CDISC SDTM Beginner's Workshop
24/03/201429/08/2012
SDTM fundamentals
Biometristi Italiani Associati – Seminario BIAS –
CDISC SDTM and ADaM: Moving from theory to
practice SAS Institute – Milano 14/03/2014
Nicola Tambascia
Statistical Programming
Accovion GmbH
What is SDTM?
àStudy
àData
àTabulation
àModel
SDTM fundamentals
SDTM defines a standard structure for study data
tabulations that are to be submitted as part of a
product application to regular authority such as
FDA. It based on material prepared by the
Submissions Data Standard (SDS) Team of the
CDISC.
SDTM IG
Version
SDTM Version Accepted by FDA
3.1.1
1.1
3.1.2
1.2
3.1.3
1.3
3.2
1.4
2
1
CDISC SDTM Beginner's Workshop
24/03/201429/08/2012
Application/Interfaces of CDISC Standards
SDTM fundamentals
3
SDTM - Basic Concepts
Domains
Domain Classes
Domain Variables
Controlled Terminology (CT)
Metadata (define.xml)
SDTM fundamentals
4
2
CDISC SDTM Beginner's Workshop
24/03/201429/08/2012
Domains
Domain = collection of observations with a common topic
• 1 Domain = 1 submission dataset (e.g., LB.XPT)
• Vertical structure
Laboratory
Test
Standard
Unit
Result in
Standard
Unit
Hemoglobin
mg/dl
10
Hematocrit
%
35
Hemoglobin
(mg/dL)
Hematocrit
(%)
10
35
NOT:
• 2 character domain name (e.g., LB)
à
2 character variable name prefix (e.g. LBTEST)
• CDISC standard domains vs. custom domains
• Domain is assigned to 1 General Observation Class
à
Common variables per class (e.g., --TEST for Findings)
SDTM fundamentals
5
Findings Variables Extract from SDTM 1.3
SDTM fundamentals
6
3
CDISC SDTM Beginner's Workshop
24/03/201429/08/2012
Domain Classes
Interventions
Events
Findings, Findings About
Special Purpose/Other
• Demographics
• Comments
• Subject Elements
• Subject Visits
Trial Design
Relationships
• Related Records to express relationships between domains
• Supplemental Qualifiers
7
SDTM fundamentals
Classes in SDTM V1.3
§ 3 general observation classes and 1 special purpose class
provide the general structure for study data.
Classes
EVENTS
§ Planned in protocol
milestones or
independent
occurrences taking
place during the
course of a trial
§ Prior to trial, e.g.
medical history
§ Pre-planned data, e.g.
randomization or
study completion
§ Unplanned incidents,
e.g. adverse events
SDTM fundamentals
INTERVENTIONS
FINDINGS
SPECIAL PURPOSE
§ Defined in protocol
and applied during
trial or preceding
coincident to the
study assessment
§ Observations
resulting from
planned evaluations
that are captured
during the trial
§ Additional subject
level data that do
not conform to one
of the 3 general
observation classes
§ Investigational/
therapeutic
treatments
§ Observations from
planned/unplanned
evaluations
§ Subject elements and
visits
§ Study/concomitant
medications
§ Pharmacokinetics
§ Substances that are
self administered by
the subject, e.g.
tobacco or alcohol
§ Comments
§ Demographics
§ Laboratory tests/ECG
tests
§ Questionnaires
For more information, please refer to the SDTM IG 3.1.3; sections 2.3, 2.4, 4, 5, 8.6.1
8
4
CDISC SDTM Beginner's Workshop
24/03/201429/08/2012
Domains in SDTM IG V3.1.3
Classes
Domains
Interventions
Exposure (EX)
Concomittant
Medication (CM)
Substance Use
(SU)
Events
Adverse Events
(AE)
Disposition (DS)
Medical History
(MH)
Protocol
Deviations (DV)
Clinical Events
(CE)
Inclusion/
Exclusion (IE)
Subject
Characteristics
(SC)
Microbiology
Specimen (MB)
Questionnaires
(QS)
Microbiology
Susceptibility
Test (MS)
Laboratory Data
Physical
(LB)
Examination (PE)
Findings
Vital Signs (VS)
PK
Concentrations
(PC)
PK Parameters
(PP)
ECG (EG)
Drug
Accountability
(DA)
Findings about Events
and Interventions (FA)
Special Purpose
Demogrphics
(DM)
Elements
Comments (CO) Subject (SE)
Trial Design
Trial Elements
(TE)
Trial Arms (TA)
Relationship
Related Records
(RELREC)
Supplemental
Qualifiers
(SUPP--)
Trial Summary
(TS)
Subject Visits
(SV)
Trial Visits (TV)
Trial Inclusion/
Exclusion (TI)
9
SDTM fundamentals
Class – Interventions
Investigational, therapeutic and
other treatments/substances administered
CDISC Standard Domains
• Exposure (EX)
à Study
Drug Treatment
• Concomitant Medication (CM)
• Substance Use (SU)
à E.g.,
SDTM fundamentals
Tobacco, Alcohol
10
5
CDISC SDTM Beginner's Workshop
24/03/201429/08/2012
Sample EX Domain
SDTM fundamentals
11
Class – Events
Incidents independent of planned study evaluations
Protocol milestones (e.g., randomization, completion)
CDISC standard domains
• Adverse Events (AE)
• Disposition (DS)
à
Protocol milestones (e.g., informed consent, study completion)
• Medical History (MH)
• Deviations (DV)
à
Major/minor Protocol Deviations
SDTM fundamentals
12
6
CDISC SDTM Beginner's Workshop
24/03/201429/08/2012
Sample AE Domain
13
SDTM fundamentals
Class – Findings
Observations resulting from planned evaluations
CDISC Standard Domains
• ECG (EG)
• Questionnaires (QS)
• Incl/Excl Exceptions (IE)
• Subj. Characteristics (SC)
• Labs (LB)
• Vital Signs (VS)
• Physical Exam (PE)
• Drug Accountability (DA)
• PK Concentrations (PC)
• PK Parameters (PP)
Majority of data in findings structure
• Test Code (- -TESTCD) , Name (- -TEST), Result (- -ORRES)
SDTM fundamentals
14
7
CDISC SDTM Beginner's Workshop
24/03/201429/08/2012
Sample VS Domain
SDTM fundamentals
15
Class - Findings About
Findings about Events or Interventions
Use Cases
• Different timing than event/intervention
à
Details collected in Findings" structure (occurrence at a fixed date/visit)
• Details on events/interventions which have qualifiers of their own
à
E.g., units, methods
• Details on underlying disease when not collected as MH
à
FAOBJ=<underlying disease>
• Related symptoms or therapies
• Occurrence of pre-specified AEs
à
AEOCCUR=‘N‘ not allowed in AE domain
Combination of FAOBJ and FATESTCD
à
à
FAOBJ: Name of clinical event or condition
FATESTCD: characteristic of event/condition
SDTM fundamentals
16
8
CDISC SDTM Beginner's Workshop
24/03/201429/08/2012
Findings About (FA) - Sample
CE
FA
Express relationship in RELREC
SDTM fundamentals
17
Class – Special Purpose
Demographics (DM)
Comments (CO)
• Separate comment page + comments on domain pages
Subject Elements
Subject Visits
SDTM fundamentals
18
9
CDISC SDTM Beginner's Workshop
24/03/201429/08/2012
Demographics (DM)
Structure 1 record per subject
• General information on subject / treatment
SDTM fundamentals
19
Comments (CO)
Captures unstructured free-text comments
• Dedicated comments CRF page (discouraged by CDASH)
• Other CRF pages referring to a specific domain
SDTM fundamentals
20
10
CDISC SDTM Beginner's Workshop
24/03/201429/08/2012
Class – Trial Design
Purpose
Concept
Trial Arms (TA)
Trial Elements (TE)
Trial Visits (TV)
Trial Inclusion/Exclusion (TI)
Trial Summary (TS)
SDTM fundamentals
21
Trial Design Datasets - Purpose
Standardized description of trial designs
• Enables quick familiarization with trial design
• Facilitates comparison of different trials
• Enables search for trial design features in data warehouse
Linked with other SDTM domains
• Define arm and arm codes referenced in Demography (DM)
• Define visits used in findings datasets
• Define inclusion/exclusion criteria (only exceptions in IE)
SDTM fundamentals
22
11
CDISC SDTM Beginner's Workshop
24/03/201429/08/2012
Trial Design Datasets – Purpose cont.
Planned design vs. actual trial observations
• Elements: TE vs. SE, Visits: TV vs. SV
• Inclusion: TI vs. IE
Trial design datasets contain planned design per protocol
Subject observation datasets contain observation as
occurred in study
• incl. unscheduled elements/visits
• Include observed observations only
à Premature
end
23
SDTM fundamentals
Trial Design Data Sets - Concept
Epochs:
Arms:
Elements: Building Blocks
(Screen, Drug A, Drug B,
Open Drug A, Rescue)
SDTM fundamentals
Branches
24
12
CDISC SDTM Beginner's Workshop
24/03/201429/08/2012
Trial Arms (TA)
Open Drug X => Open Drug A
SDTM fundamentals
25
Trial Elements (TE)
SDTM fundamentals
26
13
CDISC SDTM Beginner's Workshop
24/03/201429/08/2012
Trial Visits (TV)
Clinical Encounters Usually Defined in Protocol
May Start and End in Different Elements
VISITNUM, VISIT, VISITDY in TV
à
Controlled terminology for planned visits in findings datasets
Sample VS Domain in MSG sample submission package: tv.xpt
SDTM fundamentals
27
Trial Inclusion/Exclusion (TI)
Criteria as defined in protocol
IE
Versioning of criteria due to protocol amendments
SDTM fundamentals
28
14
CDISC SDTM Beginner's Workshop
24/03/201429/08/2012
Trial Summary (TS)
Basic information about the study in vertical format
Terminology
Planned study characteristics as per protocol
Further details on
TSVAL (Null
Flavor, Code
value, Dictionary
and version used
29
SDTM fundamentals
Trial Summary Codes
CT to be applied
SDTM fundamentals
and even more in SDTM IG 3.1.3 Appendix C3
30
15
CDISC SDTM Beginner's Workshop
24/03/201429/08/2012
Subject Elements (SE)
Focus on handling of unplanned elements
SDTM fundamentals
31
Subject Visits (SV)
Focus on handling of unplanned visits
SDTM fundamentals
32
16
CDISC SDTM Beginner's Workshop
24/03/201429/08/2012
Class – Relationship
Relationships
• Related Records to express relationships between domains
(RELREC)
à E.g.,
concomitant medication applied due to specific AE
• Supplemental Qualifiers for parent domains
à E.g.,
Clinically Significant Flag for lab assessments
CDISCfundamentals
SDTM
SDTM Workshop
33
Expressing Relationships
Relationship of multiple records in the same domain
à
Same --GRPID value
à
E.g., mark CM records that are part of a combo treatment
Relationships between records in different domains
à
Domain RELREC
à
Relationships expressed via same RELID value
à
Includes contributing parent records identified via
USUBJID, IDVAR, IDVARVAL
à
Potential IDVARs: --SEQ, --GRPID
General relationships between domains
à
Only 2 records in RELREC
à
USUBJID, IDVARVAL are blank, --GRPID describes merge key
SDTM fundamentals
34
17
CDISC SDTM Beginner's Workshop
24/03/201429/08/2012
RELREC Samples
Relationships between single records in different domains
General relationships between domains
• Prerequisite: all concentrations used for derivation of parameters
35
SDTM fundamentals
Sample SUPPQUAL (SUPPAE)
Source: CDISC SDTM/ADaM Pilot (selected variables)
SUPP- - dataset may also be created for custom domains
SDTM fundamentals
36
18
CDISC SDTM Beginner's Workshop
24/03/201429/08/2012
Parent Domains & SUPPQUAL
37
SDTM fundamentals
Custom Domains vs. SUPPQUAL
Issue
• Placement of variables not covered by standard domains
Solution
• New topic (set of variables)
à Create
Custom Domain
-
Determine observation class (Findings, Interventions, Events)
-
New domain name (convention: X-, Y-, Z-)
-
Adapt variable names and labels accordingly
• Additional qualifier to existing domain (single variables)
à Create
SDTM fundamentals
Supplemental Qualifier Dataset
-
Naming convention: SUPP- - ( -- = Domain)
-
Link to parent domain record(s) via IDVAR, IDVARVAL
38
19
CDISC SDTM Beginner's Workshop
24/03/201429/08/2012
Creating a Custom Domain
According to SDTM IG
SDTM fundamentals
39
Domain Variables
Role Concept
Core Variable Concept
Technical Requirements
Values Controlled via Terminology
SDTM fundamentals
40
20
CDISC SDTM Beginner's Workshop
24/03/201429/08/2012
AE Domain Class Table from SDTM IG
SDTM fundamentals
41
Domain Variables – Role Concept
Identifier – Key Variables
• Study, Unique Subject ID, Domain Name, Sequence Number
Topic – Focus of Observation/Dataset
• E.g., Test Code for Findings, Verbatim Term for Events
Timing – Time Relevant Information
• Visits, Date/Times, Relative Study Days
Qualifier – Attributes/further Description of Topic
• Subcategories: Grouping, Result, Synonym, Record, Variable
• E.g., Type of Lab Test, Original Result, Test Name,
Status=NOT DONE, LL/UL of Normal Range
SDTM fundamentals
42
21
CDISC SDTM Beginner's Workshop
24/03/201429/08/2012
Core Variable Concept
Core=Required
• Variable must be included in domain
• Variable cannot be Null
Core=Expected
• Variable must be included in domain
• Variable can be Null (preferably not always)
Core=Permissible
• Variable can be excluded/included in domain as needed
• Variables not populated should be omitted
SDTM fundamentals
43
Variables – Technical Requirements
Restrictions due to SAS V5 transport format
• Variable names + length of --TESTCD <= 8 characters
• Variable labels + length of --TEST <= 40 characters
• Character variables <= 200 characters
Attributes (data type, label, ...) + (order) must match
• SDTM specifications
• Study specific metadata (define.xml)
Additional variables =>
• Custom domain or supplemental qualifiers
• (SDTM domain if variable type is included in SDTM class)
SDTM fundamentals
44
22
CDISC SDTM Beginner's Workshop
24/03/201429/08/2012
Controlled Terminology
Controlled Terminology needed for
• Description of variables with a discrete list of possible values
• Vertical structure calls for standard test names and codes
à Facilitates analysis
across submissions
Current situation
• CDISC Terminology available for some but not all variables
à E.g.,
N/Y, Vital Signs Test Codes (PULSE, TEMP, ...)
• Sponsor specific terminology in addition
• CDISC terms preferred, transition period allowed
CDISC + sponsor terminology => describe in metadata
SDTM fundamentals
45
Controlled Terminology (Continued)
http://www.cancer.gov/cancertopics/cancerlibrary/termino
logyresources/cdisc
SDTM fundamentals
46
23
CDISC SDTM Beginner's Workshop
24/03/201429/08/2012
Special Terminology: Date & Durations
ISO 8601 Date/Time Format (-DTC Variables)
ISO 8601 Durations (-DUR Variables)
SDTM fundamentals
47
Metadata – define.xml
Metadata = data about data
Human-readable description to familiarize with data
• What is included?
• Where does the data come from?
• Derivation rules, if applicable
Machine-readable metadata to facilitate ...
• SDTM data consistency check
• Upload into data warehouse
Preferred => mandatory format: define.xml
SDTM fundamentals
48
24
CDISC SDTM Beginner's Workshop
24/03/201429/08/2012
Metadata – define.xml
Excerpt from the CDISC Draft Metadata Submission Guidelines define.xml Sample
49
SDTM fundamentals
Advantages of XML – Basic Technology
<?xml version="1.0" encoding="ISO-8859-1"?>
<?xml-stylesheet type="text/xsl" href="define1-0-0.xsl"?>
<ODM
...
<ItemGroupDef OID="DM"
Name="DM"
Repeating="No"
IsReferenceData="No"
Purpose="Tabulation"
def:Label="Demographics"
def:Structure="One record per subject"
def:DomainKeys="STUDYID, USUBJID"
def:Class="Special Purpose"
def:ArchiveLocationID="Location.DM">
...
SDTM fundamentals
style sheet
reference
50
25
CDISC SDTM Beginner's Workshop
24/03/201429/08/2012
define.xml Sections/Elements
Data Metadata (TOC)
Variable Metadata
Variable Value Level Metadata
(Computational Algorithms)
Controlled Terminology/Code Lists
Annotated CRF
Optional: Supplemental Data Definition Document
Navigation via Bookmarks and Hyperlinks
SDTM fundamentals
51
Data Metadata
SDTM fundamentals
52
26
CDISC SDTM Beginner's Workshop
24/03/201429/08/2012
Variable Metadata
Origins in SDTM IG 3.1.3: CRF, Derived, Assigned, eDT, Protocol, Sponsor Defined
SDTM fundamentals
53
Variable Value Level Metadata
SDTM fundamentals
54
27
CDISC SDTM Beginner's Workshop
24/03/201429/08/2012
Controlled Terminology/Code Lists
External dictionary references incl. versions
• e.g., MEDDRA, WHODRUG
SDTM fundamentals
55
CRF Annotation
Decide where data collected have to be stored in SDTM
• Target destination is based on data content
à
Data on 1 CRF page may map to multiple SDTM domains
à
Target data structure to be considered in the mapping
• Use standard SDTM domains where possible
• Custom domains for new topics
à
Decide on appropriate observation class
• Additional qualifiers for parent domains, i.e. not included in
standard set of variables
à
Supplemental Qualifier Domain (SUPP--)
• Issues resulting from non-extensible CDISC Controlled
Terminology
Aim: Illustrate the Mapping
SDTM fundamentals
56
28
CDISC SDTM Beginner's Workshop
24/03/201429/08/2012
Annotated CRF (blankcrf.pdf)
See CDISC Metadata Submission Guidelines (SDTM)
57
SDTM fundamentals
Reading an OpenCDISC SDTM Validation Report
Structure and Layout
Case Study
• Decide on
à Issues
to be corrected
à Acceptable
issues to be explained in study data reviewer’s guide
OpenCDISC validates for adherence to structural SDTM rules
Content validation still required
• E.g., Weight data mapped to Height not spotted by OpenCDISC
SDTM fundamentals
58
29
CDISC SDTM Beginner's Workshop
24/03/201429/08/2012
OpenCDISC Validation Report - Structure
Dataset Summary
Issue Summary
Rules
Details
SDTM fundamentals
59
Remaining Issues in Study Data Reviewer‘s Guide
PhUSE CSS Study data reviewers guide:
http://www.phusewiki.org/wiki/images/2/2d/SDRG_v1.1_2013-05-13.zip
SDTM fundamentals
60
30
CDISC SDTM Beginner's Workshop
24/03/201429/08/2012
What‘s new in SDTM IG 3.2
New Presentation (PDF Portfolio of files)
New Domains
•
Interventions Class
à
Exposure as Collected (EC)
à
Procedures (PR)
•
Events class
•
Findings Class
à
à
à
à
à
à
à
Healthcare Utilization (HO)
Death Details (DD)
Immunogenicity Specimen Assessment (IS)
Morphology (MO)
Subject Status (SS)
Microscopic Findingy (MI)
Reproductive System Findings (RP)
•
Findings About
•
Trial Design
à
à
Skin Response (SR)
Trial Disease Assessments (TD)
Medical Device and Associated Persons IG
New added Variables and changes to existing domains
61
SDTM fundamentals
Questions?
Nicola Tambascia
Senior Statistical
Programmer
Accovion GmbH
Helfmann-Park 10
D-65760 Eschborn, Germany
Tel. +49 6196 7709-283
nicola.tambascia@accovion.com
www.accovion.com
SDTM fundamentals
62
31
CDISC SDTM Beginner's Workshop
24/03/201429/08/2012
Why CDISC SDTM/ADaM?
General Benefits of Standards
• Improve communication
• Time/resource savings
• Comparison across submissions
FDA‘s eStudy Data Draft Guidance (Feb 2014)
• Written official requirement that sponsors submit
study data in a standardized electronic format
• Includes HOW to submit (standard format)
• Does NOT include WHAT to submit
à
Driven by science, to be agreed with reviewing division
• Refers to study data standards resources web page
63
SDTM fundamentals
Regulatory Requirements
SDTM + ADaM referenced in eCTD‘s Study Data Specifications V2.0
Results
Clinical Study Report
Traceability
ADaM
Analysis data + define (ADaM)
Analysis results/data creation programs
Data
legacy data (optional)
Option: raw data + define in legacy format
SDTM
SDTM data + define + aCRF
SEND (optional)
Details: FDA‘s web page: Electronic Common Technical Document (eCTD), Study Data Specifications
SDTM fundamentals
64
32
CDISC SDTM Beginner's Workshop
24/03/201429/08/2012
SDTM Implementation Details
Handling of free text
Handling of clinical significance data
Handling of “Check all that apply“
Splitting of domains
Unscheduled Visits
Controlled Terminology
Adding Variables to Standard Domains
65
SDTM fundamentals
Handling of Free Text
Non Result Qualifiers
Upper Abdomen
Options for populating supplemental qualifiers
• EXLOC=OTHER, QNAM=EXLOCOTH, QVAL=UPPER ABDOMEN
• EXLOC=ABDOMEN, QNAM=EXLOCOTH, QVAL=UPPER ABDOMEN
• EXLOC=UPPER ABDOMEN; No SUPPEX
Choice driven by analysis/sponsor requirements
• Categories/grouping of locations in analysis table
SDTM fundamentals
66
33
CDISC SDTM Beginner's Workshop
24/03/201429/08/2012
Handling of Free Text (continued)
Result Qualifiers
Blueish Gray
Options for representation in SDTM
• MS Standard:
SCTEST=EYE COLOR, SCORRES=OTHER, SCSTRESC=OTHER,
EYECOTH IN SUPPSC=BLUEISH GRAY
• Other Possibility:
à SCTEST=EYE COLOR, SCORRES=OTHER, SCSTRESC=BLUEISH GRAY
à SCTEST=EYE
COLOR, SCORRES=OTHER, SCSTRESC=GRAY,
EYECOTH IN SUPPSC=BLUEISH GRAY
Choice driven by analysis/sponsor requirements
• Categories/grouping of results in analysis table
67
SDTM fundamentals
Handling of Clinical Significance Data
Assessment on record/test level
• e.g., a certain lab test result assessed a clinically
significant by investigator
• Store flag in supplemental qualifier dataset
• Entry in SUPP-- with QNAM= --CLSIG
à Can
be linked to parent record
Assessment on multiple record level
• e.g., overall assessment of ECG results at a certain
time point
• Separate record in parent domain
• --TEST = "Interpretation"
SDTM fundamentals
68
34
CDISC SDTM Beginner's Workshop
24/03/201429/08/2012
Handling of "Check All that Apply"
Options
• More specific naming and labeling
• Most relevant selection may be mapped to domain variable
• Driven by analysis/sponsor requirements
SDTM fundamentals
69
Multiple Responses for Findings
Multiple records per --TEST possible
Sample EG:
EGTESTCD=RHYRATE, EGTEST=Rhythm and Rate, EGORRES=ATRIAL FIBRILLATION
EGTESTCD=RHYRATE, EGTEST=Rhythm and Rate, EGORRES=ATRIAL FLUTTER
--SEQ or --SPID necessary to uniquely identify a record
SDTM fundamentals
70
35
CDISC SDTM Beginner's Workshop
24/03/201429/08/2012
Splitting of Domains
Possible reason: size issue (CDER <= 400 MB, CBER <=
1 GB)
Split on --CAT level
Allow for retrospective combination
• Same DOMAIN value and suffix
• Consistent variable lengths
• --SEQ unique over all split domains
• Separate SUPP---- datasets following the same concept
Sample:
QS36, SUPPQS36 with DOMAIN and RDOMAIN = QS
Ask FDA
SDTM fundamentals
71
Prespecified Flag
New qualifier variable --PRESP
• Handling / marking of pre-specified term
• For use in interventions and events classes
• Rationale: higher incidences expected for pre-specified
terms
SDTM fundamentals
72
36
CDISC SDTM Beginner's Workshop
24/03/201429/08/2012
Unscheduled Visits
Repeated Measurements
• e.g. to follow up on abnormal lab values
Multiple records for the same visit?
Repeated visit gets VISITNUM with Decimal?
• VISITNUM=1, VISITNUM=1.01
• Respective VISIT values?
à
Missing value vs. “VISIT 1 – REPEAT 1“
Generic values VISIT=UNSCHEDULED, VISITNUM=9999
for every unscheduled visit?
à
Try to avoid already at data collection time
Consistency with TV and SV to be considered
Keys in define.xml should uniquely identify a record
CDISCfundamentals
SDTM
SDTM Workshop
73
37