cdash and sdtm: what are they and why do we need both … · and uses the same controlled...
TRANSCRIPT
CDASH and SDTM: What Are They and Why do We Need Both Standards?
Kit Howard, MSSenior Education Expert CDISC
Learning Outcomes
Define SDTM and CDASH
Identify how SDTM and CDASH are the same
Discuss how and why are they different
Explain why it is best to use both
All images are public domain
10/23/2018Footer Text 3
Metadata
Standards are…
What is Metadata?
37What is this?
Number? Could be char?
Integer? Decimal places?
Prime?
Age? In weeks?
Percent?
Lab result?
Scale score?
What Are CDASH and SDTM?
Clinical Data Acquisition Standards Harmonization
What is CDASH?
A set of documents
CRF content standards for basic questions in medical research
Standards that “face” the site
Collaboration among data management, programming,
statisticians, clinical, et al
Metadata supporting data capture design for SDTM datasets
CDASH Has Two Parts
10/23/2018Footer Text 8
Model Implementation Guide
CDASH Model
10/23/2018Footer Text 9
Model
Gives us generic structures
for building standards
Provides some rules and
guidance on how to use
those structures
CDASH Implementation Guide
10/23/2018Footer Text 10
Implementation Guide
Gives us implemented
“domains” based on the
model, e.g., AEs, Con Meds,
Demographics, etc.
Also provides best practices
for CRF design &
production, information
about what conformance
means, and…
ImplementationsObservation
Class DomainCDASHIG Variable
CDASHIG Variable Label Question Text Prompt
Data Type
Events AE AETERM Adverse Event Reported Term
What is the adverse event
term?
Adverse Event
Char
Events AE AESTDAT Adverse Event Start Date
What is the adverse event
start date?
Start Date Char
CDASHIG Core
Case Report Form Completion Instructions
SDTMIG Target
Mapping Instructions Implementation NotesHR Record only one
diagnosis, sign or symptom per line …
AETERM Maps directly to the SDTMIG variable …
Can be represented either as …field to capture
verbatim terms ... or could be pre-printed …
HR Record the start date …using this format (DD-MON-YYYY).
AESTDTC … concatenate all collected CDASH START DATE and TIME
components …
N/A
Controlled Terminology
10/23/2018Footer Text 12
CodeCodelist
Code
Codelist
Extensible
(Yes/No)
Codelist
Name
CDISC
Submission
Value
CDISC
Synonym(s)CDISC Definition
C66768 No Outcome
of Event
OUT Outcome of
Event
A condition or event that is attributed
to the adverse event and is the
result or conclusion of the adverse
event. (NCI)
C48275 C66768 Outcome
of Event
FATAL 5; FATAL;
Grade 5
The termination of life as a result of
an adverse event. (NCI)
C49494 C66768 Outcome
of Event
NOT
RECOVERED/
NOT
RESOLVED
One of the possible results of an
adverse event outcome that
indicates that the event has not
improved or recuperated. (NCI)
C49498 C66768 Outcome
of Event
RECOVERED/
RESOLVED
One of the possible results of an
adverse event outcome that
indicates that the event has
improved or recuperated. (NCI)
… … … … … … …
• Consistent CRFs – questions, answers, assumptions
• Ease of use for investigative sites
• Data quality rules that enable SDTM’s quality assumptions
Data Collection & Management
• Beginning with the end in mind
• Smooth flow of data from collection to submission
• Data integrity and traceability back to the source
Supports
Study Data Tabulation Model
SDTM Also Has Two Parts
10/23/2018Footer Text 15
Model
Implementation Guide
SDTM Model
10/23/2018Footer Text 16
ModelGives us generic structures
for building standards
Provides some rules and
guidance on how to use
those structures
SDTM Implementation Guide
10/23/2018Footer Text 17
Implementation Guide
Gives us implemented
“domains” based on the
model, e.g., AEs, Con Meds,
Demographics, etc.
Provides assumptions,
examples, overall guidance,
conformance rules…
And Uses the Same Controlled Terminology
10/23/2018Footer Text 18
CodeCodelist
Code
Codelist
Extensible
(Yes/No)
Codelist
Name
CDISC
Submission
Value
CDISC
Synonym(s)CDISC Definition
C66768 No Outcome
of Event
OUT Outcome of
Event
A condition or event that is attributed
to the adverse event and is the
result or conclusion of the adverse
event. (NCI)
C48275 C66768 Outcome
of Event
FATAL 5; FATAL;
Grade 5
The termination of life as a result of
an adverse event. (NCI)
C49494 C66768 Outcome
of Event
NOT
RECOVERED/
NOT
RESOLVED
One of the possible results of an
adverse event outcome that
indicates that the event has not
improved or recuperated. (NCI)
C49498 C66768 Outcome
of Event
RECOVERED/
RESOLVED
One of the possible results of an
adverse event outcome that
indicates that the event has
improved or recuperated. (NCI)
… … … … … … …
• Logically groups data into topics
• Rules: naming conventions, data types, controlled terminology
• All clean collected data plus some derived
Defines predictable representation of clean collected data
• Data aggregation for submissions, data warehouses, registries
• Regulatory review, IRB reporting, safety surveillance
• Hypothesis generation, cross-product analyses…
Supports
SDTM
Transformation
CDASH
DatasetsData
Capture
34%
66%
Notidentical
Identical
11%
89%
Notidentical
Identical +mappings
Missing Data
Show me the data, not
lack of data
SDTM assumes that if there is no record then nothing
happened. This only works if it was checked in data
capture, which requires a question and record (e.g., Were
there any AEs?)
Absence of evidence is
not evidence of
absence: must check
that missing data is
missing
CDASHSDTM
RationaleUSUBJID AETERM AESTDTC AEENDTC
2006-34-1022 HEADACHE 2006-10-14 2006-10-20
2006-34-1022 RASH 2006-10-14 2006-10-21
2006-34-1024 TIRED 2006-09-24 2006-09
2006-34-1024 HEADACHE 2006-09-24 2006-09-24
SDTM
CDASH
Human vs Machine Readable Data
Machine-
readable:
Dates/Times: ISO
8601, 1 variable,
YYYY-MM-
DDThh:mm:ss
Duration: P6H
SDTM machine-readable formats for variables
such as dates are good for data reusability but
are not user-friendly for data capture. There is
more chance for error when people record data
in unfamiliar formats
Human-readable:
Dates/Times: 2 or
more variables,
DD- MM-YYYY,
HH:MM:SS
Duration: 6 HOURS
CDASHSDTM
Rationale
USUBJID HOTERM HOSTDTC HODUR
ABC1201 Hospitalization 2011-08-06 P6H
SUBJID HOTERM HOSTDAT HOCDUR HOCDURU
1201 Hospitalization 06-AUG-2011 6 HOURS
SDTM
CDASH
Data Organization
Each CRF should have the data that makes sense to
collect together. Per CDASH, this can mix domains if
standard variable names are used, and in the end the
data appear in the right domains.
In SDTM, data MUST appear only in the correct domain.
Data on each CRF is
driven by what is
captured together, not
necessarily by domain
CDASH
Data must be
organized into
datasets by domain
SDTM
Rationale
USUBJID SUBJID BRTHDTC SEX RACE
ABC1201 01 1948-12-13 M WHITE
ABC1202 02 1955-03-22 M BLACK
DM SDTM
USUBJID RPTESTCD RPORRES RPSTRESC
ABC1201MENOSTA
TPremenarchal Premenarchal
ABC1201 NUMLIV 2 2
RP SDTM
DM CDASH
RP CDASH
Unavailable Variables
In doubleblind studies, treatment and dose are
unknown until the blind is broken;
SDTM primary focus is pooling data for submission;
CDASH primary focus is individual studies
Some variables
cannot be collected
in the way needed
for summarization
and analysis, e.g.,
ECDOSE
CDASH
Some data needed for
summarization &
submission aren’t available
until the study is over, e.g.,
exposure in double blind
studies
SDTM
Rationale
Exposure as Collected EC CRF
USUBJID EXTRT EXDOSE EXDOSU EXDOSFRQEXSTDTC EXENDTC
ABC1201 BestDrug 350 mg BID
2011-08-
06 2011-09-01
SDTM
CDASH
Ensure asking same question using the same answer lists as analyzing
Optimize site data requirements and structure for transmission & analysis
Supports traceability back through data collection
Minimizes programming and validation resources and increases quality when transferring data from capture to tabulation
Can address the points in all the previous slides
Helps to “future-proof” the data for warehousing
Some Benefits of Using Both
Thank You!Bess [email protected]