Schlag and Froth: Argonauts Navigate Between Heavy-weight and Light-weight Standardization (Part 2 of 2)

The previous section of this article laid out the context for HL7 FHIR standard and the Argonaut project; now we can look at the current status.

The fruits of Argonaut are to be implementation guides that they will encourage all EHR vendors to work from. These guides, covering a common clinical data set that has been defined by the ONC (and hopefully will not change soon), are designed to help vendors achieve certification so they can sell their products with the assurance that doctors using them will meet ONC regulations, which require a consumer-facing API. The ONC will also find certification easier if most vendors claim adherance to a single unambiguous standard.

The Argonaut implementation guides, according to Tripathi, will be complete in late September. Because FHIR is expected to be passed in September 2017, the Argonaut project will continue to refine and test the guides. One guide already completed by the project covers security authorization using OpenID and OAuth. FHIR left the question of security up to those standards, because they are well-established and already exist in thousands of implementations around the Web.

Achieving rough consensus

Tripathi portrays the Argonaut process as radically different from HL7 norms. HL7 has established its leading role in health standards by following the rules of the American National Standards Institute (ANSI) in the US, and similar bodies set up in other countries where HL7 operates. These come from the pre-Internet era and emphasize ponderous, procedure-laden formalities. Meetings must be held, drafts circulated, comments explicitly reconciled, ballots taken. Historically this has ensured that large industries play fair and hear through all objections, but the process is slow and frustrates smaller actors who may have good ideas but lack the resources to participate.

In contrast, FHIR brings together engineers and other interested persons in loose forums that self-organize around issues of interest. The process still tried to consider every observation and objection, and therefore, as we have seen, has taken a long time. But decision-making takes place at Internet speed and there is no jockeying for advantage in the marketplace. Only when a milestone is reached does the formal HL7 process kick in.

The Argonaut project works similarly. Tripathi reports that the vendors have gotten along very well. Epic and Cerner, the behemoths of the EHR field, are among the most engaged. Company managers don’t interfere with engineer’s opinions. And new vendors with limited resources are very active.

Those with a background in computers can recognize, in these modes of collaboration, the model set up by the Internet Engineering Task Force (IETF) decades ago. Like HL7, the IETF essentially pre-dated the Internet as we know it, which they helped to design. (The birth of the Internet is usually ascribed to 1969, and the IETF started in 1986, at an early stage of the Internet. FTP was the canonical method of exchanging their plain-text documents with ASCII art, and standards were distributed as Requests for Comments or RFCs.) The famous criteria cited by the IETF for approving standards is “rough consensus and running code.” FHIR and the Argonauts produce no running code, but they seem to operate through rough consensus, and the Argonauts could add a third criterion, “Get the most important 90% done and don’t let the rest hold you up.”

Tripathi reports that EHR vendors are now collaborating in this same non-rivalrous manner in other areas, including the Precision Medicine initiative, the Health Services Platform Consortium (HSPC), and the SMART on FHIR initiative.

What Next?

The dream of interoperability has long included the dream of a marketplace for apps, so that we’re not stuck with the universally hated EHR interfaces that clinicians struggle with daily, or awkwardly designed web sites for consumers. Tripathi notes that SMART offers an app gallery with applications that ought to work on any EHR that conforms to the open SMART platform. Cerner and athenahealth also have app stores protected by a formal approval process. (Health apps present more risk than the typical apps in the Apple App Store or Google Play, so they call more more careful, professional vetting.) Tripathi is certain that other vendors will follow in the lead of these projects, and that cross-vendor stores like SMART’s App Gallery will emerge in a few years along with something like a Good Housekeeping seal for apps.

The Argonaut guides will have to evolve. It’s already clear that EHR vendors are doing things that aren’t covered by the Argonaut FHIR guide, so there will be a few incompatible endpoints in their APIs. Consequently, the Argonaut project has a big decision to make: how to provide continuity? The project was deliberately pitched to vendors as a one-time, lightweight initiative. It is not a legal entity, and it does not have a long-term plan for stewardship of the outcomes.

The conversation over continuity is ongoing. One obvious option is to turn over everything to HL7 and let the guides fall under its traditional process. A new organization could also be set up. HL7 itself has set up the FHIR Foundation under a looser charter than HL7, probably (in my opinion) because HL7 realizes it is not nimble and responsive enough for the FHIR community.

Industries reach a standard in many different ways. In health care, even though the field is narrow, standards present tough challenges because of legacy issues, concerns over safety, and the complexity of human disease. It seems in this case that a blend of standardization processes has nudged forward a difficult process. Over the upcoming year, we should know how well it worked.

About the author

Andy Oram

Andy Oram

Andy Oram writes and edits documents about many aspects of computing, ranging in size from blog postings to full-length books. Topics cover a wide range of computer technologies: data science and machine learning, programming languages, Web performance, Internet of Things, databases, free and open source software, and more. My editorial output at O'Reilly Media included the first books ever published commercially in the United States on Linux, the 2001 title Peer-to-Peer (frequently cited in connection with those technologies), and the 2007 title Beautiful Code. He is a regular correspondent on health IT and health policy for He also contributes to other publications about policy issues related to the Internet and about trends affecting technical innovation and its effects on society. Print publications where his work has appeared include The Economist, Communications of the ACM, Copyright World, the Journal of Information Technology & Politics, Vanguardia Dossier, and Internet Law and Business.