Wednesday, April 18, 2012

The April HIT Standards Committee

The April HIT Standards Committee included a comprehensive review of the Standards & Certification Criteria Notice of Proposed Rulemaking (NPRM) by each workgroup/task force/power team during a 5 hour marathon session.   The capstone of the meeting with a thematic review of the entire NPRM by the Patient Engagement Power Team.

The meeting began with a "chapter and verse" review of each NPRM standard.

Here are the section by section highlights of our discussion:

§ 170.314(b)(3) E-prescribing
The NPRM requires NCPDP Script 10.6 for content and RxNorm as the vocabulary.  We also recommended that HL7 2.x be allowed for the highly constrained use case of pharmacies located within a hospital as part of an organized healthcare arrangement, since such interfaces are widely implemented today.

§ 170.314(b)(3) Demographics
The NPRM requires OMB standards for race and ethnicity, ISO 639-1 subset of 639-2 for language and ICD10 (not ICD10-CM) for Cause of Death, which has been used by government to code cause of death since since 1999.   We noted that ICD-10 CM has been delayed and recommended it be removed entirely from the NPRM.    We recommended that the HIT Policy Committee consider the use of ISO 3166-2 if country of birth is a desirable data element.   We also considered the use of a more granular race/ethnicity code set such as CDC, but deferred this to the next edition of Meaningful Use.

§  170.314(a)(4) Vital Signs and other observations
The NPRM does not require any standards in this area currently.   We referred to the September 2011 vocabulary and code sets recommendations and suggested that ONC specify LOINC for specific study name, SNOMED-CT for appropriate findings, and UCUM for specific units of measure for all structured observations including vital signs.  We also suggested that patient experience, such as pain scales, be recorded using LOINC for assessment instruments and SNOMED-CT for appropriate responses.

§  170.314(a)(5) Problem Lists
The NPRM requires SNOMED-CT for problem lists.  We recommended that the HIT Policy Committee consider the use of a field that represents an administrative categorization of the visit/billing diagnosis to support secondary uses of data but did not make a specific standards recommendation.

§  170.314(a)(8) Clinical Decision Support
The NPRM specifies the use of the "Infobutton" standard.   We recognize that clinical decision support includes access to educational materials, alerts, and reminders.   We felt that a functional description of capabilities rather than a specific standards requirement would best achieve the policy goals.   We made "Infobutton" optional, not a certification criteria, to encourage its adoption but also enable innovation.

§  170.314(a)(17) Electronic Medication Administration Record
The NPRM does not constrain the technological approach to ensuring the right medication is given to the right patient via the right route at the right dose at the right time.  The NPRM specifies the use of NTP to synchronize clocks.   We agree that the technology for EMAR should not be over specified, enabling innovation.

§  170.314(e)(2) Clinical Summmaries
The NPRM specifies the use of Consolidated CDA and several vocabularies (OMB Race/Ethnicity, ISO 639-1 subset of ISO-639-2 for language, SNOMED-CT for problems, LOINC for labs, and RxNorm for medications).   We recommended that ICD-10 PCS be replaced with SNOMED-CT for procedures, completing our recommendation to remove ICD-10 CM and ICD-10 PCS entirely from the NPRM.

§  170.314(a)(14) Patient Lists
The NPRM provides a functional description of capabilities rather than a standard for the generation of lists of patients matching specific clinical criteria.   We noted that no current standard exists to transmit lists of patients in batch to registries and repositories, although this is desirable for the next edition of Meaningful Use.

§  170.314(a)(16) Patient Education
The NPRM requires Infobutton.   As with Decision Support, we recommended that Infobutton be an optional standard, not required for certification.

§  170.314(b)(1) and § 170.314(b)(2) Transition of Care Summaries
The NPRM specifies the use of Consolidated CDA and several vocabularies (OMB Race/Ethnicity, ISO 639-1 subset of ISO-639-2 for language, SNOMED-CT for problems, LOINC for labs, and RxNorm for medications).   It also recommends the use of the Direct implementation guide, XDR/XDM, and optionally SOAP for transport.   We had a robust discussion about the parsimonious approach to transport.    We concluded that the best approach would be to require the Direct implementation guide and encourage the use of XDR/transport standards of NwHIN Exchange by listing XDR and the S&I Framework Implementation Guide for NwHIN Exchange transport as optional, not certification criteria.   This was truly an achievement - we succeeded in specifying one transport standard requirement for every EHR.  Wes Rishel made an important point about the need to support "bilateral asynchronous upgrades" - the notion that different versions of transport Implementation Guides might be used over time by senders and receivers.   We have to be very careful to ensure backward compatibility as our transport standards evolve.

For the remaining NPRM sections, we worked by the principle of consent unless otherwise noted by standards committee members.

§  170.314(e)(1) View, download, and transmit to 3rd party.
We recommended that patient download capability be required to use Consolidated CDA as a minimum.  We noted that TLS is an example of a means to secure the transmission channel but also recognized other approaches are possible.   We recommended functional criteria for securing endpoints rather than a named standard.

§  170.314(d)(2)  Auditable events and tamper-resistance.
We noted redundancy in the NPRM - audit logs cannot be modified, but there is a need to detect modification.  We also recommended that ASTM E2147 be used as a list of audit log data elements.

§  170.314(a)(13) Family History
The NPRM does not require a specific standard.  We concur that no standard is widely implemented today, although the Surgeon General's XML for family history has been used more than the HL7 pedigree standards.

§  170.314(f)(7) and § 170.314(f)(8) Cancer Registry Reporting
We noted that the CDA Cancer Registry standard is not deployed in production.   We also noted that this refers to a menu set item, so it is not a general requirement of EHR certification. It is a reasonable early standard for oncology specific EHRs.

§  170.314(c)(1)-(3) NQF Quality Data Model
We noted that additional work on the Quality Data Model will be needed to ensure it aligns with all the other recommendations we made at today's meeting.    We also noted that QRDA Category II and III are not yet balloted standards.

The Standards Committee reaffirmed its acceptance of all these recommendations by consensus.   A remarkable achievement.

Leslie Kelly Hall then presented the guiding principles of the Consumer/Patient Engagement Power Team, a truly remarkable body of work assembled in very limited time.

Our next process step is to draft and review a final transmittal letter to ONC, which we'll do over the next few weeks.

The HIT Standards Committee is such an effective group with amazing expertise and camaraderie that today's meeting was a perfect storm moment.   This is definitely a time we'll all be telling our grandchildren about.


Lygeia Ricciardi said...

Thanks for writing up the meeting, John. I am so proud of the work you are all doing! Thanks especially to Leslie and the other members of the Consumer/Patient Engagement Power Team!

Jeremy Engdahl-Johnson said...

ICD-10 deadline to be pushed back a year? What are the implications of the delay?

David said...

Hi John, great writeup as usual. Regarding Direct SMTP as the only certified protocol, I have a concern about unintended consequences if CMS takes literally that the only exchanges that "count" for MU are those that ONC certifies. That would mean that other transitions, including those mediated by HIEs, would not "count." I blogged about the implications of this at