Monday, December 17, 2018

Common Sense in C-CDA: Comparing Carequality/CommonWell C-CDA to r2.1

The history of shared clinical documents is marked by decades of industry-wide deliberation.  In recent years, catalyzed by ONC Certification, there has been widespread adoption and refinement of the standard. 

Data EHRs were required to record and share under ONC Certification were originally called Meaningful Use Data and represented a list of basic chart elements. These were later reformulated as the more expansive Common Clinical Data Set (CCDS) in 2015 Edition Certification. The emergence of 2015 Edition criteria also removed the requirement for the encounter-based document previously required in 2014 Edition - the ambulatory Clinical Summary - making the test data for the latest edition of Health IT certification effectively patient-based.

The current ONC-sanctioned R2.1 version of the C-CDA document, upon which the sharing of these minimum data elements is based, includes long-discussed enhancements that have lead to much greater clarity. But there remain issues of reliability, relevance and provenance surrounding the standard and its implementation, particularly at the encounter level. Some of these issues, particularly with respect to provenance, are being addressed in the underlying data with the Draft U.S. Core Data for Interoperability (USCDI), which is slated to replace the Common Clinical Data Set (CCDS).

There have also been efforts provide constructive recommendations on the structure of the document itself.  Earlier this year, Carequality and CommonWell Content Work Groups released a white paper, using the C-CDA R2.1 Companion Guide as a baseline to provide "complementary, not conflicting guidance." CommonWell is a not-for-profit trade association focused on interoperability, while Carequality represents a similar industry-wide collaborative, convening healthcare stakeholders on interoperability issues.

The goal of the collaboration was a more clinically-relevant and parsimonious C-CDA. The collaborative work group tackled the following issues:
  • "Unacceptably large" C-CDA documents
  • A general absence of clinical notes 
  • Need for encounter summary support
  • Need for version management

Clinical Notes should follow the encounter. (Credit: Max Pixel)
The white paper has a wide range of common-sense recommendations, but we'll discuss some of the recommendations most relevant to implementing CCDA R2.1 under 2015 Edition Certification and USCDI (which is slated to replace CCDS).
  • Problems - only those addressed during the encounter 
  • Allergies - "only if the system can recreate the active Allergy list at the time of encounter"
  • Medications - "only if the system can recreate Medications at time of encounter"
  • Immunizations - those given during the encounter

While conditionality in general can get thorny, these recommendations make a good deal of clinical sense. If events are not touched by the encounter, updating and sending them makes little sense.

Specific to the USCDI, the work group judged the data set to contain "valuable data elements and should be exchanged to improve patient care." But they go on to say that Clinical Notes should not simply be a data dump: when they are supported, support for Encounter Summary documents should be added to ensure the note follows the encounter.

The recommendations focus on Encounter Summary documents (Progress Note Document and Discharge Summary) and preserve all sections required in the base C-CDA document template. After setting that foundation, they map out a "priority subset" of clinical data from the ONC Common Clinical Data Set (CCDS) and draft US Core Data for Interoperability (USCDI). The idea is to weed out data that is stale or irrelevant to the clinical encounter.

The prioritized sections for 'always include' have data within them that should appear conditional based on the events of the encounter:

In theory at least, the C-CDA as tested under 2015 Edition would become easier to interpret and implement. There also a few general guidelines for non-priority elements:
Systems SHOULD send a ‘No information’ assertion template if nothing is available for one of the priority subset data elements. 
Systems MAY send additional data elements, beyond the priority subset, if relevant to the encounter. For these additional data elements, systems should not send a ‘No information’ template if nothing is available.
The concept of 'no information,' while ostensibly straightforward, is crucial here. It's the "known unknown," that provides more certainly to the receiver of the C-CDA that in fact nothing clinically relevant appears under that section, as opposed to having been omitted for an unknown reason.

The recommendations also touch the world of Fast Healthcare Interoperability Resources (FHIR), which implications for how its resources are used. The CommonWell-Carequality alliance is deeply involved in FHIR workgroups and its recommendations may, through governance of the standard, help shape efforts to ensure C-CDA and FHIR resource play more nicely together. Currently, there is a significant amount of subjectivity in getting the sections of a C-CDA in and out of FHIR.

Dynamic Health IT has been reviewing its C-CDA practices and guidance to make sure we help our clients send and receive documents that are clinically-relevant and readable. True interoperability requires nothing less.

We're looking forward to seeing you at the C-CDA IAT meeting. You can find the Track Agenda here.

Friday, December 7, 2018

ONC Annual Meeting 2018 Recap

On November 29-30, 2018, Health IT policymakers and implementers met in DC for the Annual ONC Meeting. As always, the agenda was full to the brim with discussions on the state-of-the-health IT industry.

Interoperability and the future of 2015 Edition CEHRT were central themes, but the agenda content reveals just how far-reaching ONC's influence reaches - and by extension, the use of health IT systems and applications. Sessions tackled subjects as wide-ranging as care coordination, interoperability strategies, APIs, disaster response and opioid prescribing.

Image result for washington dcJeff Robbins, Dynamic Health IT President, attended the meeting. His enthusiastic takeaway was that the conference was very beneficial, providing a survey trends in health IT today and learn about innovative approaches to our most pressing challenges. While it's impossible to attend all sessions (or even to summarize them in a single blog post), we'll review a few highlights.

The breakout session “Improving Opioid Prescribing through Electronic Clinical Decision Support Tools” focused on the solving the challenges posed by the Opioid Epidemic in the Prescription Drug Monitoring Program (PDMP). As part of monitoring, prescribing physicians can be securely notified of non-fatal overdose episodes, but this is currently uncommon. By connecting providers to PDMPs more widely, we can ensure that they have timely information that can limit drug-seeking, modulate prescription behavior and point to important connections to limit the spread of opioid abuse.

While we acknowledge the depth of the challenges facing the industry currently - particularly with respect to interoperability - there were plenty of success stories in evidence. DHIT was impressed by case studies showing enthusiastic adoption of the HL7 Fast Healthcare Interoperability Resource (FHIR) data standard for provider-payer data exchange. A private sector-led initiative — the Da Vinci Project — has progressed quickly enough to enlist government sponsorship from the ONC and then spawned the P2 FHIR Taskforce. The task force is supporting FHIR development efforts and turning its focus to challenges preventing adoption.

The “FHIR Implementation and Transition Planning” session was another opportunity to discuss EHR vendor Application Programming Interfaces (APIs). The 21st Century Cures Act calls for the development of APIs to promote data-sharing and the ONC 2015 Edition Test Method includes three measures (g7, g8, g9) encompassing patient data APIs - though neither explicitly mandate FHIR. During the Q&A session, Cerner and Epic were in the hot seat with application and personal health record (PHR) developers seeking broader access to their FHIR APIs.

On Friday, a lively panel discussion on “Data and Value-Based Care" was held. Economist Mark Pauly discussed the proverbial elephant in the room - a subject that is often avoided:  cost/benefit of healthcare and the need to set dollar limits on individual healthcare plan coverage.  Mark posed the hypothetical of whether there should be marginal dollar limits for coverage.

Finally, it's worth reflecting on the future of ONC's latest health IT test method: 2015 Edition CEHRT. With funding in questions and transition in both programs, Certification and Meaningful Use have both seen grave pronouncements within the last few years. But 2015 Edition CEHRT is still required in 2019 for providers reporting MIPS. And for those providers targeting the “end-to-end bonus" for eCQM reporting, that must be done in conjunction with a system that meets 2015 Edition CEHRT.

Our challenges - and the policy initiatives designed to meet them - aren't going anywhere.