Comment

On behalf of Wayne Kubick,…

On behalf of Wayne Kubick, CTO of HL7: "Row 2 includes Pharmacogenomics and medical devices, which, to the best of my knowledge are rarely used or not used at all.  So lumping those with drug trials is inaccurate.  They should separate those out with adoption level 1.

Row 3 - same comment as above:  Therapeutic Area Standards lists Process Maturity as Final, Implementation maturity as Production.  But, in fact, this refers to a whole series of IGs, some of which are in development as draft, some provisional (not quite final).  It’s not appropriate to lump them together and adopt the most mature value, but you don’t want to separate them out either.  I would change these values to Need Feedback or to some indication the they vary.  The phrase “apply across all therapeutic areas” is not correct.  They haven’t addressed, all, only some."

On behalf of Wayne Kubick,…

On behalf of Wayne Kubick, CTO of HL7: "The adoption levels in this section are mostly reasonable.  The exception is row 2 for SDTM. While EVS terminologies for drug trials are generally mature and widely adopted for regulatory submissions, row 2 includes Pharmacogenomics and medical device terminology, which, to the best of my knowledge are rarely used or not used at all.  So lumping those with drug trials is inaccurate.  They should separate those out.  Also, as a comment shows, they don’t really link to the appropriate terminology set — they just link to EVS.  The link should be to the appropriate code set not just the tool." 

feedback

Adoption of Sentinel and PCORNet is at least 2 dots.
OMOP CDM adoption has increased in 2018 and should be changed to 4 dots.
ADaM is an analytical standard but SDTM is not. SDTM is also listed in other parts of ISA. I think SDTM should be removed from the list here completely.

Also, I fully agree with other commenter that the task "Representing Analytic Data for Research Purposes" should be defined in preconditions for considerations.
 

If they are talking about…

If they are talking about research use, then LOINC, RxNorm, SNOMED CT, ICD and CPT should also be listed they are all incuded in OMOP. Believe that LOINC and some of the others are listed are in PCORI and I2B2.

All claims data bases in the US carry CPT and ICD9/10.


While mentioning databases I2B2 sHould also be listed 

Believe OMOP (ODHSI) has the…

Believe OMOP (ODHSI) has the best of the cross institutional data models. It is in the 3rd or 4th generation. More than 100 institutions have contributed to it. It is international including data sets from the US, Korea and many others. It carries an estimated 400 million distinct patients. Google Scholar asserts 2980 papers mention OMOP.  Think it qualifies as final production and adoption of 3-4 bullets. 

The adoption level may…

The adoption level may exaggerate the reality. First, it only applies to a limited context, not analytic data nor all research purposes and FDA submission constraint is not stated. Further, in some contexts the codes are post coordinated using 3-4 fields and are not consistently included.

Regenstrief - Comment

This major section is about Vocabularies/Code Systems/Terminology standards. There’s another section for data models or transport/exchange/API standards. As such, this list should be pruned to vocabularies only and the standards for data models, data tabulation models, and other storage/transport specifications should be moved elsewhere.

The question posed last year remains. What is “analytic data” (besides the obvious “data that can be analyzed”)? How is it different from the other domains of content listed elsewhere? Surely lab data is is analyzed for researched purposes. The mention of the research network (OHDSI, PCORnet, etc) data models not intended to spur their listing here, but rather to illustrate that they have adopted the main vocabularies listed elsewhere in the ISA (e.g. LOINC, SNOMED, RxNorm) for “research purposes”. So it just isn’t clear at all what this interoperability need is really meant for and why it is here. Please clarify with a a better definition and purpose statement.

This is an example of a…

This is an example of a poorly crafted interoperability need. What is “analytic data” and which kinds of “research purposes” is this meant to cover? More importantly though is why/how is this need is different from all of the other kinds of content listed elsewhere in the ISA (e.g. Lab tests, medications, diagnoses, etc). Certainly all of the large scale observational research networks (OMOP, Mini Sentinel, OHDSI, PCORnet) have developed common data models that employ many of the same vocabularies (LOINC, SNOMED, ICD) listed elsewhere in the ISA.

CDISC may be appropriate for…

CDISC may be appropriate for clinical trials, but is inadequate for a wide variety of other Analytic research.  This area needs more work.

list non-CDISC standards

In addition to CDISC standards, AllOfUs project (and many others) are using OMOP Common Data Model to standardize data.

 

This standard should be added to the list

Specs are here

https://github.com/ohdsi/commondatamodel/wiki

 

OMOP CDM Conformance testing tool is here

https://github.com/OHDSI/Achilles

 

cost is free

adoption: 2 dots

both maturities are: production/final