V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
1 RL-1 1.2 This version is expected While this version is The following changes should be made to the last paragraph Persuasive Retained reference to
Randy Levin to be an implementation- expected to be an on page 4 to help clarify the future work for the data model Implementation guides.
ready version, although implementation-ready
additional improvements version for clinical studies
and enhancements are involving human drug
2 RL-2 2.1 likely to be made over
The implementation products, future
Any sponsor wishing to These changes should be made in the last sentence under Persuasive Needs to retain reference to
guides specifically submit human or animal "Using the Model for Regulatory Sunmissions" to allow "Core" . Added sentence
describe which variables drug product data in the regulatory authorities to create and enforce domain specific instead.
are required, expected, or standard formats should first complementary guides to further clarity requirements.I2
permissible to use in consult the implementations
specific domains based guides and appropriate
on the general regulatory agency guidance
observation classes. and specifications before
preparing a regulatory
submission based on the
SDTM.
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 1 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
3 RL-3 4 SDTM has modeled a useful way to describe relationships Persuasive The Relates structure will be
among observations, but some of the machinery is redundant, changed to use only one table.
leading to overly complex business rules. We believe that a
single mechanism can incorporate all the richness of the
current system, correct some of the current system's
shortcomings, simplify the business rules, and ease the
reviewer's burden. We first describe a 'clean' implementation
and then show how the current model can be adapted without
introducing any structural changes.
All observation record grouping is done using SQ.
Observation domain GRPID gets deprecated. There are four
immediate benefits. First, one does not need a business rule
per domain to specify which grouping concept warrants use
of the special GRPID. Second, groupings can overlap
(observations can belong to more than one group). Third,
where it is useful, you can identify the Evaluator associated
with the grouping. Fourth, it removes the incentive to append
new GRPIDxx fields.
The single relationships table, i.e., RELS, would replace
RELDS and RELREC and would have the following
structure:
STUDYID-
FDOMAIN-Source observation domain code
FUSUBJID-Source observation unique subject identifier. If
NULL, the defined relationship applies to all subjects in the
study.
4 RL-4 4 Aside from GRPID, a host of other observation fields should Not Persuasive Too difficult to implement at
be deprecated in favor of use of SQ. There are two reasons. present time. Method of
First, it allows one to specify an Evaluator. We often see representing more than one
studies in which the same qualifications (e.g., relationship of Support for Evaluator
event to study treatment) are reported by the investigator and described in IG.
sponsor). Second, it simplifies the reviewer's job by
providing a single place to look for such properties.
Fields that should be deprecated are (interventions) GRPID,
CAT, SCAT, OCCUR, CLAS, CLASCD, SPREF; (events)
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 2 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
5 RL-5 2.2.7 Various OTH comment fields should be deprecated in favor Persuasive Mod OTH fields are not comments,
of CO. but specified data. However,
these have been moved to
SuppQual (which is more
appropriate for "other specify"
collected data than comments).
6 RL-6 3 In earlier versions of the trial design datasets, elements had Persuasive Restored duration variable.
nominal duration (actually an acceptable range) specified. A Window of acceptable duration
rule could be given for an early termination, and the notion is more properly part of the
was that, in the fullness of time, such rules would be snippets Statistical Analysis Plan.
of executable code, allowing one to see whether the data
supported the decisions taken. Now it appears that the
nominal duration only appears in some syntactically complex
text (after xxx units-of-time), with no clue as to what is an
acceptable range. This would limit how software can display
an expected (planned) time line for a subject's progress
through the trial. The ISO8601 duration and duration range
values should be added back as additional fields to address
this issue.
7 RL-7 4 Study design features are needed to be added to the model Not Related Cannot be implemented at this
including planned interventions and assessments and time; will be added in the
statistical analysis plan. future.
8 RL-8 2.2.2 What is the difference between the data to be entered under Not Related MHONGO is not intended to
MHSTAT variable (Status of Disposition) vs. MHONGO be in the MH domain.
(Medical Condition Ongoing)? Do one or both variables References to the variable were
refer to whether or not a specific medical history condition is errors and have been removed.
ongoing or not? Also, in the MH domain, what is an
9 RL-9 2.2.2 appropriate variable to indicate recent treatment for specific Persuasive
Which variables should be used to identify study participants The implementation guide
who stop study drug but are still considered as participating documentation will be revised
in the study? Similarly, which variables should be used to to include expanded examples
identify subjects who complete the study? for how to represent variables
such as population flags in SQ
and to include another
disposition example. It will
also be updated to address the
gap of standardized SQ Names.
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 3 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
10 RL-10 2.2.2 What about narratives for SAE and the status of SAE reports Not Related Not in Scope, but could be
(initial or follow-up) ? handled using IDREF.
11 RL-11 2.2.2 AE: What variables indicate patient deaths and how are the Not Related AEOUT and AESDTH
narratives provided? describe this how to handle
patient deaths. This is an
implementation issue and will
be discussed in a future
implementation guide.
12 RL-12 2.2.6 What happens if the randomization date is not the same as Not Related The DS example in the IG
the screening date? Given there is one subject reference start displays how to submit both
date (RFSTDTC), how are both times captured? the screening completion date
and randomization date.
13 RL-13 2.2.1 How is the reason why the Study Drug was discontinued Persuasive The DS example in the IG
represented? Is it related to TERM in the DS domain. displays how to submit the
reason study drug was
discontinued. DSTERM is
used to show the verbatim
reason and DSDECOD
displays the reason using
controlled terminology.
14 RL-14 2.2.1 How is dosage form captured? Not Persuasive --DOSFRM
15 RL-15 2.2.2 Sometimes adverse events are verified. How is this Not Persuasive Could be addressed in
information captured? Supplemental Qualifiers with
an Evaluator.
16 RL-16 2.2.2 How is the reason for discontinuation captured? Not Persuasive --DECOD
17 RL-17 2.2.3 How is the method of collection of ECG captured? Not Persuasive EGSCAT
18 RL-18 2.2.2 How are the criteria for defining seriousness of AE captured. Not Related Described in ICH and FDA
Guidance
19 RL-19 2.2.3 How are genomics and pharmacokinetic data collected? Not Related Will be addressed in a future
version.
20 RL-20 2.2.3 How are lab site identifiers captured? Not Persuasive LBNAM
21 RL-21 2.2.3 How are the algorithm for ECG interval methods captured? Not Persuasive Described in DEFINE
metadata document
22 RL-22 2.2.3 How is the medical history severity and frequency captured? Not Related Not typically collected for MH;
could be represented as a prior
AE event or in
Inclusion/Exclusion criteria.
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 4 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
23 RL-23 2.2.1 How are the reasons why a treatment was not taken is Persuasive Mod CO of SuppQual; will be
captured? addressed in a new Compliance
domain in future.
24 RL-24 2.2.5 Should there be an ENDTC (end date/time of disposition Not Persuasive Model convention is always to
rather than STDTC (start date/time of disposition)? use --STDTC when only one
date is relevant.
25 RL-25 2.2.3 How are population information captured? Persuasive Added Appendix 10.4 to
address this question.
Population flags will be stored
in SUPPQUAL. This appendix
contains the controlled
terminology for SUPPQUAL
Name codes.
26 dc Arm, epoch, and element don’t handle all the required Not Related Trial design is intended to
ADaM Team situations. E.g., there isn’t a way of recording a describe plan only. Exposure
mistreatment (subject was supposed to get one dose, but gets Interventions provide the
another). information to identify
mistreatments. Identification of
records associated with
mistreatments is best addressed
in analysis datasets.
27 dc The analysis files need Planned Treatment and Actual Not Related Trial Design ARM describes
Treatment variables (both Code and Group). ARM and plan only. Exposure records
ELEMENT can’t be used because these words aren’t actual treatments. We believe
commonly used and understood in the world of clinical trials that description of actual
statistical analysis, plus there is no consistent way across trial treatment is best addressed in
designs to use ARM and ELEMENT to describe planned analysis datasets.
treatment and actual treatment).
28 CT-01 Sec. The interventions domain is meant to hold info for surgical Not Related This is an implementation
Christine 2.2/p. 6 interventions. For surgeries, all the "dosing"variables are issue and will be addressed in
Tolk wasted, and, particulary for surgeries planned as part of the the implementation guide. We
protocol, doesn't provide for collection of variables that believe the existing model
might might really be important. Can a discussion be added could support this scenario.
on how this dataset will represent surgical procedures
pertinent to the protocol? Also, would information on
surgeries gathered as part of routine Medical History be
stored in MH or in an interventions domain?
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 5 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
29 CT-02 Sec For concomitant medications the instructions say to fill -- Future Implementation issue;
2.1..1/p. DECOD with the drug's generic name. This is easy for Addressed in IG.
7 single ingredient drugs, but what should be entered for
multiple ingredient drugs?
30 Aventis 3 Section 1.1 Section 1.2 Refers to incorrect section of document Persuasive Corrected to 1.2
Elaine Job
31 Aventis Model doesn't specify that a value that is used in the analysis Not Persuasive This issue is discussed in
(especially if derived), needs to be a row in the dataset. Implementation Guide for --
DRVFL
32 Aventis 6/2.1 Bullet 3 - "The data Type Suggest that a better indicator is needed for when a discrete Persuasive The document has been
. . ." set of values is used. Need to define this concept more, clarified to refer readers to the
when is it a published list, versus a sponsors internal list, etc Implementation Guide for
and when are the codes given in the CDISC comments meant details on which terminology is
to be an example versus Industry standard. sponsor-controlled or
standardized.
33 Aventis 5/2.1 last 2 character domain code How will this list be maintained and by whom? We suggest Not Persuasive 2 characters is maximum as
paragraph to keep a naming convention but perhaps increase to 3 bite. prefix. Maintenance will be
addressed in other documents.
34 Aventis 6/2.2 last While the choice of What does this mean? It is that if you use a timing variable it Persuasive Mod Changed to "The meaning of a
sentence timing variable. . . has to be used in every domain or if used the format must be timing variable should be
of 2nd the same. consistent within a
paragraph submission."
35 Aventis 7/2.2.1 CLAS, CLASD Use of ATC Code is a confusing example as there isn't Persuasive Modified note to limit use to
always a one to one correspondence to the WHO-DRL codes. drugs of a single class..
36 Aventis 6/2.1 last must not be added, must not be added, and Persuasive Reworded.
paragraph renamed, or modifed for existing variables must not
novel usage. be renamed, or modified for
novel usage.
37 Aventis Relates model makes it very hard to do the define.pdf and the Persuasive Mod Has been clarified in SDTM.
annotations. Has to be done very carefully in order to put IG has been modified and
data back together. Not enough guidance is provided to make expanded with more examples.
this practical to set up or QC.
38 Aventis 8/2.2.2 OCCUR and STAT Definition makes these seem redundant. Persuasive STAT Description revised to
'Used to indicate when a
question about the occurrence
of an event is not answered'
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 6 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
39 Aventis 8/2.2.2 DECOD Variable name is misleading. More of a standard dictionary Not Persuasive --DECOD is necessary for
term. Usually decode is how you code the Modify. backward compatibility; label
and description clarifies use.
40 Aventis 8/2.2.2 TRTEM Is this just a Y/N? Then add to the description as they have Persuasive This variable, which is not
done in other variables. typically collected from the
investigator, is an analysis
variable and should not be part
of the tabulation. Can be
placed in SuppQual. Will be
removed from SDTM.
41 Aventis General If it is acceptable to code N/Y and M/F then why not any Not Persuasive SDS rule is that any code that
other controlled terminology. We will have to decode and is standardized and self-evident
then change the code back to just "M/F" and "N/Y". Suggest does not require a full decode
to just submit the full decoded value.
42 Aventis 8/2.2.2 REL Why has this changed from "N/Y" and added codes from Persuasive The word "Examples" will be
E2B. In description it is not clear whether these are added for consistency. Other
examples or have to be used. details are in the IG.
43 Aventis 8/2.2.2 RELOTH Why doesn't this have a controlled terminology. Persuasive Renamed to AERELNST and
corrected description. Intended
to be a free text field, but
sponsor can define a list of
controlled values if
appropriate. See IG.
44 Aventis 8/2.2.2 PATT Is this just for collected pattern or a derived pattern. Persuasive Fixed note to specify
"collected"
45 Aventis 8/2.2.2 PATT - Pattern of e Pattern of Event Persuasive Corrected
Event
46 Aventis 8/2.2.2 SER - Is this is a serious Is this a serious event? Persuasive Corrected
event?
47 Aventis 8/2.2.2 OUT Description - Is this mandated controlled terminology or an Persuasive Added "examples"
example. Description of Not Recovered/Not Resolved is
confusing and not mutually exclusive relative to other
categories.
48 Aventis 9/2.2.2 Should include text that the TOXGR and SEV are not used Not Persuasive Implementation issue.
together.
49 Aventis General Should have the Required variables indicated in this Not Persuasive Will vary by type of data.
document even though the Mandatory and Permissible
change per domain.
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 7 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
50 Aventis General What is the order of the qualifiers? Page 12 of the Persuasive The description of order in IG
Implementation Guide suggests one order but then the has been revised. Order of
domain models don't follow that order. domains has been revised to
follow description.
51 Aventis 9/2.2.3 Is the finding model synchronized with the lab transmission Not Persuasive Models are already consistent;
model? There are a lot of additional mandatory variables in many Lab variables not
the transmission model. necessary for submission.
52 Aventis 9/2.2.3 Does OCCUR need to also be in the Findings model as well? Not Persuasive Addressed by --STAT.
Some findings can also be defined and need to be checked
whether they occur or not.
53 Aventis 10/2.2.4 Suggest putting the Unique Identifier and Timing variables Not Persuasive Author believes current order is
& 2.2.5 before the individual domains. more appropriate in SDTM.
54 Aventis 12/2.2.6 RFSTDTC and Format should be Character Persuasive Corrected
REENDTC
55 Aventis 12 Date/Time when subject Reference Date/Time for the Subjects are considered to have started the trial as of Persuasive Corrected
RFSTDT was determined to have subject. Usually equivalent Informed Consent date as per EU Directive - not appropriate
C and started the trial to date/time of first intake of to arbitrarily set this differently from regulations or company
REENDT drug. conventions for the sake of a model.
C
56 Aventis 13/2.2.6 DMDY Format should be Number. Persuasive Corrected
57 Aventis 13 Sequence numbering and cross referencing, link sequencing Not Persuasive Learning curve issue.
has to be done manually and this will be labor intensive and
prone to errors.
58 Aventis General Need assumption that if a common variable is used across Not Persuasive Need clarification; SDS-IG
domains, that the item name will be the same and not use the 4.1.2.2 describes rule for
domain prefix. assigning prefixes.
59 Aventis General Add to assumption that Element will not use the domain Not Persuasive Implied by lack of hyphens.
prefix.
60 Aventis 20 TICRIT Why are we including an algorithm in the table? This is the Persuasive Will modify text for
only place that we have this concept in the model. consistency. Note Rule is used
in other Trial Design tables.
61 Aventis 13/2.2.7 Why is this the only place where you can add an item? Persuasive Removed reference in SDTM.
Should we just add a COGRPID to group the multiple Implementation issue discussed
records that has the continuation of the comment. in IG. Retained for SAS XPT
compatibility only in this case.
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 8 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
62 Aventis Split domains are not well defined how to do this and how to Persuasive Clarified Relationships section.
put this data back together. The Relates concepts will have
to be done manually and will be labor intensive.
63 Aventis 19/3.3 It states that a subject Arm is not needed as it is the Not Persuasive ARM is the planned arm.
Demographics, but it doesn't state in Demo that this is an
actual. If it is actual, where do we put the planned?
64 Aventis general It would be burdensome to produce SDTM files with no Not Persuasive We believe there will be
obvious benefit to sponsors in terms of satisfying either the benefits to both sponsor and
statistical or clinical reviewers of our submission package. It FDA to have a standard
would certainly be an additional effort to produce and structure to define the trial
reconcile such a file with raw data and with analysis data sets design.
and has no internal utility for the sponsor.
65 Aventis general The "relates" component of the model is not mature or Persuasive Relationships section has been
specified well enough to ensure that the data could be put updated.
back together in the relational form that is necessary to make
it comprehensible - especially for domains that require
splitting under this model.
66 Aventis general There is extensive referencing of the need for standard Persuasive Mod Standard values for
vocabularies to be identified for most of the model items, but SUPPQUAL name code and
the vocabularies are not provided. Without these EG and VS Test Codes has
vocabularies, the model can not be used as a standard (no been added to the
two organizations/individuals could be expected to Implementation Guide as
intuitively code or organize item values the same way). Appendix 10.4. Other standard
vocabularies will be addressed
in the future in a separate
document.
67 Aventis general The model is not consistent with the previously approved Not Persuasive Duplicate.
CDISC Lab model and no guidance for reconciling these
differences in provided.
68 Aventis 12/2.6.8 Set the order of variables The Domain models in Section 6 do not follow this order; Not Persuasive Refers to IG document, not
(Identifier variables, they have qualifier variables before timing variables. SDTM.
Topic variable, critical
Timing variables, critical
Qualifier variables and
then other Timing and
Qualifier variables).
69 Aventis 37/6.1.1 Shouldn't CMSTAT and CMREASND be added to the Persuasive Added CMSTAT and
domain? CMREASND to CM domain
70 Aventis 46/6.1.3 SUREASND is not in the domain. Should it be? Persuasive SUREASND has been added to
Substance Use domain
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 9 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
71 Aventis 48/6.1.3. SUSTAT can be used to This contradicts page 46, where SUSTAT is used to indicate Persuasive Incorrect/confusing text has
0.7.b describe the status or that the question was not asked or the measurement was not been deleted.
past or current substance done.
use.
72 Aventis 52/6.2.1 SDSSDSAEENDTC AEENDTC Typographical error in variable name. Persuasive Corrected.
73 Aventis 52/6.2.1 Shouldn't AESTAT and AEREASND be added to the Persuasive No. Only events that actually
domain? occurred or a list of solicited
events should be submitted.
74 Aventis 52/6.2.1 There should not be a case If AEENRF is in the domain why shouldn't AESTNF also be Not Persuasive It is likely that events will be
where AESTRF is in the domain? ongoing at the end of the
unknown, but can be added reference period, whereas it is
as as timing variable. unlikely that many events with
unknown start date prior to the
start of the reference period
would be reported. However,
AESTRF may be added as a
timing variable at the sponsor’s
discretion.
75 Aventis 53/6.2.1. Same example given in 6a is repeated on page 53 and 54. Persuasive The assumption has been
0.6.a clarified.
76 Aventis 67/6.3.2 For IEORRES, CDISC If only inclusion/exclusion criteria exceptions are included in Persuasive Inclusion criteria would be "N"
notes are "Inclusion or the IE domain, then isn't IEORRES always 'Y'? and Exclusion would be "Y".
Exclusion criterion met?"
and controlled
terminology is Y,N
77 Aventis 82/6.3.6. Subject characteristics is Should the fact that it should not change during the study be Persuasive Childbearing and allergy status
0 for data not collected in added to this sentence? Also, if that is the case, should removed from text and
other domains that is childbearing status and allergy status be given as examples examples in section 9.
subject-related. since they might change during the study?
78 Aventis 133/9.2.3 Use of Figure 1, Table 1, Table 2, etc. does not match with Persuasive Corrected.
examples given, i.e. following Figure is 9.2.3.1 not Figure 1.
79 Aventis 141/9.3.2 In table of sample data, what do rows 14-16 represent? Persuasive Typographical errors have been
DSTERM, DSDECOD, VISITNUM and EPOCH are corrected and the example has
missing. been expanded and clarified.
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 10 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
80 P&G 1 2.1 We feel that if sponsors are using standard variable names Persuasive Document updated to state that
and definitions, and these have defined stanadard roles, then Role need not be submitted
the Role attribute can be populated in the metadata after it is except for sponsor extensions.
received by the regulatory agency. The Role is not used
internally by most sponors for any purpose. It makes more
sense to have one application in one location (e.g., the FDA)
to populate this column than to have every sponsor doing so.
81 P&G 2 2.2.4 The use of SEQ is good in that there is one key which Persuasive Model now allows use of --
confers uniqueness to a dataset. However, this convention SPID as an optional unique
has a downside in that it results in a loss of structural identifier.
information, since every dataset now becomes one record per
sequence number, and other variables such as visit, test, and
location are no longer primary keys
82 P&G 3 2.2.6 AGE Description: If AGE is collected on the The standard states that durations are not to be included in Not Persuasive Age is an essential selection
Usually derived as Case Report Form, it should submission datasets unless they are collected. Age is a variable for clinical data
RFSTDTC-BRTHDTC, appear here. If AGE can be duration (of life) and should be treated as such. review, whether derived or
but BRTHDTC may not derived from RFSTDTC collected, and should be
be available in all cases and BRTHDTC, then it included.
(due to subject privacy should not appear here.
concerns).
83 P&G 4 2.2.8 The reference to and use of SEQ to relate a comment to a Persuasive Relationships has been revised
parent record presents a problem since the SEQ variable will to address this. Details are
not be populated in many cases until after datasets are described in the IG.
complete. It will be an artificial value that does not
necessarily exist in the sponsor's database.
It would be easier to allow the sponsor to define what
variable is used (as allowed by IDVAR in RELDS)
84 GS-DT-1 Tabulatio I don't see how this is an HL7 specification. Is this supposed Not Related Not intended to be a v2 ORU
Gunther n "model" to be used with HL7 v2 ORU messages? Many many of message or serve as a HL7
Schadow these "observations" really beg for a structure, a model that model. Describes a conceptual
describes the study. Much of this stuff also has a model for sending submission
representation in the RIM. I may be totally confused, but if data to FDA based on current
this is supposed to be considered an HL7 clinical trial FDA practice and requirements
"model" this is just not acceptable. If it is not that, if it is and will be mapped to RIM in
only some codes that are useful in some special legacy the future. A complete clinical
technology, SAS tables, etc. then that should be clearly stated trials model is being developed
right up-front. separately, but will take time to
complete.
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 11 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
85 GS-DT-2 Tabulatio I doubt that this is in any way interoperable. 90% of the Not Related Current datatypes are
n "model" columns are of data type "Char"even if they are identifiers or consistent with SAS XPT
concept codes or something that should be a formal representation and current FDA
expression. This is not robust. I stronglys uggest that if this requirements. More robust
is to go forward without v3 interaction, then it should be dataypes will be added in a
made an HL7 v2 profile over ORU messages. And then at future version as the HL7 D-
least the HL7 v2 data types for coded value (CE) should be MIM for clinical trials evolves.
used, and all the others.
86 Mark Suggestion, overall: There is no mapping from the contents Not Related A statement will be added that
Shafarman-1 of this document to the v3 RIM: even as an informative mapping will be addressed in a
CDISC document, if it’s being balloted in HL7, such a future version.
mapping should appear in the appendix, and an explicit v3
RMIM and HMD and MT and the corresponding XML ITS
expressions of the HMD and MT should be included in the
appendix. At the very least, a statement should be added that
the next version of this document will contain such a
mapping, or that one will be published shortly as an
informative appendix.
87 MCP000 section Note that current types The more descriptive datatypes mentioned in the "Existing Not Related More descriptive datatypes will
Michael 2.1 page are restricted to character Wording" are required to use the STDM with CDISC's be supplied in a future version.
Palmer 6 and number for Operational Data Model, in for instance, the "Content
compatibility with SAS; Description for Regulatory Submissions of Clinical Trials
it is expected that Data," a.k.a. "define.xml" and passed by RCRIM as an
additional, more informative ballot in the last ballot cycle. define.xml is
descriptive datatypes intended to work with STDM That is, the current STDM
(e.g., integer, float, date, ballot does not contain enough information to create
date/time) will be used in define.xml instances for the STDM. To support define.xml,
the future. the STDM should specify the more descriptive datatypes and
simply indicate that, for instance, integer, float, date, and
date/time have the "Numeric" datatype for SAS.
88 MCP010 Tables add a "Length" column to all The "Content Description for Regulatory Submissions of Not Related Lengths will be address in a
2.2.1 to the tables Clinical Trials Data" informative ballot passed by RCRIM in future version.
2.2.7 and the last ballot cycle and a.k.a. "define.xml." requires lengths
following for all fields and STDM provides no lengths. define.xml is
tables intended to work with STDM. In order to create a generic
until the define.xml based on the STDM, the STDM needs to supply
end of the either default lengths or maximum lengths.
document
.
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 12 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
89 WEF-1 2.2.2 variable = TRTEM Remove from standard This is a derived variable. It is not collected on CRFs and is Persuasive This variable, which is not
Bill Friggle not definable without considering the analysis plan. It typically collected from the
should addressed in the ADaM models. investigator, is an analysis
variable and should not be part
of the tabulation. Can be
placed in SuppQual. Will be
removed from SDTM.
90 WEF-2 2.2.3 variables = BLFL & Remove from standard These are derived variables. They are not collected on CRFs Not Persuasive These variables are considered
DRVFL and are not definable without considering the analysis plan. essential for regulatory review
They should be addressed in the ADaM models. For and should be provided in
population flag(s), we would propose implementing an tabulations.
ADaM Covariate dataset containing, at a minimum,
STUDYID, USUBJID, SAFETY, ITT and PPE.
91 WEF-3 4.4 "...such as a population Remove from standard Describing a derived variable - should be removed and Not Persuasive These variables are considered
flag that classifies addressed in ADaM models. essential for regulatory review
subjects or subject data and should be provided in
according to their tabulations.
evaluability for efficacy
analysis."
92 WEF-4 3 The Trial Design Model Remove from standard We would like to suggest a no vote to this aspect of the Not Persuasive FDA has attested to value and
section standard. Specifically, we would like to see: value has been acknowledged
- evidence of successful testing of these concepts by other CDISC participants;
- A written statement documenting the value of submitting CDISC is currently testing the
this information and why it is considered tabulation data, as model.
in our experience it is not data collected on CRFs.
Previously we were able to identify periods, phases and
cycles in 3 columns in our submissions, which seemed very
straightforward. Trial design seems like a very complex
replacement for simple concepts, and appears to require a
manual review of the protocol in order to populate the Tx
tables and programming effort to populate the Sx tables.
93 Dubman-1 Page 4 Page 4 - "elimination of the concept of a separate Date/Time Not Related Explanation is intended for
Precision variable for each date/time variable" - it is not clear readers familiar with earliver
to me what this means. I think we need the ability to have version of the model.
different precision for different dates. Precisions are now addressed
by ISO 8601.
94 Dubman-2 Page 7 Page 7 - DOSFRQ labeled "Dosing Frequency per Interval" Not Persuasive DOSFRQ interval may differ
DOSTOT labeled "Total Daily Dose" from total daily dose; used for
This implies that the frequency "interval" is a "day". If so, different purposes
change labels to be consistent.
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 13 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
95 Dubman-3 Page 8 Page 8 - there are many variables that are of type "Char*" Not Persuasive Need clarification regarding
which indicates controlled terminology. Some of these (like what change is requested. Yes,
TRTEM & SDTH) that look like booleans or Y/N. many controlled term variables
are Y/N.
96 Dubman-4 Page 9 Page 9 - The Findings Observation Class Topic Variable is a Not Persuasive The Findings class is unique
code, while all others are verbatim terms. It is qualified by because it describes planned
the verbatim term, while others are qualified by the code. assessments. The short coded
term is often used as a column
header and CRF question, and
the coded term is preferred as a
key.
97 Dubman-5 Page 11 Page 11 - VISITDY (and other Day variables) refer to Study Not Persuasive While Course/Cycle is more
Day, while Course/Cycle Day seems more common. common for Oncology; it does
not apply to other types of
studies.
98 Dubman-6 Page 13 Page 13 - SEX variable name is dated. Gender is the more Not Persuasive Rationale for using SEX
appropriate name. instead of GENDER is
DMDY - Study Day of Collection is type Char, yet described in the IG. IOM
description says it is measured in integer days report defines SEX as
appropriate term. --DMDY
datatype corrected to NUM.
99 Diane p7 STDM SDTM Typo Persuasive Corrected.
Weatherall / section
Quintiles 1.4, point
1
100 Michael e-mail from Michael: "I've got a comment on your first point Not Related Will address in future version.
Palmer below about datatypes. Mapping XML schema datatypes into
the two SAS datatypes is easy because "text" in schema
becomes "Char" in SAS and everything else in schema
becomes "Num" in SAS. Going the other way is impossible
because, for instance, "Num" could be float, integer, date,
etc. If the SDS documentation has the XML schema
datatypes in the metadata tables, a footnote could explain
that "text" becomes "Char" in SAS XPT and everything else
101 Rob 5/2.1 Define the think that would handle the SAS XPT
is "Num". Ivalue of the sub-classes of qualifiers. legacy Persuasive Document updated to clarify
Sarate/Kathy Role does not have to be
Giordano/Me submitted by sponsors.
rck
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 14 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
102 Rob 6/2.1 We would like to have the ability to retain expected and Persuasive IG has been updated to indicate
Sarate/Kathy permissable variables that are null to lessen standard that this is a sponsor decision.
Giordano/Me programming effort The concept is not discussed in
rck SDTM.
103 Rob 6/2.1 Can we iterate response and timing variables, or do we group Not Persuasive Unclear; may be addressed by
Sarate/Kathy and sequence or put in supplemental qualifiers? Relationships.
Giordano/Me
rck
104 Rob 12/2.2.6 RFSTDTC - need the ability to have Day 0 for older studies Not Persuasive Should be discussed with FDA
Sarate/Kathy review division.
Giordano/Me
rck
105 Rob 3 General comment - examples are inconsistent with other Persuasive Variable names have been
Sarate/Kathy examples (i.e. doesn't use variable names). If not using added to column headers in
Giordano/Me variable names, should at least use agreed-to column label. tables of examples.
rck
106 Rob 3 Study Design should be deferred to a later version, until it Not Persuasive Trial Design has received
Sarate/Kathy can be piloted, it's clearly flushed out, and stable. The enough testing so that we
Giordano/Me currently utilized critical concepts of phase/period/cycle are believe it is stable and ready
rck more clearcut and should be represented in the domains the for implementation. The
way they are reported on/subsetted/represented in the model will be updated as
database by the sponsor. These new concepts will be needed based on
confusing for the industry to interpret which will result in implementation experience.
inconsistencies or noncompliance. The data will not be able
to be subsetted utilizing the SASXPTs in the same manner
they are reported on by the sponsor. Implementation of
Study Design would require complex programming and/or
manual effort to populate the domains per protocol and
subject (minimizing reusability). Has the ADAM team
looked at this?
107 Rob 4.4.1/23 Does the FDA require submission of adjudication data in Not Related Discuss with review division.
Sarate/Kathy SDS format?
Giordano/Me
rck
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 15 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
108 Rob 2.5/11 Need clarification on which variables a sponsor can drop (i.e. Not Persuasive Addressed above.
Sarate/Kathy only permissable?)
Giordano/Me
rck
109 Rob 2.5/12 For Merck, analysis is not done from the SAS transport files. Not Related PK/PD will be addressed in the
Sarate/Kathy Would drug metabolism data now need to be sent as a future.
Giordano/Me tabulation dataset?
rck
110 Rob 3.2/16 Why are the structure, purpose and key variables needed in Persuasive Revised the IG to address this.
Sarate/Kathy the metadata?
Giordano/Me
rck
111 Rob 3.2.3/18 Including all Required Include all required There really isn't a "core" anymore. Persuasive Updated IG.
Sarate/Kathy core variables variables
Giordano/Me
rck
112 Rob 4.1.2.4/2 Why do we need to define case sensitivity? Persuasive Updated the IG to indicate that
Sarate/Kathy 0 stating case sensitivity is a
Giordano/Me sponsor decision.
rck
113 Rob 6.2.3/61 There isn't a method to reflect ongoing, since MHONGO Persuasive MHENRF has been added to
Sarate/Kathy was removed. No --ENRF as a choice of timing variables. MH as a permissible variable.
Giordano/Me Would prefer MHONGO, as that is how we collect the data, See IG.
rck rather than a derivation, which should be left to ADAM.
114 Rob 7.2.2/88 8 character ARMCD too restrictive. Also, concerned about Not Persuasive Arm description can also be
Sarate/Kathy maintenance across a submission. used for longer text. This
Giordano/Me restriction may be relaxed
rck when XML is used.
115 Rob 7.2.3/90 Inconsistency between CDISC notes and Core for TVSTRL Persuasive TVENRL has been made
Sarate/Kathy and TVENRL. Notes indicate only used under certain permissable, rather than
Giordano/Me circumstances, however this is a required variable. required.
rck
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 16 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
116 Rob 7 Same issue as SDTM. General comment - examples are Persuasive Variable names have been
Sarate/Kathy inconsistent with other examples (i.e. doesn't use variable added to column headers in
Giordano/Me names). If not using variable names, should at least use tables of examples.
rck agreed-to column label.
117 Centocor 8/2.2.2 STAT Description - Used to indicate that the Persuasive Corrected.
Used to indicate that a question was not asked.
planned intervention was
not performed.
118 Centocor 9/2.2.3 If --TESTCD uses controlled terminology, shouldn't --TEST Persuasive Corrected.
also use controlled terminology? All examples in the IG
appear to be controlled terminology.
119 Centocor 10/2.2.3 TOXCAT: not clear from the Description what type of data Persuasive Description revised.
is expected. No examples provided in IG.
120 Centocor 13/2.2.6 ARMCD Label: Arm Label: Planned Arm Code; Make consistent with IG to convey concept of Persuasive Added "Planned"
Code; Description: Short Description: Numeric planned/assigned.
8-character version of version of ARM, used for
ARM, used for sorting (Formerly TRTCD).
programming. (Formerly
TRTCD)
121 Centocor 13/2/2/6 ARM Label: Description Label: Description of Make consistent with IG to convey concept of Persuasive Added "Planned"
of Arm or Treatment Planned Arm; Description: planned/assigned.
Group; Description: Name given for the arm the
Name given to an arm or subject was assigned to
treatment group (Formerly TRTGRP).
(formerly TRTGRP).
122 Centocor 15/3.1 Is the level of granularity described in the trial design Not Persuasive Sponsor Decision; additional
datasets strictly at the sponsor's discretion? For example, in examples will be supplied in
an emergency setting where screening might be completed in future.
a matter of minutes, might the sponsor fold the screening
period into the next element?
123 Centocor 17/3.2 What does "All Variables" refer to at the start of each table? Persuasive Changed to "All Observations"
124 Centocor 17/3.2.1 Have you considered providing a glossary of commonly-used Persuasive Mod Standard values for
elements to attempt to achieve consistent usage across SUPPQUAL name code and
sponsors? EG and VS Test Codes has
been added to the
Implementation Guide as
Appendix 10.4. Other standard
vocabularies will be addressed
in the future in a separate
document.
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 17 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
125 Centocor 19/3/3/1 What does ** refer to in ELEMEND and ETCD Persuasive Removed **
Desctriptions?
126 Chris Tolk Add more info about NOT DONE rule. See 4/19 e-mail from Persuasive Implementation issue
Chris. addressed in revised IG.
127 Gail Stoner 3/1.1 Other specific changes to Other specific changes to Persuasive Corrected
the model . . . below in the model . . . Below in
section 1.1 section 1.2
128 Gail Stoner 16/3.1 Vertical lines in example are not aligned properly Persuasive Corrected
129 Gail Stoner A suggestion for consideration: we're using a variety of Persuasive Corrected
phrases to refer to the Define document in the SDTM and
IG. We should perhaps come up with a common phrase. I
like "Define document" myself.
130 Chris Tolk Add instructions for how to handle ABNORMAL with no Persuasive A new assumption has been
descriptive text. added on how to handle.
131 Procter & 3 / 1.1 Other specific changes to Other specific changes to Correction of Section number. Persuasive Corrected
Gamble the model content are the model content are
Pharmaceutic described below in described below in Section
als Section 1.1 1.2
132 Procter & 5 / 2.1 First group of bullets - A A common set of Identifier Clarification. Persuasive Corrected
Gamble common set of Identifier variables that uniquely
Pharmaceutic variables identify an observation.
als These include the
Subject….
133 Procter & 5 / 2.1 Bullet on Qualifier It was requested that an example be given. Topic and Timing Persuasive Corrected
Gamble variables variables include text "such as…", but Qualifiers does not.
Pharmaceutic
als
134 Procter & 6 / 2.1 Second to last bullet - We feel that if sponsors are using standard variable names Persuasive Corrected
Gamble The Role of the variable and definitions, and these have defined standard roles, then
Pharmaceutic (e.g., Identifier, Topic, the Role attribute can be populated in the metadata after it is
als Timing, or Qualifier), received by the regulatory agency. The Role is not used
which describes how the internally by most sponsors for any purpose. It makes more
variable is used in the sense to have one application in one location (e.g., the FDA)
dataset to populate this column than to have every sponsor doing so.
135 Procter & 7 / 2.2 Last bullet - (including a (including a general Add reference for the Comments domain. Persuasive Corrected.
Gamble general Comments Comments domain, Section
Pharmaceutic domain), 2.2.7),
als
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 18 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
136 Procter & 7 / 2.2 Consistency throughout Persuasive Corrected.
Gamble document of parenthesis
Pharmaceutic in some places & not
als others, e.g. (see Section
2.2.6)
137 Procter & 7 / 2.2 Most variables in the It would be helpful to have examples/controlled terminology Persuasive Have added where possible.
Gamble tables of this section. for all variables in this table. Some variables have them while
Pharmaceutic others do not, and there is no apparent logic to this.
als
138 Procter & 8 / 2.2.2 AESER, AESCAN, This variable needs to be better defined. What is considered Persuasive Serious adverse events are
Gamble AESOD serious? AESCAN and AESOD are no longer considered as defined in ICH guidance. A
Pharmaceutic "serious" by the FDA. note has been added explaining
als that ASCAN and AESOD are
included for old studies
conducted under earlier
definitions.
139 Procter & 8 / 2.2.2 RELOTH What data should appear in RELOTH? Is this intended to be Persuasive Corrected
Gamble a classification of a relationship (e.g., "related"), a
Pharmaceutic specification of the "other" (e.g., hit by car), or both?
als
140 Procter & 8 / 2.2.2 Pattern of e Event Pattern of Event Typo Persuasive Corrected
Gamble
Pharmaceutic
als
141 Procter & 9 / 2.2.2 Correct line formatting between variables TOXGR and Persuasive Corrected
Gamble SPREF. Appears to be a publication error.
Pharmaceutic
als
142 Procter & 10 / 2.2.3 High and Low range Reorder the following variables throughout the document so Persuasive Corrected
Gamble variable order that Lower Limit always appears before Upper Limit:
Pharmaceutic ORNRLO, ORNRHI, STNRLO, STNRHI
als
143 Procter & 11 / 2.2.4 It is stated that SEQ is a number used to insure uniqueness. Persuasive Has been redefined in the IG.
Gamble Does this need to be continuous, or can there be gaps (e.g., 1, Can now include omissions as
Pharmaceutic 2, 4, 6, 7)? long as it's unique.
als
144 Procter & 11 / 2.2.4 The use of SEQ is good in that there is one key which Persuasive The IG has been revised to
Gamble confers uniqueness to a dataset. However, this convention include metadata to describe
Pharmaceutic results in a loss of structural information, since every dataset the natural key structure as well
als now becomes one record per sequence number, since other as the standard SEQ key
variables such as visit, test, and location are no longer structure..
primary keys.
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 19 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
145 Procter & 11 / 2.2.5 VISITNUM It's not clear whether VISITNUM can be negative. If Not Persuasive Sponsors would need to
Gamble VISITNUM is allowed to be negative, and subevents are to address this in their operational
Pharmaceutic be indicated by a decimal place (e.g., 1.1) , the sorting database, and should be aware
als becomes an issue, and sponsors would need to realize this. of numeric sort order.
For example, if there was a Visit -1, and then an unscheduled
visit after that, it could not be Visit -1.1, since this would
sort before Visit -1. The unscheduled visit would have to be -
0.9 or a value between -1 and zero so that it would sort after
Visit -1. The alternative is to capture subevents in a separate
field.
146 Procter & 12 / 2.2.5 --STRF should only be --STRF should only be Some people were not sure what values were to be used to Persuasive Corrected
Gamble populated when a start populated when a start date populate these variables, the ones at the beginning of the
Pharmaceutic date is not collected but is not collected. If note (not in quotes) or the ones at the end (in quotes). The
als information such as information such as suggested wording change makes it more clear that the
"PRIOR", "ONGOING", "PRIOR", "ONGOING", or values in quotes are from a CRF.
or "CONTINUING" was "CONTINUING" was
collected. This collected, this information
information should be should be translated into --
translated into --STRF. STRF.
--ENRF should only be --ENRF should only be
populated when a end populated when an end date
date is not collected but is not collected. If
information such as information such as
"PRIOR", "ONGOING", "PRIOR", "ONGOING", or
or "CONTINUING" was "CONTINUING" was
collected. This collected, this information
information should be should be translated into --
translated into --ENRF. ENRF.
147 Procter & 12 / 2.2.6 Have Table 2.2.6 Demographics Identifier variables listed in Persuasive Corrected
Gamble the same order as Table 2.2.4 Unique Identifiers.
Pharmaceutic
als
148 Procter & 13 / 2.2.6 AGE Description: If AGE is collected on the Durations are not to be included in submission datasets Not Persuasive If AGE is collected on the Case
Gamble Usually derived as Case Report Form, it should unless they are collected. Age is a duration (of life) and Report Form, it should appear
Pharmaceutic RFSTDTC-BRTHDTC, appear here. If AGE can be should be treated as such. here. If AGE can be derived
als but BRTHDTC may not derived from RFSTDTC from RFSTDTC and
be available in all cases and BRTHDTC, then it BRTHDTC, then it should not
(due to subject privacy should not appear here. appear here.
concerns).
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 20 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
149 Procter & 13 / 2.2.6 COUNTRY Clarify that COUNTRY variable is for country of the Persuasive Corrected
Gamble investigational site and not the subject's residence.
Pharmaceutic
als
150 Procter & 14 / 2.2.7 SEQ in Comments The use of SEQ is good in that there is one key which Persuasive This issue has been addressed
Gamble confers uniqueness to a dataset. However, this convention by changes to the Relates
Pharmaceutic has a downside in that it results in a loss of structural model and the Logical Key
als information, since every dataset now becomes one record per Order in 3.2.
sequence number, and other variables such as visit, test, and
location are no longer primary keys
151 Procter & 16 / 3.1 Correct formatting in Arm, Drug, Cycle chart just before Persuasive Corrected
Gamble "Epochs" subsection.
Pharmaceutic
als
152 Procter & 17 - 19 / Make Tables 3.2.1, 3.2.2, 3.2.3, and 3.3.1 consistent in Persuasive Corrected
Gamble 3.2.1, listing of variables Codes and Descriptions. In TE, the
Pharmaceutic 3.2.2, ELEMENT appears before ETCD, but in TA, ARMCD
als 3.2.3, appears before ARM., while in TV, ARM appears before
3.3.1 ARMCD.
153 Procter & 17 / 3.2.1 TENRL TEENRL Typo Persuasive Corrected
Gamble
Pharmaceutic
als
154 Procter & 20 / 3.4 Make Table 3.4.1 Trial Inclusion/Exclusion Table fall into Not Persuasive SDS team felt this table was in
Gamble the Trial Tables prior to the Subject tables. This would keep the proper position; and that
Pharmaceutic all Trial tables together. the Arm/Element/Visit tables
als belonged together.
155 Procter & 10/ 2.4 Surgery is mentioned as an intervention example. It would be Persuasive Corrected
Gamble beneficial to reserve (Section 10.3) and define a domain for
Pharmaceutic this.
als
156 Procter & 9/2.2 Correct line spacing in last two lines of the paragraph Persuasive Text is being corrected.
Gamble immediately above the bullets.
Pharmaceutic
als
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 21 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
157 Procter & 10/ 2.4 The Interventions class The Interventions class Would prefer the definition from the SDTM (2.2) be used in Persuasive Corrected
Gamble captures investigational captures investigational the IG. As written, the IG text includes concomitant
Pharmaceutic treatments, therapeutic treatments, therapeutic medications under the umbrella of "intentionally
als treatments, and surgical treatments, and surgical administered to the subject".
procedures that are procedures that are
intentionally intentionally administered to
administered to the the subject (usually for
subject (usually for therapeutic purposes) either
therapeutic purposes) as specified by the study
either as specified by the protocol (e.g., ―exposure‖),
study protocol (e.g., coincident with the study
'exposure') or coincident assessment period (e.g.,
to the study assessment ―concomitant
period (e.g., 'concomitant medications‖), or other
medications') substances self-administered
by the subject (such as
alcohol, tobacco, or
caffeine).
158 Procter & 19/ Third bullet. Minor, but "A" and the space following it should not be Persuasive Corrected
Gamble 4.1.1.5 underlined in the third bullet in order to be consistent with
Pharmaceutic the first two bullets.
als
159 Procter & 42-43/ EXDOSU, All have references to "CM" variables in the Notes column. Persuasive Corrected
Gamble 6.1.2 EXDOSFRM,
Pharmaceutic EXROUTE
als
160 Procter & 90/ 7.3 Should there be some explanation that Subject Arms are Persuasive Section 7.3 was updated to
Gamble captured in Demographics and therefore a Subject Trial Arm explain that subject ARM
Pharmaceutic Table is not required. Also relates to STDM Section 3.3. information is included in the
als Demographic domain.
161 Procter & 33/5.5.1 ETHNIC. COUNTRY Add asterisk in front of text in Controlled Terms or Format Persuasive Corrected
Gamble Column.
Pharmaceutic
als
162 Procter & 35/5.1.2 STUDYID Remove errant "-" from Controlled Terms or Format Persuasive Corrected
Gamble Column.
Pharmaceutic
als
163 Procter & 138 / SUDOSTXT variable - add NON for Row 1. Persuasive Corrected
Gamble Example
Pharmaceutic 4
als
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 22 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
164 G. 89/7.2.3 One Record per Visit per EPSEQ is not shown as a variable in TV. Persuasive Corrected
Amato/Bayer Arm per Epseq (header
of table)
165 G. 87/7 It seems it would be useful to have the EPOCH variable in Future This may be added in the
Amato/Bayer the TV dataset, to show to which epoch each visit belongs. future.
166 S. 7 / 2.2.1 For concomitant medications the instructions say to fill -- Future See duplicate comment in CT-
Drost/Bayer DECOD with the drug's generic name. This is easy for 02 above.
single ingredient drugs, but what should be entered for
multiple ingredient drugs?
167 T.Petrowitsch 9/2.2.2 the border in the table between variables --TOXGR and -- Persuasive Corrected
/Bayer SPREF is missing
168 T.Petrowitsch 11/2.2.5 Need to apply timing variables to all event domains where Not Persuasive Any subject date can be
/Bayer there are start and stop dates. We need timing variables for associated with an epoch as
the end dates as well as the start dates, for example, epoch long as Subject Elements data
related to start date and epoch related to end date. is submitted. It is more
appropriate to let the agency
calculate start and end epoch
variables, where they wish,
than for sponsors to submit
these derived variables.
169 T.Petrowitsch 11/2.2.5 Why are no variables for 'relative days to end of treatment ' Not Persuasive These variables are considered
/Bayer created? Like AESTDY and AEENDY analysis variable that aren't
included in the tabulations
datasets.
170 T.Petrowitsch 110/7.4.2 See Section 7.4.2 for a See Section 7.7.2 for a Persuasive Corrected
/Bayer discussion of unplanned discussion of unplanned
Elements Elements
171 T.Petrowitsch 111 / .... Identify the moment .... Identify the moment Persuasive Corrected
/Bayer 7.6.2 where one visit stops and where one visit stops and
another ends. another starts.
172 T. 89/ 7.2.2 Add * to ELEMENT, EPOCH and ARMCD, etc Persuasive Added.
Westermeier/
Bayer
173 C. 11/ 2.2.5 Numeric version of Numeric version of planned Persuasive Corrected
Tolk/Bayer planned time point used time point used in sorting.
in sorting; may be a
sequence number or an
elapsed time.
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 23 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
174 C. 20/3.4.1 The trial The trial Persuasive Corrected
Tolk/Bayer Inclusion/Exclusion Inclusion/Exclusion Domain
Domain (TI) contains… (TI) is not subject oriented.
It contains one record for
each of the inclusion….
175 C. 12/ 2.6 In creating a new domain, can you choose from the Persuasive Added 3d to this section.
Tolk/Bayer interventions class which variable to use? Where does it
explicitly state that?
176 C. 50/ 6.2.1 Origin AEPATT: CRF Origin AEPATT: CRF Persuasive Corrected
Tolk/Bayer or Sponsor Defined
177 C. 52/ 6.2.1 Correct variable name to AEENDTC Persuasive Corrected
Tolk/Bayer
178 C. 99/ 7.4.2 Add outside borders to table Persuasive Corrected
Tolk/Bayer
179 C. 115/ 7.9 The trial The trial Persuasive Corrected
Tolk/Bayer Inclusion/Exclusion Inclusion/Exclusion Domain
Domain (TI) contains… (TI) is not subject oriented.
It contains one record for
each of the inclusion….
180 C. 116/ 7.10 Identify the inclusion and Identify the inclusion and Persuasive Corrected
Tolk/Bayer #9 exclusion criteria. exclusion criteria to be able
to populate the TI domain.
181 C. 123/ 8.4 For the variable EVAL, remove the single paren after Persuasive Corrected
Tolk/Bayer DATABASE and before ADMINISTRATOR
182 C. 124/ Row with Domain QS, remove the "?" after all Persuasive Corrected
Tolk/Bayer 8.4.2
183 C. 162/ 10.1 Change from Bayer to Bayer HealthCare Persuasive Corrected
Tolk/Bayer
184 C. 163/ 10.2 Add Domain and Dataset. Provide link to CDISC Web site Persuasive Corrected
Tolk/Bayer for glossary.
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 24 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
185 Fima Page 6, In the description of the Intervention class of observations, Persuasive Corrected
Gershteyn. Section prior medications are not covered, as they are not addressed
Takeda 2.2, first in the three cases given: ―… that are intentionally
Global Bullet. administered to the subject (usually for therapeutic purposes)
Research and either as specified by the study protocol (e.g., ―exposure‖), -
Development (prior medications are not specified by study protocol),
(TGRD) coincident with the study assessment period (e.g.,
“concomitant medications”) – they are not necessarily
concomitant, or other substances self-administered by
the subject (such as alcohol, tobacco, or caffeine) –
they are not necessarily self-administered in the
sense implied.
186 Fima Page 8, A fifth type of variable role ―Rule‖ is introduced; however it Persuasive Corrected
Gershteyn. 1st is not mentioned in the Study Data Tabulation Model
TGRD paragraph (Release 2) document.
after the
fourth
bullet
187 Fima Page 9, The metadata attribute Role of the variable does not seem to Persuasive Document updated to state that
Gershteyn. 6th add substantial value to the variable description, especially as Role need not be submitted
TGRD bullet. multiple role designations are allowed. In our opinion, except for sponsor extensions.
requirement to classify a variable in four categories and five
types of Qualifier category adds artificial complexity to the
description without real benefits.
188 Fima Page 11, How does this statement relate to Trial Design Domains and Persuasive Added "data" before domain.
Gershteyn. 1st Special-Purpose Relationship Data Sets, as this information Decision on whether to submit
TGRD Sentence is not collected, but rather derived from the study and data TD must be discussed with
after the set designs? Is the submission of the Trial Design Domains FDA.
last bullet considered as a part of a regular study data submission, as it
– does not necessarily pertain to Case Report Tabulations?
―Sponsor
would
only
submit
the
domains
that are
actually
collected.
‖
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 25 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
189 Fima Page 37 Concomitant Medications Domain. Inclusion of an indicator Not Persuasive CMSTRF and CMENRF are
Gershteyn. distinguishing between concomitant and prior (ended before used to indicate prior
TGRD the reference point) medications would be very useful. medications.
190 Fima Page 58. Medical History Domain. Inclusion of an indicator Not Persuasive MHSTRF and MHENRF are
Gershteyn. distinguishing between the events, which have ended before used to indicate timing of
TGRD the study reference point and those which were present at events relative to the study
that point, would be very useful. reference point.
191 Apotex Demogra ETHNIOTH Missing variable name ETHNIOTH for Other Ethnicity Not Persuasive SUPPQUAL should be used
Research Inc. phics- for Other Ethnicity.
DM
192 Apotex AE - page SDSSDSAEENDTC AEENDTC Incorrect Variable Name Persuasive Corrected
Research Inc. 52
193 Apotex Medical Medications Events CDISC Notes for variable AEENRF should be changed from Persuasive Corrected
Research Inc. History- "…Medications that are ongoing at the end of the
MH reference period…" to "…Events that are ongoing at
the end of the reference period…"
194 Apotex MH MHONGO MHONGO referred to on page 61 is missing from table on Persuasive MHONGO is not intended to
Research Inc. pages pages 58-59. be in the MH domain.
58,59,61 References to the variable were
errors and have been removed.
195 Apotex PE CRF CRF or Derived Should the Origin for PETESTCD be changed from "CRF" Persuasive Corrected
Research Inc. to "CRF or Derived"?
196 Apotex PE PEDECOD Variable Name PEDECOD is missing. Not Persuasive PEDECOD is not necessary in
Research Inc. PE, because the
TESTCD/TEST is not
intended to be a verbatim term,
but a question.
197 Apotex PE PETESTCD PETEST CDISC Notes for variable PETEST should be changed from Persuasive Corrected
Research Inc. "…The value in PETESTCD cannot.." to "The value in
PETEST cannot.."
198 Apotex PE - page Should PESTAT contain NORMAL, ABNORMAL in Not Persuasive PESTAT is intended to be used
Research Inc. 75 addition to NOT DONE in Controlled Terms or Format? (see only to indicate NOT DONE.
page 75)
199 Apotex PE Core is Req for VISITNUM on page 75, however on Not Related VISITNUM is correct as is.
Research Inc. page 68 Core is Perm for the same variable. Clarify? VISITNUM is not usually
collected for IE, but often
expected with PE.
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 26 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
200 Apotex PE Core is Req for PEDTC on page 76, however on page Not Persuasive Correct as is. Not always
Research Inc. 59 Core is Perm for MHDTC. Is this correct? collected for MH.
201 Bernd 8 / 2.2.2 This issue/question also applies to the IG as well all variables Persuasive Further definition of what can
Doetzkies / with a discrete set of values. When a specific list of discrete be sponsor-defined is being
Purdue values are specified in the documents for a variable, are those provided in the IG. The
Pharma L.P. values suggestions / examples and open for the sponsor to SDTM can not specify this
define (when not dictated by an external published source), information since it varies by
or are those the expected values? For example, in the domain, and the SDTM
description for variable --REL and --OUT, specific values are addresses general classes, not
listed. Our current standard practice does not collect the specific domains.
exact number of categories nor the exact wording as
specified in the document. Our recommendation is to have
all discrete sets of values be sponsor defined unless an
industry and/or regulatory prescribed set of values are
required.
202 Bernd 9 / 2.2.3 This issue/question also applies to the IG. Although there is Persuasive There are many instances
Doetzkies / a variable (--LOINC) permitted to store the LOINC Code for where the sponsor may choose
Purdue lab tests, there is no generic variable to store company lab to have an internal code list
Pharma L.P. test codes. Our current standard practice to is to associate (generic test code variable)
internally defined lab test codes for purposes of grouping and
sorting lab test results within our standard reporting system. Use TEST and TESTCD for
Our recommendation is to have a generic Lab Test Code sponsor defined codes.
variable and then use the Comment variable in the data
definition document to specific the source of the lab test
codes. This would be consistent with how other domains are
handled.
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 27 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
203 Bernd 12 / 2.2.6 This issue / question also relates to the IG, page 26 section Not Persuasive This issue is addressed by trial
Doetzkies / 4.1.4.7. By defining Subject Reference Start and End dates / design elements or disposition
Purdue times in the Demographics domain, it is not clear how events.
Pharma L.P. multiple reference start and end points would be defined
within the SDTM. Also, this static approach to handling start
/ end points could be problematic when sponsor companies
submit data at various predetermined cut-off points. For
example when a study is defined to have a Core (double-
blind) and Extention (open-label) phase, subject data could
be submitted for the Core phase as part of NDA, and then
later re-submitted (e.g., ISS) with the reference end-date now
extended beyond the Core phase (double-blind). There are
references in the document that multiple start and end points
can be specified, but it is not clear how that would be set up
within the SDTM. Our recommendation is to either remove
the RFSTDTC and FRENDTC from the Demographics
domain and create a data structure specifically for reference
time points, or to provide additional documentation and
examples within the IG.
204 Bernd 18 / 3.2.3 How would values for VISITDY be calculated when visits / Not Persuasive This is sponsor defined. You
Doetzkies / clinical events are defined to occur within a time window? would not have to use
Purdue For example, a follow up visit is scheduled to occur between VISITDY in this case.
Pharma L.P. day 3-5 post last dose.
205 Bernd 21 / We currently use mixed case for some of the controlled Persuasive Assumption 4.1.3.2. has been
Doetzkies / 4.1.3.2 terminology associated with variables because those values modified to reflect the decision
Purdue are used to generate in-text tables within the Clinical Study that mixed case can be used.
Pharma L.P. Report. We would like to recommend that mixed case be
permissible so that multiple decode lists do not have to be
maintained.
206 Bernd 22 / It is not clear why --CLAS would be not be populated when Not Persuasive Already addressed in 4.1.3.5;
Doetzkies / 4.1.3.5 multiple classes are available. Current standard practices is to multiple classes can be
Purdue select / specify the drug class because it is used in reporting / submitted in analysis datasets.
Pharma L.P. analyses. Our recommendation is to keep the values in as
coded by the sponsor companies.
207 Bernd 22 / 4.3.7 Our standard practice is to store the full "Yes / No" values in Not Persuasive Y and N has been adopted as
Doetzkies / the database. Our recommendation is that the full value for standardized terminology for
Purdue variables with controlled terms be acceptable. consistency across all sponsors.
Pharma L.P.
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 28 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
208 Bernd 24 & 25 / Nominal time is usually represented as durations and not Not Persuasive Please clarify: if referring to
Doetzkies / 4.1.4.2 & actual times (e.g., 15-30 minutes post dose, 31-60 minutes TPT, can use a text description
Purdue 4.1.4.3 post dose, etc.). It is not clear from the table on page 24 like 15-30 minutes post dose.
Pharma L.P. where examples of Nominal Date/Time are given and the You'd only use ISO8601 to
table on page 25 where examples of durations are given, if represent actual times or
nominal times collected as durations can be specified in a durations with a start or end
T15M/T30M type format. Please clarify. time.
209 Bernd 22 / It is not clear why multiple levels are not allowed in the Not Persuasive The FDA currently would like
Doetzkies / 4.1.3.5 SDTM particularly since regulatory agencies other than the to see only the single level.
Purdue FDA might be interested in having them included. Is the
Pharma L.P. expectation that a non-standard dataset be created and then
linked back to the standard dataset via the mechanism within
the model used to represent relationships among datasets and
records. This seems like it would be substantially more work
to set up.
210 Bernd 22 / If a company does not use the default path (e.g. MedDRA) Not Persuasive This is not represented in the
Doetzkies / 4.1.3.5 how and where would that be documented and stored? SDTM because it was not
Purdue required by FDA.
Pharma L.P.
211 Bernd 87 / 7.1 This applies to the SDTM sections covering Trial Design Not Persuasive Sponsor can define what is
Doetzkies / Datasets as well. The definition and use of "visits" does not meant by clinical encounter
Purdue seem very meaningful for Phase I trials. (which is what VISIT
Pharma L.P. represents).
212 Bernd 22 / The sponsor may be Is the expectation that either the regulatory agencies will Not Persuasive This issue should be discussed
Doetzkies / 4.1.3.5 required to submit the have licenses for all dictionaries or will sponsors have to with the FDA.
Purdue dictionary if it is not account for this within contracts with vendors?
Pharma L.P. already available to the
reviewer.
213 Bernd 22 / … If the verbatim term --MODIFY is not the best suffix to identify modified Not Related This is a good suggestion, but
Doetzkies / 4.1.3.6 is modified to facilitate variable text since it is usually used to flag a record-level we don't want to change these
Purdue coding, the modified text change (such as in legacy data conversion) and not an variable names due to
Pharma L.P. is stored in --MODIFY. individual variable-level change. Also, use naming backward compatibiilty issues
conventions that allow intuitive associations between with prior versions.
variables. For example, use AETERM and AEMTERM (not
AETERM/AEMODIFY), CMTRT and CMMTRT (not
CMTRT/CMMODIFY), MHTERM and MHMTERM (not
MHTERM/MHMODIFY). Reserve the --MODIFY suffix
for record level changes.
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 29 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
214 Bernd 49/6.2.1; Variable name: Use intuitive variable naming. Change AEDECOD to Not Persuasive This suggestion is not
Doetzkies / 38/6.1.1; AEDECOD (pg 49), AEPT, CMDECOD to CMSYN, and MHDECOD to MHPT consistent with the concept
Purdue 58/6.2.3 CMDECOD (pg 38), to reflect actual dictionary conventions. If MHTERM is behind the SDTM general
Pharma L.P. MHDECOD (pg 58). coded to the CRF page terms, then store this information in a classes.
variable separate from MHPT. The suffix --COD indicates a
code value; this is incorrect since the data being stored is
derived text description (see Variable Label description).
215 Bernd 50/6.2.1 Variable name: Use intuitive variable naming. Change to AESOC to reflect Not Persuasive This suggestion is not
Doetzkies / AEBODSYS. actual dictionary conventions. AEBODSYS is residual consistent with the concept
Purdue conventions from Costart. behind the SDTM general
Pharma L.P. classes.
216 Bernd 37 / 6.1.1 Domain Model: Although it is documented that the intent of this domain is to Not Persuasive Therapies would be presented
Doetzkies / Concomitant store data on Concomitant Medications and Therapies, the in a separate domain; a
Purdue Medications standard clearly is bias toward Medications. A "Concomitant Surgical interventions domain
Pharma L.P. Therapy" designation and model structure would be clearer will be defined in the future.
and easier to process. For example, only one set of variables
are defined for encoding, however two or more dictionaries
may be used to encode the values collected on the CRF (e.g.,
WHO for medications, MedDRA for other therapies). The
comments field for the Data Definition Document will have
to list the multiple dictionaries used for a single set of
variables and from a technical perspective, how would a
system alternatively link back to the corresponding
dictionary depending on the values contained within
variable. Current commercially software does not support the
linking of multiple dictionaries to a single variable. This
would have to be handled as part of an
Extract/Transformation/Load process in creating the SDTM
files.
217 Bernd General PKPD Data Are there any plans to develop standards to support PKPD Not Related Yes, PK/PD domains will be
Doetzkies / data, or is the SDTM sufficiently robust to support this developed in the future.
Purdue already? There is the Systems Biology Markup Language,
Pharma L.P. which is an XML standard, that could be used to transfer
PKPD data (see www.sbml.org/docs/ for additional
information).
218 Bernd 17 & 18 / Are there any plans for future releases to structure the rules Not Related Yes.
Doetzkies / 3.2 (TESTRL, TENRL, TABRANCH, TATRANS, TVSTRL,
Purdue TVENRL) so that they could be used programatically?
Pharma L.P.
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 30 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
219 Alain Smits / TPT and TPTNUM variables are important keys like other Not Related Taken under advisement; May
JJPRD timing variables VISIT, VISITNUM. These variables should define a Trial Design dataset to
not have prefix. model planned TPTs.
220 Sandy Lei / Open-Label extension trial with different studyid, what is the Not Persuasive Sponsor Defined. Could use
JJPRD recommendation for RFSTDTC in terms of first intake of Relates to use reference dates
drug (is this based on USUBJID or based on STUDYID)? for both studies.
Same concept can be applied to ARM?
221 Centocor 9/2.2 Each dataset is Each dataset is Persuasive Corrected
distinguished by a distinguished by a unique,
unique, two-character two-character DOMAIN
DOMAIN code that code that should be used
should be used consistently throughout the
consistently throughout submission. This DOMAIN
the submission. code is used as the dataset
name, the value of the
DOMAIN variable within
that dataset, and as a prefix
for most variable names in
the dataset.
222 Centocor 15/3.1 The wording regarding the permissibility of null values for Persuasive Corrected
and expected variables is slightly different in these 2 sections. If
19/4.1.1. a sponsor does not collect or derive an expected variable,
5 then all values will necessarily be null. Based on 3.1, this
situation is OK; based on 4.1.1.5, it is not. It does not seem
reasonable to "force" sponsors to collect qualifier or timing
data that they would not ordinarily collect in order to
populate an "expected" variable and thus be in compliance
with the model.
223 Centocor 20/4.1.1. Only variables specified Only qualifier variables Persuasive Corrected
5 in a domain model are specified in a domain model
permissible for that are permissible for that
domain. domain.
224 Centocor 17/3.2.3 Can CDISC offer any guidance on how to handle datasets Not Persuasive There are no longer limits on
that exceed the FDA 50 MB size limitation? the file size.
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 31 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
225 Centocor 17/3.2.3 Can CDISC offer any guidance on how to annotate the CRF Future This may be added in the
for variables that have been significantly transformed from future.
their original appearance (e.g., an Ongoing checkbox
changed to --ENRF=AFTER)?
226 Centocor 20/4.1.2. The Define data Prefix paragraph at left CDISC implies, but does not clearly recommend that all Persuasive Corrected
4 and definition document with: It is recommended that textual data be submitted as upper case other than the
21/4.1.3. should indicate as a textual data be submitted in exceptions noted in 4.1.3.2.
2 general note or upper case text.
assumption whether case
sensitivity applies to text
data for any or all
variables in the dataset.
If case sensitivity differs
across variables, it
should be documented at
the variable level in the
define document.
227 Centocor 21/4.1.3. Perhaps this will be covered in future guidance, but if Persuasive When examples of controlled
1 CDISC provides specific controlled terms that are not terminology are provided
labeled as examples (e.g., M, F, and U for SEX), is a sponsor within the domains, those
required to use those values in submissions? values are expected to be used.
A new assumption has been
added to clarify.
228 Centocor 25/4.1.4. Please clarify use of nW. Is it to be used in place of Persuasive See ISO 8601 documentation
3 nYnMnDTnHnMnS or just the nYnMn portion? The table to see how to use nW for
shows an example of '2W3D'. durations.
229 Centocor 29/4.1.5. In line 7 of the laboratory examples, why isn't <10,000 Persuasive The sponsor must choose what
1 shown as the value for LBSTRESC? value to put in this field. Some
my choose 10,000, 9,999 or
other values.
230 Centocor 6/many SPREF variables are referencing SDTM 2.4 but --SPREF is Persuasive Variable name has been
datasets not discussed in section 2.4 (but probably should be) changed to --SPID and has
been added to the table in
2.2.4.
231 Centocor 6/many The CDISC Note for many --DTC, --STDTC, and --ENDTC Persuasive "Use if collected" has been
datasets variables is "Use if collected." but this is often inconsistent deleted.
with the Core category of "Req" or "Exp." Suggest that "Use
if collected" be deleted.
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 32 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
232 Centocor 37/6.1.1 CM.xpt, Concomitant CM.xpt, Concomitant Meds As per assumption 1b. Persuasive Corrected
Meds — Interventions, — Interventions, Version
Version 3.1, March 15, 3.1, March 15, 2004. One
2004. One record per record per medication
medication per subject, intervention episode per
Tabulation subject, Tabulation
233 Centocor 38/6.1.1 CMOCCUR CDISC Used when the use of Persuasive Corrected
Note: Used when the use specific medications is
of specific medications is solicited. Values are Null
solicited. Values are Null for spontaneously reported
for spontaneously medications.
solicited events.
234 Centocor 42/6.1.2 Should EXTRT use controlled terminology? Not Persuasive EXTRT is sponsor defined; it
may used controlled
terminology but that is not a
requirement.
235 Centocor 43/6.1.2 EXLOC CDISC Note: Specifies anatomical Persuasive Corrected
Can be used to specify location of administration.
where drug was Example: LEFT ARM for a
administered. Example: topical application.
LEFT ARM for a topical
application.
236 Centocor 44/6.1.2. Please provide an assumption clarifying the phrase "constant Persuasive Text has been added to
0 dosing interval." Does this mean to submit a single record as Assumptions to explain that
long as there were no missed doses from the planned granularity is sponsor defined.
regimen? Is daily dosing implied? How would a regimen of
1 dose every 8 weeks for 2 years be submitted - a record for
each dose or a single record for the entire 2 year period as
long as no doses were missed? What if individual dose times
are captured?
237 Centocor 45/6.1.3 Dataset description indicates 1 record per substance type per Persuasive Added 'per visit' to domain
subject, but assumptions discuss use in efficacy evaluations description
implying multiple records per substance.
238 Centocor 46/6.1.3 SUOCCUR CDISC Used when the use of Persuasive Corrected
Note: Used to indicate specific substances is
existence of a SU event solicited. Values are Null
when no additional detail for spontaneously reported
information is available. substances.
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 33 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
239 Centocor 47/6.1.3. In item 3c, what is WHODRL? Persuasive Corrected
0
240 Centocor 48/6.1.3. Description of SUSTAT in item 7b is inconsistent with Persuasive Description revised.
0 CDISC Notes for this variable.
241 Centocor 50/6.2.1 AEOCCUR CDISC Used when the occurrence Persuasive Corrected
Note: Used when the of specific adverse events is
occurrence of specific solicited. Values are Null
adverse events is for spontaneously reported
solicited. Values are Null events.
for spontaneously
solicited events.
242 Centocor 58/6.2.3 MHOCCUR CDISC Used when the occurrence Persuasive Corrected
Note: Used when the of medical history events is
occurrence of medical solicited. Values are Null
history events is for spontaneously reported
solicited. Values are Null events.
for spontaneously
solicited events.
243 Centocor 59/6.2.3 MHONGO has been dropped. Should MHSTREF and/or Persuasive MHENRF has been added to
MHENREF be added as permissible vars? Also see 6.2.3.0, MH as a permissible variable.
item 5 on page 61. See IG.
244 Centocor 62/6.3 Why are --DTC variables required on Findings domains Persuasive In IE and SC, --DTC have been
rather than expected? If the test was not done, then no date changed to permissible.
will be available. IE and SC are collected 1x per study like
DM - DMDTC is permissible.
245 Centocor 70/6.3.3 CDISC Note for LBSPCCND indicates that free or Persuasive * has been removed in
standardized test can be used, but controlled terminology is Controlled Terms or Format
indicated. column.
246 Centocor 72/6.3.3 Provide sample values for LBTOXCAT and/or assumption Persuasive Added text "Example: NCI
describing use. CTCAE short name." And
changed --TOXGR to
LBTOXGR.
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 34 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
247 Centocor 75/6.3.4 Each record should have If there are findings for a Persuasive Corrected
a value in this variable. If body system, then either the
there are findings for a dictionary preferred term (if
body system, then either findings are coded using a
the dictionary preferred dictionary) or PEORRES (if
term (if findings are findings are not coded)
coded using a dictionary) should appear here. If
or PEORRES (if PEORRES is Null,
findings are not coded) PESTRESC should be Null.
should appear here. If
PEORRES is Null,
PESTRESC should be
Null.
248 Centocor 80/6.3.6 SC.xpt, Subject SC.xpt, Subject Persuasive Corrected
Characteristics — Characteristics — Findings,
Findings, Version 3.1, Version 3.1, March 15,
March 15, 2004. One 2004. One record per
record per subject characteristic per subject,
characteristic, Tabulation Tabulation
249 Centocor 81/6.3.6 Why would SCSTRESN and SCDTC be expected and Persuasive SCSTRESN and SCDTC have
required rather than permissible? If the only subject been changed to permissible.
characteristics collected were eye color and initials, neither Added to assumption that the
of these vars would be very meaningful. Also, the data is collected once per
assumptions do not mention the concept that this is data that subject.
is collected once per subject and does not change during the
trial as noted in the example in section 9.4.6.
250 Centocor 84/6.3.7 Why is VSPOS expected? Not relevant for all tests (e.g., Persuasive Relevant for blood pressure;
BMI, temperature) typically included with Vitals;
so it would be expected for at
least some of the rows in VS.
251 Centocor 117/8 Why must --SEQ values be sequential with no numbers Persuasive Implementation issue
omitted within a USUBJID/Domain? If --SEQ values addressed in revised IG.
originate in the CDMS and records are deleted during data Numbers may now be omitted.
cleaning, there could be gaps. Why would there be a
requirement to renumber the records as long as they are
uniquely identified?
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 35 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
252 Centocor 124/8.4.2 Why would information about lab specimen quality be added Persuasive Example will be deleted.
as an additional test when the LBSPCCND variable is
available?
253 Centocor 125/8.5 Comments related to Comments related to Persuasive Corrected
specific SDS domain specific SDS domain
records for a subject records for a subject would
would populate the populate the DOMAIN and
DOMAIN, SEQ, and either SEQ or GRPID with
GRPID with values values from the parent
from the parent record(s).
record(s).
254 Centocor 137/9.2.3 Example 4, row 1 does not have any indication that subject Persuasive The example has been
1234005 was a non-smoker. How is this to be modeled? modified to include a value of
"non" for non-smokers
255 Centocor 140- DS example data is mixed up. Julie - Gail corrected in Word Persuasive Corrected
141/9.3.2 version of IG and sent to Events Team for review.
256 Centocor 141/9.3.3 Not all variables Not all variables specified in MHSTDTC is an expected variable. Need to add to sample Persuasive Corrected
specified in the CDISC the CDISC MH domain are data (with null values).
MH domain are present present on the CRF (e.g.,
on the CRF (e.g., MHENDTC). Since they are
MHSTDTC, not used, the variables are
MHENDTC). Since they not present on the MH
are not used, the dataset. However, all
variables are not present required variables are
on the MH dataset. present and populated.
However, all required
variables are present and
populated.
257 Centocor 143/9.4 Please provide example for IE. Persuasive Example has been added
258 Centocor 146/9.4.2 Suggest that a sample row be included for categorical/ordinal Persuasive Example has been added.
test result (e.g., Urine Protein with value of 1+).
259 Thomas 11/2.2.5 Applies to IG as well.This section describes timing variables Persuasive Additional statement added to
Schneider / which are document.
Roche to be added to all classes of data sets. However, templates in
IG do not include all variables e.g. --ENDRF. It should be
explicitly stated in either document whether the Sponsor is
supposed to add these variables if needed or not.
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 36 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
260 Thomas 11/2.2.3 Applies to IG as well. Sponsor reference number, --SPREF, Persuasive Corrected; now --SPID
Schneider / is (formerly --SPREF) is Char
Roche char in SDTM, but number in IG. This inconsistency should throughout.
be resolved. In general, identifiers of records in data sources
should be character since many database systems use non-
numeric keys. The same applies to landmarks on CRFs.
261 Norbert Fritz 11/2.2.4 The field --SEQ is defined as sequence number to ensure Not Persuasive --SEQ is used for sorting and
/ Roche uniqueness of records within dataset for a subject. If this thus should remain as a
field, --SEQ, is used to ensure uniqueness only and is not numeric.
used for sorting during data display it could be defined as
character as well. This data type is preferred since it eases
the generation of this parameter during data extraction from
Clinical Database systems which use character rather than
numeric values to identify records.
262 Chryl Kubsch Generally speaking, a comprehensive Future This may be added in the
/ Roche example is needed. The example should begin with a future.
"typical study" CRF annotated with the SDS V3.1 domain
names and variable names. Data for each domain should be
provided especially to show how to relate records from
different domains to one another. A narrative should walk
the reader through examining the special cases for the data in
each domain with references back to the SDTM section
which documents that domain. For instance pointing out
that, in the adverse event dataset, patient 12 had AE
headache and took ibuprofen for the headache.
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 37 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
263 Gerry Page Controlled terminology Controlled terminologies in Controlled terminologies should be Future Not possible at this time.
Petratos / 22/Sectio or text should be used HL7 version 3 are handled similarly to how the HL7 version 3 RIM handles Being addressed by a separate
Roche n 4.1.3.4 instead of, or in addition accommodated by the controlled terminologies. terminology group.
to, arbitrary number Reference Information
codes in order to reduce Model (RIM) and their
ambiguity for submission representation of various
reviewers. For example, dictionaries, including the
for concomitant coded terms used, should be
medications, the concluded. For instance,
verbatim term and/or unique identifiers from
dictionary term should be various controlled
presented, rather than terminologies should be
numeric used in addition to textual
codes. Separate coded representations to reduce
values should not be ambiguity for submission
included in the reviewers. The dictionary to
Tabulation dataset, but be referenced by a particular
may be submission dataset, whether
necessary in analysis it is in a Tabulation or
datasets. Analysis dataset, should be
provided to the reviewers.
Since the preferred term is
often not enough to
uniquely identify a medical
concept (e.g. as in the case
of drug terms in WHO-Drug
which require multiple
additional fields like generic
name, salt, and ATC code)
the unique identifier of the
term should be
264 Ole Jeppesen 43 systematically included. Regarding the EXLOT variable: This ought to be a character Not Persuasive The variable is now CHAR.
/ LEO variable as I have come across numrerous lot numbers that
Pharma contained characters, in our present FDA filing.
265 Richard 32 Regarding the INVID variable: In our clinical trials we often Not Persuasive This is a Sponsor operational
Foster / LEO have several investigators at a study site: Could the updated decision. One typical approach
Pharma v. 3.1. accomodate for this. is to use a common SITEID
and separate INVIDs for each
investigator.
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 38 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
266 Richard Would a new CRT be created for Drug accountability? Our Not Related This will be covered in the
Foster / LEO CRF collects infromation on number and weight of study forthcoming Compliance
Pharma medication tubes returned to sponsor and investigator. This domain.
does not seem to fit into the Exposure domain.
267 R.Meinert, 8 / 2.2.2 Pattern of e Event Pattern of Event Typo Persuasive Corrected.
A.
Eimermacher
/ KKS-Mainz
268 R.Meinert, 21 / Format of Heading is inconsistent Persuasive Corrected.
A. 4.1&4.2
Eimermacher
/ KKS-Mainz
269 R.Meinert, 52 SDSSDSAEENDTC AEENDTC Typo Persuasive Corrected.
A.
Eimermacher
/ KKS-Mainz
270 R.Meinert, 61 MHONGO MHENRF Persuasive Corrected.
A.
Eimermacher
/ KKS-Mainz
271 R.Meinert, 67 criteria criterion In cols "Variable label" and "CDISC" notes i rows Persuasive Corrected.
A. "IETECTCD", "IETEST", IESCAT"
Eimermacher
/ KKS-Mainz
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 39 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
272 R.Meinert, 123 SUPPQUAL dataset SUPPQUAL dataset records Persuasive Revised RELATES
A. records related to all related to all subject records accordingly.
Eimermacher subject records in all in all domains, such as the
/ KKS-Mainz domains, such as the Intent
Intent To Treat To Treat population flag
population flag (ITT), (ITT), the Per Protocol
the Per Protocol Evaluation population flag
Evaluation population (PPE), and Safety
flag (PPE), and Safety population flag (SAFETY),
population flag will have DOMAIN, SEQ
(SAFETY), will have and GRPID set to Null, as
both SEQ and GRPID the only key needed to
set to Null, as the only identify the
key needed to identify relationship/association to
therelationship/associatio all subject records (for that
n to all subject records subject) in all domains is
(for that subject) in all USUBJID
domains is USUBJID
273 R.Meinert, "ITT", "PPE", amd "SAFETY" seem to be controlled Persuasive Added Appendix 10.4. This
A. terminology but this is not explicitly stated anywhere. appendix contains the
Eimermacher controlled terminology for
/ KKS-Mainz SUPPQUAL Name codes.
274 R.Meinert, We suggest to include the derived variable AERELYN Not Persuasive The sponsor may include their
A. (Relationship to study treatment (Yes/No)) in domain AE attribution for relationship to
Eimermacher because its derivation might be sponsor-specific, depending study treatment (Y/N) in the
/ KKS-Mainz on the categories of AEREL. (If for AEREL compulsary controlled SUPPQUAL dataset.
terminology was defined then AERELYN would be superflous.)
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 40 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
275 R.Meinert, We would prefer to include all 5 MedDRA levels in the Not Persuasive This is not represented in the
A. domain AE SDTM because it was not
Eimermacher required by FDA.
/ KKS-Mainz
276 R.Meinert, 164 We are surprised by the large number of domain codes, Not Persuasive This suggestion is not
A. particularely for special efficacy findings. We think that 2 consistent with the concept
Eimermacher more general domains "Efficacy Events" and "Efficacy behind the SDTM; it is
/ KKS-Mainz Findings" would have been sufficient. valuable to reviewers to
separate domains by topicality.
277 R.Meinert, In domain MH a variable indicating whether a medical Not Persuasive MHOCCUR is used for this
A. condition was solicited, or whether a medical condition was purpose.
Eimermacher reported spontaneously would be helpful to distinguish
/ KKS-Mainz solicited conditions that have missing values from
spontaneously reported conditions. (Variable MHSTAT with
values (TICKBOX, FREE WORDING) could be used for this purpose)
278 R.Meinert, We miss a thorough description of a clear concept Not Related Repeated observations are
A. how to handle (identify) planned repeated addressed by using different
Eimermacher observations within a visit (e.g. domains VS) and values for timing visits and
/ KKS-Mainz unplanned repeated observations within a visit (e.g. different SEQ. Additional
repeated lab-test because of a quality problem or an examples will be provided in
abnormal finding). (--SEQ is useful to uniquely identify the future.
records in the database but it is not very useful for
sorting because its values are arbitrary and have no
contextual interpretation.)
279 R.Meinert, Whereever "controlled terminology" is mentiond it is not Persuasive Assumption 4.1.3.1 has been
A. clear whether this terminology is compulsary (controlled) for modified to explain the
Eimermacher all sponsors or only controlled within a specific submission differences in the 2 types of
/ KKS-Mainz of a certain sponsor. controlled terminology.
280 Sanofi 3/1.1 What is the definition for study data tabulation exactly? It Persuasive The term submission datasets
seems to me that this is the same as submission datasets refers to all datasets submitted,
because we are referring Submission Data Standards including analysis datasets.
Implementation Guide for implementation details. If we The Study Data Tabulations do
think they are the same, why not call it submission data not include analysis datasets.
tabulation? The name of the
Implementation Guide has
been changed to Study Data
Tabulations Model
Implementation Guide.
281 Sanofi 3/1.1 SEND - should be spelled out first Persuasive Corrected
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 41 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
282 Sanofi 7/2.2.1 If one of the exclusion criteria for a study is "Subject used Not Persuasive Does not belong to a class;
Viagra", which class does this record belong, Interventions goes in TI because it's a Trial
(CM) or Findings (IE)? Design criteria question.
283 Sanofi 8/2.2.2 sound like all domains that fall into Events class include the Not Persuasive The IG describes the variables
variables referenced in Section 2.2.2, but actually this for each specific domain. In
example is specific to AEs and wouldn't apply to disposition, the SDTM, the general class
for example. Either include only variable applicable to all must list all variables for all
domains in a class or reference the dataset may look like the Events domains.
following for the AE domain (for example).
284 Sanofi 8/2.2.2 Pattern of e Event Pattern of Event Persuasive Corrected
285 Sanofi 11/2.2.5 Element and Epoch should be footnoted to refer to the trial Persuasive For TAETORD and EPOCH,
design model on Page 15 and 16. references to Section 3.2.2
have been added.
286 Sanofi 16/3.1 The illustration of cycles need to be fixed. Persuasive Corrected
287 Sanofi 16/3.1 It seems that time intervals for epoch between arms are the Persuasive The Implementation Guide
same. If time intervals are not the same between arms, can includes a discussion of issues
this be called epoch? affecting whether Epochs can
be defined.
288 Sanofi 17/3.2.1 TENRL TEENRL Persuasive Corrected
289 Sanofi 17/3.2.2 table seems truncated but appears on next page. Persuasive Corrected
290 Sanofi 5/1.1 US Food and Drug US Food and Drug Persuasive Corrected
Authority Administration
291 Sanofi 16/3.2.1 CM - 1. One record per medication per subject may not be Persuasive Changed text to read ―one
appropriate as medication can be given in different cycles as record per medication
well. 2. CM - this dataset also can be used for prior intervention episode per
medication as well as post medication. It might be more subject.‖
accurate to call it Medication and use MD as domain Wording of CM assumptions
abbreviation. are currently sufficient –
already mention Prior Meds
292 Sanofi 16/3.2.1 For EX - There may exist more than one record per constant Persuasive Corrected
dosing interval per subject for combination therapy. Suggest
to use One record per constant dosing interval per study
medication per subject.
293 Sanofi 16/3.2.1 Not sure the structure for SU, DS, MH, LB, PE, QS, SC and Persuasive DS and MH structure clarified.
VS is correctly identified.
294 Sanofi 17/3.2.2 Normalized and denormalized term usage - are these Persuasive We have revised the references
referring to database logical design terms? We are not sure to these terms.
these are accurately used.
295 Sanofi 21/4.1.3. form from Persuasive Corrected
1
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 42 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
296 Sanofi 22/4.1.3. For those who use WHODRUG or WHODRL, what to Persuasive The Assumptions state that in
5 include in the dataset to reflect multiple classes? cases where a plethora of
classes could apply, xxCLAS
will not be used. In those cases,
the FDA reviewer can refer to
the standard dictionary if class
attribution is desired.
Providing xxCLAS is a
convenience since either the
sponsor dictionary or reference
to a standard dictionary will be
provided, so xxCLAS is not a
critical data value (note that it
is Permissible, not Required).
297 Sanofi 27/4.1.4. What is the connection for check boxes on CRF Not Persuasive Yes.
7 ("PRIOR","ONGOING",or "CONTINUING") with the
definition of --STRF and --ENRF? Are we deriving--STRF
and --ENRF?
298 Sanofi 49/6.2.1 propose the following derived variable be removed, and that Not Persuasive These variables are necessary
they should be part of the analysis datasets.__DUR, for the FDA Safety Review.
AETRTEM, __BLFL, __DRVFL
299 Sanofi 52/6.2.1 SDSSDSAEENDTC AEENDTC Persuasive Corrected
300 Sanofi 61/6.2.3. Did not see MHONGO in the model Persuasive MHONGO is not intended to
0 be in the MH domain.
References to the variable were
errors and have been removed.
301 Sanofi 72/6.3.3 LBTOXCAT LBTOX Propose to change LBTOXCAT to LBTOX. LBTOX is the Persuasive Corrected
toxicity name such as hypercalcemia or hypocalcemia. Also
change the label to Toxicity.
302 Sanofi 72/6.3.3 Toxicity Grade Category Toxicity Persuasive Corrected
303 Sanofi 72/6.3.3 Indicates whether Standard toxicity name such Persuasive Corrected
LBTOXGR applies to as hypercalcemia or
low or high value hypocalcemia.
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 43 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
304 Sanofi 87/7 What is the purpose of the trial datasets? Most of this Not Persuasive We believe there will be
information is in text format and not programming friendly. benefits to both sponsor and
This information could be better illustrated with a diagram or FDA to have a standard
study scheme. Element seems to be naming things that aren't structure to define the trial
Elements. Elements should start with randomization, not design.
screening. All patients are screened and then randomization
occurs for patients meeting eligibility. Therefore diagrams
would illustrate this better like the Figures in the section.
How are the subject visit datasets handled for patients who
drop-out after a visit, but have all their information recorded
on the last CRF visit page? (i.e. a patient drops out after visit
5, but instead of a visit 6 this information is recorded on the
visit 8 pages?)
305 Sanofi 118/8.2 seq is continous within a domain and USUBJID. When Persuasive This is being reworded so that
relating records in separate domains, seq is used in RELREC gaps are allowed.
dataset for linking. There could be a problem When a record
in a dataset got deleted because deleting a record will lead to
--seq to shift and result in RELREC seq mismatch. We
would like to recommend for --seq to allow gaps.
306 Sanofi 121/8.3.2 What the value of LSREFID and LXREFID will be? Could Persuasive The RELATES chapter has
these be used as merging keys? been revised; it no longer limits
merge keys to SEQ and
GRPID. Future examples are
forthcoming.
307 Sanofi 131 Veritcal line is missing between EXSEQ and EXGRPID Persuasive Corrected
308 ADaM Team p 24 IG – p 24 Note for Analysis Datasets Only: Don’t Persuasive Corrected
understand why it’s there since it’s for analysis, and most of
other references say that this document doesn’t apply to
analysis and ADaM may not use this format. This section
should be removed.
309 ADaM Team a. Issue for IG – The name of the document should be SDTM Persuasive The document has been
IG, not SDS IG because Submissions Data Standards is renamed.
broader than what the IG covers. POST MEETING NOTE:
4/21 - Dave and Wayne agreed on the following: Broader
term for both Study Data Tabulations (SDT) and Statistical
Analysis Datasets (SAD) is proposed to be ―Regulatory Data
Submissions‖. This will include the SDT V3.1 that the SDS
committee developed and the SAD models being developed
by ADaM.
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 44 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
310 ADaM Team p 25 a. Issue for IG – p 25 – The duration calculation may not be Not Persuasive This is an analysis issue.
appropriate for all possible acceptable ISO8601 values. E.g.,
if both values are dates with no time. It should be end – start
+ 1 (Ray H.)
311 ADaM Team a. Issue for IG – Analysis population flags and derived flags Not Persuasive These variables are necessary
should not be in these models. Where do these derived for the FDA Safety Review.
variables come from? If these are imputed flags, they may The IG now specifies standard
not match with the analysis files. Or if they get them from flags to use in SUPPQUAL;
the analysis files, then it’s a kind of a circular reference (get additional guidance on use will
the values from SDT for analysis, get the SDT derived be forthcoming.
variables from analysis – the sequence of processing is
usually create SDT . . . there is a concern that variables
derived in the SDT datasets may not match similar variables
in the analysis datasets.) The SDT variables may not be
adequate for analysis.
312 Genentech The goal of CDISC to develop consistent structures and Not Related Further guidance on analysis
naming conventions for clinical trial data is an excellent one datasets will be forthcoming,
in that it has the potential to reduce the time to approval and and we expect the FDA will
the resources needed to submit an application to the FDA. accept submissions in this
format for many years to come.
The particular standards in the v3.1 model are most suited to
collecting and storing data, rather than to drawing
conclusions from data. As such, these structures are
amenable to data sharing across companies, FDA archival,
and inputting into tools designed specifically for reporting
from such structures on a subject-by-subject basis.
Summaries and statistical analyses are generally best
programmed using horizontal structures. Therefore, it is very
beneficial that the v3.1 documentation indicates that analysis
datasets, as well as datasets in v3.1 format, will be allowed
and even encouraged for a complete submission. This will
313 Genentech It would be best to remove derived data from the v3.1 Not Persuasive Most sponsors involved in the
structures. If these structures contain derived data, then the development of SDS V3.1 do
efficient approach of extracting from an operational not consider this an issue.
database, converting to v3.1, and then converting from v3.1
to analysis datasets will not work because the analysis
datasets, or the code that creates them, will be needed to
create the v3.1 datasets. Recursive algorithms and redundant
code can lead to inconsistency.
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 45 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
314 Genentech · Using the ISO8601DT format for date/time variables is Not Persuasive Tha nks for the feeback on the
workable. It will require an additional conversion to be date / time variable.
usable in SAS for calculation, but the ISO standard does
address the partial date issue without the need for a precision
variable.
315 Genentech · A CRF page number variable is often used for mapping a Not Persuasive The link to the annotated CRF
data point to its source CRF and for preparing annotated can be expressed in the
CRFs. It is not clear how to implement this. define.xml file. Currently not
required for submissions.
316 Genentech Events Class (Adverse Events): · It would be more efficient Not Persuasive This is not represented in the
to submit all 5 coding levels for MedDRA-coded data to SDTM because it was not
alleviate FDA staff from having to add on other levels for the required by FDA.
reviewers.
317 Genentech Events Class (Adverse Events):· Collection of adverse events Future Sponsor decision, but any
(AEs) on a visit rather than log basis, whereby AEs can be timing variable can be added to
continuing from a prior visit, does not have an obvious a domain. A null end date
implementation. would communicate this
information. We will consider
whether to add an --ONDTC
variable in the future to
accommodate this specific
situation.
318 Genentech Trial Design Datasets: This is a good concept although the Not Persuasive The Trial Design datasets were
recommended implementation is not straightforward. designed to be as simple as
possible while still fulfilling
the needs of most studies.
319 Genentech Keyword/Fragment Table: · This table is very useful. An Persuasive Added Appendix 10.4. This
additional table with standard test name code values e.g., appendix contains the
(―SBP‖ for ―Systolic Blood Pressure‖) and standard controlled terminology for
attributes (e.g., ―numeric‖) would also be very useful, ECG and Vital Signs Test
enabling better standardization when converting v3.1 codes.
datasets to horizontal structures.
320 Genentech Death: There appears to be no obvious place to record death Not Persuasive SuppQual could be used. An
information, particularly cause of death. Autopsy domain model and
further examples are planned.
321 Bob Section For --STDNRC field, why not have a low and high range as Not Persuasive The SDS team felt like no
Warnock/GS 2.2.3 with the other variable ranges for the dataset? value is added in breaking
K them out.
322 Bob Section EPOCH Should add footnote to reference Section 3.1 as it is not well- Persuasive For TAETORD and EPOCH,
Warnock/GS 2.2.5 described prior to section 2.2.5 references to Section 3.2.2
K have been added.
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 46 June 2004
V3.1 Comments and Responses
This information supplied by Commenter This info supplied by SDS Team
Submitter
No. Tracking ID Section Existing Wording Proposed Wording Comments Disposition Disposition Comment
323 Bob Section Extra . after "analysis" on 3rd line of paragraph Persuasive Corrected
Warnock/GS 2.2.6
K
324 Review all domains to remove --SPREF and apply --SPID Persuasive Corrected
and --REFID
325 Review all examples. Persuasive Reviewed.
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Comments] 47 June 2004
Ballot Submission/Resolution Instructions
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Instructions] August, 2002
Ballot Submission/Resolution Instructions
Column Headers
Commenter (sections in lavender)
Submitter Tracking ID #Internal identifier (internal to the organization submitting the comments). This should be a meaningful
number to the organization that allows them to track comments. This can be something as simple as the
reviewer‟s initials followed by a number for each comment, i.e. JD-1, or even more complex such as
„001XXhsJul03‟ where „001‟ is the unique item number, „XX‟ is the reviewer's initials, „hs‟ is the company
initials, and „Jul03‟ is the date the ballot was released. If additional rows are added, please do so after the
last row in the comment spreadsheet and ensure that the sequential numbers are maintained.
Section Section of the document, e.g., 3.1.2. Note: This column can be filtered by the committee, for example, to
consider all comment line items reported against section 3.1.2.
Existing Wording Copy and Paste from review materials.
Proposed Wording Denote desired changes.
Reason for the Change. In the case of proposed wording, a note indicating where the changes are in the
Comments
proposed wording plus a reason would be beneficial for the committee reviewing the comment.
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Instructions] August, 2002
Ballot Submission/Resolution Instructions
SDS Team Resolution (sections in turquoise)
Disposition The instructions for selecting dispositions were too large for this section and have been moved to the
worksheet titled "Instructions Cont.."
Disposition Comment Enter a reason for the disposition as well as the context. Some examples from the CQ committee include:
20030910 CQ WGM: The request has been found Not Persuasive because....
20031117 CQ Telcon: The group agreed to the proposed wording.
20031117 CQ Telcon: Editor recommends that proposed wording be accepted.
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Instructions] August, 2002
Ballot Submission/Resolution Instructions
Instructions continued...
For the column titled "Dispostion" please select one of the following:
Applicable to Comments:
1. Persuasive. The team has accepted the comment as submitted and will make the appropriate change in the next publication cycle.
2. Persuasive with Mod. The team believes the comment has merit, but has changed the proposed solution given by the commenter. Example
scenarios include, but are not limited to;
-The TC has accepted the intent of the ballot comment, but has changed the proposed solution
-The TC has accepted part of the ballot comment, and will make a change to the standard; the other part is not persuasive
-The TC has accepted part of the ballot comment, and will make a change to the standard; the other part may be persuasive but is out of scope
The standard will be changed accordingly in the next publication cycle. The nature of, or reason for, the modification is reflected in the Disposition
Comments.
3. Not Persuasive. The team does not believe the ballot comment has merit or is unclear. A change will not be made to the standard or proposed
standard. Example scenarios include, but are not limited to:
- the submitter has provided a recommendation or comment that the committee does not feel is valid
- the submitter has not provided a recommendation/solution; the submitter is encouraged to submit a proposal for a future ballot
- the recommendation/solution provided by the submitter is not clear; the submitter is encouraged to submit a proposal for a future ballot
4. Not Related. The team has determined that the comment is not relevant to the domain at this point. Example scenarios include, but are not limited
to;
- the submitter is commenting on a portion of the standard, or proposed standard, that is not part of the current ballot
- the submitter's comments may be persuasive but beyond what can be accomplished at this point in the without creating potential controversy.
- the submitter is commenting on something that is not part of the domain
5. Future - The team has determined that the comment has merit and may be considered in the future.
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Instructions Cont..] August, 2002
Ballot Submission/Resolution Instructions
2bc5aac8-484b-4bcb-ace8-36f553ed6716.xls [Instructions Cont..] August, 2002