CVM GFI #197 Documenting Electronic Data Files and Statistical Analysis Programs

Administrative guidance, not enacted legislation. This captured publisher edition preserves the listed document and attachments; later developments are not independently consolidated.

#197
Documenting Electronic Data Files
and Statistical Analysis Programs
Guidance for Industry
This version of the guidance replaces the version made available December 2015. The
document has been revised to clarify existing language, remove recommendations that are no
longer applicable, and provide additional details on the README file.
Submit comments on this guidance at any time. Submit electronic comments to
https://www.regulations.gov. Submit written comments to the Dockets Management Staff
(HFA-305), Food and Drug Administration, 5630 Fishers Lane, Rm. 1061, Rockville, MD
20852. All comments should be identified with the docket number FDA-2009-D-0052.
For further information regarding this document, contact AskCVM@fda.hhs.gov.
Additional copies of this guidance document may be requested from the Policy and Regulations
Staff (HFV-6), Center for Veterinary Medicine, Food and Drug Administration, 7500 Standish
Place, Rockville MD 20855, and may be viewed on the Internet at either
https://www.fda.gov/animal-veterinary or https://www.regulations.gov.
U.S. Department of Health and Human Services
Food and Drug Administration
Center for Veterinary Medicine (CVM)
November 2020

Contains Nonbinding Recommendations
2
Table of Contents
I. 	Introduction .......................................................................................................................... 3
II. 	Background ........................................................................................................................... 3
III. 	Structure and Content of README Files ......................................................................... 4
A. The README File ......................................................................................................... 4
B. Structure of README Files.......................................................................................... 5
1. 	Electronic Data Files ................................................................................................ 5
2. 	Audit Trail Files ........................................................................................................ 7
3. 	Data Analysis Programs........................................................................................... 9
IV. 	Data and Analysis Programs: Additional Comments ..................................................... 11
Appendix ...................................................................................................................................... 12

Contains Nonbinding Recommendations
3
Documenting Electronic Data Files
and Statistical Analysis Programs
Guidance for Industry
This guidance represents the current thinking of the Food and Drug Administration (FDA or
Agency) on this topic. It does not establish any rights for any person and is not binding on FDA
or the public. You can use an alternative approach if it satisfies the requirements of the
applicable statutes and regulations. To discuss an alternative approach, contact the FDA staff
responsible for this guidance as listed on the title page.
I. 	Introduction
This guidance is provided to inform sponsors of recommendations for documenting electronic
data files and statistical analyses submitted to the Center for Veterinary Medicine (CVM) to
support new animal drug applications. These recommendations are intended to reduce the
number of revisions that may be required for CVM to effectively review data submissions. They
are also intended to simplify submission preparation for sponsors by providing a suggested
documentation framework, including a sample structure on how to describe and organize the
information regarding the electronic data files and statistical analysis programs.
The determination of what data and analysis are needed to support a new animal drug application
may vary depending on many factors and is outside the scope of this document. You should
refer only to those portions of this guidance that are applicable to your particular submission.
In general, FDA’s guidance documents do not establish legally enforceable responsibilities.
Instead, guidances describe the Agency’s current thinking on a topic and should be viewed only
as recommendations, unless specific regulatory or statutory requirements are cited. The use of
the word should in Agency guidances means that something is suggested or recommended, but
not required.
II. 	Background
For new animal drug applications, FDA requires full reports of investigations which have been
conducted to show a drug is safe and effective for use [section 512(b)(1)(A) of the Federal Food,
Drug, and Cosmetic Act (the FD&C Act)]. Additionally, section 512(n)(1)(E) of the FD&C Act
requires that abbreviated applications for the approval of a new animal drug contain information
to show that the new animal drug is bioequivalent to the approved new animal drug.
Submissions to CVM in support of new animal drug applications generally include a Final Study
Report (FSR)1 for one or more studies. For each study that includes electronic data files, CVM
1 For the purpose of this guidance, supporting study reports (e.g., statistical or bioanalytical phase reports) appended
to the FSR are considered part of the FSR.

Contains Nonbinding Recommendations
4
needs information regarding the process for data generation and the statistical analysis conducted
to review submissions and verify that there is sufficient quality and detail of evidence to support
an animal drug application. An adequately documented submission should include readable
electronic data files, a description of how the data were processed, and a description of the
statistical analyses employed to support your conclusions.
Your submission should clearly describe the entire process by which the data were collected
(e.g., transcribed or electronic data capture), including a record of all changes to the data, starting
from the electronic data files created from transcribed case report forms, or from electronic data
capture (EDC) systems, to the completed statistical analyses which form the basis for your
study’s conclusions. To understand the process by which you compiled the data and conducted
the statistical analysis, CVM needs to understand the contents of each electronic data file, the
computer programs that processed all the electronic data files for analysis, and the programs that
implemented the statistical analyses. The information that should be submitted to CVM,
together with the FSR and electronic data files, are described in sections III. Structure and
Content of README Files and IV. Data and Analysis Programs: Additional Comments. Section
III describes recommendations for how README files should be organized and completed to
describe the data files and analyses, and section IV contains additional recommendations
regarding statistical analysis programs.
III. 	Structure and Content of README Files
Data submissions to CVM via eSubmitter2 typically include data in Statistical Analysis System
(SAS) transport XPORT (XPT) or eXtensible Markup Language (XML) format, analysis
programming files in XML format, and documents in Portable Document Format (PDF). A
manual or instructional resource that contains important information about the electronic data
files in the submission should be submitted to CVM in a README file. This section describes
the information that should be in the README file for CVM to evaluate the electronic data files.
Electronic data files, in the context of this guidance, refer to the XPT and XML files included in
a submission. If any of the information described below is contained within the FSR it does not
need to be repeated in the README file. Instead, the README file may reference the
corresponding location of the information in the FSR.
A. The README File
An overview of electronic data files, documentation, and programming files included in the
submission helps CVM assess the submission and the files for review. The README file
contains information about the electronic data files and programming files in a submission.
The README file explains how the data files are organized and describes the programs
used for data file generation and data analysis. An effective README file should quickly
orient the user to crucial information needed to understand the electronic data files in a
submission.
2 For current eSubmitter file specifications, see CVM Recommended File Specifications in the CVM eSubmitter
Resource Center: https://www.fda.gov/industry/fda-esubmitter/cvm-esubmitter-resource-center.

Contains Nonbinding Recommendations
5
The README file is typically a PDF file with the filename README.pdf. The README
file should not contain data, interpretation of the data, literature, references, notes to file,
protocol-associated documents, communication records, personnel records, information not
needed to interpret the data provided in the submission, or other information needed for the
technical section.
A submission may contain one or more README files depending on the organization and
complexity of the submission. README file(s) should be separate from the FSR.
B. Structure of README Files
The README file should include a brief introduction that includes the study number or
other identifier, and study descriptor along with general orientation, background, and other
information relevant to analyzing and interpreting the data for each study. A description of
how the data were processed for analysis could also be included, such as how data were
captured and merged to derive the analysis data files or audit trail processes.
The following describes the README file and a sample structure.
1. 	Electronic Data Files
a. 	List of Electronic Data Files
In this section, you should provide a table listing the electronic data files submitted
in XML or XPT file format with brief descriptions. This table should include the
file name, a brief description of contents, name of data collection form(s), if
applicable, and information on how the data were collected or any reference to
location of information in the FSR that is needed to interpret the data (e.g.,
reference ranges or scoring definitions). See Example Table 1a. Listing of XML and
XPT Data Files.

Contains Nonbinding Recommendations
6
Example Table 1a. Listing of XML and XPT Data Files.
File Name1 	File contents 	Name of Data
Collection Form
(e.g., Case Report
Form)
Comments2
ClinObs.xml 	Daily clinical
observations during
hospitalization
Observation Form 	Clin Obs were collected
via [name of data
capture system]
OwnerObs.xpt 	Owner observation
results for individual
animals
Owner Diary 	Recorded manually and
transcribed
PlateletMan.xpt 	Secondary variable:
Manual platelet count
Clinical Pathology
Form
Reference ranges
available in Appendix
G of Final Study Report
1 File name extension included in order to identify the file as XML or XPT.
2 Information included on how data were collected (paper data collection forms or EDC
system) or any reference(s) to location of information in the FSR (e.g., reference ranges or
scoring definitions) that are needed to interpret the data file.
b. 	Electronic Data File Contents
For each data file submitted, you should provide a table that includes the variable
names, the abbreviations used in the file, variable label or description, and
additional details (e.g., description of coded values, unit of measure, formatting
information), if applicable. See Example Table 1b. Contents of
DataFileName.XML. Results from the CONTENTS procedure in SAS are not
sufficient. You may choose to use a standardized internal file format [e.g., Clinical
Data Interchange Standards Consortium-Study Data Tabulation Model (SDTM) and
Standard Exchange for Nonclinical Data (SEND)]3 to be submitted along with audit
trail information if data were collected using EDC (described below).
3 See Study Data Standards Resources: https://www.fda.gov/industry/fda-resources-data-standards/study-data-
standards-resources.

Contains Nonbinding Recommendations
7
Example Table 1b. Contents of DataFileName.XML (number of observations=,
number of variables=).
Variable
Name
Variable Label or
Description
Additional Details
(e.g., descriptions of coded values,
units of measure, formatting
information)
Animal 	Animal Identification
Dot 	Day of treatment 	Study day formatted as mm/dd/yyyy
Trt 	Treatment 	T01=Placebo; T02=Drug
Dscore 	Daily Depression Score 	0=normal; 1=slight; etc.
Overallscore 	Overall Depression Score Sum of Daily Depression scores
(Day 1 to 5)
2. 	Audit Trail Files
This section only applies to studies that include audit trail files as part of the controls
that ensure the authenticity and integrity of records in EDC systems.4 The audit trail is
a portion of the raw data5 that includes the original recorded data value and any changes
to the data value, the identification of individuals that entered or changed data in the
EDC system, the date and time the data value was entered or changed, the reason for
each change, and the date when the database was locked.6 If you are submitting audit
trails on specified variables, the electronic audit trails should be submitted in an
eSubmitter acceptable file format.
CVM prefers to receive one copy of the EDC study database that includes the final
variable observations (e.g., the electronic data files (see footnote 2) described in
section III.B.1 Electronic Data Files above) and associated audit trail information,
described in this section, together in one electronic data file or set of data files. If you
submit these combined data files, you should also submit the programs needed to create
data files that contain the final observations suitable for review and statistical analysis.
If you are unable to submit one copy of the EDC study database with associated audit
trail information, submission of separate data files for analysis and electronic audit trail
files is acceptable.
4 Electronic Data Capture (EDC) or Electronically Captured Data (ECD) refers to data that was electronically
captured at first observation (e.g., web-based software or analytical instrument).
5 As defined in 21 CFR 58.3(k) and in FDA Guidance for Industry #85 (VICH GL9), “Good Clinical Practice”
(https://www.fda.gov/regulatory-information/search-fda-guidance-documents/cvm-gfi-85-vich-gl9-good-clinical-
practice).
6 See Guidance for Industry, “Computerized Systems Used in Clinical Investigations,” dated May 2007.
(https://www.fda.gov/regulatory-information/search-fda-guidance-documents/computerized-systems-used-clinical-
investigations)

Contains Nonbinding Recommendations
8
a. 	Audit Trail File Listing
If you are submitting the electronic audit trail separately from the data files for
review and statistical analysis, you should provide a table listing the audit trail files
submitted. This table should include the file name, the description of the file
including EDC system name, and any information necessary for review. See
Example Table 2a. Listing of Audit Trail Files.
Example Table 2a. Listing of Audit Trail Files.
File Name 	Description 	Comments
Site_A_Audit.xml 	Audit trail of Site A 	[name of EDC system]
Site_B_Audit.xml 	Audit trail of Site B 	[name of EDC system]
b. 	Audit Trail File Contents
For each audit trail file submitted, you should provide a table that includes the
variable names, variable label or description (e.g., description of coded values, unit
of measure, formatting information), and other information necessary for review.
See Example Table 2b. Contents of AuditTrailFile.XML. Each audit trail file
submitted should include every original and updated value, operator identification
and date and time stamps for each data entry and any change, and reason for each
change. Additional columns may be added as needed.
Example Table 2b. Contents of AuditTrailFile.XML (number observations=,
number of variables=).
Variable Name 	Description
animalid 	Animal identification number
formname 	Name of the data collection form
entryfield 	Name of entry field from data collection form
entrytype 	Type of entry field (e.g., numeric, text, radio button, or
checkbox)
entry 	Data entry [include explanation if applicable (e.g., “on” =
radio button was selected, “off” = radio button not selected; if
blank, no data entered)]
operatorid 	Individual entering data
entrydate 	Date/time stamp when the data were entered
reason 	Reason for change, if applicable
lockdate 	Date and time when the database was locked

Contains Nonbinding Recommendations
9
3. 	Data Analysis Programs
a. 	Program File Listing
In this section, you should provide a table listing the programs used to perform
randomization, process the data, generate summaries, and perform the statistical
analysis. This table should include the file name, the purpose of the program, the
electronic data files accessed and generated by each program, a list of any results
(tables/graphs) generated, and software used, if applicable. See Example Table 3.
Listing of Program Files.
Example Table 3. Listing of Program Files.
Program File
Name
Purpose 	Data File(s)
Used in
Program
(Input)
Data File(s)
Created
(Output)
Results or
Tables/
Graphs
Created
Comments
setup.xml 	Sets up
needed
libraries and
data formats.
Creates all
temporary
data files
used in the
analysis
Effect_data.xpt
AE_all.xpt
Clinobs_all.xpt
Effect.sas7bdat
AE.sas7bdat
ClinObs.sas7bdat
None 	Run this
program
first before
running any
analysis
programs
s_clinobs.xml 	Summarizes
clinical
observations
ClinObs.sas7bdat None 	Summary
of clinical
observa-
tions
Profile
plots
Table X
provided in
Final Study
Report
effect_prog.xml Summarizes
effectiveness
variables and
determine
success/
failure
Effect.sas7bdat 	Success.sas7bdat 	None 	Details of
success/
failure
evaluation.
Table Y
provided in
Final Study
Report.
prim_anlys.xml Conducts
primary
analysis
Success.sas7bdat None 	Primary
analysis
results
b. 	Overview of Data and Analysis Flow
You should provide a general overview of the data and analysis process used in the

Contains Nonbinding Recommendations
10
study and submission. For example, you should describe whether data files were
directly downloaded from data management systems and whether any processing
(merging, validation, editing) were performed prior to analysis.
For electronic databases that do not store or directly export data in XPT or XML
formats, conversion from another file format (e.g., XLSX, CSV, or SAS7bdat) will
be needed in order to submit to CVM via eSubmitter (see footnote 2). If conversion
is necessary, CVM expects that electronic data files will be converted into XPT or
XML format without modification (e.g., no column changes or calculations). The
Appendix has examples of SAS programs that convert data files to acceptable
formats. The program used to make the conversion should be provided to CVM in
XML format. If a system limitation precludes the creation of XPT or XML
formatted files, you should contact CVM. Additionally, CVM encourages you to
evaluate/analyze the data using the same file that is submitted to CVM. For
example, if the study data are submitted to CVM in file “ALLDATA.XML,” the
programs submitted should use this file for analysis.
Submitted data files should be unmodified after export from the database. If a data
file was modified for analysis (e.g., addition or edit of variable names, variables or
animals excluded, or coded information reformatted as text), a description of the
changes and the computer program used to create the changes should be included,
but the modified data files should not be submitted. The logic or coding that was
used in creating the modified data files should be documented in the computer
program and clearly detailed in the submission so that CVM may recreate and verify
the modifications.
c. 	Instructions for Running Programs
You should describe the sequence of program calls needed for CVM to run your
programs. Starting with the first program to be run, you should describe calls to
other programs and custom functions.
So that CVM can understand and verify your analysis process, for each program you
should document all directories and files referenced to access or store data,
including directory and file names, locations, and aliases, if used. Describe
programs defining custom styles or formats or, if such styles or formats are
predefined, you should provide instructions for their installation. If the programs
were designed to call other programs or access data in specific folders or directory
structures, you should describe this structure so that CVM can verify your analysis
process.
d. 	Randomization Programs
Electronic data files of all randomization schedules used (e.g., animal to treatment,
procedure order) should be included in the submission and listed in the README
file. These files should contain all variables specified in restrictions to

Contains Nonbinding Recommendations
11
randomization, for example, any stratification variables (e.g., weight, sex) or if
treatments were balanced by one or more factors (e.g., room, cohort). If available,
programs used to generate random assignments, including the randomization
seed(s), should be included in the submission and listed in the README file.
Detailed descriptions of all randomization processes should be included in the FSR.
IV. 	Data and Analysis Programs: Additional Comments
Submission of a separate document (e.g., Statistical Report) as an appendix in the FSR that
provides details on the statistical analysis as well as additional analysis results and summaries is
acceptable. As is the case with all README file content described in this document, if the FSR
contains the information regarding program files and program execution as described in
section III.B.3. Data Analysis Programs above, then the information does not have to be
repeated in the README file. Instead, the README file can refer to the corresponding
section(s) of the FSR.
CVM’s verification of the analysis process will be facilitated by submission of well-documented
analysis programs. You should describe the general purpose of each program in the beginning of
the statistical program. Within the program, you should include sufficient comments to explain
complex sections, for example, if a program will call certain macros or subroutines.
You should not submit electronic files generated by the statistical analysis, such as tables of
descriptive summaries or least squares means, in data format (e.g., XML or XPT). Documents
containing these summaries may be included or appended to the FSR as appropriate. For
example, a listing of subject success/failure outcomes should be included in the FSR, but the
secondary electronic data file derived from the initially collected data should not be submitted.
Descriptions of derived variables and corresponding formulas or complex transformations should
be described in the FSR or contributing scientist report and documented in the analysis program.
If values were computed, derived, or transformed from other variables, the equation(s) for each
variable, descriptions of calculations, and a table of the calculated values should be in the FSR.
recFormulas for standard conversions (e.g., lb to kg; mg to kg) do not need to be provided.
Additionally, you should not submit outputs from running the analysis, for example, the outputs
and log files produced when running SAS.

Contains Nonbinding Recommendations
12
Appendix
Examples of SAS codes that convert data files to acceptable formats
1. 	SAS code to generate XML files
libname in 'file location';
libname out xml 'file location\filename1.xml';
data out.dataset1; set in.filename; run;
2. 	SAS code to generate XPT files
libname in 'file location';
libname out xport 'file location\filename1.xpt';
data out.filename1; set in.filename; run;