2015-09-17 CV Discussion Meeting notes

 AGENDA:  The CV Imaging TA team is wondering why the CV Terminology team agreed to create anatomy-specific test concepts in the current Cardiovascular Test/Code codelists such as Left Atrium Dimension; Left Ventricle Dimension, End-Diastole, etc. We are seeing similar new concepts in the CV Imaging CDE (i.e. Left Ventricular Diameter, End-Systole) and would appreciate the terminology team's opinion on how to better model and handle this type of data (whether the anatomy part should be captured in LOC instead of being part of the CVTEST). For example, for the CVTEST of Left Atrium Dimension, could the test be simply Dimension, and the LOC variable can be populated with Left Atrium or Right Atrium?

The table below was sent out before the meeting:

 The meeting discussions showed that there were 2 major different opinions:

1 - The CDISC SDTM has multiple variables for holding the additional information for the test name (i.e. anatomical location, specimen, method, unit, etc.).  A generic test can then be used multiple times and if put together with the additional variables would make the test distinct.  This is a common practice in the CDISC SDTM.  

The terminology rule for left and right is that when the structure and function are exactly the same, then the left/right can be split off and stored in the SDTM laterality variable.  The CV Imaging TA team agreed to leave the left and right as part of the anatomical location because the structure and function are NOT exactly the same.

The CV Imaging TA team wants all of the CV tests to be broken out across multiple variables:  Test and Anatomical Location.

2 - The existing CV terms in the above table should not be broken out:

Jimmy Tcheng:  The rationale for the naming of these CV Tests as specific unique entities, rather than as a single entity with sidedness, is that the concept of sidedness in the context of the human body refers to symmetrical, mirror-image structures which are present on both the right and left side of the body.  This concept applies to virtually all of the structures of the body EXCEPT those of the thorax and abdomen. Examples – the right and left eye, ear, nostril, teeth, the bony structures / muscles / tendons / ligaments of the arms and legs, etc.

The right and left heart, on the other hand, are NOT mirror images of each other.  They are embryologically, anatomically, morphologically, geometrically, and functionally separate entities.  The right heart is shaped like a warped Frisbee, while the left heart is shaped like a football.  The techniques used to assess cardiac measurements such as ejection fraction and volumes are specific and unique to each chamber of the heart.  Even a given measurement (e.g., end-diastolic volume, ejection fraction) have different clinical meanings, implications, and reference ranges between the two “sides” of the heart.   And there are any number of measurements on one side of the heart that do not have a corresponding measurement on the other side of the heart (e.g., left ventricle end-diastolic diameter has no right ventricle equivalent; similarly, right ventricle end-diastolic dimension has no left ventricle equivalent).

Including the right / left side of the entity also creates specificity at the level of the Test name, which is likely critical given that the use of these cardiac Tests is not balanced (50-50) as one would expect for mirror-image structures.  For example, >95% of the time the term “ejection fraction” means left ventricular ejection fraction.  If right ventricular ejection fraction were data being submitted, then it would be obvious to the submitter not to submit the value using a Test named “left ventricular ejection fraction”.

DIAMETER versus DIMENSION test name: The methodologies for calculating LV and RV volumes are distinctly different given the different morphologies. This is because the LV is shaped like a football, while the RV is shaped like a concave disc. And the RV doesn't have an RV "diameter" - there isn't a round dimension to it that would describe a diameter. Instead, there is a clinically comparable concept called the "right ventricle end diastolic dimension" (not diameter). And it is determined along a different axis than the LVEDD.

3 - Some people felt that a combination of some specific test names and some generic test names should be used.  Each test should be evaluated on its own.

4 - Some people felt that a choice should be made to either break out or not break out so we could have one consistent approach.  They were against the some specific and some generic approach.

5 - Pam Douglas:  I tend to agree with Dr Tcheng that the description of ejection fraction, or LV volume, or LV dimension, etc., is specific and unique to the chamber.

6 - David Kong email:  So if you had a concept of "End-diastolic volume" and "End-systolic volume" and had a descriptor in anatomic location of "Left ventricle" you would have the most general solution. The debate likely comes from the fact that this generality allows for future creation of concepts that in reality will be used only rarely.... at the expense of every user who needs the common terms having to reference the anatomic location variable to ensure that the descriptor is what they want. Kind of like the Baltimore subway construction project - a lot of people struggled with inconvenience for a long time for an end product that isn't used much.

7 - Mary Beth Sabol email:  I don’t see much of a difference between the two approaches.  Clinically, there is a difference in the value of each in how they contribute to an individual’s health.  But that is not the question here.  To me, it is just a “how would folks like to capture the data”?  Ultimately the person reviewing data that is coded to location left ventricle will need to understand clinically what that means.  Whether it is a consolidated definition or broken out into a more generic approach with location pulled out, is really simply a preference of coding style and either approach works for me.  That probably does not help much in terms of choosing an approach, but should help in knowing that ultimately either approach is acceptable.  It’s simply how do folks want to categorize data…..how would that help someone in stats down the road if they were trying to manipulate large data sets?  Does it matter?  That would seem the driver at this point.

 Table updated during the meeting:

The terminology teams are consensus, decision-making teams.  This discussion had subject matter experts divided.  There was no consensus. 

This was a quick poll at the end of the meeting:

     Both Generic and Specific:  Amy, Linda

     Generic:  Erin, Jordan, Abby

     Specific:  Maria, Becky, Jimmy, Pamela, Steve, Bernice

     Choose one way or the other but be consistent:  Fred

Emails from the people who could not attend showed the following:

     Mary Beth Sabol – could go either way, she sees merit in both

     Specific:  Dr. James Topping, Dr. David Kong

Fred Wood suggested that we ask the FDA medical reviewer how they want to receive the CV data and review it.  Action assigned to Bernice Yost to contact the FDA.

Post-Meeting Note:  Bernice requested permission from Sam Hume/Nicole Harmon to contact Norman Stockbridge.  Permission Received.

                                 Fred Wood also contacted Diane Wold about contacting the FDA.

                                  Norman Stockbridge suggested to contact Karen Hicks, and she suggested to talk to Fatima Frye and Mitral Rocca.

                                   Waiting for response from Fatima Frye who returns to the office on 29Sep2015. 

 The CV Imaging team was given an action item of creating an example table of how to store the data in the SDTM where the data is broken out across multiple variables.  Action assigned to Amy Palmer.

Post- Meeting Note:  Example Table Completed