Document text
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 1 of 237
NON -INTERVENTIONAL (NI) STUDY CONCEPT PROTOCOL
Title Post-Emergency Use Authorization Active
Safety Surveillance Study among Individuals
in the Veteran’s Affairs Health System
Receiving Pfizer -BioNTech Coronavirus
Disease 2019 (COVID -19) Vaccine
Protocol number C4591012
Protocol version identifier Final Version 12.0
Date of last version of protocol 27 January 2021
EU Post Authori zation Study (PAS)
register number To be registered before the start of data
collection EUPAS39779
Active substance 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) system
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 e vents of interest following
receipt of the Pfizer -BioNTech
COVID-19 vaccine;
• To assess whether sub -cohorts of
interest (i.e., immunocompromised,
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077536
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 2 of 237 elderly, 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 system experience increased
risk of safety events of interest
following receipt of the Pfizer -
BioNTech COVID -19 vaccine.
Secondary study objective :
• To characterize utilization patt erns 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,
demographi cs 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
This document contains confidential information belonging to Pfizer. Except as otherwise agreed to in writing,
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.
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077537
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 3 of 237 1.TABLE OF CONTENTS
1. TABLE OF CONTENTS ................................ ................................ ................................ 3
2. LIST OF ABBREVIATIONS ................................ ................................ .......................... 5
3. RESPONSIBLE PARTIES ................................ ................................ .............................. 8
4. ABSTRACT ................................ ................................ ................................ ................... 9
5. AMENDMENTS AND UPDATES ................................ ................................ ............... 26
6. MILESTONES ................................ ................................ ................................ .............32
7. RATIONALE AND BACKGROUND ................................ ................................ ...........33
8. RESEARCH QUESTION AND OBJECTIVES ................................ ............................. 34
9. RESEARCH METHODS ................................ ................................ .............................. 35
9.1. Study Des ign ................................ ................................ ................................ ....35
9.1.1. Self -Controlled Risk Interval (SCRI) Design with Post -Vaccination
Control Interval ................................ ................................ .......................... 35
9.1.2. Active Comparator Design ................................ ................................ .....39
9.1.3. Additional Study Designs in the Signal Evaluation Phase ....................... 40
9.1.4. Study Period ................................ ................................ .......................... 40
9.2. Setting ................................ ................................ ................................ ..............40
9.2.1. Inclusion Criteria ................................ ................................ ................... 40
9.2.2. Exclusion criteria ................................ ................................ .................. 40
9.2.3. Subgroups ................................ ................................ ............................. 40
9.3. Variables ................................ ................................ ................................ ..........42
9.3.1. Exposure of Interest ................................ ................................ ............... 42
9.3.1.1. Pfizer -BioNTech COVID -19 Vaccine Groups of Interest ........43
9.3.2. Baseline Characteristics ................................ ................................ .........43
9.3.3. Outcomes ................................ ................................ .............................. 45
9.4. Data Source ................................ ................................ ................................ ......67
9.5. Study Size ................................ ................................ ................................ .........68
9.5.1. Power ................................ ................................ ................................ ...68
9.6. Data Management ................................ ................................ ............................. 71
9.6.1. Case report forms (CRFs)/Electronic data record ................................ ....71
9.6.2. Record retention ................................ ................................ .................... 71
9.7. Data Analysis ................................ ................................ ................................ ....72
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077538
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 4 of 237 9.7.1. Baseline Characteristics ................................ ................................ .........72
9.7.2. Vaccine Utilization Patterns ................................ ................................ ...73
9.7.3. Safety Signal Analyses ................................ ................................ ..........73
9.7.3.1. Signal Detection ................................ ................................ ....75
9.7.3.2. Signal Evaluation ................................ ................................ ..79
9.7.3.3. Signal Verification ................................ ................................ 83
9.7.4. Seasonality -Adjusted Cases -Centered Method ................................ ........84
9.7.5. End -of-Season and End -of-Surveillance Analyses ................................ ..84
9.7.6. Subgroup Analysis ................................ ................................ ................ 86
9.7.7. Incidence Rates and Time to Safety Event of Interest Analysis ................ 86
9.7.8. Prioritized Safety Analysis of Myocarditis/Pericarditis ........................... 86
9.8. Quality Control ................................ ................................ ................................ .88
9.9. Strengths and Limitations of the Research Methods ................................ ............88
9.10. Other Aspects ................................ ................................ ................................ ..90
10. PROTECTION OF HUMAN SUBJECTS ................................ ................................ ....90
10.1. Patient Information ................................ ................................ .......................... 90
10.2. Patient Consent ................................ ................................ ............................... 91
10.3. Institutional Review board (IRB)/Independent Ethics Committee (IEC) ............91
10.4. Ethical Conduct of the Study ................................ ................................ ...........91
11. MANAGEMENT AND REPORTING OF ADVERSE EVENTS/ADVERSE
REACTIONS ................................ ................................ ................................ ............... 91
12. PLANS FOR DISSEMINATING AND COMMUNICATING STUDY RESULTS .......93
13. REFERENCES ................................ ................................ ................................ ...........94
14. LIST OF TABLES ................................ ................................ ................................ ....100
15. LIST OF FIGURES ................................ ................................ ................................ ...100
16. ANNEX 1. LIST OF STAND ALONE DOCUMENTS ................................ ..............101
17. ANNEX 2. ENCEPP CHECKLIST FOR STUDY PROTOCOLS ............................... 101
18. ANNEX 3. ADDITIONAL INFORMATION ................................ ............................ 101
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077539
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 5 of 237 2. LIST OF ABBREVIATIONS
Abbreviation Definition
ACIP Advisory Committee on Immunization Practices
ADEMACOS Acute disseminated encephalomyelitis 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
CIDP Chronic inflammatory demyelinating polyneuropathy
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
TDapTdap Diphtheria, tetanus and (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 hypothesis
HBV Hepatitis B virus
HCPCS Healthcare Common Procedure Coding System
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077540
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 6 of 237 Abbreviation Definition
HCV Hepatitis C virus
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 System
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 syndrome in 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 analysis 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 syndrome
SPEAC Safety Platform for Emergency vACcines
SRSS Subcommittee on Research Safety and Security
TEN Toxic epidermal necrolysis
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077541
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 7 of 237 Abbreviation Definition
TM Transverse myelitis
TTS Thrombosis with thrombocytopenia syndrome
UK United Kingdom
US United States
VA Department of Veterans Affairs
VAIRRS VA Innovation and Research Review System
VAERS Vaccine Adverse Event Reporting System
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 Safety Datalink
VTE Venous thromboembolism
WHO World Health Organization
WOC Without compensation
YRR Your Reporting Responsibilities
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077542
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 8 of 237 3. RESPONSIBLE PARTIES
Name, degree(s) Job Title Affiliation Address
Principal Investigator:
Yinong Young -Xu,
ScD, MA, MS Director, Clinical Epidemiology
Program Veterans Affairs
(VA) Medical
Center 163 Veterans Drive,
White River Junction,
VT 05009
Cynthia de Luise,
PhD, MPH Senior Epidemiologist /Safety
Surveillance Research Scientist ;
Risk Management and Safety
Surveillance Research Pfizer, Inc. 235 East 42nd Street,
New York, NY 10017
Mei Sheng Duh,
ScD, MPH Managing Principal and Chief
Epidemiologist
Visiting Scientist, Department of
Biostatistics Analysis Group,
Inc.
Harvard T. H.
Chan School of
Public Health 111 Huntington Ave
14th Floor
Boston, MA 02199
677 Huntington Ave
Boston, MA 02115
Maral DerSarkissian,
PhD
Vice President and Senior
Epidemiologist
Adjunct Assistant Professor Analysis Group,
Inc.
Fielding School of
Public Health,
University of
California, Los
Angeles 333 South Hope Street
27th Floor
Los Angeles, CA
90071
650 Charles E Young
Drive South
Los Angeles, CA
90095
Rachel Bhak, MS
Manager and Senior Biostatistician
Analysis Group,
Inc.
111 Huntington Ave
14th Floor
Boston, MA 02199
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077543
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 9 of 237 4. ABSTRACT
Title: Post-Emergency Use Authorization Active Safety Surveillance Study among
Individuals in the Veteran’s Affairs Health System Receiving Pfizer -BioNTech Coronavirus
Disease 2019 (COVID -19) Vaccine
Protocol Version: 12.0; Date of Protocol : 27 January31 Aug 2021
Authors: Yinong Young Xu, ScD, MA, MS , Veterans Affairs Medical Center; Cynt hia de
Luise, PhD, MPH, Pfizer, Inc.; Mei Sheng Duh, ScD, MPH, Analysis Group, Inc.
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 was first identified by public health officials in China
in December 2019.1 The COVID -19 pandemic presents an unprecedented public health
crisis. As of January 7 , 2021, over 21.4 million 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
individuals (NC T04368728). The Food and Drug Administration (FDA) reviewed the
available safety data from 37,586 participants 16 years of age and older and did not identify
any specific safety concerns. In addition, the analysis 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 days 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 vacci ne and
placebo groups, respectively).3,4 Based 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 years 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.6 On December 21, 2020, the
European Medicines Agency (EMA) granted 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 st ates.7
As required by 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
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077544
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 10 of 237 pre-determined safety events of in terest (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, elderly, and those with
specific comorbidities ).4 Pfizer in collaboration with the US Veterans Health Administration
(VHA) and Analysis Group herein propose post -EUA active safety surveillance of safety
events of interest based primarily on the Priority List of Adverse Events of Special Interest
from the Brighton Collaboration’s Safety Platform for Emergency vACcines (SPEAC)
Project, and from the FDA and thepreliminary list of safety events of interest presented at the
September 22 , 2020, meeting of Centers for Disease Control and P revention’s ( CDCCDC’s)
Advisory Committee on Immunization Practices (ACIP) on the enhanced safety monitoring
recommendation .of COVID -19 vaccines .8,9 This safety surveillance study will identify and
evaluate rapid, near real-time potential safety signals associated with the Pfizer -BioNTech
COVID-19 vaccine in the large -scale VHA electronic medical record (EMR) database. The
observed safety event of int erest 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 Safety Monitoring (PRISM) program for the H1N1
vaccine.810 This non-interventional study is designated as a Post -Authorization Safety Study
(PASS) commitment to the US FDA and is a Category 3 commitment in th e EU Risk
Management Plan .
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, 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 followin g receipt of the Pfizer -BioNTech COVID -19 vaccine;
• To assess whether sub -cohorts of interest (i .e., immunocompromised, elderly,
individuals with specific comorbidities, individuals receiving only one dose of the
Pfizer-BioNTech COVID -19 vaccine, and indivi duals 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.
Secondary study objective:
• 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 healt h histories of
recipients, overall and among the sub-cohorts of interest .
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077545
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 11 of 237 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 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 pre or post-vaccination non -risk intervals (“prevaccination control
interval” and “ post-vaccination control interval”) in the same individual.
• An active comparator design will be used to sequentially m onitor occurrence of safety
events of interest 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 (SCCS)
and comparison to unvaccinated contemporary control s. Additionally, s ignal 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 interest that occur among all individuals receiving the Pfizer -BioNTech
COVID-19 vaccine in the period from December 11, 2020 t o present. Individuals will be
included if they have a record of at least one dose of Pfizer -BioNTech COVID -19 vaccine .
Individuals who receive at least one dose of COVID -19 vaccine from a manufacturer other
than Pfizer -BioNTech will be identified and reportedsummarized, but they will be excluded
from further analysis . All individuals will be required to be enrolled in and not disenrolled
from VHA benefits during the 1 year prior to vaccination date ( i.e., baseline period).
Depending on the attrition rate, the leng th of the baseline period may be modified to 6
months.
The influenza vaccine comparator cohort will be identified based on a record of at least one
dose of seasonal influenza vaccine during prior flu seasons, from 2014/2015 through
2018/2019 .
Variables :
• Exposures: Administration of Pfizer -BioNTech COVID -19 vaccine post -EUA
approval will be identified based on the following : (see Appendix Table 3 for
additional details) :
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077546
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 12 of 237 o Current Procedural Terminology (CPT) code 91300 (Severe acute respiratory
syndrome coronavirus 2 (SARS CoV2) (coronavirus disease [COVID 19])
vaccine, mRNALNP, spike protein, preservative free, 30 mcg/0.3mL dosage,
diluent reconstituted, for intramuscular use) and associated vaccine
administration Healthcare Common Procedure Coding System ( HCPCS)
codes corresponding to the first dose: 0001A (ADM SARS CoV2 30
mcg/0.3mL 1st), and the sec ond dose: 0002A (ADM SARS CoV2 30
mcg/0.3mL 2nd);9,10; OR
o 10 and 11 -digit National Drug C odes (NDCs ) 5926710001 (corresponds to
first dose), 59267 100001 (corresponds to second dose);9); OR
o Immunization records that contain data on vaccine code descriptor, vaccine
manufacturer ( i.e., Pfizer), lot number, injection site, and date(s) of
immunization;911
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 Table 3 for additional details) :
o CPT codes
o 90654 (Influenza virus and associated vaccine, trivalent (IIV3), spli t virus,
preservative free, for intradermal use); administration HCPCS codes ; OR
▪ 90656 (Influenza virus vaccine, trivalent (IIV3), split virus,
preservative free, 0.5 mL dosage, for intramuscular use); OR
▪ 90658 (Influenza virus vaccine, trivalent (IIV3), split virus, 0.5 mL
dosage, for intramuscular use); OR
o 10 and 11 -digit NDCs; OR
o Immunization 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 Table 1 and Appendix
Table 2Appendix Table 2) are 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 ( CDC’s ACIP) enhanced
safety monitoring recommendations .
The list of safety events of interest may be revised over the course of the study, and i f
unanticipated potential safety events of interest are identified during the course of
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077547
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 13 of 237 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 (see Table 1). Outpatient (including , emergency
department [(ED])), and/or inpatient settings will be used to identify safety events of
interest depending on the type of event. The specific encounter setting to be
considered for each safety event of interest is summarized in Table 1 and can be
assigned to 1) the risk interval following vaccination Pfizer-BioNTech COVID -19
vaccination, 2) the prevaccination self control interval, 3) the post-vaccination self -
control interval, or 43) 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 followi ng a specified
clean window ( i.e., the occurrence -free baseline period used to 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 Table 2) in order to rule out pre -existing events.
• Key Covariates : Baseline demographic ( i.e., age, sex, race/ethnicity, stateservice
region) and clinical characteristics ( i.e., smoking, body mass index [ BMI], history of
anaphylaxis/allergic reactions, previous anaphylaxis to vaccine component, history of
hospitalizations, frailty index, Charlson Comorbidity Index [CCI], select ed
comorbidities , and concurrent immunizations )1112 will 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 , elderly, in dividuals with specific
comorbidities, (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 chemotherapy or immune modulators; individuals
diagnosed with rheum atologic/inflammatory conditions and administered systemic
corticosteroids; individuals who were administered chemotherapy, immune
modulators, or systematic steroids for at least 14 days) ,13 elderly, individuals with
specific comorbidities,12 those receiving only 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 . Analyses will also
be performed among individuals enrolled in the VHA with dual coverage who are
also identified in linked Centers for Medicare & Medicaid Services (CMS) Medicare
administrative claims data.
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077548
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 14 of 237 Data source : The VHA is the largest integrated health ca re system 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.1214 This study will use data
from VHA’s Corporate Data Warehouse (CDW), which is an integrated electronic medical
record (EMR) system with a centralized data warehouse that is updated on a daily basis. The
CDW does not include information on any 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’ own homes.13 15 In a subgroup analysis 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 evaluation of the care an indivi dual 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 analyses. 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 analysis : A stepwise approach, illustrated in the diagram , will be performed for signal
detection, evaluation, and verification.
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077549
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 15 of 237
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077550
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 16 of 237
Notes:
[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 control intervals selected for the SCRI analysis for each safety event of i nterest 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 inpatien t 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 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 inte rest occurs in the
clean 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
1) Signal detection : The goal is to provide rapi d-cycle, near real -time safety surveillance. In
the signal detection phase, the SCRI analysis will only include prevaccination control
intervals as the post-vaccination control intervals willfor certain safety events of interest that
require a longer time to accumulate and will be used COVID-19 diagnosis (i.e., severe
COVID-19, multisystem inflam matory syndrome in the signal evaluation phase.adults [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 will be
applied. For comparison with individuals who received seasonal influenza vaccination, the
Poisson-based MaxSPRT will be applied . for all other safety events of interest .
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 .116 Signals will be detected if the
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077551
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 17 of 237 critical values are reached via the SCRI or active comparator analysis. Critical values wil l 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 power. Incidence rates will also be calculated and Kaplan Meier methods wi ll be used to
analyze time to safety event of interest
2) Signal evaluation: If signals are detected for safety events of interest 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 medi cal records, checking 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 regression to account for baseline differences between
Pfizer-BioNTech COVID -19 vaccinated and active comparator cohorts. SCRI analyses using
the post-vaccination control intervals and SCCS using post -vaccination contr ol time periods
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 weighting (IPTW). The assessment of temporal clustering will also be conducted.
Incidence rates will also be calculated and Kaplan -Meier methods wi ll be used to analyze
time to safety event of interest. Lastly, the assessment of temporal clustering will also be
conducte d. Signal evaluation a nalyses will be conducted every six months.
3) Signal verification: diagnostic validation of the detected safety events of interest via
adjudication of medical records by VHA clinicians for outcome verification 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 ,,13 individuals
with specific comorbidities ,,12 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 -
vaccination serology, those receiving care regularly at VA facilities , and lastly 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 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 .17 To provide additional context to the investigation conducted by CDC,
separate safety analyses will be prioritized and performed to assess the risk of
myocarditis/pericarditis following Pfizer -BioNTech COV ID-19 vaccination . These analyses
will be conducted to align with the rapid -cycle analysis performed by the Vaccine Safety
Datalink (VSD).18 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 ),
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077552
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 18 of 237 gender, and race/ethnicity, respectively. Incidence rate ratios will be summarized to compare
the rate of myocarditis/pericarditis events between vaccinated individuals who se event occurs
in a pre-specified risk interval versus vaccinated individuals who se event occurs in a
comparison interval on the same calendar day. Myocarditis/p ericarditis events will also be
adjudicated via chart review and validated using the Brighton Collaboration’s case
definitions.19 Risk factor analysis may also be conducted among confirmed cases. Lastly,
additional da ta surrounding risk factors, clinical course, and sequelae of identified
myocarditis/pericarditis event up to 365 days following the event will be collected and
summarized .
Milestones :
• Registration in the EU PAS register : To be registered before the start of data
collection;
• VHA Cooperative Research and Data Agreement ( CRADA) and execution: 8
January 2021 ;
• Determination of Institutional Review Board ( IRB) approvals (estimated) :exemption:
10 February 2021 ;
• Determination of Research Safety and Security exemption: 17 February 2021 ;
• Approval by Designated Member Review: 26 February 2021 ;
• Registration in the EU PAS register: 5 MarchApril 2021;
• Start of data collection (estimated planned date for starting data extraction for
analysis):: 11 May 2021;
• Interim reports: 30 June 2021; 31 December 2021; 30 June 2022, 31 December 2022;
• End of data collection (estimated planned date for final data cut): 10: 30 June 2023;
• Final study report: 31 December 2023
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077553
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 19 of 237 SUMMARY
Objective Primary 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
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.
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 occurrence of safety events of interest
while controlling for time -invariant confounders. This design allows inclusion of either a pre vaccination
control interval or a post-vaccination control interval , depending on the safety event of interest (e.g., post -
vaccination control intervals are used for outcomes where there is concer n for bias due to indication or
contraindication); .
• An active comparator design will be used to sequentially monitor occurrence of safety events of interest 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 h ealth 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 (SCCS) and comparison of v accinated 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) .
Study population 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
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077554
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 20 of 237
Objective Primary 1 Primary 2 Secondary
• 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
• At least 1 year of enrollment in and no disenrollment from VHA benefits ( i.e., the base line 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 other than Pfiz er-BioNTech will be identified and reportedsummarized , 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 : (see Appendix Table 3 for additional details) :
• Current Procedural Terminology (CPT) code 91300 (Severe acute respiratory syndrome coronavirus 2
(SARSCoV2) (coronavirus disease [COVID 19]) vaccine, mRNALNP, spike protein, preservative free, 30
mcg/0.3mL dosage, diluent reconstituted, for intramuscular use) and associated vaccine administration
HCPCS codes corresponding to the first dose: 0001A (A DM SARS -CoV-2 30 mcg/0.3mL 1t), and the second
dose: 0002A (ADM SARS -CoV-2 30 mcg/0.3mL 2d);; OR
• 10 and 11 -digit National Drug Codes (NDCs ) 59267-1000-1 (corresponds to first dose), 59267 -1000-01
(corresponds to second dose ); OR
• Immunization records that contain data on vaccine code descriptor, vaccine manufactur er (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 : (see Appendix Table 3 for additional details) :
• CPT codes
• 90654 (Influenza virus and associated vaccine, trivalent (IIV3), split virus, preservative -free, for intradermal
use); administration HCPCS codes ; OR
o 90656 (Influenza virus vaccine, trivalent (IIV3), split virus, preservative free, 0.5 mL dosage, for
intramuscular use); OR
o 90658 (Influenza virus vaccine, trivalent (IIV3), split virus, 0 5 mL dosage, for intramuscular use);
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.
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077555
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 21 of 237
Objective Primary 1 Primary 2 Secondary
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 (including , 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
pre-vaccination self -control interval, 3) the post-vaccination self -control interval, or 43) 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) will be included; this means that if a safety event is identified but
diagnosis 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 :
• Generalized convulsions/seizures;
• Guillain-Barré syndrome (GBS);
• Aseptic meningitis ;
• Encephalitis/encephalomyelitis ;
• Other acute demyelinating diseases ;
• Transverse myelitis (TM);
• Multiple sclerosi s (MS);
• Optic neuritis (ON);
• 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)
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077556
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 22 of 237
Objective Primary 1 Primary 2 Secondary
• Optic neuritis (ON)
• Other acute demyelinating diseases
• Transverse myelitis (TM)
Immunologic :
• Anaphylaxis ;
• Vasculitides ;
• Arthritis and arthralgia/joint pain ;
• Multisystem inflammatory syndrome in adults (MIS -A);
• Kawasaki disease (KD);
• Fibromyalgia ;
• Autoimmune thyroiditis
• Fibromyalgia
• Kawasaki disease (KD)
• Multisystem inflammatory syndrome in adults (MIS -A)
• Vasculitides
Cardiac :
• Myocarditis;
• Pericarditis;
• Acute myocardial infarction (AMI)
• Arrhythmia
• Coronary artery disease (CAD)
• Heart failure and cardiogenic shock
• Microangiopathy
• Myocarditis
• Pericarditis
• Stress cardiomyopathy
Hematologic :
• Cerebrovascular hemorrhagic stroke
• Chilblain -like lesions
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077557
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 23 of 237
Objective Primary 1 Primary 2 Secondary
• Th o bocytope i ;
• Disseminated intravascular coagulation (DIC)
• COVID19 (for all COVID 19related safety events of interest listed below, an inpatient diagnosis of COVID
19 will be re quired in combination with the codes or laboratory values specified in Appendix Table 2 ; in
addition, COVID -19 related safety events of interest will only be evaluated using data from 2020 onward
using the SCRI design ):
• Se e e COVID 19 dise se ;
• Microangiopathy ;
• Heart failure and cardiogenic shock ;
• Stress cardiomyopathy ;
• Coronary artery disease (CAD);
• Arrhythmia ;
• Deep vein thrombosis (DVT );)
• Pulmonary embolus;
• Cerebrovascular hemorrhagic stroke; Hemolytic anemia
• Ce eb o sc l non-hemorrhagic stroke;
• Limb ischemia ;
• Hemorrhagic disease ;
• Limb ischemia
• Pulmonary embol ism (PE)
• Acute kidney injury ;
• Liver injury ;
• Chilblain like lesions ;
• Single organ cutaneous vasculitis ;
• Thrombocytopenia
• Thrombosis with thrombocytopenia syndrome (TTS)
Other:
• Acute kidney injury
• Appendicitis
• Death
• Erythema multiforme
• Liver injury
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077558
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 24 of 237
Objective Primary 1 Primary 2 Secondary
• Other
• De th;
• Narcolepsy / and cataplexy ;
• Non-anaphylactic allergic reactions ;
• Appendicitis Severe COVID -19 disease
• Stevens-Johnson syndrome (SJS)/Toxic epidermal necrolysis (TEN)
Data source The VHA Corporate Data Warehouse (CDW) database will be used . and may be supplemented with Medicare
administrative claims data from the Centers for Medicare & Medicaid Services (CMS) .
Data analysis A stepwise approach will be performed for signal detection, evaluation, and verification.
1) Signal detection: The goal is to provide rapid -cycle, near real -time safety surveillance. In the signal detection phase,
the SCRI analysis will only include pre -vaccination control intervals as SCRI analyses using the post-vaccination
control intervals will be conducted for certain safety events of interest that require a longer time to accumulate and will
be used in the signal evaluation phase. COVID-19 diagnosis (i.e., severe COVID -19 illness, MIS -A). To account for
multiple testing and bi -weekly review of the data, the m aximized 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 MaxSPRT will 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 power. 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 intervals and the SCCS design with post -vaccination control time period will be
conducted as an additional inferential analysis once enough post -vaccination time has accumulated. Lastly, theTo
address potential period effects, a comparison to contemporary unvaccinated controls will also be performed with
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077559
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 25 of 237
Objective Primary 1 Primary 2 Secondary
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 a nalysis (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 -
vaccination serology, those receiving care regularly at VA facilities, and lastly 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 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/pericarditis following Pfizer -BioNTech COVID -19 vaccination . These anal yses 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 will be summarized to compare the rate of
myocarditis/pericarditis events between 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/pericard itis event up to 365
days following the event will be collected and summarized .
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077560
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 26 of 237 5. AMENDMENTS AND UPDATE S
None.
Amendment number Date Protocol
section(s)
changed Summary of amendment(s) Reason
1 31 August
2021 6 Updated the Milestones 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
2021 9.1.1 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 Center
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 analysis if there is no
overlap between the risk intervals
for the two doses.
1 31 August
2021 9.1.1 Added that a dditional doses of the
Pfizer-BioNTech COVID -19 vaccine
may be included in the analysis. To address the potentia l approval of
additional doses. Details for this
analysis will be further described in
the statistical analysis plan.
1 31 August
2021 9.1.1, 9.3.3,
9.7.3, 9.7.5, 9.9 Removed SCRI design with pre -
vaccination control interval and added
SCRI design with post -vaccination To address CBER request to
remove the pre -vaccination control
interval as its comparison to the risk
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077561
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 27 of 237 Amendment number Date Protocol
section(s)
changed Summary of amendment(s) Reason
control interval for 2 safety events of
interest (severe COVID -19,
multisystem inflammatory syndrome
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. interval may introduce bias and
reduce the probability of subsequent
vaccination. Note additional and
more robust analyses 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
2021 9.2.3, 9.4 Added clarification for the
identification of subgroups who are
immunocompromised s and
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
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077562
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 28 of 237 Amendment number Date Protocol
section(s)
changed Summary of amendment(s) Reason
with linked Medicare data has been
added.
1 31 August
2021 9.3.1 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 wel l as
additional codes identified at the time
of the data analysis. To update the protocol with all
relevant CPT/HCPCS/NDC codes,
while maintaining concise language
in the main text.
1 31 August
2021 9.3.1, 18 Added Appendix Table 4 in Section
18 regarding the LOINC codes used to
identify COVID -19 RT-PCR Test
among the study 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
2021 9.3.2, 18 Added frailty 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 vacc ination as frailty
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077563
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 29 of 237 Amendment number Date Protocol
section(s)
changed Summary of amendment(s) Reason
may be a prognostic factor for
safety events of interest.
1 31 August
2021 9.3.3, 18 Added four additional safety events of
interest: thrombosis with
thrombocytopenia syndrome,
convulsions/seizures in individuals
with controlled epilepsy, Steven -
Johnson syndrome/Toxic epidermal
necrolysis, and hemolytic anemia
(increasing the number of safety
events of interest from 42 to 46).
Reclassified COVID -19-related safety
events of interest to be measured
independently of the pa tient’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). To consider new safety events
based on emerging research and
align with codes from the FDA
CBER COVID-19 Vaccine Safety
Surveillance : Active Monitoring
Master Protocol.20,21,22
The COVID -19-related safety
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), and
therefore are defined independent of
a COVID -19 diagnosis, with the
exception of “s evere COVID -19
disease” and “MIS -A” which
requires a concurrent COVID -19
diagnosis.
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077564
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 30 of 237 Amendment number Date Protocol
section(s)
changed Summary of amendment(s) Reason
Extending the clean window will
address the reduction in healthcare
resource utilization during the
pandemic to more accurately
identify incident events.
1 31 August
2021 9.7.3.2 Clarified in the Signal Evaluation
section that the signal evaluation
analyses will be conducted every six
months. To provide additional detail on the
timing of the signal evaluation
analyses.
1 31 August
2021 9.1.3, 9.7.3.2.5 Added self-controlled case series
(SCCS) design with full post-
vaccination peri od as an additional
analysis in the Si gnal Evaluation
analysis. Added that Signal
Evaluation analyses 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).23 SCCS analysis 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 analyses may also be
conducted based on signals detected
in external sources or based on
regulatory request even if such
analyses were not first id entified in
the Signal Detection phase of this
study.
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077565
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 31 of 237 Amendment number Date Protocol
section(s)
changed Summary of amendment(s) Reason
1 31 August
2021 9.7.3.2.6 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
comparator design that uses
historical controls of influenza
vaccinated individuals.
1 31 August
2021 9.7.8 Added new section on
myocarditis/pericarditis safety
analysis and risk factor analysis. To include a separate analys is
focused on myocarditis/pericarditis
based on emerging evidence
regarding th is event in association
with mRNA COVID -19 vaccines.17
1 31 August
2021 9.9 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 analyses.
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077566
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 32 of 237 6. MILESTONES
Milestone Planned date
VHA CRADA execution, Determination of IRB &
Research Safety and Security exemptions,
Approval by Designated Member Review[1-
3]Registration in the EU PAS register To be registered before the start of
data collection January - February
2021
Registration in the EU PAS register VHA CRADA
and IRB approvals (estimated) 5 March April 2021
Start of data collection (estimated) 11 May 2021[14]
Interim reports 30 June 2021
31 December 2021
30 June 2022
31 December 2022
End of data collection (estimated) 1030 June 2023[25]
Final study report 31 December 2023
Abbreviations: ACOS, Associate Chief of Staff; COVID -19, Coronavirus disease 2019; CRADA,
Cooperative Research and Data Agreement; IRB, Institutional Review 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, Subcommittee 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 Adm inistration ; US, United States .
Notes:
[1] Start of data collection is the planned [1] IRB exemption determination was granted in accordance with 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 verification phase for chart review, a second IRB review application will be submitted for an
expedited or full review. The two -stage IRB application process is to expedite the initiation of the project.
[2] Research Safety and Security exemption determination was granted by the Subcommittee on Research
Safety and Security (SRSS), VA Innovation and Research Review System (VAIRRS).
[3] Approved by Associate Chief of Staff for Research and Deve lopment (ACOS/R&D) and R&D
Committee of the Northern New England Research 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 will include the includes Pfizer-BioNTech COVID -19 vaccine exposure since exposures
from December 11, 2020 , (the EUA approval date by the US FDA .) to March 12, 2021 (the data cutoff date).
[25] End of data collection is the planned date on which after the Pfizer -BioNTech COVID -19 vaccine
exposure data reached 30 mont hs post-EUA approval .
and the last day of the month that the study will be completed.
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077567
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 33 of 237 7. RATIONALE AND BACKGROUND
In March 2020, the World Health Organization (WHO) declared a global pandemic for the
coronavirus disease 2019 (COVID -19) due to the severe acu te respiratory syndrome
coronavirus 2 (SARS -CoV-2), which was first identified by public health officials in China
in December 2019 .1 The COVID -19 pandemic presents an unprecedented public health
crisis. As of January 7, 2021, over 21.4 million COVID -19 cases and 364,000 deaths have
been reported in the United States (US) alone.2 To date, the incidenc e of COVID -19 has
continued to rise, largely affecting the elderly 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 lu ng disease).15,1624,25 SARS-CoV-2 is a
well-adapted highly infectious human pathogen with a case fatality rate that ranges between
0.5% and 20%, based on the individual’s age, gender, race, and como rbidites.1726
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). To this end, Pfizer is conducting a Phas e 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 immunogenicity of
two mRNA vaccine candidates ( i.e., BNT162b1, BNT162b2) at various dose levels,
candidate BNT162b2 was selected for advancement to a pivotal Phase 2/3 safety and efficacy
evaluation due to its milder systemic reactogenicity profile, especially in older adults.1827 The
study was initiated in July 2020 with a target enrollment of 43,998 individuals .1928
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. In 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, especially adverse events and cases of
severe COVID -19 in vaccinated study subjects.20,2129 The FDA reviewed the available safety
data of the Phase 1/2/3 trial from 37,586 participants 16 years of age and older and did not
identify any specific safety concerns. In addition, the analysis of available efficacy data from
36,523 participants 12 years o f age and older without evidence of prior SARS -CoV-2
infection at least 7 days 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 grou ps, respectively).3,4 Based 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 a nd potential benefits of the vaccine outweighed the known
and potential risks for the prevention of COVID -19 in individuals 16 years 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 wo rld to grant temporary authorization for
emergency use of the Pfizer -BioNTech COVID -19 vaccine.6 On December 21, 2020, the
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077568
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 34 of 237 European Medicines Agency (EMA) granted 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 by 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 safety 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, elderly, and those with
specific comorbidities ).4 Post-authorization safety evalua tions are important for identifying
rare, serious safety 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 po st-trial. Pfizer in collaboration with the US Veterans Health
Administration ( VHA) of the Department of Veterans Affairs (VA) and Analysis Group
herein propose post -EUA active safety surveillance of safety events of interest based
primarily on the Priority List of Adverse Events of Special Interest from the Brighton
Collaboration’s Safety Platform for Emergency vACcines (SPEAC) Project , and from the
FDA and thepreliminary list of safety events of interest presented at the September 22 , 2020,
meeting of Centers for Dise ase Control and Prevention’s ( CDCCDC’s) Advisory Committee
on Immunization Practices (ACIP) on the enhanced safety monitoring recommendation .of
COVID-19 vaccines .8,9 This safety surveillance study will identify and evaluate rapid, near
real-time potential safety signals associated with the Pfizer -BioNTech COVID -19 vaccine in
the large-scale VHA electronic medical record ( EMR) database. The observed rates of safety
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
approach es previously used by the Post -Licensure Rapid Immunization Safety Monitoring
(PRISM) program for the H1N1 vaccine .810 This non-interventional 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 .
8. RESEARCH Q UESTION 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 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, elderly, with
specific comorbidities, individuals receiving only o ne 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.
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077569
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 35 of 237 Secondary 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-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 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). In addition, safety events of interest associated with Pfizer -
BioNTech COVID -19 vaccinations will be sequentially monitored and compared to
recipients of influenza vaccine in the VHA between 2014/2015 to 2018/2019 .8,22 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 vaccin ation) to compare the risk interval following vaccination to pre or
post-vaccination non -risk intervals (“prevaccination control interval” and “ post-vaccination
control interval”) in the same individual.23 Whether a pre or postvaccination control
interval is used will depend on the clinical nature, seasonality, and frequency of the safety
event of interest , as described in greater detail below.31 A length of 42 days has been used to
define the risk interval i n SCRI design studies for signal detection to ascertain the safety
profile of the H1N1 vaccine.8,2210,30 The same length of risk interval is proposed here, subject
to further modification based on clinica l input, clinical trial data, biologic plausibility, and
published literature. The day of vaccination will only be included in the risk period for those
safety events of interest for which a same -day occurrence is biologically plausible
(e.g., anaphylaxis) .
As some individuals may choose to decline or delay Pfizer BioNTech COVID 19
vaccination soon after an illness (known as the “healthy vaccinee effect”),24 the
prevaccination control interval will exclude the 14 day period before vaccination.25 While
using a prevaccination control period allows for timely analysis, especially pertinent for
rarer safety events of interest , aA post-vaccination control interval would be more
appropriate and will be used for certain safety events of interest for the following reasons: (1)
a recent prior safety event of interest might preclude vaccination ( i.e., anaphylaxis), (2)
individuals might have an underlying conditi on that is also a contraindication for vaccination
(i.e., seizure disorder), or (3) safety event of interest and vaccination may be seasonal in
nature.2632 The time between the risk and control intervals will be determined based on the
biological mechanism of action for each safety events of interest assessed, and may be
subject to change based on further clinical input . Examples of the SCRI design with a pre
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077570
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 36 of 237 vaccination control interval and a post-vaccination control interval (in an individual who only
receives the first dose of vaccine) is presented in Figure 1 below.
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,
Showing Both Pre andwith Post-vaccination Control Intervals *
*The risk interval may include day 0, date of Pfizer -BioNTech COVID -19 vaccination, for some of the safety
events of 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 1 while Figure 2 represents an example where the complete course with 2 doses are received.
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077571
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 37 of 237 Two doses of the Pfizer -BioNTech COVID -19 vaccine are recommended 3 weeks apa rt.
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
doses 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).
For Given the risk intervals for specific safety events of interest range from 1 day to 90 days
(please see Table 1 in 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 event after the first
dose. See Figure 2 below for SCRI design examples where a safety event with a 42 risk
interval window (e.g., Bell’s palsy ; Table 1 in Section 9.3.3 ) is assessed in hypothetical
individual s who receive two doses of the vaccine, two separate control intervals will be
defined to correspond to the risk interval associated with each dose (regardless of whether
pre or postvaccination control intervals are used). See Figure 2 below for an example in an
individual who receives two doses of Pfizer-BioNTech COVID-19 vaccine ,: Figure 2A
shows the SCRI design with the second dose received 21 days after the first. Safety (i.e., the
risk interval for dose 1 overlaps wi th the risk interval for dose 2) , while Figure 2B shows the
SCRI design with the second dose received 60 days 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), the 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 2)Figure 2A) may be flagged for separate analyses to
discern the additive effect of Pfizer -BioNTech COVID -19 vaccine dose 1 and dose 2. For the
second scenario ( Figure 2B), events will only be measured during the risk intervals, ignoring
the gap between the end of the risk interval for dose 1 and dose 2 initiation.
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077572
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 38 of 237 For each analysis, control intervals corresponding to the risk interval s 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).
Figure 2. Example of SCRI Design with Overlapping for Assessment of a Safety Event
of Interest with a 42-day Risk Intervals when Two Doses of Pfizer BioNTech
COVID 19 Interval in an Individual who Receives Two Vaccine are
Administered, Showing a Pre andDose s, with Post -vaccination Control
Interval Intervals
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077573
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 39 of 237
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 with the event frequency amo ng 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 a nd under-reporting of medical events. The same risk interval length
(e.g., 42 days) will be used to evaluate safety events of interest following vaccination with
Pfizer-BioNTech COVID -19 vaccine and to assess safety events of interest occurring after
vaccination for seasonal influenza in prior seasons. The observed number of safety events of
interest for Pfizer -BioNTech COVID -19 vaccine will be compared to the expected number
calculated for the influenza vaccine in past seasons.8.10
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077574
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 40 of 237 9.1.3. Additional Study Design s 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 analyses in the signal detection phase . These include
analyses using self-controlled case series (SCCS) and comparison of vaccinated unvaccinated
contempora ry controls. Additionally, signal evaluation analyses may also be conducted for
signals detected in external sources or based regulatory request (e.g.,
myocarditis/pericarditis) . These analyses are further detailed in Section 9.7.3.2.
9.1.3. 9.1.4. 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.
9.2. Setting
The study population will be kept as broad as possible in order to capture safety events of
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 per iod 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 year of enrollment in and no disenrollment from VHA benefits ( i.e., the
baseline period) pr ior 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 reportedsummarized , but they will be excluded from further
analysis.
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 chemotherapy or immune modulators;
individuals diagnosed with rheumatologic/inflammatory conditions and administered
systemic cortico steroids; or individuals who were administered chemotherapy,
immune modulators, or systematic steroids for at least 14 days ;13
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077575
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 41 of 237 • Different age groups , with a focus on the elderly (e.g., < 35, 35 - < 45, 45 - < 55, 55 -
< 65, 65 - < 75, > 75);
• Individuals with specific comorbidities;
• 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 Syndrome, cardiovascular conditions [e.g., heart failure, coronary
artery disease, or cardiomyopathies], 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 encounters in the one year prior to vaccination . The encounters must be
separated by > 30 days (for inpatient, by admission date), and at least one must be
within six months prior to the date of vaccination . This will ensure that individuals
have ongoi ng health care encounters, particularly near the vaccination date, and
regularly receive their healthcare from VHA facilities, rather than outside facilities
that would not be captured in the CDW;VHA’s Corporate Data Warehouse ( CDW);
• Individuals who are in the VA priority group 1 Veteran. These individuals have either
the highest levels of service connected disability (> (≥50% disabling), are considered
unemployable, or have received the medal of honor .27.33 Individuals categ orized as
priority group 1 are the highest priority 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 & Medicaid 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 health care encounters .
Additional subgroups of interest will be assessed as ad ditional information becomes
available from ongoing clinical trials, Vaccine Adverse Event Reporting System (VAERS),
and other sources that will inform the Pfizer -BioNTech COVID -19 vaccine safety profile.
Given that VA population has a median age of over 46 years for females and is comprised of
approximately 90% males , the evaluation of the Pfizer-BioNTech COVID -19 vaccine safety
during pregnancy, including fetal death and infant outcomes, may have poor feasibility and
will therefore not be conducted.
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077576
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 42 of 237 9.3. Vari ables
9.3.1. Exposure of Interest
Administration of Pfizer -BioNTech COVID-19 vaccine post-EUA approval will be identified
based on the following : (see Appendix Table 3 for additional details) :
• Current Procedural Terminology (CPT) code 91300 ( Severe acute respiratory
syndrome coronavirus 2 (SARS CoV2) (coronavirus disease [COVID 19]) vaccine,
mRNALNP, spike protein, preservative free, 30 mcg/0.3mL dosage, diluent
reconstituted, for intramuscular use ) codes and associated vaccine administration
HCPCS codes corresponding to the first dose: 0001A (ADM SARS CoV2 30
mcg/0.3mL 1st), and the second dose: 0002A (ADM SARS CoV2 30 mcg/0.3mL
2nd);9,10; OR
• 10 and 11 -digit National Drug Codes (NDCs ) 5926710001 (corresponds to first
dose), 59267 100001 (corresponds to second dose ); OR
• Immunization records that contain data on vaccine code descriptor, vaccine
manufacturer ( i.e., Pfizer), lot number, injection site, and date(s) of immunization.911
Relevant codes will be continuously reviewed and amended if new codes are added.
Person-time at-risk exposure to the first dose only, overlapping first and second doses, and
second dose only will be analyzed separately.
Administration of the seasonal influenza vaccine during 2014/2015 through 2018/2019 flu
seasons wil l be identified based on the following : (see Appendix Table 3 for additional
details):
• CPT codes
o 90654 (Influenza virus vaccine, trivalent (IIV3), split virus, preservative free,
for intradermal use) ; OR
o 90656 (Influenza virus vaccine, trivalent (IIV3), split virus, preservative free,
0.5 mL dosage, for intramuscular use) ; OR
• 90658 (Influenza virus vaccine, trivalent (IIV3), split virus, 0.5 mL dosage, for
intramuscular use );; 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.
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077577
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 43 of 237 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 PRISM safety surveillance
program of H1N1 vaccine safety:810
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 days prior to COVID -19 vaccination during the
same flu season in which COVID -19 vaccination occurred;
Cohort C: Individuals vaccinated with Pfizer -BioNTech COVID -19 vaccine who received
the seasonal influenza vaccine within 42 days 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, ICD10CM Current
Procedural Terminology (CPT), CPT, or Healthcare Common Procedure Coding System
(HCPCS) procedure codes, and generic drug names, as appropriate (Appendix Table 1). The
following demographic and clinical characteristics will be assessed:
Demographic s:
• Age
• Sex
• Race/ethnicity
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077578
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 44 of 237 • State
• VHA service area
Clinical characteristics:
• Smoking status
• Body mass index ( BMI)
• History of anaphylaxis/ allergic reactions
• Previous anaphylaxis of vaccine component
• History of hospitalizations
• Frailty index
• Charlson comorbidity index (CCI)
• Selected c omorbidities
o Autoimmune d isease
o Asthma
o Bleeding diathesis or condition associated with prolonged bleeding
o Cancer
o Cardiovascular conditions
o Chronic kidney disease/dialysis
o Chronic obstructive pulmonary disease (COPD)/COPD/interstitial lung
disease
o Diabetes mellitus
o Down syndrome
o Sickle cell disease
o Hepatitis B virus (HBV)
o Hepatitis C virus (HCV)
o Human immunodeficiency virus (HIV)
o HIV
o Hyperlipidemia
o Hypertension
o Liver disease
o Neurological disease
o Other immune deficiencies
o Solid organ transplant
o Venous thromboembolism (VTE)
• Concurrent immunizations
o Seasonal influenza vaccine
o Tetanus di phtheria and pertussis (Tdap or Td)
o Chickenpox (varicella)
o Shingles (herpes zoster recombinant and/or live)
o Human papillomavirus (HPV)
o Pneumococcal conjugate
o Pneumococcal polysaccharide
o Hepatitis A
o Hepatitis B
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077579
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 45 of 237 o Meningococcal conjugate (MenACWY) and serogrou p B meningococcal
(MenB)
o Haemophilus influenza type b
Specific covariates of interest for the prioritized analysis of myocarditis/pericarditis are
described in Section 9.7.8 .
9.3.3. Outcomes
The 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 Centers for Disease Control and
Prevention ( CDC) enhanced safety monitoring recommendations.288,299 Endpoints of special
interest in signal detection, as noted by the FDA and CDC’s Advisory Committee on
Immunizati on Practices ( ACIP) are denoted in italics .29. 9 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 . See Appendix Table 2 for 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-CM codes). Outpatient (including , ED), and/or inpatient setting s will be
used to identify safety events of interest, depending on the type of event. The specific
encounter setting considered for each safety 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 safety events of interest will be assessed:
Neurologic :
• Generalized convulsions/seizures
• Guillain Barré syndrome (GBS)
• Aseptic meningitis
Encephalitis/encephalomyelitis
• Bell’s palsy
• 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 demyelinating diseases
• Transverse myelitis (TM)
Immunologic :
• Anaphylaxis
• Arthritis and arthralgia/joint pain
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077580
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 46 of 237 • Autoimmune thyroiditis
• Fibromyalgia
• Kawasaki disease (KD)
Multiple sclerosis (MS) Optic neuritis (ON) Bell’s palsy
Immunologic :
Anaphylaxis
• Vasculitides
• Arthritis and arthralgia/joint pain
• Multisystem inflammatory syndrome in adults (MIS -A)
• Vasculitides
Kawasaki disease (KD)
• Fibromyalgia
Autoimmune thyroiditis
Cardiac :
• Myocarditis
• Pericarditis
• Acute myocardial infarction (AMI)
• Arrhythmia
• Coronary artery disease (CAD)
Hematologic :
Thrombocytopenia
• Disseminated intravascular coagulation (DIC)
COVID 19 (for all COVID 19related safety events of interest listed below, an inpatient
diagnosis of COVID 19 will be required in combination with the codes or laboratory values
specified in Appendix Table 2; in addition, COVID 19 related safety events of interest will
only be evaluated using data from 2020 onward using the SCRI design ):
• Severe COVID 19 disease
Microangiopathy
• Heart failure and cardiogenic shock
• Microangiopathy
• Myocarditis
• Pericarditis
• Stress cardiomyopathy
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077581
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 47 of 237 Hematologic :
• Cerebrovascular hemorrhagic stroke
• Chilblain -like lesions
• Disseminated intravascular coagulation (DIC)
Coronary artery disease (CAD)
Arrhythmia
• Deep vein thrombosis (DVT)
• Pulmonary embolus
• Hemolytic anemia
Cerebrovascular hemorrhagic stroke
• Cerebrovascular non hemorrhagi c stroke
• Limb ischemia
• Hemorrhagic disease
• Limb ischemia
• Pulmonary embol us (PE)
• Acute kidney injury
Liver injury
Chilblain like lesions
• Single organ cutaneous vasculitis
• Thrombocytopenia
• Thrombosis with thrombocytopenia syndrome (TTS)
Erythema multiforme
Other :
• Acute kidney injury
• Appendicitis
• Death
• Erythema multiforme
• Liver injury
Death
• Narcolepsy / and cataplexy
• Non-anaphylactic allergic reactions
• Appendicitis
• Severe COVID -19 disease
• Stevens-Johnson syndrome (SJS)/Toxic epidermal necrolysis (TEN )
The risk and control intervals selected for the SCRI analysis for each safety event of interest
are based on biological plausibility a nd 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 prevaccination
control interval, 3) the post-vaccination control interval , (self-controlled designs) , or 43) the
risk interval for the active co mparators receiving seasonal influenza vaccine . (active
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077582
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 48 of 237 comparator design) . 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
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 safety event of interest is identified but diagnosi s codes (or
laboratory values in the case of select safety 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 safety e vents of interest in order to rule
out pre-existing events. This approach is consistent with the FDA’s COVID -19 Vaccine
Safety Surveillance Project.116 Additionally, the length of the clean window may be
extended (e.g., 2 ye ars) 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:
• If a safety event of interest occurs in the individual’ s prevaccination control risk
interval and there are no other diagnosis codes for the same safety event of interest in
the clean window (e.g., 1 -year prior to thatvaccination date), the safety event of
interest should be assigned to the prevaccination control risk interval.
o If a safety event of interest occurs in the pre vaccination control interval but
another diagnosis code for the same safety event of interest is identified
during the risk interval, then the safety event of interest will not be assig ned to
the risk interval and will only be assigned to the pre vaccination control
interval as it will have occurred in the required clean window preceding the
risk interval. However, if an outpatient safet y event of interest occurs in the
clean 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 in
order to capture event exacerbation.
o If a safety event of int erest occurs in the risk interval and another diagnosis
code for the same 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
o If a safety event o f interest occurs in the post -vaccination control interval and
there are no other diagnoses for the same safety event of interest in the risk
interval and clean window (e.g., oneyear prior to this date), which also
includes the pre -vaccination control interval , then the safety event of interest
will be assigned to the riskpost-vaccination control interval.
• The same appr oach will be applied for the post vaccination control intervals.
• The risk intervals for outcome evaluation for the active comparators who received
seasonal influenza vaccination will be the same as for the individuals who received
Pfizer-BioNTech COVID -19 vaccine .
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077583
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 49 of 237 • However, it is possible that some safety events of 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 delayed than anticipated. In these cases,
misspecification of the risk (and control) intervals could result 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 varying risk intervals (longer as well as shorter) in order to
increase the likelihood that the safety risk is detected accurately. Additionally, if
further refinement and evaluation is necessary, temporal scan statistics may be used
to empirically identify the at-risk time interval by evaluating clusters of safety even ts
of interest. This will be further described in the SAP.
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077584
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 50 of 237 Table 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 Prevaccination
control interval
(days) Ri
sk
int
er
val
(d
ay
s) Post-vaccination control
interval (days)
Neurologic
Aseptic meningitis IP only16 6 months16 1-
42
34 43-8434
Bell’s palsy16Ge e li ed
co lsio /sei es8 IP or OP8 6 months N/
A1
-
42 43-840
14 1529
GBS8,22 Cerebrovascular non -
hemorrhagic stroke16 IP onlyIP, primary
position only14 1 year N/A 1-
42
28 438429-56
Aseptic
meningitis30Convulsions/seizures in
individuals with controlled
epilepsy35 IP or OP-EDIP
only14 1 year N/A 1-
42
90 91-1804384
Encephalitis/encephalomyelitis816 IP only14 6 months 1
year 56 through 15 1-
42 43-84N/A
Other acute demyelinating
diseases8Guillain-Barré Syndrome
(GBS)16 IP or OP8IP, primary
position on 1 year 98 through 15 1-
42 43-84N/A
Generalized convulsion/seizures10 IP or OP-ED 6 months 0-
14 15-29 Deleted Cells
Deleted Cells
Deleted Cells
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077585
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 51 of 237 Table 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 Prevaccination
control interval
(days) Ri
sk
int
er
val
(d
ay
s) Post-vaccination control
interval (days)
TMaMultiple sclerosis (MS)10,30 IP only14or OP 1 year 98 through 15 1-
42 43-84N/A
MS8,22Optic neuritis (ON)10,30 IP or OP8 1 year 98 through 15 1-
42 43-84N/A
ON8,22Other acute demyelinating
diseases10,30 IP or OP8 1 year 98 through 15 1-
42 43-84N/A
Bell’s palsy8,22Transverse myelitis
(TM)16 IP or OP1-ED 1 year 56 through 15 1-
42 43-84N/A
Immunologic
Anaphylaxis8,22 IP or OP1-ED16 6 months 1
month16 N/A 0
20
-
116 797-810,30
VasculitideseArthritis and
arthralgia/joint pain0 IP onlyor OP 1 year N/A 1-
28
42 43-842956
Arthritis and arthralgia /joint
paincAutoimmune thyroiditis0 IP or OP 1 year N/A 1-
42 43-84
MISAbFibromyalgia0 IP or OPIP only14 1 year N/A 1-
42 43-84
KD31Kawasaki disease (KD)36 IP only31 1 year N/A 1-
28 29-56 Deleted Cells
Deleted Cells
Deleted Cells
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077586
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 52 of 237 Table 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 Prevaccination
control interval
(days) Ri
sk
int
er
val
(d
ay
s) Post-vaccination control
interval (days)
FibromyalgiacMultisystem
inflammatory syndrome in adults
(MIS-A)16 IP or OP-ED 1 year N/A 1-
42 43-84
Autoimmune
thyroiditiscVasculitides0 IP or OPonly 1 year N/A 1-
42
28 29-564384
Cardiac
Acute myocardial infarction (AMI)16 IP only 1 year 1-
28 29-56
Arrhythmiac IP only 1 year 1-
42 43-84
Coronary artery disease (CAD)c IP only 1 year 1-
42 43-84
Heart failure and cardiogenic shockc IP only 1 year 1-
42 43-84
Microangiopathy0 IP only 1 year 1-
28 29-56
Myocarditis8,2216 IP or OP14 1 year 56 through 15 1-
42
d1
42 43-84N/A Deleted Cells
Deleted Cells
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077587
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 53 of 237 Table 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 Prevaccination
control interval
(days) Ri
sk
int
er
val
(d
ay
s) Post-vaccination control
interval (days)
Pericarditis8,2216 IP or OP14 1 year 56 through 15 1-
42
d1
42 N/A43-84
AMIdStress cardiomyopathyc IP only14 1 year 56 through 15 1-
42 N/A43-84
Hematologic
Thrombocytopenia30 IP or OP14 1 year N/A 142 4384
DICeCerebrovascular hemorrhagic
stroke16 IP only14 1 year N/A 1-
42
28 29-564384
COVID 19 (for all COVID 19related safety events of interest listed below, an inpatient diagnosis of COVID 19 will be required in
combination with the codes or laboratory values specified in Appendix Table 2; in addition, COVID 19 related safety events of
interest will only be evaluated using data from 2020 onward using the SCRI design ):
Severe COVID 19
diseasebChillblain -like lesions0 IP onlyor OP 1 year N/A 1-
42
28 29-564384
MicroangiopathyeDisseminated
intravascular coagulation (DIC)16 IP onlyor OP-ED 1 year N/A 1-
42
28 29-564384 Deleted Cells
Deleted Cells
Deleted Cells
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077588
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 54 of 237 Table 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 Prevaccination
control interval
(days) Ri
sk
int
er
val
(d
ay
s) Post-vaccination control
interval (days)
Heart failure and cardiogenic
shockdDeep vein thrombosis
(DVT)16 IP onlyor OP 1 year 1-
28 29-56
through
15 142 N
/
A
Stress cardiomyopathydHemolytic
anemiae IP onlyor OP 1 year 56 through 15 1-
42 43-84N/A
CADdHemorrhagic disease0 IP only 1 year 1-
28 29-56
through
15 142 N
/
A
ArrhythmiadLimb ischemia0 IP only 1 year 1-
28 29-56
through
15 142 N
/
A
DVTe IP or OP1 1 year N/A 142 4384
Pulmonary emboluse (PE)16 IP or OP14OP 1 year N/A 1-
42
28 29-564384
Cerebrovascular hemorrhagic stroke8 IP only14 1 year N/A 142 4384
Cerebrovascular non hemorrhagic
stroke8 IP only1 1 year N/A 142 4384
Limb ischemiaeSingle organ
cutaneous vasculitis0 IP only 1 year N/
A1
-
28 29-561
42 4384 Deleted Cells
Inserted Cells
Deleted CellsDeleted Cells
Deleted Cells
Inserted Cells
Deleted Cells
Deleted Cells
Deleted Cells
Deleted Cells
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077589
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 55 of 237 Table 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 Prevaccination
control interval
(days) Ri
sk
int
er
val
(d
ay
s) Post-vaccination control
interval (days)
Hemorrhagic diseasee IP only 1 year N/A 142 4384
Acute kidney injuryg IP only 6 months N/A 142 4384
Liver injurygThrombocytopenia16 IP or OP 1 year N/A 1-
42 43-84
Chillblain like lesionseThrombosis
with thrombocytopenia syndrome
(TTS)e IP or OP 1 year N/A 1-
42 43-84
Single organ cutaneous vasculitise IP only 1 year N/A 142 4384
Erythema multiformef IP only 6 months N/A 12 89
Other
Acute kidney injuryf IP only 6 months 1-
42 43-84
Appendicitis16 IP or OP-ED 1 year 1-
42 43-84
Death IP or OP 1 year 0-
42 43-85
Erythema multiformeg IP only 6 months 1-
2 8-9
Liver injuryf IP or OP 1 year 1-
42 43-84
Narcolepsy and cataplexya IP or OP1416 1 year16 98 through 15 1
42N/A43-84 Deleted Cells
Deleted Cells
Deleted Cells
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077590
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 56 of 237 Table 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 Prevaccination
control interval
(days) Ri
sk
int
er
val
(d
ay
s) Post-vaccination control
interval (days)
1-
42
16
Non-anaphylactic allergic
reactions810,2230 IP or OP8 6 months N/A 1-
2 8-9
Appendicitis32Severe COVID -19
diseaseh IP only14 1 year 6
months N/A 01
-
42 43-84
Stevens-Johnson syndrome
(SJS)/Toxic epidermal necrolysis
(TEN)g IP only 6 months 1-
2 8-9
*Safety events of interest are 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.
Notes:
a Published risk and control intervals for demyelinating diseases and cranial disorders were applied to TM and narcolepsy/catap lexy
b Published setting, clean window, and As severe COVID 19 ranges from severe pneumonia, acute respiratory distress syndrome, and multisystem organ
failure/MIS A, a 142 day risk interval was appli ed in order to capture the 14 day incubation period of the disease and 4 5 day period from exposure to
symptom onset
a. c Published risk and control intervals for autoimmune disorders were applied to similar autoimmune rheumatic conditions (i.e., arthritis and arthralgia/joint
pain, fibromyalgia and autoimmune thyroiditis).
d Published setting, clean window, and risk and control intervals for myocarditis DVT, pulmonary embolus and pericarditis DIC were applied to other
cardiovascular conditions (ie, heart failure and cardiogenic shock, stress cardiomyopathy, CAD, arrhythmia, AMI) Deleted Cells
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077591
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 57 of 237 Table 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 Prevaccination
control interval
(days) Ri
sk
int
er
val
(d
ay
s) Post-vaccination control
interval (days)
b. e Similar risk and control intervals were applied to all cardiovascular and hematological disorders characterized by damage to the blood vessels and/or
arteries and clotting (i.e., microangiopathy, DVT, pulmonary embolus, limb ischemia, hemo rrhagic disease, DIC, 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. f Published Published setting, cle an 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 myocardi tis/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.
c.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 ).
g Risk intervals of 42 days were applied for acute kidney injury and liver injury to be consistent with other COVID -19 related safety events of interest .
d.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 appli ed in order to capture the 14 -day incubation period of the disease and 4 -5 day period from exposure to symptom onset
Deleted Cells
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077592
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 58 of 237 9.4. Data Source
The VHA is the largest integrated health care system 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.1214 VHA’s health care delivery system is
organized regionally 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 System (VISN 1) covers
VHA facilities in Massachusetts, Connecticut, New Hampshire, Maine, and Rhode Island,
while the VA Heart of Texas Health Care Network (VISN 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.3337 Among active and convalescent cases , approximately
145,000 are Veterans and approximately 15,000 are employees (with an estimated 630 as
Veteran employees ).33).33 While African American Veterans make up approximately 12% of
the VHA,3438 the burden of COVID -19 cases are skewed, with African American Veterans
comprising approximately 20% of all COVID -19 cases.3337 Approximately 7,099 COVID-
19-infected VA patients have died, an estimated 2,738 in VHA hospitals.3337
The objectives of this study will be addressed using data from VHA’s Corporate Data
Warehouse ( CDW),, which is an integrated EMR system with a centralized data warehouse
that is updated on a daily basis. The CDW stores data in separate databases, one for each type
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 identif ication number 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 15 secondary diagno ses), 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 provider, 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 early data on the safety 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.3539 The VA offers eligible Veterans
long-term care services ranging from nursing homes and assisted -living centers to caregiver
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077593
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 59 of 237 support in the Veterans’ own homes.1315 Secondly, and relatedly, VH A data are refreshed
daily and would thus enable early and rapid data analysis. Third, the VHA population is on
average older than the general US population.3640 Of these, about 30% (roughly
1,000,000 individuals) use VHA health se rvices almost exclusively ( i.e., those with a priority
group of 1 or 4 ; Veterans assigned to Priority group 4 are either accepting VA assistance or
housebound benefits, or have been determined to be “catastrophically disabled” by the
VA.2733), 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.37,3841,42 These priority
groups include Veterans with the highest levels of service -connected disability and are
therefore, the highest priority for VHA care.2733 Finally, the VHA population ha s, on average,
more comorbid conditions than the general population, which also indicat es that these
individuals may be at higher risk of COVID -19.3943 While the VHA pop ulation is
predominantly male (approximately 90%), and thus lacks generalizability to females, it will
still provide a useful setting to examine re al-world vaccine safety.
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 servic es received in the inpatient and outpatient 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 numbe r of recipients of Pfizer -BioNTech
COVID-19 vaccine within the VHA database during the study period , which will increase
over time with subsequent analyses. 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. Specifically, 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 applicable safety events of interest ) may have only partially
elapsed in some cases. To account for this, we will use methods adopted in previous
studies,810,2544,045 whereby risk intervals 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 cycle analysis (RC A) approaches proposed for safety event
of interest signal detection will be conducted according to the methods of Kulldorff et
al.41,4246,47 Table 2 illustrates the estimated power for the RCA approach using the Poisson -
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077594
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 60 of 237 based maximized sequential probability ratio test (MaxSPRT), and provides an overview of
the power required to detect varying 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 (Table 2 and Table 3). Power of ≥ 80% is typically desirable in drug safety research.
Usually the FDA views a RR of > 3 as meaningful, so this has been used for power
calculations here. 4348 As an example, as shown in Table 2, the surveillance system would
have sufficient power (80.0%) to detect an increased risk of safety e vents of interest
associated with the Pfizer -BioNTech COVID -19 vaccine by 3 fold when the expected
number of safety events of interest reaches 6 events.
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077595
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 61 of 237 Table 2. Estimated Statistical Power for the Poisson -based MaxSPRT1MaxSPRT46
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077596
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 62 of 237 9.6. Data Management
Data for this study wil l 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 ( see Section 9.7.3.3). The completed
original CRFs should not be made available in any form to third parties, except for
authorized rep resentatives 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 after
reviewing data in the completed CRFs. Analysis Group shall ensure that the CRFs are
securely stored on VHA servers in an encrypted electronic and/or paper] form and will be
password protec ted or secured in a locked room to prevent access by 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 any 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 ad judication committee members to
attest that the data contained on the forms are true and 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 physician'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 records, which includes study documents and
deliverables such as the protocol, SAP, aggregated results tables, SAS Institute (SAS)
programming files , and study report. The records shou ld be retained by Analysis Group
according to local regulations or as specified in the vendor contract, whichever is longer.
Analysis Group must ensure that the records continue to be stored securely for so long as
they are retained.
If Analysis Group beco mes unable for any reason to continue to retain study records for the
required period, Pfizer should be prospectively notified. The study records must be
transferred to a designee acceptable to Pfizer.
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077597
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 63 of 237 Study records must be kept for a minimum of 15 years a fter completion or discontinuation of
the study, unless Analysis Group and Pfizer 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 by applicable local re gulations.
Analysis Group must obtain Pfizer's written permission before disposing of any records, even
if retention requirements have been met.
9.7. Data Analysis
Detailed methodology for summary and statistical analyses of data analyzed in this study will
be documented in a statistical analysis plan ( SAP),, which will be dated, filed , and
maintained by the sponsor. The SAP may modify the plans outlined in the protocol; any
major modifications of primary endpoint definitions or their analyses would be ref lected in a
protocol amendment. T he SAP will also provide additional detail regarding the evaluation of
a threshold of excess risk for each of the safety events of interest. Consistent with the
approach of Kulldorff et al ., this 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.4146 This 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 safety event of in terest 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 analyses will be conducted using SAS Enterprise Guide version 7.1 (SAS Institute Inc.,
Cary, NC) or R Version 3.5.3 or its latest version (R Core Team, Vienna, Austr ia). In
addition, SaTScan will also be used to conduct specific temporal analyses.
9.7.1. Baseline Characteristics
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 standard 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 calculat ed between Pfizer BioNTech COVID -19 vaccine
recipients and active comparators who received seasonal influenza vaccination to evaluate
whether there are any major differences in individuals’ baseline characteristics. Standardized
differences < 10% will indi cate that matching has appropriately balanced the charac teristics
between recipients of the Pfizer -BioNTech COVID -19 vaccine and seasonal influenza
vaccine.
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077598
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 64 of 237 9.7.2. Vaccine Utilization Patterns
Descriptive statistics will also be used to summarize vaccine utilizat ion 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 Pfizer -BioNTech COVID -19
vaccine will be reportedsummarized .
9.7.3. Safety Signal Analyses
Several analyses corresponding to the designs discussed previously will be co nducted to
detect safety signals associated with Pfizer -BioNTech COVID -19 vaccine. Analyses 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 +C will 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 only one dose vs. two doses.
A stepwise process, illustrated below, will be performed for signal detection, evaluation, and
verification ( Figure 3). This approach has been adapted from the Active Monitoring Protocol
of the FDA’s COVID -19 Vaccine Safety Surveillance Project.1416 The 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.
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077599
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 65 of 237 Figure 3. Steps in Signal Detection, Evaluation , and Verification
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077600
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 66 of 237
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 safety event of interest occurs in the clean window, a repeat
occurrence will not be counted in the risk interval). However, event wor sening 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
occurrence for the same type of safety event of interest occurs in the risk interval, the inpatient occ urrence 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
because historical controls would not meet the criteria of having a COVID -19 diagnosis.
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077601
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 67 of 237 9.7.3.1.1. Sequential Testing - SCRI Design using the Binomial -based MaxSPRT for
Comparison to PrePost-vaccination Control Intervals For the Two Safety Events
Requiring COVID -19 Diagnosis
The goal is to provide rapid -cycle, near real -time safety surveillance. In the signal detection
phase, the SCRI analysis will only include pre vaccination control intervals as the with post-
vaccination control intervals will require a longer time to accumulate and thus will not
allowbe used for timely analysis. 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 during the signal evaluation phase (see Section 9.7.3.2), to allow for additional
observation time to accrue as well as to more deeply investigate potential signals. This will
allow for timely RCA without the need to wait for data to accumulate for safety events of
interest with post vaccination control intervals 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 hypothesis (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 demyelinating disease), meaning
a RR of 1 is specified under H 0.2230 The one-sided composite alternative hypothesis (H a)
assumes that the risk of a safety event of interest during the risk interval is greater than the
risk of the same safety event of interest developing during the control interval, accounting for
differences in interval duration ( i.e., RR > 1, Ha is applicable across a range of RRs).146
Specifically, for the Pfizer -BioNTech COVID -19 vaccine, let x represent the total count of
safety events of interest in the control in terval (Figure 4), let y represent the total count of
safety events of interest in the risk interval, and let r represent the ratio of y to 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 𝑦𝑟
𝑥.2544 The RR and corresponding 99%
confidence intervals (CIs) will be calculated.
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077602
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 68 of 237 Figure 4. Example of SCRI Design for a Safety Event of Interest with a 42-day Risk
Interval and a PrePost-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
H0.146 This ratio is calculated whenever new data are received to account for the conti nuous
data stream until the full 42 -day risk period is complete.
𝐿𝐿𝑅 =ln𝑃(𝑦 | 𝐻𝑎)
𝑃(𝑦 | 𝐻0)
Once the LLR test statistic reaches a pre -specified critical value , a signal is detected.
Specifically, 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 reach 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.2544
For each safety event of interest (and specific to each age group, if age -stratified analyses 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.2544 Upper
limits will be determined based on the expected number of safety events of interest under the
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077603
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 69 of 237 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. Sequential Testing - Poisson -based MaxSPRT for Comparison to Active
Comparators who Received Seasonal Influenza Vaccination
For compariso n 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 probability distribution. In the Poisson MaxSPRT
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 of safety events of
interest in the risk interval after seasonal influenza vaccination in five prior seasons , ranging
from 2014/15 through 2018/19. This approach is particularly important for extremely rare
safety events of interest (i.e., less than 50 anticipated based on historical influenza vaccine
rates of safety events of interest ).2230 Poisson MaxSPRT is used to monitor very rare safety
events of interest as binomial MaxSPRT may not detect a signal, despite a clinically
meaningful RR.2544 This will also allow for more timely analysis using historical da ta, as well
as improved power and sample size.
GBS is of particular interest relative to the safety profile of Pfizer -BioNTech
COVID-19 vaccine. 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 safety 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 3 based on calculations
conducted by Kulldorff et al 2011.4146 For example, assuming T = 6 (number of expected
events under the null) and RR = 3, which corresponds to a power of 80.0% (See
Section 9.5.1), 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 separately to
determine a critical value based on background incidence, alpha, power, and clinically
meaningful RR. These details will be addressed in the SAP.
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077604
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 70 of 237 Table 3. Critical Values for Poisson -based MaxSPRT
Multiple types of alpha spending fun ctions can be employed to calculate the cumulative rate
at which Type 1 error (alpha) probability is spent during sequential testing.4449 To achieve
optimal expected time -to-signal, especially when historical Poisson data are used with
surveillance data, a power-type convex alpha spending shape will be used based on published
literature.49 Additionally, ρ = 1.5 is referenced as a “rule of thumb” as it i s suggested to be
appropriate in most applications.
9.7.3.2. Signal Evaluation
Signals are detected when the event frequency of a safety event of interest during the risk
interval following vaccination with Pfizer -BioNTech COVID -19 vaccine is significantly
increased compared to the event frequency of the same safety events of interest in the control
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077605
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 71 of 237 comparator (i.e., the critical value is achieved and surpassed). If signals are indeed detected
for safety events of interest based on the analysis described above, furt her evaluation is
warranted to refine and confirm such detections. This will includeconsist of the following
additional analyses to assess the robustness of the findings 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 (see Section 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 by 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 via active comparison,
additional analyses comparing to prepost-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, C 4591012) 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 vaccinated and active comparator populations), a
multivariate Poisson regression analysis 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.).8 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 safety surveillance must allow for sufficient type I error probability for rapid
detection of safety events of interest , and statistically significant signals must be studied
further to ensure that a true association is present.4550 Therefore, the presence of temporal
clusters will be assessed using the software SaTScan to calculate temporal scan statistic in
order to further refine safety signals detected from the signal detection analyses.2230 A
temporal scan statistic accounts for multiple testing present during overlapping risk intervals.
The null hypothesis assumes that there is no association between the safety events of interest
and immunization, and safety events of interest are assumed to be di stributed independently
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077606
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 72 of 237 and uniformly during a period of time subsequent to Pfizer -BioNTech COVID -19
vaccination.2230 A temporal scan statistic will be generated by 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 hypothesis.646
9.7.3.2.4. Sequential Testing - SCRI Design using the Binomial MaxSPRT for
Comparison with Post -Vaccination Control Intervals
Similar to the SCRI design Any safety events of interest with signals detected and not
already analyzed during the signal detection phase with the SCRI design using the binomial -
based MaxSPRT will be analyzed during the signal evaluation phase using SCRI design
using the binomial -based MaxSPRT method for prevaccination control intervals, sequential
testing analyses will be conducted using the post-vaccination control intervals as appropriate
for specific safety events of interest . 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 for the pre 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 safety events occurring in the risk interval
following vaccination with the incidence of safety events occurring during all other times
post-vaccination in the same individual until the earliest of 183 days af ter 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 desig n in that instead of having
fixed post-vaccination control intervals will be applied. of the same duration as the risk
interval, it has a time-varying post-vaccination control time period that include s 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 5 below for
an example of an individual who receives two doses of Pfizer -BioNTech COVID -19 vaccine ,
where the safety event of interest has a 42 -day risk interval window (e.g., Bell’s palsy ; 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 SCCS design with the second dose received 60 days
after the first (i.e., with gaps between the end of dose 1 risk interval and dose 2 ). The post-
vaccination control time period is displayed below as shading with gray lines .
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077607
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 73 of 237 Figure 5. Example of SC CS Design for Safety 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
Compared to the SCRI design, the SCCS design with post -vaccination control time period
will have increased statistical power, which is especially useful for the study of rare safety
events of interest. A conditional Poisson regression model will be used to compare the rates
of safety even ts of interest in the risk interval vs post -vaccination control time period. From
this model we will report rate ratios and 95% CIs that will be interpretated as the rate ratio
for the safety event of interest in the risk interval compared to the control interval.
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077608
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 74 of 237 9.7.3.2.6. 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 CO VID-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 wh o 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 by > 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 (IPTW) will be used to ensure comparability
between the Pfizer -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.51 IPTW is defin ed 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 weight will be
applied for both doses .23 Initial 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 vaccine and 1 - Pr (Pfizer-BioNTech COVID -19 = 1) / (1 -PS) for the contemporary
unvaccinated c ontrols. 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 t he risk of safety events of interest between cohorts.
Hazard ratios and corresponding 95% CIs will be summarized .
9.7.3.3. Signal Verification
If a signal persists after conducting signal evaluation , signal verification through medical
records review may be conducted .
9.7.3.3.1. Medical Records Review
As part of the signal evaluation process, diagnostic validation of the detected safety events of
interest (i.e., cases) via a djudication of patient medical records by VHA clinicians for
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077609
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 75 of 237 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 inter est detected, such
that all cases may be reviewed for safety events of interest where a small number of events
result 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 .4752 For rare events,
potentially all cases may be adjudicated. An adjudication charter will be developed to govern
signal evaluation and medical records review. Specifically, validation of detected safety
events of interest will be performed through patient medical chart review in collaboration
with an adjudication committee comprised of the treating or trained healthcare
professionals.7 52
9.7.4. Seasonality -Adjusted Cases -Centered Method
A case-centered analysis for specific safety events of interest for which signals were detected
may also be conducted in order to a ccount for bias caused by seasonality of safety events of
interest and vaccination.23 3053
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 safety event of interest cases that were vaccinated inside versus outside a pre -
specified risk interval , as of 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.2544 Specifically, the association of vaccination with risk of safety
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 on the date of the safety event of
interest occurrence. In this way, risk sets are anchored to calendar dates, and confounding by
seasonality of the safety events of interest and vaccination is addressed .853 Note that other
confounders may also be adjusted for by restri cting 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 analyses (over the course
of the 30-month period) and an end -of-surveillance analysis ( i.e., at 30 months , after the end
of surveillance) will be conducted. Similar methodology will be applied for the end -of-
surveillance analysis and end -of-season analysis conducted for seasonal influe nza vaccine in
order to adjust for the seasonality of both disease and vaccine administration.810 This
approach will be able to define the true risk intervals after each dose and estimate the risk for
potential safety events of i nterest after both dose 1 and 2 of the Pfizer -BioNTech COVID -19
vaccine, as well as the ability 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 di stinct 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 . In order to monitor
the safety after the first and full course of the vaccine, the number of potential safety events
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077610
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 76 of 237 of interest occurring in three separate risk intervals (P 1, P2, P3) will be estimated ( Figure
5).Figure 6). P1 represents the risk interval after the first dose only, excluding any overlap in
risk intervals with the second dose. P 2 represents the overlapping risk intervals for first and
second dose of the vaccine. P 3 represents the risk interval of the second do se of the vaccine,
excluding the overlapping risk interval already 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% CIs will be calculated around the
RR in order to ascertain whether the Pf izer-BioNTech COVID -19 vaccine is associated with
safety events of interest.
Figure 6. Example of Risk (P1, P2, P3) and PreAggregate Post-vaccination Control
Intervals for the SCRI End -of-surveillance Analyses of 1 or 2 Doses of
Pfizer-BioNTech COVID -19 Vaccine
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077611
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 77 of 237
In Figure 6A, P1 + P2 + P3 represent the risk intervals where a safety event of interest may
occur. In Figure 6B, there is no overlapping risk interval so that P 1 + P3 represent the risk
intervals where a safety event of interest may occur. The timing of the risk and control
intervals may be adjusted for in order to control for the effect of seasonality across the
intervals assessed.
9.7.6. Subgroup Analysis
Separate analyses of baseline characteristics, vaccine utilization patterns, signal detection,
signal evaluation, and signal verification in subgroups of interest may be conducted based on
feasibility, sample size, and data available.
9.7.7. 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 analyses. Kaplan -Meier methods will be used to analyze time -to-event (i.e., time to
safety event of interest ). If individuals do not experience the safety events of interest , they
will be censored at the end of the risk interval. Median time to safety event of interest and
corresponding CI s will be reportedsummarized .
9.7.8. Prioritized Safety Analysis of Myocarditis /Pericardi tis
Notably, CDC recently investigated the occurrence of myocarditis/pericarditis following
mRNA COVID -19 vaccinations.17 Therefore, separate safety analyses will be prioritized and
performed to assess the risk of myocar ditis/pericarditis following Pfizer -BioNTech COVID -
19 vaccination, to provide additional context to the CDC investigation and address regulatory
requests for further information on this safety event . Therefore, separate analyses will be
prioritized and conducted to better understand the risk of myocarditis/pericarditis following
Pfizer-BioNTech COVID -19 vaccination in the VHA. This analytical approach is intended to
align with the methodology used by the Vaccine Safety Datalink (VSD) and preliminary
findings of myocarditis/pericarditis published by ACIP on June 23, 2021.17,18 The VSD
protocol defines myocarditis/pericarditis (ICD-10-CM codes B33.22, B33.23, I30, I40)
events as the first ev ent in 60 days 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 days prior to or on the day of the event. This analysis will follow the outcome
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077612
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 78 of 237 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 analysis described in this protocol , but will facilitate comparison with the results
presented by ACIP.16,17
This analysis will include all individuals in the primary analysis who were vaccinated with
the Pfizer -BioNTech COVID -19 vaccine. The n umber 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.
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 interv al and vaccinees who are concurrently, 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 myocarditis/pericarditis
events between those individ uals who were in a risk interval versus those individuals who
were in a comparison interval on the same calendar day. Data will be analyzed 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 for adjustment variables (i.e., age group, sex,
race/ethnicity, and VHA service area). Thus , the number of myocarditis/pericarditis events in
a risk or comparison interval on a calendar day will be modeled as a function of whether the
stratum’s vaccinees are 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
contributing 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 be associated with
risk of post -vaccination myocarditis/pericarditis, to account for changes COVID -19 and other
viruses circulating and other ecologic factors, analyses may also be stratified by calendar
time, for example in 6 months increments.
In addition to analyzing codified data, Case confirmation for myocarditis/pericarditis events
identified in the codified data will be conducted based on medical chart review.
Myocarditis/pericarditis cases will be confirmed and validated using the Brighton
Collaborati on’s case definitions .19 Risk factor analysis will also be conducted via logistic
regression among confirmed cases of myocarditis/pericarditis to further evaluate variables
associated with the event ; additional detail s will be provided in the SAP .
Additional data surrounding risk factors, clinical course, and sequelae of identified
myocarditis/pericarditis events up to 365 days 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, comorbid immunocompromising conditions and systemic immune -
mediated diseases, demographics, and medicat ion history); time between Pfizer -BioNTech
COVID-19 dose (first and second) and onset of myocarditis/pericarditis; echocardiogram
information; lab troponin information; symptoms (e.g., chest pain, shortness of breath,
weakness or fatigue, arm or shoulder p ain, heart palpitations cough, swelling in abdomen or
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077613
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 79 of 237 legs, fever); treatments received for myocarditis/pericarditis (e.g., non-steroidal anti -
inflammatory drugs ( NSAIDs), colchicine, corticosteroids, pericardectomy); healthcare
resource utilization follow ing the event, and long -term sequelae for up to one year following
the event (for myocarditis: recovery, sudden cardiac death, heart failure cardiogenic shock,
fulminant myocarditis, inflammatory cardiomyopathy, heart transplant, arrhythmia; for
pericardit is: recovery, chronic pericarditis, restrictive pericarditis, recurrent pericarditis).
9.8. Quality Control
Data for the study will be extracted from electronic databases in the CDW of the VHA. Each
data content area in the CDW is subjected to similar checks, f rom 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 by codes defining the diagnosis
• Percentages, rat es, are as expected (check ranges and for missing)
• Both inpatient and outpatient diagnosis codes are captured. Referenced variables exist
and are of appropriate length and type
Data retrieval will be coordinated by an experienced programmer/analyst. The an alyst will
write programming for retrieval of each data element from the electronic databases. Double
programming will be performed for the first iteration of the analyses; results/datasets will be
compared, and if any discrepancies are identified, both pr ogrammers will determine a
resolution, bringing in a third programmer if needed. Subsequent iterations of analyses (i.e.,
re-runs of the analyses) will be audited 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 only 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 inherently adjusts for within-individual confounders, such as
age, sex, and confounding by indicat ion. Additionally While control intervals can be defined
both pre- and post- vaccination , the inclusion of current study will only use a post-vaccination
control period will account because individual s may be more vigilant for increased detection
bias from stimulated the reporting of possible safety events of interest due to heightened
vigilance on COVID 19 vaccines 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.4954 Specifically, safety events of interest may be more likely to be reported or sought
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077614
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 80 of 237 care for after vaccination with Pfizer -BioNTech COVID -19 vaccine than before (i.e., during
the prevaccination control interval) , which may result in bias against the Pfizer -BioNTech
COVID-19 vaccine . Lastly, SCRI allows for near real -time monitoring of safety 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 yield s a more
interpretable result than other planned analyses using SCRI and active comparators who
receive seasonal influenza vaccination (i.e., the increased risk of experiencing a specific
safety event due to Pfiz er-BioNTech COVID -19 vaccination). The potential for selection bias
(i.e., confounding by indication, healthy user bias) will be mitigated by comparing baseline
demographic and clinical characteristics among the unvaccinated controls. Unvaccinated
controls will be required to have similar healthcare -seeking behaviors as Pfizer-BioNTech
COVID-19 vaccinees, including a t least 1 year of enrollment in and no disenrollment from
VHA benefits prior to their match date. This design is also not limited to assumption s
required by 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 vaccination, and the refore 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 CDW also includ es 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 evaluation verification and case validation,
as needed). Importantly, the VHA CDW retains electronic immunizat ion records that include
manufacturer 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 daily basis, enabling near real -time rapid mon itoring of potential safety signals.
However, there are several limitations when relying 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 individu als receive the Pfizer -BioNTech COVID -19 vaccine
outside of a VHA facility, this information will not be captured in the VHA EMR system.
Similarly, individuals may have also received past seasonal influenza vaccinations outside of
the VHA system, and thus would be misclassified as not having received vaccine in the
current analysis. For example, veterans with secondary insurance or veterans who are
65 years of age or older who have Medicare may receive health care services outside of VHA
facilities. One stu dy 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.5055 Another 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.5156 Hence, it is
important to note that data on vaccination status ma y be incomplete. However, this limitation
will be addressed by examining subgroups of individuals who receive care regularly at VHA
facilities, as well as those with Priority group 1 status, to ensure that their healthcare data are
complete to the extent p ossible in the CDW. SecondThe results from these subgroup analyses
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077615
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 81 of 237 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 safety events
of interest in the comparative analyses. However, if there are discrepancies that su ggest data
are not missing at random and could bias results, subgroup analyses 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 patien t level to
ensure a more comprehensive evaluation of the care an individual receives. Linking variables
are available in the data to allow for patient -level linking of the two data sources. Given the
older age of many veterans, it is likely that these indi viduals 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 predominantly male Veterans (approximately 90% male),
findings from this study may not be generalizable to women in the US.
9.10. Other Aspects
Not applicable.
10. PROTECTION OF HUMAN SUBJECTS
10.1. Patient Information
All parties will comply with all applicab le 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 directly identifiable data in any
reports, publications, or other disclosures, except where required 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, any patient names will be removed and will be replaced by 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, Pfi zer will
maintain high standards of confidentiality 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 River Junction VA Medical Center will conduct this
safety surveillance study with sponsorship from Pfizer and assistance from Analysis Group,
Inc. The project will be led by the VA, with Dr. Yinong Young -Xu, Director of CEP, serving
as the Principal Investigator. Data access will be granted through VA Informatics and
Computing Infrastructure (VINCI). VHA data will not be provided to Pfizer or Analysis
Group. Rather, only VA employees, including those with research servic e without
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077616
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 82 of 237 compensation (WOC) employee status, who have completed necessary VA training and have
proper clearance will access and analyze data on secure VA servers and behind necessary
firewalls, under the direction and supervision of Dr. Young -Xu. Given th e sensitive nature of
healthcare data, comprehensive security 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 pri vacy laws according to applicable legal
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, pro tocol amendments, and their
relevant documents from the relevant IRBs/IECs. All correspondence with the IRB/IEC must
be retained. Copies of IRB/IEC approvals must be forwarded to Pfizer. The study protocol
will be reviewed by the IRB of the VA Medical Cent er, 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 generally accepted research practices
described in G uidelines for Good Pharmacoepidemiology Practices (GPP) issued by the
International Society for Pharmacoepidemiology,5357 the FDA Guidance for Industry and
FDA Staff: Best Practices for Conducting and Reporting, Pharmacoepidemiol ogic Safety
Studies Using Electronic Healthcare Data5258 and Good Epidemiological Practice (GEP)
guidelines issued by the International Epidemiological Association (IEA).5459
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 conver ted to
structured form during the implementation 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 pos sible to link (i.e.,
identify a potential association between) a particular product and medical event for any
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 physician 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
informatio n (defined per the patient population and study period specified in the protocol).
Explicit attribution is not inferred by a temporal relationship between drug administration
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077617
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 83 of 237 and an AE, but must be based on a definite statement of causality by a healthcar e provider
linking drug administration to the AE.
The requirements for reporting safety 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 wit h 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 Safety
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, w ithin 24 hours of awareness, to Pfizer Safety using the NIS 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.
All the demographic fields on the NIS 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 captur ed in the narrative field of
the form) will be reported on the NIS 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 stateme nt “A 35-year-old female...” or “An elderly
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/year (mmm/yyyy) format rather than identifying the
actual date of occurrence within the month /year 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 by
the trainee) as a record o f completion of the training, which must be kept in a retrievable
format. Copies of all signed training certificates must be provided to Pfizer.
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077618
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 84 of 237 Re-training must be completed on an annual basis using the most current Your Reporting
Responsibilities train ing materials.
12. PLANS FOR DISSEMINATING AND COMMUNICATING STUDY RESULTS
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 Authorisati on
Safety (EU PAS) Register and may be submitted for publication in a peer reviewed medical
journal.
In the event of any prohibition or restriction imposed (e.g., clinical hold) by an applicable
competent authority in any area of the world, or if the inves tigator is aware of any new
information which might influence the evaluation of the benefits and risks of a Pfizer
product, Pfizer should be informed immediately.
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077619
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 85 of 237 13. REFERENCES
1. World Health Organization (WHO). WHO Director -General’s opening remarks at th e
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 Cen ter. Accessed November 10,
2020. https://coronavirus.jhu.edu/. Accessed November 10, 2020.
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, 2 020.
4. U.S. Food &and Drug 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 Au thorization 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. Accessed January 4, 2021.
7. Pfizer Inc. Pfizer and BioNTech Receive Authoriz ation 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. Accessed December 28, 2020.
8. Law B. SO2 -D2.1.2 Priority List of COVID-19 Adverse Even ts 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 sa fety monitoring for COVID -19 vaccines in early phase
vaccination . National Center for Immunization & Respiratory Diseases. 8September
22, 2020. https://www.cdcgov/vaccines/acip/meetings/downloads/slides -2020-
09/COVID -03-Shimabukuro.pdf
10. Yih WK, Lee GM, Lieu TA, et al. Surveillance for adverse events following receipt
of pandemic 2009 H1N1 vaccine in the Post -Licensure Rapid Immunization Safety
Monitoring (PRISM) System, 2009 -2010. Am J Epidemiol. 2012;175(11):1120 -1128.
911. 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/system/files/2020 -
11/covid-19-immunizations -appendix-q-table.pdf. Accessed January 12, 2021.
10. Centers for Medicare & Medicaid Services. COVID 19 Vaccines and Mon oclonal
Antibodies. 2020; https://www.cms.gov/medicare/medicare partbdrugaverage
salesprice/covid 19vaccinesandmonoclonal antibodies. Accessed January 14,
2021.
090177e197e9b182\Final\Final On: 26-Aug-2021 15:45 (GMT)
FDA-CBER-2021-5683-1077620
BNT162b2 (COVID -19 vaccine)
C4591012 NON-INTERVENTIONAL STUDY PROTOCOL
Final Version 12.0, 27 January 31 Aug 2021
PFIZER CONFIDENTIAL
Page 86 of 237 1112. Center for Disease Control (CDC). People with Certain Medical Conditions.
December 29, 2020 ;. Accessed January 4, 2021.
https://www.cdc.gov/coronavirus/201 9-ncov/need -extra-precautions/people -with-
medical-conditions html. Accessed January 4, 2021.
12. U.13. Patel M, Chen J, Kim S, et al. Analysis of MarketScan Data for
Immunosuppressive Conditions and Hospitalizations for Acute Respiratory Illness,
United States. Emerging Infectious Disea ses. 2020;26(8):1720 -1730.
doi:10.3201/eid2608.191493.
14. U.S. Department of Veterans Affairs. Veterans Health Administration. 2020 ;
https://www.va.gov/health/. . Accessed November 10, 2020.
https://www.va.gov/health/
15. 13. U.S. Department of Veterans Affai
…[truncated]