View source for Subject Data Story Boards

== Links ==
[[Stage IB Project Page|Stage 1B page]] | [[Study Design, Study Participation and Subject Data|CDISC HL7 Project Page]] | [[Stage 2 Project Page|Stage 2 Page]]

== Useful Guides ==
For those wanting assistance in editing wiki pages have a look at these pages

* http://meta.wikimedia.org/wiki/Help:Editing
* http://en.wikipedia.org/wiki/Wikipedia:Cheatsheet

==Sponsor Planned Data ==
Data collected in accordance with the plan.  The Study Design message conveys (among other things) the plan for data collection, the Study Participation message includes information on the subjects in the study, and thus those two messages imply an expected set of data to be collected. The subject data message needs to be able to convey that data. 
===Story Board 1: Submission of SDTM and ADaM Data===
Study A1234 is complete and Acme Pharmaceuticals now wants to send to the FDA all the observations recorded for each subject during the study as part of their study report submission. Acme uses the CDISC-HL7 subject data message to provide all the recorded observations, as well as all the key derived parameters resulting from those observations, as defined by the CDISC SDTM and ADaM standards. The message contains all important relationships, such as the relationship between an observed and planned assessment (or lack thereof), the relationship between unplanned assessments and other observations (i.e. finding of jaundice led to a bilirubin measurement), and the relationship between the derived parameters and the collected data. 

Those observations that were previously reported in a spontaneous adverse event report (ICSR) need to use common identifier, see story board 4

''Scope question: Do we want ADaM in here for the first version?''
''How should we define "Key derived parameters"?""
''Definition is in ADaM IG.''

''ACTION: Need to have someone from the ADaM team review and provide feedback.''
''2009-02-12 - ADaM representative attended.''

* Data must be linked to plan (protocol)
* Convey all the data we do today (e.g. SDTM domains)
* Be able to conduct a safety review per "Conducting a Clinical Saftey Review of a New Product Application and Preparing a Report on the Review" CDER 2005 document 
* Think about non SDTM data (e.g. analysis, eHR. audit trail)
* Present as an SDTM view
* We need to support define.xml (include all metadata held in define)
* What about the blank CRF (part of define.xml)
* Extra data entered on the CRF page (e.g. comments at all levels, page and/or variable)

'''Stage II to define.xml:'''
*planned and unplanned observation
*linking
*blank CRF
*extra data entered on the CRF page
*Ensure that RPS has address this

===Story Board 2: FDA Completeness Check===
The FDA has received the data for Study A1234 and wishes to assess the level of completness of the data submitted by Acme Pharmaceuticals. The reviewer accesses the Janus data warehouse and runs a check to assess if all planned activities were performed. The reviewer should be presented with a report that provides a detailed view of the missing observations.

''This requires access to the plan held within the Study Design message and needs to allow for all paths to be evaluated at a high level of detail definition (sufficient to allow for a machine to perform the check). Note that this is a check of what is missing against the plan and does not consider additonal data that may have been collected.''

''The check needs to run at a granular (data point) level but the story boards make no statement on how the results are to be presented''

===Story Board  3: Periodic Submission ===

Acme Pharmaceuticals study XYZ123 being conducted and has some potential toxicity issues. The FDA requested that all subject data be submitted quarterly while the trial was ongoing.  
Subject data was submitted on 5 occasions while the trial was ongoing including updates to previously submitted data points.
After the trial concluded, all of the subject data was sent to the FDA as a final transmission.

* ''Question 1: We are seeing a need for an update facility (transactional mechanism). We do we send all of the data again at the end? YES''
* ''Question 2: Do we need study status information for such things as "finishing early", "Last message"'' - ACTION check that this is in Study Participation/Registration Message
* ''Question 3: How does Subject Data in this use case relate to the use of Study Participation and the status info carried there?'' - ACTION Make sure there is no overlap between Study Participation and Subject Data

''Protocol amendments were discussed but the group concluded that this should be deferred to a later release''

'''Stage II to define.xml'''
*Need to address action items for questions 2 and 3.

==Cross Reference to ICSR==
Data collected in connection with ICSRs.  It should be easy to make connections between data in the ICSR and other data conveyed in the Subject Data message.  At minimum, this probably means that the Subject Data message needs to include unique identifiers for ICSRs. 
===Story Board  4: Non-duplicative Adverse Event ===
ABC Pharmaceuticals is running study 123A.  One of their sites PharmaCRO reports an SAE. ABC Pharmaceuticals collects all the relevant information and promptly submits a report via the ICSR to the FDA (referenced by HL7 identifier 2.16.840.2.113883.4.125)

A year later, ABC pharmaceuticals is preparing their submission for study 123A to the FDA utilizing the Subject Data message.  As part of their submission process, where information has previously been provided via ICSR there is a link in the subject data message which points back to the ICSR. Note the data can also be included in addition to the link.

The message must be able to identify previously submitted data. 

''Question 1: This needs further thought given the possible impact on the sponsor company given the current separation of CDMS-type systems and Safety systems?''

== Class 3 ==
===Story Board 5. FDA Initiated Query of Subject healthcare Data from an EHR===
The FDA team is reviewing the Study 123A which was submitted by ABC Pharmaceuticals.
Several patients have experienced SAEs for which the study design has pre-defined treatment strategies. FDA reviewers notice a wide disparity in the outcome for patients with similar disease levels.

The FDA requests to see a complete set of health records for the affected patients from ABC Pharmaceuticals, who in turn makes a request to all of their sites. Both use RPS as the information request and fulfillment mechanism.

''Note: Request is from FDA -> Sponsor -> Site''

The site constructs an EHR HL7 message containing the requested information and sends it to ABC Pharmaceuticals who in turn forwards the information to the FDA.

The FDA discovers that some sites have availed themselves of a new treatment procedure available for the treatment for these  worrisome SAEs and that use of this procedure correlates back to improved outcomes.

''This story board was discussed but the group concluded that this should be deferred to a later release''

===Story Board  6: Sponsor Initiates Additional Data Collection ===
Use Case: Sponsor initiates the collection of additional data from investigators and then provides an update to the FDA. All of the data reported would be in addition to the plan. 
Study A1234 is complete and Acme Pharmaceuticals has provided the data to the FDA using the CDISC-HL7 subject data message to provide all the recorded observations, as well as all the derived parameters resulting from those observations, as defined by the CDISC SDTM and ADaM standards. Acme is aware of the need to collect some additional observations. The sponsor initiates the collection of the data from the sites and provides the new data to the FDA using the Subject Data message. This data needs to be linked to the data already provided to the FDA but is considered additonal to the plan as defined within the Study Design message. In some cases the Study Design Message may need to be updated.

==Class 4 ==
Subject Data may be conveyed in multiple stages, rather than as a monolithic data transfer.  Therefore, the message must be capable of conveying both new and updated data.
===Story Board 7: Periodic SubjectData messages (blinded) and then data is unblinded===
See the above use cases
# Since Acme Pharmaceuticals study XYZ123 is being conducted in a vulnerable population, FDA requested that subject data be submitted quarterly while the trial was ongoing.  
# Subject data was submitted on 5 occasions while the trial was ongoing.  In data on study drug administration,  the name of study drug was given as “Blinded product” and dose was recorded with units of “tablets.”
# After the trial completed and was unblinded, data on study drug administration was updated to include all subjects, and to provide actual study drug data (Drug A, Drug B, or Placebo) for all subjects and dose information (20 mg for Drug A, 50 mg for Drug B) for subjects who received active product.  

===Story Board 8: Periodic sending of SubjectData messages and Clarification process results in a data point being changed ===
# Since Acme Pharmaceuticals study XYZ123 is being conducted in a vulnerable population, FDA requested that subject data be submitted quarterly while the trial was ongoing.  
# On the second occasion when data was submitted, lab data included an ALT value of 526 for subject 145 at Visit 3.
# This abnormal ALT value triggered a query.  The response to the query was that there was a transcription error, and the true value was 256.
# In the third submission of periodic data to the FDA, the message included the corrected ALT value of 256 for subject 145 at Visit 3.

Note: Audit trail is discussed in a different Story Board. Sending the query and the response to the query is not in scope for this use case.

== Others ==

===Story Board 9. Determine if patients met inclusion criteria===
The NCI-sponsored Study RTOG 93-09 is a randomized, unblinded, multicenter, two arm parallel design study comparing Chemotherapy + Radiation Therapy (CT+RT) vs. Chemotherapy + Radiation Therapy + Surgery (CT+RT+S) for the treatment of Stage IIIa non-small cell lung cancer. A key inclusion criterion requires the presence at screening of a single, newly diagnosed, primary lung parenchymal lesion of stage IIIA (T1, 2 or 3) with ipsilateral positive mediastinal nodes. The reviewer wants to ensure that only subjects meeting this criterion were enrolled. The Study Design message contains the plan to collect these screening data, and the reviewer is able to locate and analyze the collected screening data for each subject from the Subject Data message to verify that this criterion was met. 

Use Case: A study is conducted in  order to determine if a product is safe and effective in a sub-population of patients.  The inclusion criteria are constructed so that only patients in the sub-population of interest are enrolled in the study.  The reviewer wants to ensure that only patients who met the inclusion criteria were enrolled in the study. 

26-March Discussion: 
- Everything in the storyboard looks ok except the statement "Requirement on Study design message to indicate those observations used to support inclusion / exclusion processing": this is not currently done in pharmaceutical companies.  So this is a Study Design requirement, not a Subject Data requirement.
- The following statement should be moved over to the Study Design requirements:
Requirement on Study Design message to indicate those observations used to support inclusion / exclusion processing.
- The current study design message doesn't include the data described in the above statement.  Some items that are collected at screening don't indicate that they are being collected for IE processing.
- clinical statement contains why you are collecting certain data
- clinical stmt may include data that isn't currently being done in pharmaceutical companies
- It's an FDA policy issue about what's going to be required in the message
- There is concern that it might not be possible to build a structurally valid message if you didn't collect the data that is required for the message.  It also may not be practical to provide some required data.
- One way to look at a complicated IE inclusion and how it is linked back to the planned assessment is as an analysis plan, which should be deferred to a future release.
- For this first release, we are concerned only about collecting the data.
- The Storyboard makes sense without the Study Design requirement.

'''Stage II to define.xml:'''
*Does study design allow this kind of connection?

== Audit Trail Use Cases ==
'''Stage II question: How does HL7 represent audit trial?'''

===Story Board  11. Original CRF Value Changed===
Use Case: Sometimes the original CRF value is changed by the sponsor (could be the investigator in the case of correcting a data entry error).  The full audit trail is provided in the subject data message.

1. ABC Pharmaceuticals has completed Study123A and this study is part of a submission for a drug that lowers blood pressure.

2. Data was collected via paper CRFs.

3. Subject data for BOB123 has a first data value for a concomitant medication of “Atenolol” and a last concomitant medication value of “Aspirin”.  It is highly suspicious since “Atenolol” is a prohibited medication. The reason for the change is “Dose correction” on the last Aspirin medication record.

4. The reviewer inspects the full audit trail for BOB123 concomitant medications which has 6 modifications for the record of first value of “Atenolol” and last value of “Aspirin”.

===Story Board 12. Audit Trail Information===
Study A1234 is complete and FDA wishes to audit the study data collection process. In order to do so, the reviewer reviews the subject data for the following audit trail information about each recorded observation that is associated with the observation and which has been submitted to the agency: 

• Who originated or changed the data point (e.g. study coordinator, data entry clerk, investigator, laboratory, imaging facility, biomedical device) for that subject
• Date and time the data point was originated or changed 
• Reason for that modification 
• Date and time the investigator signed off


The following are specific examples: 

{| border="1"
|Observation
|Result
|Audit Trail
|-
|Patient ID
|AB0012	
|Who: randomization algorithm in central computer
Date: 2008-06-01T15:00
Subject #:AB0012
Sign off: Dr. R. Smith
Date: 2009-03-09T09:00
|-
|Sex
|Male
|Who: MS. E. Clerk
Date: 2008-06-01T11:00
Subject #: AB0012
|-
|Hemoglobin
|15.3 g/L



15.5 g/L 
(modification)
|Origin: Co-op Labs
Date: 2008-06-02T12:00
Subject #:AB0012


Origin: Dr. B. Green
Date: 2008-07-02T14:00
Subject #: AB0012
Reason: Lab error reported by Co-op Labs due to standardization problem; sample retested
|-
|Chest X-ray
|Right Upper Lobe Consolidation	
|Origin: Dr. P Brown
Date: 2008-06-01T16:00
Subject #: AB0012
|-
|Blood Pressure
|124/88
|Origin: Cardiodiagnostics serial #7834 
Date: 2008-06-01T11:00
Subject #: AB0012
|-
|Concomitant Medication
|Lasix
|Origin: Dr. R. Smith
Date: 2008-06-01T11:00
Subject #:AB0012
|}

== Required Association Between Data and Study Design ==
===Story Board 13. ===
Estimate mean and variance of subject response in a study cell, and functions of these means and variances.
A reviewer wishes to estimate the mean and variance of a continuous response variable (e.g. blood pressure) at one or more times (e.g. visit) in one or more study cells, and calculate functions of these means and variances.

Rosie Reviewer is interested in understanding how blood pressure is affected, over time, by the treatment strategies being evaluated in the Acme 9999 study.  In order to do this, she must know, for each measurement, the subject's treatment strategy and the length of time on that strategy at the time of the measurement.  She must also be able to identify the baseline measurement for that treatment strategy.  In order to evaluate the treatment effect at various timepoints relative to the start of treatment, she must select measurements to be included in evaluation of that timepoint.  This will involve decisions about which observations are close enough to the timepoint to be included in the analysis, selecting from among multiple "close enough" observations, and deciding whether and how to impute values for subjects with no "close enough" measurements.  Once observations have been identified, she will calculate estimate of relevant statistics (means, variances, changes from baseline, etc.) for each treatment strategy and timepoint and also estimate differences between treatment strategies.

March 26 discussion:

- need to know which data was collected during which treatment strategies
- question about how cells are associated within a clinical statement: care plans should have corresponding care records once the plan is executed
- we need to continue this discussion at the F2F meeting

March 31 discussion:

-the message needs to identify a baseline

===Story Board 14. ===
Estimate mean survival time for subjects in a study cell
A reviewer needs to estimate the mean survival time to an event (e.g. heart transplant) in a study cell.  In order to calculate the mean, the reviewer needs to know if the event happened, and if the happened, when the event happened.

Rosie Reviewer wants to compare the survival times for two treatment strategies intended to delay or avoid the need for heart transplant.  The information needed, for each subject, is when the subject had a transplant, or, if they did not have a transplant, when the last contact with the subject occurred (i.e., when they were censored).  The time of transplant or of censoring must be expressed as time from the randomization/start of treatment.  Once this data has been derived, Rosie estimates survival times for each treatment strategy and tests for a difference in survival between treatment strategies.

March 31 discussion:

- first part is statictical analysis plan 
- need to ensure that data is there so analysis can be done
- no additional work needed for this story board

===Story Board 15. ===
Estimate the baseline value of a subject response:

An analyst wants to estimate the pretreatment value of a patient outcome (e.g. blood pressure).  Estimation of this value will be based upon one or more values of the attribute in a study cell prior to the study cell containing study treatment, or from patient history data.
Must be able to link observations to the study design. Should be able to convey historical data.

===Story Board 16. ===
Test that a function of the data in one or more study cells is equal to, less than, or greater than a constant.
Calculate an analysis of covariance for a continuous outcome measure for study cells in the second epoch of the study. The value at visit 3 is the response variable, and the sponsor-defined baseline score is the covariate. 
Message needs to allow for linking of the data to the study plan.

===Story Board 17. ===
Test that a function of the data in one or more study cells is equal to, less than, or greater than a constant.
Calculate a Mantel-Haenszel test for study cells in the second epoch of the study. The response variable is categorical (e.g. presence or absence of an adverse event, seriousness of an adverse event).  Stratification needs to be done by site, age, sex, and race.