulmonary disease
[COPD], Down Sy ndrome, cardiovascular conditions [e.g., heart failure, coronary
artery disease, or cardiomy opathies], immunocompromised state from solid organ
transplant, obesity [body mass index (BMI) of 30 kg/m2 or higher but < 40 kg/m2],
severe obesity [BMI of 40 kg/m2or higher], sickle cell disease, smoking, type 1 and 2
diabetes mellitus);12
Individuals receiving only one dose of Pfizer-BioNTech COVID-19 vaccine;
Individuals with prior SARS -CoV-2 infection based on medical history or
pre-vaccination serology (Appendix Table 4 );
Individuals with regular use of VHA medical care, defined as at least two outpatient
(excluding emergency department [ED], as ED visits may not be considered regular)
or inpatient enco unters in the one y ear prior to vaccination .The encounters must be
separated b y >30days (for inpatient, by admission date), and at least one must be
within sixmonths prior to the date of vaccination .This will ensure that individuals
have ongoing healt h care encounters, particularl y near the vaccination date, and
regularly receive their healthcare from VHA facilities, rather than outside facilities
that would not be captured in the VHA’sCorporate Data Warehouse ( CDW);
Individuals who are in the VApriority group 1Veteran. These individuals have either
the highest levels of service connected disability (≥50% disabling), are considered
unemploy able, or have received the medal of honor.33Individuals categorized as
priority group 1 are the highest priorit y for VHA care. This will ensure that the
individual is more likely to receive all of their care from a VA facility .
Individuals enrolled in the VHA with dual coverage who are also identified in the
Centers for Medicare & Med icaid Services (CMS) Medicare administrative claims
data, which will be linked to the CDW, in order to supplement CDW data for a more
complete evaluation of healthcare encounters .
Additional subgroups of interest will be assessed as additional information becomes
available from ongoing clinical trials, Vaccine Adverse Event Reporting Sy stem (VAERS),
and other sources that will inform the Pfizer -BioNTech COVID -19 vaccine safet y profile.
Giventhat VA population has a median age of over 46 y ears for females and is comprised of
approximately 90% males , the evaluation of the Pfizer-BioNTech COVID -19 vaccine safet y
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077377
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 38of 194during pregnancy , including fetal death and infant outcomes, may have poor feasibility and
will therefore not be conducted.
9.3.Variables
9.3.1.Exposure of In terest
Administration of Pfizer -BioNTech COVID -19 vaccine post-EUA approval will be identified
based on the following (seeAppendix Table3for additional details) :
Current Procedural Terminology (CPT) code sand associated vaccine administration
HCPCS codes; OR
10 and 11- digit National Drug Codes (NDCs); OR
Immunization records that contain data on vaccine code descriptor, vaccine
manufacturer ( i.e., Pfizer), lot number, injection site, and date(s) of immunization.11
Relevant codes will be continuously reviewed and amended if new codes are added.
Person-time at-risk exposure to the first dose onl y, overlapping first and second doses, and
second dose onl y will be anal yzed separatel y.
Administration of the seasonal influenza vaccine during 2014/2015 through 2018/2019 flu
seasons wil l be identified based on the following (seeAppendix Table3for additional
details):
CPT codes; OR
10 and 11- digit NDCs; OR
Immunization records that contain data on vaccine code descriptor, vaccine
manufacturer, lot number, injection site, and date(s) of immunization.
9.3.1.1.Pfizer-BioNTech COVID -19 Vaccine Groups of Interest
While the primary vaccination group of interest is all in dividuals receiving Pfizer- BioNTech
COVID-19 vaccine (irrespective of receipt of seasonal influenza vaccination), additional
subsets of the study population will be studied, similar to the PRI SM safety surveillance
program of H1N1 vaccine safet y:10
Cohort A: Individuals vaccinated with Pfizer -BioNTech COVID- 19 vaccine who did not
receive the influenza vaccine during the flu season in which COVID -19 vaccination
occurred;
Cohort B: Individuals vaccinated with Pfizer -BioNTech COVID-19 vaccine who received
the seasonal influenza vaccine at least 42 day s prior to COVID -19 vaccination during the
same flu season in which COVID -19 vaccination occurred;
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077378
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 39of 194Cohort C: Individuals vaccinated with Pfizer -BioNTech COVID-19 vaccine who received
the seasonal influenza vaccine within 42 day s before or any time after COVID -19
vaccination during the same flu season in which COVID- 19 vaccination occurred;
Cohort D: Individuals vaccinated with both Pfizer -BioNTech COVID -19 vaccine and the
seasonal influe nza vaccine on the same day .
The following sub -cohorts will be assessed for each of the Cohorts A -D:
Individuals vaccinated with only 1 dose (i.e., incomplete course) of Pfizer -BioNTech
COVID-19 vaccine;
Individuals vaccinated with 2 doses ( i.e., complete course) of Pfizer- BioNTech
COVID-19 vaccine.
9.3.2.Baseline Characteristics
The following data elements regarding baseline demographic and clinical characteristics will
be assessed based on a 1-year baseline period prior to the date of vaccination with Pfizer -
BioNTech COVID -19 vaccine and date of seasonal influenza vaccination for active
comparator s.Depending on the attrition rate, the length of the baseline period may be
modified to 6 months. All diagnoses, procedures, and medications will be identified by the
International Classification of Diseases, Tenth Revision, Clinical Modification ( ICD-10-CM)
diagnosis codes, ICD -10-PCS (procedure coding system) codes, CPT, orHealthcare
Common Procedure Coding S ystem (HCPCS) procedure codes, and generic drug names, as
appropriate (AppendixTable1). The following demographic and clinical characteristics will
be assessed:
Demographic s:
Age
Sex
Race/ethnicit y
VHA service area
Clinical characteristics:
Smoking status
BMI
History of anaphylaxis/allergic reactions
Previous anaph ylaxisof vaccine component
History of hospitalizations
Frailty index
Charlson c omorbidity index (CCI)
Selected c omorbidities
oAutoimmune disease
oAsthma
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077379
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 40of 194oBleeding diathesis or condition associated with prolonged bleeding
oCancer
oCardiovascular conditions
oChronic kidney disease/dialy sis
oCOPD/interstitial lung disease
oDiabetes mellitus
oDown syndrome
oSickle cell disease
oHepatitis B virus ( HBV)
oHepatitis C virus ( HCV)
oHIV
oHyperlipidemia
oHypertension
oLiver disease
oNeurological disease
oOther immune deficiencies
oSolid organ transplant
oVenous thromboembolism (VTE)
Concurrent immunizations
oSeasonal influenza vaccine
oTetanus diphtheria and pertussis (Tdap or Td)
oChickenpox (varicella)
oShingles (herpes zoster recombinant and/or live)
oHuman papillomavirus (HPV)
oPneumococcal conjugate
oPneumococcal pol ysaccharide
oHepatitis A
oHepatitis B
oMeningococcal conjugate (MenACWY) and serogroup B meningococcal
(MenB)
oHaemophilus influ enza type b
Specific covariates of interest for the prioritized analy sis of myocarditis/pericarditis are
described in Section 9.7.8 .
9.3.3.Outcomes
The safety events of interest for active surveillance wereidentified based on the Priority List
of Adverse Events of Special Interest from the Brighton Collaboration’s SPEAC Project, the
FDA and CDC enhanced safet y monitoring recommendations.8,9Endpoints of special interest
in signal detection, as noted by the FDA and CDC’s ACI P are denoted in italics.9If
unanticipated potential safety eventsof interest are identified during the course of
surveillance, they will be added to the list and included in the anal yses.See
Appendix Table2for the operational definitions of the outcome variables based on ICD -10-
CM diagnosis codes, which may be refined as the study progresses based on additional
available information and the published literature ( e.g., frequency of ICD-10-CMcodes).
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077380
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 41of 194Outpatient ,ED,and/orinpatient setting s will be used to identify safety events of interest,
depending on the t ype of event. The specific encounter setting considered for each safet y
event of interest is summarized in Table 1.Any record of death will be captured, regardless
of whether the individual died in a healthcare or non -healthcare setting. The following safet y
events of interest will be assessed:
Neurologic :
Aseptic meningitis
Bell’s pals y
Cerebrovascular non -hemorrhagi c stroke
Convulsions/seizures in individuals with controlled epilepsy
Encephalitis/encephalomyelitis
Guillain-Barré Syndrome (GBS)
Generalized convulsion/seizures
Multiple sclerosis (MS)
Optic neuritis (ON)
Other acute dem yelinating diseases
Transverse myelitis (TM)
Immunologic :
Anaphylaxis
Arthritis and arthralgia/joint pain
Autoimmune thy roiditis
Fibromyalgia
Kawasaki disease (KD)
Multisystem inflammatory syndrome in adults (MIS-A)
Vasculitides
Cardiac:
Acute myocardial infarction (AMI)
Arrhythmia
Coronary artery disease (CAD)
Heart failure and cardiogenic shock
Microangiopath y
Myocarditis
Pericarditis
Stress cardiom yopathy
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077381
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 42of 194Hematologic :
Cerebrovascular hemorrhagic stroke
Chilblain-like lesions
Disseminated intravascular coagulation (DIC)
Deep vein thrombosis (DVT)
Hemolytic anemia
Hemorrhagic disease
Limb ischemia
Pulmonary embolus (PE)
Single organ cutaneous vasculitis
Thrombocytopenia
Thrombosis with thrombocy topenia sy ndrome (TTS)
Other:
Acute kidney injury
Appendicitis
Death
Erythema multiforme
Liver injury
Narcolepsy and cataplexy
Non-anaphylactic allergic reactions
Severe COVID -19 disease
Stevens-Johnson sy ndrome (SJS)/Toxic epidermal necrol ysis (TEN)
The risk intervals selected for each safety event of interest are based on biological
plausibility and precedents in the published literature ( Table 1). A safety event of interest
will only be counted if it can be assigned to 1) the risk interval following Pfizer- BioNTech
COVID-19 vaccination (all designs) , 2) the post -vaccination control interval (self-controlled
designs), or 3) the risk interval for the active comparators receiving seasonal influenza
vaccine(active comparator desig n). Events outside the intervals will not be counted. Onl y the
individual’s first instance of a safety event of interest following a specified clean window
(i.e., the occurrence -free baseline period used to define incident outcomes during which
individuals enter the study cohort only if the safety event of interest did not occur during this
period) will be included ; this means that if a safet y event of interest is identified but
diagnosis codes (or laboratory values in the case of select safet y events of interest)
corresponding to the safety event of interest are also observed during the clean window, it
will not be counted. The duration of the pre -specified window will differ by safet y events of
interest in order to rule out pre -existing events. This approa ch is consistent with the FDA’s
COVID-19 Vaccine Safety Surveillance Project.16Additionally , the length of the clean
window may be extended (e.g., 2 y ears) given the reduction in healthcare resource utilization
since the start of the pandemic. By way of example, a safety events of interest for the SCRI
design can be considered in the following ways:
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077382
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 43of 194If a safety event of interest occurs in the individual’s risk interval and there are no
other diagnosis codes for the same saf ety event of interest in the clean window (e.g.,
1-year prior to vaccination date ), the safet y event of interest should be assigned to the
risk interval.
oHowever, if an outpatient safety event of interest occurs in the clean window
and an inpatient occurrence for the same t ype of safety event of interest occurs
in the riskinterval, the inpatient occurrence will be counted in order to capture
event exacerbation.
oIf a safety event of interest occurs in the risk interval and another diagnosis
code for the sa me safety event of interest is identified during the post -
vaccination control interval, then the safety event of interest will only be
assigned to the risk interval
oIf a safety event of interest occurs in the post -vaccination control interval and
there are no other diagnoses for the same safety event of interest in the risk
intervaland clean window, then the safety event of interest will be assigned to
the post-vaccination controlinterval
The risk intervals for outcome evaluation for the active comparato rs who received
seasonal influenza vaccination will be the same as for the individuals who received
Pfizer-BioNTech COVID-19 vaccine.
However, it is possible that some safety eventsof interest do not have a precise time
interval from which to evaluate risk, for example if biological plausibility is unknown
or the diagnostic time window is more delay ed than anticipated. In these cases,
misspecification of the risk (and control) intervals could re sult in misclassification
and introduce bias, often toward the null. For instance, the assumption of a longer risk
interval than is true may result in “washing out” the signal, and an erroneously short
risk interval may similarly result in underestimation of effect when using post -
vaccination time intervals for self -control. To address this, sensitivity analyses may
be conducted with vary ing risk intervals (longer as well as shorter) in order to
increase the likelihood that the safet y risk is detected accur ately. Additionally , if
further refinement and evaluation is necessary , temporal scan statistics may be used
to empirically identify the at-risk time interval b y evaluating clusters of safety events
of interest. This will be further described in the SAP.
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077383
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 44of 194Table 1.Outcome algorithms for SCRI analysis, with risk and control intervals
Safety Event of Interest Setting
(Inpatient [IP], Outpatient
[OP], Emergency
Department [ED])Clean window Risk interval
(days)Post-vaccination
control interval
(days)
Neurologic
Aseptic meningitis IP only166 months161-423443-8434
Bell’s pals y16IP or OP 6 months 1-42 43-84
Cerebrovascular non -hemorrhagic
stroke16IP only 1 year 1-28 29-56
Convulsions/seizures in individuals with
controlled epileps y35IP or OP-ED 1 year 1-90 91-180
Encephalitis/encephalomyelitis16IP only 6 months 1-42 43-84
Guillain-Barré Syndrome (GBS)16IP, primary position on 1 year 1-42 43-84
Generalized convulsion/seizures10IP or OP-ED 6 months 0-14 15-29
Multiple sclerosis (MS)10,30IP or OP 1 year 1-42 43-84
Optic neuritis (ON)10,30IP or OP 1 year 1-42 43-84
Other acute dem yelinating diseases10,30IP or OP 1 year 1-42 43-84
Transverse m yelitis (TM)16IP or OP-ED 1 year 1-42 43-84
Immunologic
Anaphylaxis IP or OP-ED161 month160-1167-810,30
Arthritis and arthralgia/joint painaIP or OP 1 year 1-42 43-84
Autoimmune thy roiditisaIP or OP 1 year 1-42 43-84
FibromyalgiaaIP or OP 1 year 1-42 43-84
Kawasaki disease (KD)36IP only 1 year 1-28 29-56
Multisystem inflammatory syndrome in
adults (MI S-A)16IP or OP-ED 1 year 1-42 43-84
VasculitidesbIP only 1 year 1-28 29-56
Cardiac
Acute myocardial infarction (AMI)16IP only 1 year 1-28 29-56
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077384
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 45of 194Table 1.Outcome algorithms for SCRI analysis, with risk and control intervals
Safety Event of Interest Setting
(Inpatient [IP], Outpatient
[OP], Emergency
Department [ED])Clean window Risk interval
(days)Post-vaccination
control interval
(days)
ArrhythmiacIP only 1 year 1-42 43-84
Coronary artery disease (CAD)cIP only 1 year 1-42 43-84
Heart failure and cardiogenic shockcIP only 1 year 1-42 43-84
Microangiopath ybIP only 1 year 1-28 29-56
Myocarditis16IP or OP 1 year 1-42d43-84
Pericarditis16IP or OP 1 year 1-42d43-84
Stress cardiom yopathycIP only 1 year 1-42 43-84
Hematologic
Cerebrovascular hemorrhagic stroke16IP only 1 year 1-28 29-56
Chillblain -like lesionsbIP or OP 1 year 1-28 29-56
Disseminated intravascular coagulation
(DIC)16IP or OP-ED 1 year 1-28 29-56
Deep vein thrombosis (DVT)16IP or OP 1 year 1-28 29-56
Hemolytic anemiaeIP or OP 1 year 1-42 43-84
Hemorrhagic diseasebIP only 1 year 1-28 29-56
Limb ischemiabIP only 1 year 1-28 29-56
Pulmonary embolus (PE)16IP or OP 1 year 1-28 29-56
Single organ cutaneous vasculitisbIP only 1 year 1-28 29-56
Thrombocy topenia16IP or OP 1 year 1-42 43-84
Thrombosis with thrombocytopenia
syndrome (TTS)eIP or OP 1 year 1-42 43-84
Other
Acute kidney injuryfIP only 6 months 1-42 43-84
Appendicitis16IP or OP-ED 1 year 1-42 43-84
Death IP or OP 1 year 0-42 43-85
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077385
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 46of 194Table 1.Outcome algorithms for SCRI analysis, with risk and control intervals
Safety Event of Interest Setting
(Inpatient [IP], Outpatient
[OP], Emergency
Department [ED])Clean window Risk interval
(days)Post-vaccination
control interval
(days)
Erythema multiformegIP only 6 months 1-2 8-9
Liver injuryfIP or OP 1 year 1-42 43-84
Narcoleps y and cataplexy IP or OP161 year161-421643-84
Non-anaphylactic allergic reactions10,30IP or OP 6 months 1-2 8-9
Severe COVID -19 diseasehIP only 1 year 1-42 43-84
Stevens-Johnson sy ndrome (SJS)/Toxic
epidermal necrol ysis (TEN)gIP only 6 months 1 -2 8-9
Notes:
a.Published setting, clean window, and risk and control intervals for autoimmune disorders were applied to similar autoimmune r heumatic conditions (i.e.,
arthritis and arthralgia/joint pain, fibromyalgia and autoimmune thyroiditis).
b.Published setting, clean window, and risk and control intervals for DVT, pulmonary embolus and DIC were applied to other cardiovascular and
hematological disorders characterized by damage to the blood vessels and/or arteries and clotting (i.e., microangiopathy, lim b ischemia, hemo rrhagic
disease, chilblain -like lesions, single organ cutaneous vasculitis and vasculitides ).The published risk and control intervals for KD were applied to
vasculitides given that KD is a type of medium and small -vessel vasculitis.
c.Published setting, clean window, and risk and control intervals for myocarditis and pericarditis were applied to other cardiovascular conditions (i .e., heart
failure and cardiogenic shock, stress cardiomyopathy, CAD, arrhythmia).
d.For the prioritized safety analysis of myocarditis/pericarditis, additional risk intervals (i.e., 1 -7 days and 1 -21 days) will be examined and are described in
Section 9.7.8.
e.Published setting, clean window and risk and control intervals for thrombocytopenia were applied to hem olytic anemia and TTS.
f.Risk intervals of 42 days were applied for acute kidney injury and liver injury to be consistent with other similar safety ev ents of interest.
g.Published setting, clean window, and risk and control intervals for non -anaphylactic aller gic reactions were applied to hypersensitivity disorders (i.e.,
erythema multiforme and SJS/TEN).
h.As severe COVID -19 ranges from severe pneumonia, acute respiratory distress syndrome, and multisystem organ failure/MIS -A, a 1-42 day risk interval
was applied in order to capture the 14 -day incubation period of the disease and 4 -5 day period from exposure to symptom onset.
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077386
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 47of 1949.4.Data Source
The VHA is the largest integrated health care s ystem in the US, providing both inpatient and
outpatient clinical care to over 9 million Veterans enrolled at more than 170 medical centers
and 1,074 community -based outpatient clinics.14VHA’s health care delivery system is
organized regionall y around 18 Veterans Integrated Service Networks (VISNs) across the
US. Each VISN is responsible for health care planning and resource allocation in a particular
geographical region. For example, the VA New England Healthcare S ystem (VISN 1) covers
VHA facilities in Massachusetts, Connecticut, New Hampshire, Maine, and Rhode I sland,
while the VA Heart of Texas Health Care Network (VI SN 17) oversees the facilities in
Texas.
The VHA also maintains i ts own mortality data where 99% of enrollees’ deaths are reported
within one month of occurrence .As of January 7, 2021, the VHA has had over 174,000
confirmed COVID -19 cases.37Among active and convalescent cases , approximately 145,000
are Veterans and approximately 15,000are employees (with an estimated 630 as Veteran
employees).33While African American Veterans make up approximately 12% of the VHA,38
the burden of COVID -19 cases are skewed, with African American Veterans comprising
approximately 20% of all COVID -19 cases.37Approximately 7,099COVID-19-infected VA
patientshave died, an estimated 2,738in VHA hospitals.37
The objectives of this study will be addressed using data from VHA’s CDW, which is an
integrated EMR system with a centralized data warehouse that is updated on a daily basis.
The CDW stor es data in separate databases, one for each t ype of clinical information ( e.g.,
inpatient medication, inpatient admission, outpatient medication, outpatient visit). Individual
demographic information such as date of birth and gender are also available. Immunization
records include information on manufacturer, lot number, injection site, and concurrent
immunizations. The CDW does not include information on any care received outside of a
VHA facility .
Each individual is assigned a unique identification numbe r to allow for longitudinal
follow-up as well as to cross -reference to the various separate databases. For example, in
each inpatient admission record, there is information on the primary discharge diagnosis (and
as many as 15secondary diagnoses), date of admission, date of discharge, and length of stay .
This record can then be linked to other information of that inpatient stay located in other
files, including procedures that the patient underwent during the hospitalization, medical
specialty of the provi der, and prescriptions dispensed. Other files are similarly structured,
and therefore may be linked together to provide comprehensive information about the patient
and his/her medical encounters.
The VHA database is an appropriate data source to evaluate the safety of the Pfizer -
BioNTech COVID-19 vaccine for the following reasons. First, as the vaccine will be
distributed through government facilities (including VHA) as part of initial distribution ,
analysis of VHA data will provide earl y data on the safet y of the vaccine. Veterans living in
long-term care facilities and Veteran s who are healthcare workers will be prioritized in the
first wave of Pfizer -BioNTech COVID -19 vaccinations.39The VA offers eligible Veterans
long-term care services ranging from nursing homes and assisted- living centers to caregiver
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077387
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 48of 194support in the Veterans’ own homes.15Secondly, and relatedl y, VHA data are refreshed dail y
and would thus enable early and rapid data anal ysis. Third, the VHA population is on
average older than the general US population.40Of these, about 30% (roughly
1,000,000 individuals) use VHA health services almost exclusively (i.e., those with a priority
group of 1 or 4; Veterans assigned to Priorit y group 4 are either accepting VA assistance or
housebound benefits, or have been determined to be “catastrophicall y disabled” b y the
VA.33), which lends itself to having complete, longitudinal healthcare data for such
individuals who may be at higher risk of COVID -19 due to older age.41,42These priorit y
groups include Veterans with the highest levels of service- connected disability and are
therefore, the highest priority for VHA care.33Finally, the VHA population ha s, on average,
more comorbid conditions than the general population, which also indicates that these
individuals may be athigher risk of COVID -19.43While the VHA pop ulation is
predominantly male (approximately 90%), and thus lacks generalizability to females, it will
still provide a useful setting to examine real -world vaccine safet y.
Since it is possible that individuals may not have all their health encounters within the VHA,
(especially older veterans who are also covered by Medicare ), additional subgroup analyses
will be conducted in which the CDW data will be supplemented with data from CMS, linking
Medicare administrative claims data at the patient level to ensure a more comprehensive
evaluation of the care an individual receives. Medicare data will include eligibility files and
claims for services received in the inpatient and outpatien t setting, as well as skilled nursing
facilities, hospice, and home health agencies, and will cover the US primarily among those
aged 65 years or older.
9.5.Study Size
The sample size achieved will depend on the number of recipients of Pfizer -BioNTech
COVID-19 vaccine within the VHA database during the study period, which will increase
over time with subsequent analy ses. The population size will increase with each bi -weekly
analysis as the Pfizer -BioNTech COVID-19 vaccine becomes more readily available and a
greater number of individuals are vaccinated. Specificall y, the data will be refreshed on a
biweekly basis and a continuous sequential test procedure will be used to reevaluate data
according to this schedule. As of January 21, 2021, 112,201 doses of Pfizer -BioNTech
COVID-19 vaccine have been administered within the VHA (based on CPT code 91300) to a
total of 107,458 patients.
As a result of the ability to perform near -real-time analysis, the risk interval (and post -
vaccination control interval, for applicab le safety events of interest ) may have only partially
elapsed in some cases. To account for this, we will use methods adopted in previous
studies,10,44,45whereby risk in tervals will be scaled (or truncated) in order to ensure an
equivalent length (or a fixed ratio) of time is assessed between the control and risk intervals.
9.5.1.Power
Power calculations for the rapid cy cle analysis (RCA) approaches proposed for safety event
ofinterestsignal detection will be conducted according to the methods of Kulldorff et al.46,47
Table2illustrates the estimated power for the RCA approach using the Poisson -based
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077388
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 49of 194maximized sequential probability ratio test (MaxSPRT), and provides an overview of the
power required to detect vary ing relative risk (RR ) estimates with an alpha level of 0.01. T
denotes the expected number of safety events of interest to occur during the risk interval of
interest (Table2and Table 3). Power of ≥80% is ty pically desirable in drug safet y research.
Usually the FDA views a RR of > 3 as meaningful, so this has been used fo r power
calculations here.48As an example, as shown in Table2, the surveillance system would have
sufficient power (80.0%) to detect an increased risk of safet y events of interest associated
with the Pfizer -BioNTech COVID- 19 vaccine b y 3 fold when the expected number of safety
events of interest reaches 6 events.
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077389
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 50of 194Table2.Estimated S tatistical Power for the Poisson -based MaxSPRT46
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077390
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 51of 1949.6.Data Management
Data for this stud y will be stored and extracted from the VHA database (p reviously described
in Section 9.4) that contain information about patient demographics, vaccinations,
procedures, diagnoses, and death.
9.6.1.Case report forms (CRFs)/Electronic data record
As used in this protocol, the term CRF should be understood to refer to either a paper form or
an electronic data record or both, depending on the data collection method used in this study .
A CRF is required and should be completed for each included patient in the signal
verification phase that requires EMR and chart review ( seeSection9.7.3.3). The completed
original CRFs should not be made available in an y form to third parties, except for
authorized representatives of Pfizer or appropriate regulatory authorities, without written
permission from Pfizer. The CRF will consist of two parts: (1) a database CRF that will be
populated based on a direct extraction of data from the VA CDW for review by the
adjudicators; (2) an adjudication page that will be completed by an adjudicator afte r
reviewing data in the completed CRFs. Analysis Group shall ensure that the CRFs are
securely stored on VHA servers in an encry pted electronic and/or paper] form and will be
password protected or secured in a locked room to prevent access b y unauthorized third
parties.
Analysis Group has ultimate responsibility for the collection and reporting of all clinical,
safety, and laboratory data entered on the database CRFs and an y other data collection forms
(source documents) and ensuring that they are accurate,authentic/original, attributable,
complete, consistent, legible, timely (contemporaneous), enduring, and available when
required. The adjudication page must be signed by the adjudication committee members to
attest that the data contained on the formsare trueand accurate based on their review of the
EMR data . Any corrections to entries made in the CRFs or source documents must be dated,
initialed, and explained (if necessary) and should not obscure the original entry .
The source documents are the hospital or the ph ysician's chart. In these cases, data collected
on the CRFs must match those charts.
9.6.2.Record retention
To enable evaluations and/or inspections/audits from regulatory authorities or Pfizer,
Analysis Group agrees to keep all study -related record s, which includes study documents and
deliverables such as the protocol, SAP, aggregated results tables, SAS Institute (SAS)
programming files, and study report. The records should be retained b y Analysis Group
according to local regulations or as specifie d in the vendor contract, whichever is longer.
Analysis Group must ensure that the records continue to be stored securel y for so long as
they are retained.
If Analysis Group becomes unable for any reason to continue to retain study records for the
requiredperiod, Pfizer should be prospectivel y notified. The study records must be
transferred to a designee acceptable to Pfizer.
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077391
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 52of 194Study records must be kept for a minimum of 15 years after completion or discontinuation of
the study, unless Analy sis Group and Pfi zer have expressly agreed to a different period of
retention via a separate written agreement. Record must be retained for longer than 15 years
if required b y applicable local regulations.
Analysis Group must obtain Pfizer's written permission before dis posing of any records, even
if retention requirements have been met.
9.7.Data Analysis
Detailed methodology for summary and statistical anal yses of data analyzedin this study will
be documented in a SAP, which will be dated, filed, and maintained b y the spons or. The SAP
may modify the plans outlined in the protocol; any major modifications of primary endpoint
definitions or their anal yses would be ref lected in a protocol amendment. The SAP will also
provide additional detail regarding the evaluation of a thres hold of excess risk for each of the
safety events of interest. Consistent with the approach of Kulldorff et al., t his will be
determined based on background incidences for each event ( e.g., based on historical
influenza vaccinated active comparator cohort data to be evaluated during the study ), in
addition to pre -specified significance level ( e.g., alpha=0.01 or 0.05) and power.46This
information, in conjunction with a clinically meaningful RR ( e.g., 2 or 3) and the expected
upper limit of events under the null hypothesis will allow for the calculation of critical values
of each safet y event of interest using the MaxSPRT method. Greater power (e.g. , 80%) is
also a natural criterion to use when selecting the upper limit on the length of surveillance,
and in turn, the expected number of events to occur, although there is ultimately a tradeoff
between that power and the time allowed to identify the expected number of events to occur.
Data analy ses will be conducted using SAS Enterprise Guide version 7.1 (SAS I nstitute Inc.,
Cary, NC) or R Version 3.5.3 or its latest version (R Core Team, Vienna, Austria). In
addition, SaTScan will also be used to conduct specific temporal anal yses.
9.7.1.Baseline Characteristi cs
Baseline demographics and clinical characteristics for individuals receiving Pfizer -BioNTech
COVID-19 vaccine and individuals who received seasonal influenza vaccination will be
summarized using descriptive statistics, consisting of the mean and standar d deviation (SD)
and median (interquartile range [IQR]) values for continuous variables and frequency
distributions for categorical variables. Incidence rates ( i.e., per-patient per -month) for prior
hospitalizations may be calculated as the number of events divided by person-time of
observation since the length of the baseline period may vary between individuals.
Standardized differences will be calculated between Pfizer BioNTech COVID -19 vaccine
recipients and active comparators who received seasonal influ enza vaccination to evaluate
whether there are an y major differences in individuals’ baseline characteristics. Standardized
differences <10% will indicate that matching has appropriatel y balanced the charac teristics
between recipients of the Pfizer- BioNTech COVID -19 vaccine and seasonal influenza
vaccine.
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077392
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 53of 1949.7.2.Vaccine Utilization Patterns
Descriptive statistics will also be used to summarize vaccine utilization patterns, including
proportion of individuals receiving vaccine, 2 -dose completion rate ,distribution of time gaps
between the first and second dose , and care setting where immunization was received ( e.g.,
outpatient clinic, pharmacy , inpatient ward). Counts of individuals who received a COVID -
19 vaccine from a different manufacturer in addition to the Pf izer-BioNTech COVID-19
vaccine will be summarized .
9.7.3.Safety Signal Analyses
Several analy ses corresponding to the designs discussed previousl y will be conducted to
detect safet y signals associated with Pfizer -BioNTech COVID -19 vaccine. Analy ses will be
conducted among all individuals receiving the vaccine, individuals who received
Pfizer-BioNTech COVID-19 vaccine without seasonal flu vaccine (Cohort A will be used for
SCRI; Cohort B +Cwill be used for active comparator analyses), and individuals receiving
Pfizer-BioNTech COVID- 19 vaccine and seasonal flu vaccine on the same day (Cohort D),
along with sub- cohorts receiving onl y one dose vs. two doses.
A stepwise process, illustrated below, will be performed for signal d etection, evaluation, and
verification ( Figure 3).This approach has been adapted from the Active Monitoring Protocol
of the FDA’s COVID -19 Vaccine Safet y Surveillance Project.16The statistical approach
described below may be modified further based on data availability , additional clinical input,
and for consistency or to complement similar studies of Pfizer -BioNTech COVID-19
vaccine.
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077393
BNT162b2 (COVID- 19 vaccine)
C4591012 NON -INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 54of 194Figure 3.Steps in Signal Detection, Evaluation , and Verification
Notes:
[1] List of safety events of interest and corresponding definitions may be refined as the study progresses based
on additional available information.
[2] The risk and control intervals selected for the SCRI analysis for each safety event of interest are based on
biological plausibility and precedents in the literature. Only the individual’s first instance during the specified
clean window (i.e., the interval used to define incident outcomes) will be included. Note that only the first
inpatient or outpatient occurrence of a safety event of interest following the clean window will be used to
identify incident events ( e.g., if an inpatient saf ety event of interest occurs in the clean window , a repeat
occurrence will not be counted in the risk interval). However, event worsening will be counted as a safety event
of interest. For example, if an outpatient safety event of interest occurs in the cl ean window and an inpatient
occurrence for the same type of safety event of interest occurs in the risk interval, the inpatient occurrence will
be counted as a safety event of interest.
9.7.3.1.Signal Detection
Signal detection will rely on SCRI design with comparison to post -vaccination control
intervals
for the two safety events that require COVID -19 diagnosis (i.e., severe COVID -19
disease, MIS -A) and active comparator design for the remaining safety events. While the
active comparator design will be the main analysis for signal detection because it can be
performed the fastest, it cannot be used for safety events that require COVID -19 diagnosis
becausehistorical controls would not meet the criteria of having a COVID -19 diagnosis.
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077394
BNT162b2 (COVID- 19 vaccine)
C4591012 NON -INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 55of 1949.7.3.1.1. Sequential Testing -SCRI Desi gn using the Binomial -based MaxSPRT for
Comparison to P ost-vaccination Control Intervals
For the Two Safety Events Requiring
COVID-19 Diagnosis
The goal is to provide rapid- cycle, near real -time safet y surveillance. In the signal detection
phase, the SCRI analysis withpost-vaccination control intervals will be used for certain
safety events of interest ( i.e., severe COVID -19 disease , MIS-A). All other safety events of
interest will be assessed in the signal detection phase using the active comparator design. The
post-vaccination control period will be assessed once enough post -vaccination time has
accumulated.
To account for multiple testing and bi -weekly review of the data, the MaxSPRT using a
binomial probability model will be applied. The null hy pothesis (H 0) assumes that the risk of
a safety event of interest during the risk interval is equivalent to the risk of the same safety
event of interest developing during the control interval, accounting for differences in interval
duration as needed ( e.g.,for safety events of interest such as dem yelinating disease), meaning
a RR of 1 is specified under H 0.30The one-sided composite alternative h ypothesis (H a)
assumes that the risk of a safet y event of interest during the risk in
terval is greater than the
risk of the same safety eventof interest developing during the control interval, accounting for
differences in interval duration ( i.e., RR>1, Hais applicable across a range of RRs).46
Specifically, for the Pfizer- BioNTech COVID -19 vaccine, let xrepresent the total count of
safety events of interest in the control interval (Figure
4), let y represent the total count of
safety events of interest in the risk interval, and let rrepresent the ratio of yto x under the
null hypothesis. Thus, when the total control interval du ration and total risk interval duration
are equal, r will be 1. The RR is estimated by ..
.44The RR and corresponding 99%
confidence intervals (CIs) will be calculated.
Figure 4.Example of SCRI Design for
a Safety Event of Interest with a 42-day Risk
Interval and a Post -vaccination Control Interval
For the binomial model, the log -likelihood ratio (LLR) is calculated as the log probability of
observing this distribution of y under H a, divided by the probability of this occurring under
H0.46This ratio is calculated whenever new data arereceived to account for the continuous
data stream until the full 42- day risk period is complete.
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077395
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 56of 194 =ln ( | )
( | )
Once the LLR test statistic reaches a pre -specified critical value , a signal is detected.
Specificall y, the null hypothesis will be rejected if the LLR exceeds the critical value. The
null hypothesis will not be rejected if the LLR does not r each or exceed the critical value, if
the total number of safety events of interest reaches a pre -specified upper limit, or if
surveillance ends without reaching this upper limit.44
For each safety event of interest (and specific to each age group, if age -stratified anal yses are
conducted), the critical value of the LLR will be determined based on the safety event of
interest specific upper limit of expected safety events of interest and alpha level.44Upper
limits will be determined based on the expected number of safety events of interest under the
null hypothesis, assuming the risk after Pfizer -BioNTech COVID- 19 vaccination is no
greater than the risk of safety events of interest after seasonal influenza vaccination.
Therefore, upper limits will be chosen such that they would not usually be reached.
9.7.3.1.2. S equential Testing -Poisson-based MaxSPRT for Comparison to Active
Comparators who Received Seasonal Influenza Vaccination
For comparison with active comparators who received seasonal influenza vaccination, the
Poisson-based MaxSPRT will be applied, following the same statistical approach as
described above, but using a Poisson probabilit y distribution. In the Poisson MaxSP RT
approach, the event frequency of safety events of interest in the risk interval after Pfizer-
BioNTech COVID -19 vaccination will be compared to a background rate ofsafety events of
interest in the risk interval after seasonal influenza vaccination in fi ve prior seasons, ranging
from 2014/15 through2018/19. This approach is particularl y important for extremely rare
safety events of interest (i.e., less than 50 anticipated based on historical influenza vaccine
ratesof safety events of interest ).30Poisson MaxSPRT is used to monitor very rare safet y
events of interest as binomial MaxSPRT may not detect a signal, despite a clinically
meaningful RR.44This will also allow for more timely anal ysis using historical data, as well
as improved power and sample size.
GBS is of particular interest relative to the safety profile of Pfizer -BioNTech
COVID-19vaccine. As GBS is an extremely rare safety event of interest , the primary RCA
proposed will focus on Poisson MaxSPRT and apply an alpha of 0.05. The Poisson
MaxSPRT has increased power to detect a signal with fewer occurrences of the safet y event
of interest. However, this method cannot fully control for confoundin g by indication.
9.7.3.1.3. Critical Values and Alpha Spending
Critical values for the LLR test statistic are shown below in Table 3based on calculations
conducted b y Kulldorff et al 2011.46For example, assuming T =6 (number of expected
events under the null) and RR =3, which corresponds to a power of 80.0% (See
Section9.5.1), the critical value would be 5.14 using alpha of 0.01 for the Poisson -based
MaxSPRT. As noted previously , each safety event of interest will be evaluated separatel y to
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077396
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 57of 194determine a critical value based on background incidence, alpha, power, and clinically
meaningful RR. These details will be addressed in the SAP.
Table 3. Critical Values for Poisson- based MaxSPRT
Multiple ty pes of alpha spending functions can be employ ed to calculate the cumulative rate
at which Ty pe 1 error (alpha) probabilit y is spent during sequential testing.49To achieve
optimal expected time -to-signal, especiall y when historical Poisson data are used with
surveillance data, a power-typeconvex alpha spending shape will be used based on published
literature.49Additionally , ρ =1.5 is referenced as a “rule of thumb” as it i s suggested to be
appropriate in most applications.
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077397
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 58of 1949.7.3.2.Signal Evaluation
Signals are detected when the event frequency of asafety eventof interest during the risk
interval following vaccination with Pfizer -BioNTech COVID -19 vaccine is significantl y
increased compared to the event frequency of the same safety events of interest in the control
comparator (i.e., the critical value is achieved and surpassed). If signals are indeed detected
for safety events of interest based on the anal ysis described above, furt her evaluation is
warranted to refine and confirm such detections. This will consist of the additional analy ses
describe in the following sections, which will be conducted every six months.
9.7.3.2.1. Post-Signal Quality Assurance
Quality assurance will first be con ducted in order to assess the quality of the data and
analysis that produced the signal. While quality control measures will be conducted during
the signal detection phase (seeSection 9.8), post-signal quality assurance will also be
performed during the signal evaluation phase. This will include a comprehensive quality
assurance (for example, check for possible duplications of claims or medical record s,
checking for unusual clustering in claim or medical record accrual b y service date for
potential coding issues, check for geographical distribution of cases that may be related to lot
numbers or diagnostic practice) . In addition, for signals detected vi a active comparison,
additional anal yses comparing to p ost-vaccination control intervals may be formed to check
for consistency . Signals will also be confirmed across all of the safety studies planned to be
performed (i.e., C4591008, C4591011, C4591012) to confirm that specific data sources are
not biased.
9.7.3.2.2. Multivariate Adjustment using Poisson Regression
If signals are detected and persist after conducting quality assurance, further evaluation via
statistical measures are warranted. Specifically , to investigate whether potential signals
identified via Poisson MaxSPRT for the comparison to active comparators with seasonal
influenza vaccination are not confounded ( i.e., to take into account baseline differences
between the Pfizer BioNTech COVID -19 vaccina ted and active comparator populations), a
multivariate Poisson regression anal ysis will be conducted to compare the incidence rates of
the safety events of interest occurring within the risk intervals. T he predictor would be
whether the individual had received the Pfizer -BioNTech COVID -19 vaccine or had received
the influenza vaccine during historical seasons. Analyses will be adjusted for relevant
baseline and/or clinical characteristics ( e.g., age, sex, race, CCI and/or specific comorbidities
of interest, state, etc.).10
If the signal remains, based on an IRR > 3 with a p- value <0.01 from the adjusted Poisson
regression, further evaluation may be considered via signal verification.
9.7.3.2.3. Assessment of Temporal Clusters
Vaccine safet y surveillance must allow for sufficient ty pe I error probability for rapid
detection ofsafety events of interest , and statistically significant signals must be studied
further to ensure that a true association is present.50Therefore, the presence of temporal
clusters will be assessed using the software SaTScan to calculate temporal scan statistic in
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077398
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 59of 194order to further refine safety signals detected from the signal detection analy ses.30A temporal
scan statistic accounts for multiple testing present during overlapping risk intervals. The null
hypothesis assumes that there is no association between the safet y eventsof interest and
immunization, and safety events of interest are assumed to be distribut ed independentl y and
uniformly during a period of time subsequent to Pfizer -BioNTech COVID- 19 vaccination.30
A temporal scan statistic will be generated b y moving a time interval of fixed length across
the risk interval, comparing the number of observed versus expected safety events of interest
within the time interval under the null hy pothesis.46
9.7.3.2.4. Sequential Testing -SCRI Design using the Binomial MaxSPRT for
Comparison with Post -Vaccination Control Intervals
Any safety events of interest with signals detected and not alread y analyzed during the signal
detection phase with the SCRI design using the binomial -based MaxSPRT will be anal yzed
during the signal evaluation phase using SCRI design using the binomial- based MaxSPRT
method for post -vaccination control intervals. This will be conducted during the signal
evaluation phase in order to allow time to accumulate during the post -vaccination control
period. The same statistical methodology as described above will be applied.
9.7.3.2.5. SCCS Design using Conditional Poisson Regression for Comparison with
Post-Vaccination Control Time Period
Similar to the SCRI design with post-vaccination control intervals, SCCS design with post -
vaccination control time period will include cases (i.e., individuals vaccinated with the
Pfizer-BioNTech COVID- 19 vaccine who experience safety events of interest following
vaccination) to compare the incidence of safet y events occurring in the risk interval
following vaccination with the incidence of safet y events occurring during all other times
post-vaccination in the same individual until the earliest of 183 day s after the Pfizer-
BioNTech COVID -19 vaccination, disenrollment, death, end of data availability .This
analysis will be conducted for all safety events of interest with signals detected in the signal
detection phase . The SCCS design differs from the SCRI design in that instead of having
fixedpost-vaccination control intervals of the same duration as the risk interval, it has a time -
varying post-vaccination control time period that includes all-non risk interval time from
Pfizer-BioNTech COVID -19 vaccination date until the earliest of 183 days after Pfizer-
BioNTech COVID -19 vaccination , disenrollment, death, end of data availabilit y.23
For individuals who receive two doses of the vaccine, the post-vaccination control time
period may include time before and after Pfizer-BioNTech COVID -19 vaccine dose 2 or
solely include time after Pfizer-BioNTech COVID- 19 vaccine dose 2. See Figure 5below for
an example ofan individual who receives two doses of Pfizer-BioNTech COVID- 19 vaccine ,
where the safet y event of interest has a 42 -day risk interval window (e.g., Bell’s pals y; Table
1 in Section 9.3.3). Figure 5A demonstrates the SCCS design with the second dose received
21 days after the first (i.e., the risk interval for dose 1 overlaps with the risk interval for dose
2), while Figure 5B demonstrates the SC CS design with the second dose received 60days
after the first (i.e.,
with gaps between the end of dose 1 risk interval and dose 2). The post -
vaccination control time peri od isdisplayed belowasshading with gray lines.
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077399
BNT162b2 (COVID- 19 vaccine)
C4591012 NON -INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 60of 194Figure 5.Example of SC CS Designfor Safety Event of Interest with a 42 -day Risk
Intervalwith Post -vaccination Control Intervals when Two Doses of Pfizer -
BioNTech COVID -19 Vaccine are Administered
Compared to the SCRI design, the SCCS design with post -vaccination control time period
will have increased statistical power, which isespecially useful
forthe study of raresafety
events of interest. A conditional Poisson regression model will be used to compare the rates
of safety events of interest in the risk interval vs post -vaccination control time period. From
this model we will report rate ratios and 95% CIsthat will be interpretated as the rate ratio
for the safety event of interest in the risk interval compared to the control interval.
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077400
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 61of 1949.7.3.2.6. Comparison with Contemporary Unvaccinated Controls
To address period effects that could impact the appropriateness of using the historical
comparator cohort, a nalyses will also be performed comparing individuals who received the
Pfizer-BioNTech COVID- 19 vaccine to individuals who were not vaccinated at that point in
time.The unvaccinated controls will be assigned an index date matched to a corresponding
Pfizer-BioNTech COVID -19 vaccinee’s vaccination date ; these individuals can later receive
the Pfizer -BioNTech COVID- 19 vaccine and enter the vaccination group if all inclusion and
exclusion criteria are met . To address possible selection bias due to health seeking behaviors,
the unvaccinated controls will be selected from a population of patients who have regular use
of VHA medical care, defined as at least two outpatient (excluding ED, as ED visits may not
be considered regular) or inpatient encounters in the one year prior to vaccination. The
encounters must be separated b y > 30 days (for inpatient, by admission date), and at least one
must be within six months prior to index date. This approach is consistent with the Center for
Biologics Evaluation and Research ( CBER) Surveillance Program, Draft Master Protocol
Assessment of Risk of Safety Outcomes Following COVID -19 Vaccination.23
Inverse probability treatment weighting (I PTW)will be used to ensure comparability
between the Pfi zer-BioNTech COVID -19 vaccinated cohort and contemporary unvaccinated
controls. The IPTW approach uses weights to create a “pseudo -population” in which the
distribution of covariates is on average the same in each cohort.51IPTW is defined as the
inverse of the individual’s probability of receiving the first dose of Pfizer-BioNTech
COVID-19 vaccine, conditional on their demographic and clinical characteristics. This
approach assumes that an individual’s probability of receiving Pfizer -BioNTech COVID -19
vaccination is constant for the first and second doses of the vaccine , as the wei ght will be
applied for both doses .23Initial inverse probability weights will be calculated as 1 /
propensity score (PS) for individuals who received the Pfizer -BioNTech COVID- 19 vaccine
and 1/ (1- PS) for individuals with no record of COVID -19 vaccination. To avoid extreme
weights, each individual’s weight will be stabilized by the marginal probability of being in
their assigned cohort. Therefore, the stabilized weights will be calculated as Pr (Pfizer -
BioNTech COVID -19 = 1) / PS for individuals who received the Pfizer-BioNTech COVID -
19 vaccineand 1 -Pr (Pfizer-BioNTech COVID -19 = 1) / (1- PS) for the contemporary
unvaccinated controls. The distribution of weights will be examined to check for extreme
values, and truncation will be considered if necessary .
Weighted Cox regression with robust standard errors to account for within -subject
correlation will be conducted to compare the risk of safet y events of interest between cohorts.
Hazard ratios and corresponding 95% CIs wi ll be summarized .
9.7.3.3.Signal Verification
9.7.3.3.1. Medical Records Review
As part of the signal evaluation process, diagnostic validation of the detected safet y events of
interest (i.e.,cases) via a djudication of patient medical records by VHA clinicians for
outcome verification in a representative sample of cases will be conducted. The total number
of charts to be reviewed will depend on the number of safety events of interest detected, such
that all cases may be reviewed for safety events of interest where a small nu mber of events
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077401
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 62of 194result in signal detection and a representative sub -sample may be reviewed for safety events
of interest where a larger number of events results in signal detection .52For rare events,
potentially all cases ma y be adjudicated. An adjudication charter will be developed to govern
signal evaluation and medical records review. Specificall y, validation of detected safet y
events of interest will be performed through patient medical chart review in collaboration
withan adjudication committee comprised ofthe treating or trained healthcare
professionals.52
9.7.4.Seasonality- Adjusted Cases -Centered Method
A case-centered anal ysis for specific safety events of interest for which signals were detected
may also be conducted in order to account for bias caused b y seasonality of safety events of
interest and vaccination.30.53
This method will use data on all safety event of interest cases that occur after vaccination
with Pfizer -BioNTech COVID -19 vaccine. Logistic regression will be used to compare the
number of safet y event of interest cases that were vaccinated inside versus outside a pre-
specified risk interval , asof the date of the safety events,where the total number of
vaccinations given inside versus outside the risk interval (in the population of all vaccinees)
is used as the offset term.44Specifically , the association of vaccination with risk of safet y
events of interest will be estimated from a logistic regression model that includes
summarized data with one record per risk set. The key independent variable will be the
proportion of the risk set who were in the risk interval o n the date of the safety event of
interestoccurrence. In this way , risk sets are anchored to calendar dates, and confounding b y
seasonality of the safet y eventsof interest and vaccination is addressed .53Note that other
confounders may also be adjusted for b y restricting risk sets to vaccinees similar with respect
to select characteristics ( i.e., through stratification).
9.7.5.End-of-Season and End -of-Surveillance Analyses
For any safety event of interest with signals detected, end -of-season anal yses (over the course
of the 30- month period) and an end -of-surveillance anal ysis (i.e., at 30 months, after the end
of surveillance) will be conducted. Similar methodology will be applied for the end -of-
surveillance anal ysis and end-of-s eason anal ysis conducted for seasonal influenza vaccine in
order to adjust for the seasonality of both disease and vaccine administration.10This approach
will be able to define the true risk intervals after each dose and estima te the risk for potential
safety events of interest after both dose 1 and 2 of the Pfizer -BioNTech COVID- 19 vaccine,
as well as the abilit y to discern whether or not one or two doses of seasonal influenza vaccine
were administered during the same period.
The number of events in the sum of three distinct risk intervals will be compared to the
control interval, adjusting for potential differences in interval length, to estimate the RR of
Pfizer-BioNTech COVID-19 vaccine compared to the influenza vaccine . Inorder to monitor
the safety after the first and full course of the vaccine, the number of potential safety events
of interest occurring in three separate risk intervals (P 1, P2, P3) will be estimated ( Figure 6).
P1represents the risk interval after the first dose only , excluding any overlap in risk intervals
with the second dose. P 2represents the overlapping risk intervals for first and second dose of
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077402
BNT162b2 (COVID- 19 vaccine)
C4591012 NON -INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 63of 194the vaccine. P 3represents the risk interval of the second dose of the vaccine, excluding the
overlapping risk interval alread y captured in P 2. This design will allow for the assessment of
risk during the appropriate periods, regardless of the time interval between vaccine doses. As
multiple endpoints will be assessed, 99% CI swill be calculated around the RR in order to
ascertain whether the Pfizer -BioNTech COVID -19 vaccine is associated with
safety events
of interest.
Figure 6.Example of Risk (P 1, P2, P3) and Aggregate Post-vaccination Control
Intervals for the SCRI End -of-surveillance Analyses of 1 or 2 Doses of
Pfizer-BioNTech COVID -19 Vaccine
In Figure 6A, P1+ P2+ P3represent the risk intervals where a safet y event of interest may
occur. In Figure 6B,there is no overlapping risk interval so that P 1+ P3represent the risk
intervals where a safety event of interest may occur.The timing of the risk and control
intervals may be adjusted fo
r in order to control for the effect of seasonalit y across the
intervals assessed.
9.7.6.Subgroup Analysis
Separate anal yses of baseline characteristics, vaccine utilization patterns, signal detection,
signal evaluation, and signal verification insubgroups of interest may be conducted based on
feasibility , sample size, and data available.
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077403
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 64of 1949.7.7.Incidence Rates and Time to Safety Event of Interest Analysis
Incidence rates (and corresponding CIs) will be calculated from safety event of interest signal
detection anal yses. Kaplan -Meier methods will be used to anal yze time-to-event (i.e., time to
safety event of interest ). If individuals do not experience thesafet y events of interest , they
will be censored at the end of the risk interval. Median t ime to safety event of interest and
corresponding CI swill be summarized .
9.7.8.Prioritized Safety Analysis of Myocarditis/ Pericarditis
Notably, CDC recentl y investigated the occurrence of myocarditis/pericarditis following
mRNA COVID -19 vaccinations.17Therefore, separate safet y analyses will be prioritized and
performed to assess the risk of my ocarditis/pericarditis following Pfizer -BioNTech COVID-
19 vaccination, to provide additional context to the CDC investigation and addr ess regulatory
requestsfor further information on this safety event. Therefore, separate analy ses will be
prioritized and conducted to better understand the risk of my ocarditis/pericarditis following
Pfizer-BioNTech COVID- 19 vaccination in the VHA. This a nalytical approach is intended to
align with the methodology used bythe Vaccine Safet y Datalink (VSD)and preliminary
findings of m yocarditis/pericarditis published by ACIP on June 23, 2021.17,18The VSD
protocol defines m yocarditis/pericarditis (ICD-10-CM codes B33.22, B33.23, I 30, I40)
events as the first event in 60 day s identified through an ED or inpatient encounter, without a
first diagnosis of COVID -19 (i.e., COVID -19 diagnosis code or positive COVID -19 lab test)
in the 30 day s prior to or on the day of the event. This analy sis will follow the outcome
definition used in the VSD and uses three distinct risk intervals following vaccination (i.e., 1 -
7 days, 1-21 days, and 1-42 days). This definition and the statistical approach differ from the
primary analysisdescribed in this protocol , but will facilitate comparison with the results
presented by ACIP.16,17
This analy sis will include all individuals in the primary anal ysis who were vaccinated with
the Pfizer -BioNTech COVID- 19 vaccine. The number of m yocarditis/pericarditis events in
the risk interval will be identified, and incidence rates per million doses will be summarized .
Subgroup anal yses will also be performed, stratified by age ( e.g., 12-39 years, 40-49 years,
50-64 years, 65+ y ears), gender, and race/ethnicity , respectively.
In addition, vaccinated concurrent comparators will be selected among individuals who
received the Pfizer -BioNTech COVID -19 vaccine, and then events will be compared
between vaccinees who are in their risk interval and vaccinees who are concurren tly, on the
same calendar date, in their comparison interval. Poisson regression will then be used to
calculate incidence rate ratios and 95% CIs to compare the rate of m yocarditis/pericarditis
events between those individuals who were in a risk interval v ersus those individuals who
were in a comparison interval on the same calendar day. Data will be analy zed at the stratum
level for each calendar day and will include strata for the independent variable of interest
(i.e., risk vs. comparison interval) and f or adjustment variables (i.e., age group, sex,
race/ethnicity , and VHA service area). Thus , the number of m yocarditis/pericarditis events in
a risk or comparison interval on a calendar day will be modeled as a function of whether the
stratum’s vaccinees ar e in a risk versus comparison interval on that calendar day , controlling
for age, sex, race/ethnicity , and VHA service area. The log of the number of individuals
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077404
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 65of 194contributing data to each stratum on each calendar day will be included as an offset term in
the Poisson model. Additionally , if it is suggested that calendar time may beassociated with
risk of post -vaccination my ocarditis/pericarditis, to account for changes COVID- 19 and other
viruses circulating and other ecologic factors, analyses may alsobe stratified by calendar
time, for example in 6 months increments.
In addition to analyzing codified data, C ase confirmation for my ocarditis/pericarditis events
identified in the codified data will be conducted based on medical chart review.
Myocarditis/peric arditis cases will be confirmed and validated using the Brighton
Collaboration’s case definitions .19Risk factor analysis will also be conducted via logistic
regression among confirmed cases of myocarditis/pericarditis to further evaluate variables
associated with the event ; additional details will be provided in the SAP.
Additional data surrounding risk factors, clinical course, and sequelae of identified
myocarditis/pericarditis events up to 365 day s following the event will be collected and
summarized .These will include an examination of other possible etiologies/risk factors (i.e.,
prior COVID -19 infection, prior Coxsackie infection, other prior viral infections, other
vaccines received, como rbid immunocompromising conditions and s ystemic immune -
mediated diseases, demographics, and medication history); time between Pfizer -BioNTech
COVID-19 dose (first and second) and onset of my ocarditis/pericarditis; echocardiogram
information; lab troponin i nformation; sy mptoms (e.g., chest pain, shortness of breath,
weakness or fatigue, arm or shoulder pain, heart palpitations cough, swelling in abdomen or
legs, fever); treatments received for my ocarditis/pericarditis (e.g., non-steroidal anti -
inflammatory drugs (NSAIDs), colchicine, corticosteroids, pericardectomy ); healthcare
resource utilization following the event, and long- term sequelae for up to one y ear following
the event (for my ocarditis: recovery , sudden cardiac death, heart failure cardiogenic shoc k,
fulminant my ocarditis, inflammatory cardiomy opathy, heart transplant, arrhy thmia; for
pericarditis: recovery , chronic pericarditis, restrictive pericarditis, recurrent pericarditis).
9.8.Quality Control
Data for the study will be extracted from electronic d atabases in the CDW of the VHA. Each
data content area in the CDW is subjected to similar checks, from high level variable
name/type checks, to detailed trending comparisons. As an example, the diagnostic data is
subject to the following checks:
Referenced table exists
Diagnosis type is correctly assigned b y codes defining the diagnosis
Percentages, rates, are as expected (check ranges and for missing)
Both inpatient and outpatient diagnosis codes are captured. Referenced variables exist
and are of appropri ate length and t ype
Data retrieval will be coordinated b y an experienced programmer/anal yst. The anal yst will
write programming for retrieval of each data element from the electronic databases. Double
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077405
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 66of 194programming will be performed for the first iteration o f the analyses; results/datasets will be
compared, and if an y discrepancies are identified, both programmers will determine a
resolution, bringing in a third programmer if needed. Subsequent iterations of anal yses (i.e.,
re-runs of the anal yses) will be au dited by a senior programmer. All tables will be reviewed
by the project manager and the principal investigator to evaluate for internal consistency of
counts and totals. All calculated variables will be checked against the component variables
(cross tabs) to ensure accuracy . For example, categorical age would be compared with
continuous age to confirm that each category of age contained onl y individuals of the
expected age ranges within that category .
9.9.Strengths and Limitations of the Research Methods
To identify individuals who experienced safety events of interest associated with Pfizer-
BioNTech COVID -19 vaccine, the SCRI method of signal detection offers some key
advantages. The SCRI approach inherentl y adjusts for within-individual confounders, such as
age, sex, and confounding b y indication. While control intervals can be defined both pre -and
post-vaccination, the current study will only use a post-vaccination control period because
individual smay be more vigilant for the reporting of possible safet y events after they receive
a vaccine than before vaccination, which may bias the comparison between a post -vaccine
risk interval with a pre -vaccine control interval.54Specificall y, safety events of interest may
be more likel y to be reported or sought care for after vaccination with Pfizer -BioNTech
COVID-19 vaccine than before ,which may result in bias against the Pfizer -BioNTech
COVID-19 vaccine . Lastly, SCRI allows fo r near real -time monitoring of safet y risks
associated with the Pfizer -BioNTech COVID- 19 vaccine. Similar considerations apply to the
SCCS design with post -vaccination control interval that will be used in the signal evaluation
phase.
The comparison of vaccinated to contemporary unvaccinated controls y ieldsa more
interpretable result than other planned anal yses using SCRI and active comparators who
receive seasonal influenza vaccination (i.e., the increased risk of experiencing a specific
safety event due to Pfizer -BioNTech COVID -19 vaccination). The potential for selection bias
(i.e., confounding b y indication, health y user bias) will be mitigated b y comparing baseline
demographic and clinical characteristics among the unvaccinated controls. Unvacci nated
controls will be required to have similar healthcare -seeking behaviors asPfizer-BioNTech
COVID-19 vaccinees, including a t least 1 y ear of enrollment in and no disenrollment from
VHA benefits prior to their match date. This design is also not limited to assumptions
required b y SCCS and SCRI , and can also be completed rapidly as it does not require post -
vaccination control intervals. However, it is noted that the mass vaccination campaign in the
past year has provided various channels to receive vaccin ation, and therefore unvaccinated
controls may be misclassified if they are vaccinated outside of the VHA.
The VHA CDW provides a range of benefits, including its comprehensive structure, large
number of variables, and electronic accessibility . The VHA C DW also includes EMR data
that include structured fields (which will be used for signal detection) and open fields (such
as physician notes, which will be used for signal verification and case validation, as needed).
Importantly , the VHA CDW retains electronic immunization records that include
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077406
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 67of 194manufacturer name and lot numbers, facilitating the identification of brand -specific vaccines,
such as the Pfizer -BioNTech COVID -19 vaccine. Moreover, the VHA CDW data are
updated on a dail y basis, enabling near real -time rapid monitoring of potential safet y signals.
However, there are several limitations when rely ing on VHA that should be noted. First,
there could be gaps in the data since individuals may receive healthcare services outside of
VHA facilities. As such, if individuals receive the Pfizer-BioNTech COVID- 19 vaccine
outside of a VHA facility, this information will not be captured in the VHA EMR sy stem.
Similarly, individuals may have also received past seasonal influenza vaccinations outside of
the VHA s ystem, and thus would be misclassified as not having received vaccine in the
current anal ysis. For example, veterans with secondary insurance or veterans who are
65years of age or older who have Medicare may receive health care services outside of VHA
facilities. One study on VHA enrollees in seven different states found that of all individuals
admitted to VHA hospitals in 2007, one fifth also had a non VHA hospitalization during that
year.55Another study reported that about 53% of Veterans 65 years of age and older who
were dually eligible for VHA and Medicare services in 2003 2004 used both.56Hence, it is
important to note that data on vaccination status may be incomplete. However, this limitat ion
will be addressed b y examining subgroups of individuals who receive care regularl y at VHA
facilities, as well as those with Priority group 1 status, to ensure that their healthcare data are
complete to the extent possible in the CDW. The results from t hesesubgroup analyses will be
compared to the overall population results from the VHA CDW to confirm consistent
findings such that if there are missing data for individuals in the overall population, the
missing data can be assumed to be missing at random and not biasing the results in any
direction. This will be evaluated in the context of evaluating the relative risk of safet y events
of interest in the comparative anal yses. However, if there are discrepancies that suggest data
are not missing at random a nd could bias results, subgroup anal yses will be conducted for
individuals with dual coverage in the VHA and Medicare. T he CDW data will be
supplement ed and linked with Medicare administrative claims data at the patient level to
ensure a more comprehensive evaluation of the care an individual receives. L inking variables
are available in the data to allow for patient -level linking of the two data sources. Given the
older age of man y veterans, it is likely that these individuals have secondary coverage with
Medicare.
Lastly, to the extent that the individuals in the VHA database are different from individuals
outside of the VHA, the results may not be generalizable to the broader US population. For
example, since the VHA includes predominantl y male Veterans (approximately 90% male),
findings from this study may not be generalizable to women in the US.
9.10.Other Aspects
Not applicable.
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077407
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 68of 19410. PROTECTION OF HUMAN SUBJECTS
10.1.Patient Information
All parties will comply with all applicable laws, including laws regarding the implementation
of organizational and technical measures to ensure protection of patient personal data. Such
measures will include omitting patient names or other directl y identifiable data in an y
reports, publications, or other disclosures, except where re quired by applicable laws.
To protect the rights and freedoms of natural individuals with regard to the processing of
personal data, when study data are compiled for transfer to Pfizer and other authorized
parties, an y patient names will be removed and wil l be replaced b y a single, specific,
numerical code. All other identifiable data transferred to Pfizer or other authorized parties
will be identified by this single, patient -specific code. In case of data transfer, Pfizer will
maintain high standards of co nfidentialit y and protection of individuals’ personal data
consistent with the vendor contract, and applicable privacy laws.
No personal data is planned to be transferred off the VA servers. Specifically, the Clinical
Epidemiology Program (CEP) at White R iver Junction VA Medical Center will conduct this
safety surveillance stud y with sponsorship from Pfizer and assistance from Analy sis Group,
Inc. The project will be led by the VA, with Dr. Yinong Young -Xu, Director of CEP, serving
as the Principal Investi gator. Data access will be granted through VA Informatics and
Computing I nfrastructure (VINCI). VHA data will not be provided to Pfizer or Anal ysis
Group.Rather, onl y VA employees, including those with research service without
compensation (WOC) employ ee status, who have completed necessary VA training and have
proper clearance will access and anal yze data on secure VA servers and behind necessary
firewalls, under the direction and supervision of Dr. Young -Xu. Given the sensitive nature of
healthcare data, comprehensive securit y measures will be implemented to ensure the
confidentiality , integrity, and protection of Veterans’ privacy and healthcare data.
10.2.Patient Consent
As this study does not involve data subject to privacy laws according to applicable lega l
requirements, obtaining informed consent from individuals by Pfizer is not required.
10.3. Institutional Review board (IRB)/Independent Ethics Committee (IEC)
There must be prospective approval of the study protocol, protocol amendments, and their
relevant documents from the relevant I RBs/IECs. All correspondence with the I RB/IEC must
be retained. Copies of IRB/IEC approvals must be forwarded to Pfizer. The study protocol
will be reviewed b y the IRB of the VA Medical Center, White River Junction, VT.
10.4.Ethical Conduct of the Study
The study will be conducted in accordance with legal and regulatory requirements, as well as
with scientific purpose, value and rigor and follow generall y accepted research practices
described in Guidelines for Good Pharmacoepidemiolog y Practices (GPP) issued by the
International Societ y for Pharmacoepidemiology ,57the FDA Guidance for Industry and FDA
Staff: Best Practices for Conducting and Reporting, Pharmacoepidemiologic Safet y Studies
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077408
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 69of 194Using Electronic H ealthcare Data58and Good Epidemiological Practice (GEP) guidelines
issued by the International Epidemiological Association (IEA).59
11.MANAGEMENT AND REPORTING OF ADVERSE EVENTS/ADVERSE
REACTIONS
Signal Detection and Signal Evaluation
This study involves data that exist as structured data by the time of study start or a
combination of existing structured data and unstructured data, which will be converted to
structured form during the implementatio n of the protocol solely by a computer using
automated/algorithmic methods, such as natural language processing. In these data sources,
individual patient data are not retrieved or validated, and it is not possible to link (i.e.,
identify a potential assoc iation between) a particular product and medical event for an y
individual. Thus, the minimum criteria for reporting an adverse event (AE) (i.e., identifiable
patient, identifiable reporter, a suspect product, and event) cannot be met.
Signal Verification
This study protocol requires human review of patient- level unstructured data; unstructured
data refer to verbatim medical data, including text-based descriptions and visual depictions
of medical information, such as medical records, images of ph ysician notes, neurological
scans, X-rays, or narrative fields in a database. The reviewer is obligated to report adverse
events (AEs) with explicit attribution to any Pfizer drug that appear in the reviewed
information (defined per the patient population and study period specified in the protocol).
Explicit attribution is not inferred b y a temporal relationship between drug administration
and an AE, but must be based on a definite statement of causality by a healthcare provider
linking drug administration to the A E.
The requirements for reporting safet y events of interest on the non -interventional study
(NIS) adverse event monitoring (AEM) Report Form to Pfizer Safety are as follows:
All serious and non- serious AEs with explicit attribution to any Pfizer drug that
appear in the reviewed information must be recorded on the data collection tool (e.g.,
chart abstraction form) and reported, within 24 hours of awareness, to Pfizer Safet y
using the NIS AEM Report Form.
Scenarios involving drug exposure, including exposure during pregnancy , exposure
during breast feeding, medication error, overdose, misuse, extravasation, lack of
efficacy, and occupational exposure associated with the use of a Pfizer product must
be reported, within 24 hours of awareness, to Pfizer Safet y using the NI S AEM
Report Form.
For these AEs with an explicit attribution or scenarios involving exposure to a Pfizer
product, the safety information identified in the unstructured data reviewed is captured in the
Event Narrative section of the report form, and constitutes all clinical information known
regarding these AEs. No follow-up on related AEs will be conducted.
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077409
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 70of 194All the demographic fields on the NI S AEM Report Form may not necessarily be completed,
as the form designates, since not all elements will be available due to privacy concerns with
the use of secondary data sources. While not all demographic fields will be completed, at the
very least, at least one patient identifier (e.g., gender, age as captured in the narrative field of
the form) will be reported on the NI S AEM Report Form, thus allowing the report to be
considered a valid one in accordance with pharmacovigilance legislation. All identifiers will
be limited to generalities, such as the statement “A 35- year-old female...” or “An elderl y
male...” Other identifiers will have been removed.
Additionally , the onset/start dates and stop dates for “Illness”, “Study Drug”, and “Drug
Name” may be documented in month/y ear (mmm/yyyy ) format rather than identify ing the
actual date of occurrence within the month /y ear of occurrence in the day /month/year
(DD/MMM/YYYY ) format.
All research staff members must complete the following Pfizer training requirements:
Your Reporting Responsibilities ( YRR)Training for Vendors Working on Pfizer
Studies
These trainings must be completed by research staff members prior to the start of data
collection. All trainings include a “Confirmation of Training Certificate” (for signature b y
the trainee) as a record of completion of the training, which must be kept in a retrievable
format. Copies of all signed training certificates must be provided to Pfizer.
Re-training must be completed on an annual basis using the most current Your Reporting
Responsibilities training materials.
12.PLANS FOR DISSEMINAT ING AND COMMU NICATING STUDY RESUL TS
This protocol will be posted on publicly available registers following its finalization. The
final study results will be made publicly available via the European Union Post Authorisation
Safety (EU PAS) Register and may be submitted for publication in a peer reviewed medical
journal.
In the event of an y prohibition or restriction imposed (e.g., clinical hold) by an applicable
competent authorit y in any area of the world, or if the investigator is aware of an y new
information which mig ht influence the evaluation of the benefits and risks of a Pfizer
product, Pfizer should be informed immediately.
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077410
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 71of 19413.REFERENCES
1.WorldHealth Organization (WHO). WHO Director -General’s opening remarks at the
media briefing on COVID -19 -11 March 2020.
https://www.who.int/dg/speeches/detail/who -director-general-s-opening-remarks-at-
the-media-briefing-on-covid-19---11-march-2020. Accessed November 11, 2020.
2. Johns Hopkins University. Coronavirus Resource Center. Accessed November 10,
2020. https://coronavi rus.jhu.edu/.
3. U.S. Food and Drug Administration (FDA). Vaccines and Related Biological
Products Advisory Committee Meeting: FDA Briefing Document Pfizer -BioNTech
COVID-19 Vaccine. Accessed December 10, 2020.
4.U.S. Food andDrug Administration (FDA). Pfizer COVID -19 Vaccine EUA Letter of
Authorization reissued 12-23 -20. In:December 23, 2020.
5.U.S. Food and Drug Administration (FDA). FDA Takes Key Action in Fight Against
COVID-19 By Issuing Emergency Use Authorization for First COVID -19 Vaccine.
https://www.fda.gov/news -events/press -announcements/fda -takes-key-action-fight-
against-covid-19- issuing-emergency -use-authorization -first-covid-19.
6.Pfizer Inc. Pfizer and BioNTech Achieve First Authorization in the World for a
Vaccine to Combat COVID -19. Accessed January 4, 2021.
https://www.pfizer.com/news/press -release/press- release-detail/pfizer -and-biontech-
achieve-first-authorization -world.
7.Pfizer Inc. Pfizer and BioNTech Receive Authorization in the Europe an Union for
COVID-19 Vaccine. Accessed December 28, 2020.
https://www.pfizer.com/news/press -release/press- release-detail/pfizer -and-biontech-
receive-authorization- european-union.
8.Law B. SO2 -D2.1.2 Priority List of COVID -19 Adverse Events of special interest:
Quarterly update. Brightoncollaboration.us . April 23, 2020.
https://brightoncollaboration.us/wp-
content/uploads/2021/01/SO2_D2.1.2_V1.2_COVID -19_AESI-update-23Dec2020 -
review_final.pdf
9. Shimabukuro T. Enhanced safet y monitoring for COVID -19 vaccines in early phase
vaccination. National Center for Immunization & Respiratory Diseases. September
22, 2020. https://www.cdc.gov/vaccines/acip/meetings/downloads/slides- 2020-
09/COVID-03-Shimabukuro.pdf
10.Yih WK, L ee GM, Lieu TA, et al. Surveillance for adverse events f ollowing receipt
of pandemic 2009 H1N1 vaccine in the Post- Licensure Rapid Immunization Safety
Monitoring (PRI SM) System, 2009 -2010. Am J Epidemiol. 2012;175(11):1120 -1128.
11.American Medical Association (AMA). Appendix Q: Severe Acute Respiratory
Syndrome Coronavirus 2 (SARS -CoV-2) (coronavirus disease [COVID -19])
Vaccines. Accessed January 12, 2021. https://www.ama -assn.org/sy stem/files/2020 -
11/covid-19-immunizations- appendix-q-table.pdf.
12. Center for Disease Control (CDC). People with Certain Medical Conditions.
December 29, 2020 . Accessed January 4, 2021.
https://www.cdc.gov/coronavirus/2019 -ncov/need -extra-precautions/people- with-
medical-conditions.html.
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077411
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 72of 19413.Patel M, Chen J, Kim S, et al. Anal ysis of MarketScan Data for Immunosuppressive
Conditions and Hospitalizations for Acute Respiratory Illness, United States.
Emerging Infectious Diseases . 2020;26(8):1720 -1730. doi:10.3201/eid2608.191493.
14.U.S. Department of Veterans Affairs. Veterans Health Administration. 2020.
Accessed November 10, 2020. https://www.va.gov/health/
15. U.S. Department of Veterans Affairs. VA nursing homes, assisted living, and home
health care. 2020 .Accessed January 4, 2020.https://www.va.gov/health -care/about -
vahealth-benefits/long -term-care/
16.U.S. Food and Drug Administration (FDA). COVID -19 Vaccine Safet y Surveillance :
Active Monitoring Master Protocol. February 10, 2021. Accessed April 23, 2021 .
https://www.bestinitiative.org/wp -content/uploads/2021/02/C19- Vaccine-Safety-
Protocol-2021.pdf
17.Shimabukuro T. COVID-19 Vaccine safet y updates. Center for Disease Control
(CDC) Advisory Committee on Immunization Practices (ACIP). June 23, 2021.
https://www.cdc.gov/vaccines/acip/meetings/downloads/slides -2021-06/03-COVID-
Shimabukuro-508.pdf
18.Klein, N, Donahue, J, Weintraub, E. Rapid Cy cle Analy sis (RCA) to monitor the
safety of COV ID-19 vaccines in near real- time within the Vaccine Safet y Datalink.
Center for Disease Control (CDC) . March 3, 2021. Accessed July 11, 2021.
https://www.cdc.gov/vaccinesafety /pdf/VSD -1342-COVID19 -RCA-
Protocol_FinalV1.1_508.pdf
19.Brighton Collaboration. My ocarditis/Pericarditis Case Definition. July 16, 2021.
Accessed July 26, 2021. https://brightoncollaboration.us/my ocarditis- case-definition -
update/
20. Shimabukuro T. Update: Thrombosis with thrombocy topenia sy ndrome (TTS)
following COVID -19 vaccination. Center for Disease Control (CDC) Advisory
Committee on I mmunization Practices (ACIP). May 12, 2021.
https://www.cdc.gov/vaccines/acip/meetings/downloads/slides -2021-05-12/07-
COVID-Shimabukuro -508.pdf
21.Dash S, Sirka CS, Mishra S, Viswan P. Covid ‐19 vaccine induced Steven‐Johnson
syndrome: a case report. Clinical and Experimental Dermatology . 2021 Jun 3.
22.Brito S, Ferreira N, Mateus S, et al. A Case of Autoimmune Hemoly tic Anemia
Following COVID -19 Messenger Ribonucleic Acid Vaccination. Cureus. 2021
May;13(5).
23.Tworkoski E, Wong HL, Zhou C, et al. Draft Master Protocol -Assessment of Risk
of Safety Outcomes Following COVID -19 Vaccination. U.S. Food and Drug
Administration (FDA ) Center for Biologics Evaluation and Research (CBER). March
23, 2021. Accessed July 11, 2021. https://www.bestinitiative.org/wp -
content/uploads/2021/04/COVID-19 -Vaccine-Safety-Inferential-Draft-Master-
Protocol.pdf
24. Center for Disease Control (CDC ) COVID-19 Response Team. Preliminary Estimates
of the Prevalence of Selected Underl ying Health Conditions Among Patients with
Coronavirus Disease 2019 —United States, February 12–March 28, 2020. MMWR
Morbidity and Mortality Weekly Report 2020;69(13):382 -386.
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077412
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 73of 19425.Gupta S, Hay ek SS, Wang W, et al. Factors Associated With Death in Critically Ill
Patients With Coronavirus Disease 2019 in the US. JAMA Intern Med.
2020;180(11):1 -12.
26.Williamson EJ, Walker AJ, Bhaskaran K, et al. Factors associated with COVID -19-
related death using OpenSAFELY. Nature. 2020;584(7821):430-436.
27.Walsh EE, Frenck RW, Jr., Falsey AR, et al. Safety and Immunogenicity of Two
RNA-Based Covid-19 Vaccine Candidates. N Engl J Med. 2020.
28.U.S. National L ibrary of Medicine. Study to Describe the Safet y, Tolerability,
Immunogenicit y, and Efficacy of RNA Vaccine Candidates Against COVID -19 in
Healthy Individuals. NCT04368728.
29. U.S. Food and Drug Administration (FDA) Center for Biologics Evaluation and
Research (CBER). Emergency Use Authorization for Vaccines to Prevent CO VID-
19: Guidance for Industry . Accessed November 10, 2020.
https://www.fda.gov/media/142749/download
30.Lee GM, Greene SK, Weintraub ES, et al. H1N1 and seasonal influenza vaccine
safety in the vaccine safety datalink project. Am J Prev Med. 2011;41(2):121 -128.
31.Fireman B, Lee J, L ewis N, Bembom O, van der Laan M, Baxter R. I nfluenza
vaccination and morta lity: differentiating vaccine effects from bias. Am J Epidemiol.
2009;170(5):650 -656.
32.Keren R, Zaoutis TE, Bridges CB, et al. Neurological and Neuromuscular Disease as
a Risk Factor for Respiratory Failure in Children Hospitalized With I nfluenza
Infection. JAMA. 2005;294(17):2188-2194.
33.U.S. Department of Veterans Affairs. VA priority groups. 2020.
https://www.va.gov/health- care/eligibility /priority-groups/
34.Liu CH, Yeh YC, Huang WT, Chie WC, Chan KA. Assessment of pre -specified
adverse events fol lowing varicella vaccine: A population -based self -controlled risk
interval study . Vaccine. 2020;38(11):2495-2502.
35.Manjunath R, Paradis PE, Parisé H, Lafeuille MH, Bowers B, Duh MS,et al. Burden
of uncontrolled epilepsy in patients requiring an emergency room visit or
hospitalization. Neurology . 2012 Oct 30;79(18):1908 -16.
36.Baker MA, Baer B, Kulldorff M, et al. Kawasaki disease and 13 -valent pneumococcal
conjugate vacci nation among young children: A self -controlled risk interval and
cohort study with null results. PLoS Med. 2019;16(7):e1002844.
37.U.S. Department of Veterans Affairs. COVID -19 National Summary . 2020. Accessed
November 10, 2020.
https://www.accesstocare.v a.gov/Healthcare/COVID19NationalSummary
38.U.S. Department of Veterans Affairs. Racial and Ethnic Minority Veterans. 2020.
Accessed January 18, 2021.
https://www.va.gov/HEALTHEQUITY/Race_Ethnicity .asp.
39.U.S. Department of Veterans Affairs. COVID -19 vaccines at VA. 2020;
https://www.va.gov/health- care/covid -19-vaccine/.
40. U.S. Department of Veterans Affairs. Profile of Veterans: 2014. 2016. Accessed
November 10, 2020.
https://www.va.gov/vetdata/docs/SpecialReports/Profile_of_Veterans_2014.pdf.
41.U.S. Department of Veterans Affairs. Number of Veteran Patients b y Healthcare
Priority Group: FY2000 to FY2017. 2018. Accessed November 10, 2020.
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077413
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 74of 194https://www.va.gov/vetdata/docs/Utilization/Number_of_Veteran_Patients_by _HC_P
riority_Groups_2000_2017.pdf
42.U.S. Department of Veterans Affairs. Average Expenditures Per Patient by
Healthcare Priorit y Group: FY2000 to FY2013. 2013 . Accessed November 10, 2020.
https://www.va.gov/vetdata/docs/Utilization/AvgCost_FINAL2.xlsx
43.Agha Z, Lofgren RP, Van Ruiswyk JV, Layde PM. Are patients at Veterans Affairs
medical centers sicker? A comparative anal ysis of health status and medical resource
use. Arch Intern Med. 2000;160(21):3252-3257.
44.Greene SK, Kulldorff M, Yin R, et al. Near real -time vaccine safet y surveillance with
partially accrued data. Pharmacoepidemiol Drug Saf. 2011;20(6):583 -590.
45.Li R, Stewart B, McNeil MM, et al. Post licensure surveillance of influenza vaccines
in the Vaccine Safet y Datalink in the 2013 -2014 and 2014 -2015 seasons.
Pharmacoepidemiol Drug Saf. 2016;25(8):928 -934.
46. Kulldorff M, Davis RL, Kolczak, M, Lewis E, L ieu T, Platt R. A Maximized
Sequential Probability Ratio Test for Drug and Vaccine Safet y Surveillance.
Sequential Analysis. 2011;30(1):58 -78.
47. Kulldorff M, Sil va IR. Continuous Post -Market Sequential Safet y Surveillance with
Minimum Events to Signal. arXiv preprint arXiv. 2015(1503.01978).
48.When Are Risks Real? I n. Clinical Nutrition Insight. Vol 332007:6.
49.Silva IR, Lopes WM, Dias P, Yih WK. Alpha spendin g for historical versus
surveillance Poisson data with CMaxSPRT. Stat Med. 2019;38(12):2126 -2138.
50.Silva IR. Type I error probability spending for post -market drug and vaccine safet y
surveillance with binomial data. Stat Med. 2018;37(1):107-118.
51.Hernan, M.A. and J.M. Robins, Estimating causal effects from epidemiological data.
J Epidemiol Community Health. 2006; 60(7):578-86.
52. U.S. Food and Drug Administration (FDA) Center for Biologics Evaluation and
Research (CBER). Guidance for Industry: GoodPharmacovigilance Practices and
Pharmacoepidemiologic Assessment. 2005. Accessed November 12, 2020.
53.Baker MA, Lieu TA, L i L, et al. A vaccine stud y design selection framework for the
postlicensure rapid immunization safety monitoring program. Am J Epidemiol.
2015;181(8):608 -618.
54. Hoffman KB, Demakas AR, Dimbil M, Tatonetti NP, Erdman CB. Stimulated
reporting: the impact of US food and drug administration -issued alerts on the adverse
event reporting s ystem (FAERS). Drug Saf. 2014;37(11):971 -980.
55.West AN, Charlton ME, Vaughan -Sarrazin M. Dual use of VA and non -VA hospitals
by Veterans with multiple hospitalizations. BMC Health Serv Res. 2015;15:431.
56.Petersen LA, By rne MM, Daw CN, Hasche J, Reis B, Pietz K. Relationship between
clinical conditio ns and use of Veterans Affairs health care among Medicare -enrolled
veterans. Health Serv Res. 2010;45(3):762 -791.
57.International Societ y for Pharmacoepidemiology (ISPE). Guidelines for good
pharmacoepidemiology practices (GPP). Pharmacoepidemiol Drug Sa f.
2008;17(2):200 -208.
58.U.S. Food and Drug Administration (FDA) Center for Drug Evaluation and Research
(CDER). Best Practices for Conducting and Reporting Pharmacoepidemiologic
Safety Studies Using Electronic Healthcare Data. 2013 . Accessed January 19,2021.
https://www.fda.gov/media/79922/download.
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077414
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 75of 19459. International Epidemiological Association (IEA). Good Epidemiological Practice
(GEP), IEA Guidelines for Proper Conduct of Epidemiological Research. 2007.
Accessed January 25, 2021.
https://ieaweb.org/I EAWeb/Content/IEA Publications.aspx .
60.Segal JB, Chang HY, Du Y, Walston J, Carlson M, Varadhan R. Development of a
claims-based frailty indicator anchored to a well -established frailty phenotype.
Medical care. 2017 Jul;55(7):716.
61.Menendez ME, Neuhaus V, van Dijk CN, Ring D. The Elixhauser comorbidity
method outperforms the Charlson index in predicting inpatient death after orthopaedic
surgery. Clin Orthop Relat Res . 2014; 472(9):2878-2886.
62.Centers for Medicare & Medicaid Services. 20 21 ICD-10-CM Official Guidelines for
Coding and Reporting. Accessed January 17, 2021.
https://www.cms.gov/files/document/2021 -coding-guidelines -updated-12162020.pdf .
63.Baxter R, Eaton A, Hansen J, Aukes L , Caspard H, Ambrose CS. Safet y of
quadrivalent live attenuated influenza vaccine in subjects aged 2 -49years. Vaccine.
2017;35(9):1254 -1258.
64.Johns Hopkins Vasculitis Center. Ty pes of Vasculitis. Accessed January 19,
2021.https://www.hopkinsvasculitis.org/t ypes-
vasculitis/#: ~:text=%E2%80%9CAngiitis%E2%80%9D%20and%20%E2%80%9CAr
teritis%E2%80%9D,lit'%20i%20deez%E2%80%9D.
65.Chen, R. TTS I nterim Case Definition c10.16. Brightoncollaboration.us. May 3,
2021. Accessed June 12, 2021. https://brightoncollaboration.us/wp -
content/upl oads/2021/05/TTS -Interim-Case- Definition -v10.16.3 -May-23-2021.pdf
66.OptumInsight Inc. Guide to Clinical Validation, Documentation and Coding: Acute
Kidney Injury.
https://www.optum360coding.com/upload/pdf/ECDCG14/CDCG14_v2.pdf .
Accessed January 17, 2021 .
67. Brighton Collaboration. COVID- 19 Updated AESI List. 2021. Accessed June 12,
2021. https://brightoncollaboration.us/wp- content/uploads/2021/01/COVI D-19-
updated-AESI-list.pdf
68.Forns J, Cainzos ‐Achirica M, Hellfritzsch M, Morros R, Poblador ‐PlouB, Hallas J, et
al. Validity of ICD‐9 and ICD‐10 codes used to identify acute liver injury : A study in
three European data sources. Pharmacoepidemio Drug Saf 2019 Jul;28(7):965-75.
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077415
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 76of 19414.LIST OF TABLES
Table 1. Outcome algorithms for SCRI analysis, with risk and control
intervals................................ ................................ ................................ .....44
Table2. Estimated S tatistical Power for the Poisson -based MaxSPRT46..............50
Table 3. Critical Values for Poisson -based MaxSPRT ................................ ...........57
AppendixTable1.Demographic and Clinical Characteristics Definitions ............................ 77
Appendix Table2.Operational Definitions of Safet y Events of Interest.............................. 108
Appendix Table3.COVID-19 and Seasonal I nfluenza Vaccine Exposure CPT,
HCPCS, NDC Codes ................................ ................................ ..............180
Appendix Table 4. COVID-19 RT-PCR Test L OINC................................ .......................... 193
15.LIST OF FIGURES
Figure 1. Example of SCRI Design for Assessment of a Safety Event of
Interest with a 42- day Risk Interval in an Individual who Receives
Only One Vaccine Dose, with Post -vaccination Control I ntervals* ........33
Figure 2. Example of SCRI Design for Assessment of a Safety Event of
Interest with a 42- day Risk Interval in an Individual who Receives
Two Vaccine Doses, with Post -vaccination Control I ntervals................. 35
Figure 3. Steps in Signal Detection, Evaluation, and Verification .......................... 54
Figure 4. Example of SCRI Design for a Safety Event of Interest with a 42 -
day Risk Interval and a Post -vaccination Control Interval ....................... 55
Figure 5. Example of SCCS Design for Safet y Event of Interest with a 42 -day
Risk Interval with Post -vaccination Control Intervals when Two
Doses of Pfizer -BioNTech COVID -19 Vaccine are Administered.......... 60
Figure 6. Example of Risk (P1, P2, P3) and Aggregate Post- vaccination
Control Intervals for the SCRI End-of-surveillance Anal yses of 1 or
2 Doses of Pfizer -BioNTech COVID -19 Vaccine ................................ ....63
16.ANNEX 1. LIST OF STAND ALONE DOCUMENTS
None.
17.ANNEX 2. ENCEPP CHEC KLIST FOR STUDY PROT OCOLS
N/A
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077416
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 77of 19418.ANNEX 3. ADDITIONAL INFORMATION
AppendixTable1.Demographic and Clinical Characteristics Definitions
Variable Description Operational definition
Demographic Characteristics
Age Continuous variable;
Categorical variable:
<16
16–64
65–74
≥75Age on the date of Pfizer- BioNTech
COVID-19 vaccination (and/or date of
seasonal influenza vaccination for active
comparators)
Sex Categorical variable:
Male
Female
Unknown
Race/ethnicit yCategorical variable:
White, non -Hispanic
Black
Hispanic ethnicity , any
race
Asian
Native Hawaiian or
Pacific Islander
American Indian or
Alaskan native
Two or more races
Unknown
VHA service
areaGeographic regions in the US;
Categorical variable:
South
Midwest
West
Northeast
Other
UnknownRegion associated with the most recent
healthcare encounter prior to index date
Clinical Characteristics
Smoking Status Dichotomous variable ICD-9-CM codes:
305.1, Tobacco use disorder
V15.82, History of tobacco use
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077417
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 78of 194AppendixTable1.Demographic and Clinical Characteristics Definitions
Variable Description Operational definition
ICD-10-CM codes:
F17.200, Nicotine dependence,
unspecified, uncomplicated
Z7.20, Tobacco use
Z87.891, Personal history of
nicotine dependence
Body mass
index (BMI) *Continuous variable;
Categorical variable:
Underweight (<18.5)
Normal weight (18.5 –<25)
Overweight (25– <30)
Obese (30 –<40)
Severe obesit y (≥40)
UnknownCalculated from height and weight data
(kg/m2)
History of
anaphylaxis/
allergic
reactionsDichotomous variable ICD-9-CM code:
V13.81, Personal history of
anaphylaxis
V14.0–V14.6, V14.8, V14.9,
Personal history of allergy to
drugs, medications and
biological substances, excluding
serum and vaccine
V15.0x, Other allergy
525.66, Allergy to existing dental
restorative material
995, Other anaph ylactic shock,
not elsewhere classified
995.1, Angioneurotic edema, not
elsewhere classified
995.21, Arthus phenomenon
999.27, Other drug allergy
995.3, Allergy , unspecified, not
elsewhere classified
995.6x, Anaphy lactic shock due
to food
999.41, Anaph ylactic reaction
due to administration of blood
and blood products
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077418
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 79of 194AppendixTable1.Demographic and Clinical Characteristics Definitions
Variable Description Operational definition
999.49, Anaph ylactic reaction
due to other serum
ICD-10-CM code:
Z87.892 Personal history of
anaphylaxis
Z88.0–Z88.6, Z88.8, Z88.9,
Allergy status to drugs,
medications and biological
substances, excluding serum and
vaccine
T78.00xx –T78.09xx,
Anaphylactic reaction due to
food, initial encounter,
subsequent encounter and
sequela
T78.2xxx, Anaphy lactic shock,
initial encounter, subsequent
encounter and sequela
T78.3xxx, Angioneurotic edema,
initial encounter, subsequent
encounter and sequela
T78.41xx, Arthus phenomenon
T80.51xx, Anaphy lactic reactio n
due to administration of blood
and blood products, initial
encounter, subsequent encounter
and sequela
T80.59xx, Anaphy lactic reaction
due to other serum, initial
encounter, subsequent encounter
and sequela
T88.6xxx, Anaphy lactic reaction
due to adverse effect of correct
drug or medicament properly
administered, initial encounter,
subsequent encounter and
sequela
Previous
anaphylaxis of Dichotomous variable ICD-9-CM code:
999.42, Anaph ylactic reaction
due to vaccination
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077419
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 80of 194AppendixTable1.Demographic and Clinical Characteristics Definitions
Variable Description Operational definition
vaccine
componentV14.7, Personal history of
allergy to serum or vaccine
ICD-10-CM codes:
T80.52xx, Anaphy lactic reaction
due to vaccination, initial
encounter, subsequent encounter
and sequela
Z28.04, Immunization not
carried out because of patient
allergy to vaccine or component
Z88.7, Allergy status to serum
and vaccine
History of
hospitalizations Dichotomous variable;
Continuous variableDefined b y having any hospitalizations
(dichotomous) and number of
hospitalizations (continuous)
Frailty index60Continuous variable ICD-9-CM codes available in Appendix
Table 1 of Segal et al, 2017. I CD-9-CM
codes mapped to ICD -10-CM codes.
Charlson
Comorbidity
Index (CCI )61Continuous variable ICD-9-CM codes:
410.x, 412.x, Myocardial
infarction
398.91, 402.01, 402.11, 402.91,
404.01, 404.03, 404.11, 404.13,
404.91, 404.93, 425.4 –425.9,
428.x, Congestive heart failure
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077420
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 81of 194AppendixTable1.Demographic and Clinical Characteristics Definitions
Variable Description Operational definition
093.0, 437.3, 440.x, 441.x,
443.1–443.9, 447.1, 557.1,
557.9, V43.4, Peripheral vascular
disease
362.34, 430. x–438.x,
Cerebrovascular disease
290.x, 294.1, 331.2, Dementia
416.8, 416.9, 490.x– 505.x,
506.4, 508.1, 508.8, Chronic
pulmonary disease
446.5, 710.0 –710.4, 714.0 –
714.2, 714.8, 725.x, Rheumatic
disease
531.x–534.x, Peptic ulcer disease
070.22, 070.23, 070. 32, 070.33,
070.44, 070.54, 070.6, 070.9,
570.x, 571.x, 573.3, 573.4,
573.8, 573.9, V42.7, Mild liver
disease
250.0–250.3, 250.8, 250.9,
Diabetes without chronic
complication
250.4–250.7, Diabetes with
chronic complication
334.1, 342.x, 343.x, 344.0 –
344.6,344.9, Hemiplegia or
paraplegia
403.01, 403.11, 403.91, 404.02,
404.03, 404.12, 404.13, 404.92,
404.93, 582.x, 583.0–583.7,
585.x, 586.x, 588.0, V42.0,
V45.1, V56.x, Renal disease
140.x–172.x, 174.x –195.8,
200.x–208.x, 238.6, Any
malignancy , including l ymphoma
and leukemia, except malignant
neoplasm of skin
456.0–456.2, 572.2 –572.8,
Moderate or severe liver disease
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077421
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 82of 194AppendixTable1.Demographic and Clinical Characteristics Definitions
Variable Description Operational definition
196.x–199.x, Metastatic solid
tumor
042.x–044.x, Acquired
immunodeficiency syndrome
(AIDS)/Human
immunodeficiency virus (HIV)
ICD-10-CM codes:
I21.x, I21.xx, I 22.x, I25.2,
Myocardial infarction
I09.9, I11.0, I13.0, I13.2, I25.5,
I42.0, I42.5–I42.9, I43, I43.x,
I50.x, I50.xx, Congestive heart
failure
I70.x, I71.x, I 73.1, I73.8, I 73.9,
I77.1, I79.0, I79.2, K55.1, K55.8,
K55.9, Z95.8, Z95.9, Periphe ral
vascular disease
G45, G45.x, G46.x, H34.0,
I60.x–I63.x, I60.xx–I63.xx,
I60.xxx–I63.xxx, I65.x–I69.x,
I65.xx–I69.xx, I65.xxx–I69.xxx,
Cerebrovascular disease
F00.x–F03.x, F00.xx –F03.xx,
F05, F05.1, G30.x, G31.1,
Dementia
I27.8, I27.9, J40.x –J47.x,
J40.xx–J47.xx, J40.xxx –J47.xxx,
J60.x–J67.x, J68.4, J70.1, J70.3,
Chronic pulmonary disease
M05, M05.x, M05.xx, M05.xxx,
M06, M06.x, M06.xx, M06.xxx,
M31.5, M32.x –M34.x, M32.xx –
M34.xx, M35.1, M35.3, M36.0,
Rheumatic disease
K25.x–K28.x, Peptic ulcer
disease
B18.x, K70.0– K70.3, K70.9,
K71.3–K71.5, K71.7, K73.x,
K74.x, K74.xx, K76.0, K76.2 –
090177e197e9a28e\Approved\Approved On: 26-Aug-2021 16:54 (GMT)
FDA-CBER-2021-5683-1077422
BNT162b2 (COVID -19 vaccine)
C4591012 NON- INTERVENTIONAL STUDY PROTOCOL
Version 2.0, 31 Aug 2021
PFIZER CONFIDENTIAL
Page 83of 194AppendixTable1.Demographic and Clinical Characteristics Definitions
Variable Description Operational definition
K76.4, K76.8, K76.9, Z94.4,
Mild liver disease
E10.0, E10.1x, E10.6x, E10.6xx,
E10.8, E10.9, E11.0x, E11.1x,
E11.6x, E11.6xx, E11.8, E11.9,
E12.0, E12.1, E12.6, E12.8,
E12.9, E13.0x, E13.1x, E13.6x,
E13.6xx, E13.8, E13.9, E14.0,
E14.1, E14.6, E14.8, E14.9,
Diabetes without chronic
complication
E10.2x–E10.5x, E10.2xx–
E10.5xx, E10.7, E11.2x –E11.5x,
E11.2xx–E11.5xx, E11.7, E12.2 –
E12.5, E12.7, E13.2 –E13.5x,
E13.
…[truncated]