PhUSE Wiki

A Reviewers' Guide is a document included in dataset submissions to the FDA.  While this document is not required (per regulation) it is becoming increasingly popular to include one.  The purpose of this guide is to orient the reviewer to the dataset package and provide meaningful information in a concise way.  This document should not repeat information that is covered elsewhere in the submission, but rather supplement it.

Headings will be bolded.
Instructions will be italicized.
Sample text will be in plain text.

Introduction
State the purpose of this document.

The Reviewer’s Guide was designed to familiarize the reviewer with the assumptions and business rules used in readying the data for submission, variables of particular importance in the context of the study objectives, and to describe accompanying programs.

1.1    Submission Dataset Package
State what you are submitting.  State is clearly.

This submission dataset package consists of SDTM tabulation datasets with electronic listings of the collected data, ADaM analysis datasets supporting the scientific objectives of the study, plus supporting documentation.

The submitted datasets are based on the CDISC standards SDTM X.X.X and ADaM X.X. 
Datasets and documentation were organized in a directory structure compliant with the Study Data Specifications document (SDS) version 1.5.1, published January 4, 2010.

1.1.1    Data Definition File (Define.xml)
Describe the contents of 1.1.  Define.xml and Supplemental Data Definitions document should generally be submitted.

Define.xml is the primary reference for the content and format of the Tabulation and Analysis datasets. 
This file complies with the ODM version X.X standard.
Accompanying the define.xml are style and schema files which enable these data to be viewed with the Microsoft Internet Explorer browser.  Correct rendering has been verified in browser versions IE6, IE7. 
A blank case report form (blankcrf.pdf) book has been prepared to show the mapping of each collected item to corresponding variable(s) in the SDTM datasets.  The Origin column in define XML cross references CRF page numbers with tabulation variables.

1.1.2    Supplemental Data Definitions PDF
An accompanying document details any applicable derivations too verbose to fit in the define.xml

1.2    Analysis Sets
Describe any analysis sets that are the most important for your reviewer.

The following patient sets were used in the analysis.

1.2.1    Full Analysis Set
Subjects who are randomized (and received at least one dose of study drug) are included in the Full Analysis (FA) set.  The FA set was used for XXXXXXXX.  There is no subject–level population flag for the full analysis set, because all subjects are represented in all ADaM datasets.  OR Subject–level population flag FASFL = “Y” identifies subjects and records in the full analysis set.

1.2.2  Per Protocol Set

2.    User Guide to Analysis Datasets
All Analysis Datasets were derived from the submitted SDTM tabulation datasets.

A brief description of key analysis datasets is presented.  For more details on data handling and derivations please see the SAP.  See define.xml for the full record layout. Presented below are examples replicating primary analyses from these datasets. These examples will serve as guides for the use of the specific datasets.

2.1    Subject–level Dataset (ADSL)
Describe what is important in your ADSL dataset.  There are probably 100 variables in here, what should the reviewer concentrate on?

This is the ADaM standard dataset with one record per subject.  Excepting screen failures, all subjects who were randomized are included in ADSL, whether or not they were dosed.
In addition to required variables as specified in the ADaM Implementation Guide, population indicators FASFL (full analysis set), ITTFL (intent–to–treat), and COMPLFL (completers) are provided. 

2.2    {Label} Dataset (ADXX )
For important analysis datasets briefly describe the purpose , structure, and any important considerations.

Purpose:
Why is this dataset important?

This dataset was created for XXXXXXX purpose.  The record layout conforms to the ADaM 2.0 Basic Dataset Structure (BDS).

Structure:
Who is in this dataset?  How many records per patient?

All subjects in the XXXX analysis set are represented. OR  All available data for XXXXX is included.  Therefore subjects who were dosed but did not XXXXX are not represented in ADXX. 

Important Considerations:
Is there anything that I definitely need to know about how this was created or how it should be used?

Dataset ADXX was produced from SDTMs XX, YY, ZZ by {merging}|{summarizing}|{modeling}|{wriggling}.  Relevant subject–level variables from ADSL are included.
Records are uniquely identified by subject (USUBJID), visit (VISITNUM), and functional measure (PARAMCD).  When repeat observations were made at a given visit, a calculated record with the average of the repeats is included (PARAMTYP=DERIVED, DTYPE=MEAN).
ADXX records carried forward into the tables and figures are selectable with analysis flag ANL01FL.