Skip to end of metadata
Go to start of metadata

You are viewing an old version of this page. View the current version.

Compare with Current View Page History

Version 1 Next »

VariableComment
ADURIs ADUR any different than the duration variables CRSDUR and NTDUR?
ADURN
ADURU
Is there a specific derivation or can this be sponsor defined? Some compute duration of each AE record only when both start/end dates are complete. Also if time is collected on CRF, duration unit computed as Sec, Min, Hour, Days.
AEACNOncology studies often have an action of dose delayed which is captured in AACN (Analysis Action Taken) since it is not covered by NCI CT. ACACN is 'Exp' variable in SDTM, not 'Perm'.
AEBODSYSThis is 'Exp' variable in SDTM, not 'Cond'
AEDECODThis is 'Req' in SDTM, not conditional.
AETOXGRNDoes this need to be a required variable if analysis equivalents (character & numeric) are generated? (same comment for ADAE)
APERSDY
APEREDY
Consider adding APERSDT and APEREDT. Are APERSDY and APEREDY derived as days from TRTSDT?. If so, add derivation information. Column of 'CDISC Core' and 'FDA Core' are empty, can we put 'N/A' for 'CDISC Core' column and put 'Cond' for 'FDA Core' column?
ASTDT
AENDT
Confirm that sponsor can have their own imputation rules and they would be same as ADAE.
ATOXGR
ATOXGRN
How to define the analysis grade? If there is no difference between analysis grade vs. CRF collected variables, can we keep these two variables as 'Optional'? Also, all columns like 'Type' ,'Codelist', 'CDISC Core','FDA Core' and 'Source' columns are empty.
CRSDUR
NTDUR
For the CRSDUR/NTDUR the label is 'Time to CRS/NT resolution by subject-period', would the label be better as "CRS Duration by Subject-Period", "NT Duration by Subject-Period"?
if there are multiple events, for example first event started and ended, then few days later, the last event started and ended, do we count in gaps for the time to resolution? Or we collapse events and count the longest duration from collapsed events, then choose the longest duration as the time to resolution?
Derivation does not account for unresolved events, i.e. when CRSENDY /NTENDY may be missing. In this case do we want to censor duration at last known alive date?.
CRSFLR
NTFLR
Flag variables should end in -FL.  Are record-level flags needed since FDAGT populated as 'CRS' 'NT' serves the same purpose? Can the codelist be "Y, Null"? 
CRSFLS
NTFLS
Flag variables should end in -FL. Suggest changing label to be consistent with other variables, i.e. "CRS Flag by Subject-Period" "NT Flag by Subject-Period". Can the codelist be "Y, Null"?
Is this intended to be populated for each subject within a period?.
Does "experienced a CRS (NT) event" mean CRS (NT) event overlap with the period or CRS (NT) start date within period?
CRSMXSDY
NTMXSDY
What's the reference study day? If that the first treatment day/RFSTDTC?
CRSONFL
NTONFL
Is the 'Ongoing CRS flag' and 'Ongoing NT flag' only populated on the CRS or NT event record that is ongoing? Or is it a subject-level flag to indicate the subject has CRS or NT ongoing at eos/last known alive/cutoff (add cutoff to the FDA Additional Information column)? Can the codelist be "Y, Null"?
Is this record level flag to indicate that event within FDA grouped term was unresolved?. Or is it subject level flag to indicate at least one event within FDA grouped term was unresolved?.
CRSSTDY
CRSENDY
NTSTDY
NTENDY
CRSMXSDY
NTMXSDY
If these are derived from AESTDY/AEENDY then events with partial dates will not be used to derive. Suggest adding variables ASTDY/AENDY to use all dates to derive. In addition, AEENDY can be missing if event is not resolved, derivation needs to clarify that if at least one event is unresolved then CRSENDY/NTENDY should be set to missing.
CRSSTDY
CRSENDY
NTSTDY
NTENDY
This is AESTDY of first CRS/NT event for subject in APERIOD, and AEENDY of last event for subject in APERIOD. If the first event started in Period 1 and the last event ended in Period 2, then is Period 1 <CRS/NT>ENDY=null, and Period 2 <CRS/NT>STDY=null?
DTHFLDTHFL is a SDTM variable in DM, which includes deaths after data cutoff date. Should ADSL.ADTHFL be here to replace DTHFL?
FDAGTIs the intent of FDAGT similar to a customized query? if so, use the ADaM standard variable for customized queries, CQzzNAM. For example, CQ01NAM='CRS' and CQ02NAM='NT' when the record contains an AETERM belonging to the CQ. The variables are blank for AETERMs that are not in the CQs.
Are the FDA groups available publicly? How are they versioned? Is the versioning aligned with MedDRA versions?
Column of 'CDISC Core' and 'FDA Core' are empty, can we put 'N/A' for 'CDISC Core' column and put 'Req' for 'FDA Core' column?
FDAGT
CRSFLR
NTFLR
Looks like all three variables are record-level variables, then why do we need both FDAGT and CRSFLA/NTFLR?
GeneralMany variables have  'CDISC Core' and 'FDA Core' columns as empty
Should this dataset only include CRS/NT events or all events from ADAE?. The purpose of this analysis is not clear. When changes in event such as grade, relationship, seriousness, etc occur, each change is captured as a separate event in BMS CRF. Unless we collapse contiguous events for each PT, we would not get the measure of actual duration/resolution of event. Also, it appears that analysis is conducted for each period in which case TRTEMFL is not used.
For those variables which are obtained from ADAE, values within all columns should be consistent with ADAE column values. 
Is this analysis dataset on CRS/NT events mandatory to be created ?
Does it make sense to have a more generalized dataset that can support other AESIs besides CRS and NT?
 - What analysis is planning for this type of data?
 - We suggest modeling this as a safety BDS dataset (e.g. ADSAF) to allow for this analysis.  If Survival Analysis is planned then it could/should be modeled as a Time to Event Dataset (e.g. ADSAFTTE) it would then be possible to include censoring of ongoing events at analysis cut points as necessary.  This would also allow for explicit flags of Y and N for whether an event occurred for all subjects not just those with either a CRS or NT event (depending on the denominators used for analysis)
 - Consequently the variables CRSFLS, NTFLS, CRSSTDY, CRSENDY, NTSTDY, NTENDY, CRSMAXTX, NTMAXTX, CRSMXSDY, NTMXSDY, CRSDUR, NTDUR would be better as parameters in a BDS dataset.  In an occurrence based dataset we may see duplication of content which may make analysis less transparent (especially with the addition of both subject level and record level flags)
Could record level flags be added to ADAE as well as to here to aid traceability
Please use proper case for variable labels to be consistent with CDISC labels.
TRTEMFLComments from ADAE also apply
  • No labels