Section 2.1 of the
Analysis Data Model Implementation Guide provides the
fundamental principles of the CDISC ADaM model.
-
Analysis data sets and associated
metadata must clearly and unambiguously communicate the content and
source of the data sets supporting the statistical analyses performed
in a clinical study.
-
Analysis data sets and associated
metadata must provide traceability to enable an understanding of where
an analysis value came from.
-
Analysis data sets must be readily
usable with commonly available software tools.
-
Analysis data sets must be associated
with metadata to facilitate clear and unambiguous communication. Ideally,
the metadata is machine-readable.
-
Analysis data sets should have
a structure and content that enable statistical analyses to be performed
with minimal programming. Such data sets are described as analysis-ready.
Implementation of the
CDISC ADaM 2.1 reference standard in the SAS Clinical Standards Toolkit
supports each of these principles.
The number and structure
of analysis data sets are highly dependent on the type of study, the
study objectives as defined in the statistical analysis plan, and
discussions with the reviewing authority. ADaM data sets incorporate
derived and collected data that permit analysis with little or no
additional programming. Data can be from various SDTM domains, other
ADaM data sets, or any combination thereof.
As initially released,
the CDISC ADaM 2.1 reference standard supports two types of analysis
data set structures:
-
The subject-level analysis data
set (ADSL) provides descriptive information about subjects, such as
study disposition, demographic, and baseline characteristics. The
ADSL is the primary source for subject-level variables included in
other analysis data sets, such as population flags and treatment variables.
There is only one ADSL per study, and the ADSL and its related metadata
are required in each CDISC-based submission of data from a clinical
trial, even if no other analysis data sets are submitted.
-
The ADaM Basic Data Structure (BDS)
is used for the majority of ADaM data sets, regardless of the therapeutic
area or type of analysis. Each BDS data set contains one or more records
per subject and analysis parameter. The structure of some BDS data
sets might include an analysis time point. A record in a BDS analysis
data set can represent an observed, derived, or imputed value required
for analysis. Each BDS data set contains a core set of variables that
describe the analysis parameter and the value being analyzed. A data
value can be derived from any source file, including any combination
of SDTM and ADaM data sets.
The Analysis Data Model
identifies four types of metadata that are captured and supported
by the SAS Clinical Standards Toolkit.
ADaM Metadata Types and SAS Clinical Standards Toolkit Locations
|
|
SAS Clinical Standards
Toolkit Location
|
Analysis data set metadata
|
reference_tables.sas7bdat
|
Analysis variable metadata
|
reference_columns.sas7bdat
|
Analysis parameter-value-level
metadata
|
reference_columns.sas7bdat
(parameterid column)
|
Analysis results metadata
|
Derivable from reference_columns.sas7bdat
(where table=’RESULTS’)
|
Version 1.0 of the
Analysis Data Model Implementation Guide (ADaMIG) defines
a common set of ADSL and BDS columns that can be used as templates
for any ADaM analysis data set. These columns are defined in the SAS
Clinical Standards Toolkit reference_tables and reference_columns
data sets. These SAS Clinical Standards Toolkit data sets contain
a metadata representation of the analysis results metadata. Empty
ADSL, BDS, and analysis results data sets can be derived from the
SAS Clinical Standards Toolkit global standards library using the
utility macro cst_createTablesForDataStandard.
Sample Reference_Tables Record (CDISC ADaM 2.1)
|
|
|
|
|
|
|
|
|
|
|
Subject-Level Analysis
Dataset
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Subject disposition,
demographic, and baseline characteristics
|
|
|
|
Sample Reference_Columns Record (CDISC ADaM 2.1)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
TRTP is a record-level
identifier that represents the planned treatment attributed to a record
for analysis purposes. TRTP indicates how treatment varies by record
within a subject and enables analysis of crossover and other designs.
TRT xxxxP (copied from ADSL)
may also be needed for some analysis purposes, and may be useful for
traceability and to provide context.
|
The SAS Clinical Standards
Toolkit CDISC ADaM reference standard also provides metadata and code
to validate the structure and content of the ADaM analysis data sets.
To enable validation,
supplemental files supporting ADaM validation processes include these
SAS Clinical Standards Toolkit global standards library files:
-
The Validation Master data set
in the
validation/control folder contains
the superset of checks validating the structure and content of each
analysis data set. These checks are based on version 1.1 of the CDISC
ADaM Validation Checks as prepared by the CDISC ADaM team.
-
The Messages data set in the
messages folder provides error messaging for all
Validation Master checks.
-
SAS code in the
macros folder provides code that is specific to ADaM
that augments code that is provided in the primary SAS Clinical Standards
Toolkit autocall library (
!sasroot/cstframework/sasmacro).
These supplemental files,
in whole or in part, define the SAS Clinical Standards Toolkit CDISC
ADaM reference standard.