Other topics in this post:
Real World Experience: Charles Jaffe, MD, PhD
Real World Experience: Clay Shirky
The Online Health Forum launched on October 28th on the FACA Blog to elicit comments and opinions on your Health IT adoption experience (or your plans for EHR adoption). Thank you for your active participation, with nearly 200 posts to date and over 10,000 “hits.” True to the spirit of our Open Government philosophy, many of you extended the conversation throughout a number of more prominent health IT blogs, including John Halamka, Adam Bosworth, Sean Nolan, Wes Rishel, Bob Coffield, and Brian Ahier, among others.
Your posts reflected a diverse range of views on the most effective approach to standards adoption. Standards Committee member Marc Overhage initiated a discussion thread based on his experiences operating one of the nation’s most mature health information exchange organizations. Others discussed the TLS, vocabulary semantics, network security issues and privacy challenges. Numerous physicians chimed in with stories of the challenges and benefits of adopting electronic health records capable of delivering information compliant with health IT standards.
Our original intention was to seek public input on the work of the HIT Standards Committee to “pull forward” adoption. We identified seven topics (Standards, Interoperability, Vocabularies, Privacy, Security, Quality and Implementation Case Studies) to help frame the discussion. If you would like to comment on these seven topics, please start your comment by identifying the topic to which you are responding (i.e. “Vocabularies. Drawing from my experiences …”). Your self-categorization will help us to better organize and utilize your feedback.
Thank you for your comments in each of these areas. In addition, the Standards and Policy Committees have held public hearings on the topics of privacy and network security. Moving forward, a new work group focused on vocabularies will commence. We anticipate our Clinical Quality work group to build on the discussion of quality measures, with particular emphasis on the implications of our standards effort.
The final blog postings below from Dr Charles Jaffee, HL7, and Clay Shirky, Consultant/Internet technologies, are meant to summarize and further entice discussion and discourse on our seven topics and our interest in pulling forward standards adoption. We will keep the Online Health Forum open for your comments until December 1st. Following that, the Implementation work group will consolidate our ideas and make recommendations to the HIT Standards Committee in early 2010.
Again, thank you for your time and interest. You have contributed greatly to this important component of advancing our nation’s healthcare system. We look forward to your continued input.
Real World Experience: H7 Proposal Reduces Barriers to Health Information Exchange
Health Level 7 (HL7) has taken some bold steps to reduce the barriers to healthcare information exchange. This proposal has been over a year in development but resonates even more significantly with the efforts of the HIT Policy and Standards Committees. At the same time, we recognize the relevance of legacy health information systems that serve as the backbone of our current care model.
HL7 version 2 standards enable data exchange in virtually every hospital and healthcare system in the US. For many, these specifications have been sufficient for more than two decades. As we seek to find solutions for sharing healthcare information more broadly, additional steps must be taken to ensure seamless interoperability. Certainly, this includes the application and enhancement of structured vocabularies.
For many major research institutions, and certainly Vanderbilt, Duke and Indiana come to mind, version 2 poses no impediment to information exchange. In an earlier posting, Marc Overhage made a clear case for this approach. Certainly, this family of standards has been fundamental to the deployment of VISTA for the VA system.
At the same time, many healthcare systems have successfully deployed version 3 messaging, although this has been largely advanced by the activities of the UK NHS Connecting for Health program, by Canada Health Infoway’s achievements in connecting its provinces and territories, and by the Australian NeHTA initiatives. These programs have not been without their obstacles and miscues, but we have learned a great deal from their leadership. Perhaps more important, these national organizations have created a substantial body of collateral technology and supporting tooling and implementation resources, as well as deep technical expertise.
Closer to home, the NIH (National Cancer Institute) has led an alliance of health and research groups in the adoption of RIM-based technology. Certainly, the lofty goals of caBIG (Cancer Bioinformatics Grid) to enable data exchange among cancer research centers could not be achieved without the adoption of version 3 standards. Across the US, there has been widespread implementation of the clinical document architecture (CDA). CDA/CCD has been used to facilitate data exchange very broadly. For example, Kaiser sends tens of thousands of transaction to Microsoft daily for inclusion in its HealthVault, and the CDC (Centers for Disease Control and Prevention) relies upon CDA for exchanging thousand of documents daily with state and local health departments for biosurveillance. Lastly, the FDA (Federal Drug Administration) has begun a program that will require the use of CDA for its regulatory submissions from Pharma.
In parallel, HL7 has reached out to other standards development organizations that enhance our standards and specification. Many of these relationships have been remarkably productive, both on a national and international scale. Earlier this year, for example, the HL7 EHR Functional Model became an ISO standard. Our continued work with CDISC has likewise provided a much-needed overarching model for connecting data from patient care and clinical research. We also have collaborated with both the National Quality Forum to enable quality metrics and facilitate decision support. Lastly, we have begun a development process with NCPDP that further enhances to use of pharmacy data, which will undoubtedly improve efforts to support comparative effectiveness research.
Our efforts to collaborate with the clinical community have achieved significant recognition and success. Funded by the Agency for Healthcare Research and Quality (AHRQ), this initiative has garnered the support of more than 100 clinical professional societies worldwide. Not limited to physician groups, this organization, now known as the Clinical Information Interchange Collaborative (CIIC), recognized the need for the seamless exchange of clinical information between all caregivers, including nurses, pharmacists, therapists, dieticians, and researchers. This group has been able to inform the HIT community of its needs for common vocabulary, workflow and business requirements.
Perhaps most important, HL7 saw an opportunity to further reduce the barriers to implementation. We have submitted a proposal to HHS that would allow the licensing of HL7 intellectual property (including the standards and the supporting technology) free of charge for use in the United States. We have further proposed coupling this with several initiatives to streamline standards development and standards harmonization, while retaining our open, collaborative ANSI-approved process. As this moves forward, we expect to add momentum to the process of enabling “meaningful use” for all of our stakeholders.
Charles Jaffe, MD, PhD
CEO
Health Level 7, Inc.
The key phrase here, as Congress and HHS have recognized, is meaningful use. At its core, meaningful use means relevant health care information is available to improve care because the technology enables actual sharing among health care providers who care for the same patients, between those providers and the patient, and between those providers and institutions charged with monitoring public health.
If the meaningful use rule is to motivate such sharing, then a key to broadly distributed success will be limiting the number and cost of standards that must be implemented to the minimum set necessary.
It’s tempting to believe that the more complex and speculative a set of standards is, the further it will project participants into the future. But that model has actually been a template for the most significant specification and procurement disasters in IT history, including the failed IRS, FBI, and FAA database upgrade efforts. Of particular relevance here is the Government’s catastrophic support in the early 1990s for GOSIP (FIPS 146-1), instead of internet protocols. The GOSIP effort failed despite half a decade of effort and millions of dollars spent, because pure specification is no match for “rough consensus and running code,” David Clark’s famous description of the internet engineering model.
Progress is not going to be achieved by single-shot complexity, but by incremental simplicity. A simple, minimal set of standards is a requirement for including as many of the country’s small practices as possible. Small practices with just a handful of doctors provide a large proportion of the nation’s outpatient health care. Well-provisioned institutions may require high levels of technical complexity in their dealings with one another, but the success or failure of any nationwide meaningful use strategy will be determined in no small measure by the cost, expense, and hassle of providing small practices with tools and services that allow them to engage in the simple transactions.
The higher those barriers, the smaller the percentage of the healthcare system can participate. Beneath a certain level, we wouldn’t have broad meaningful use at all; we would limited exchange between large institutions in large cities. Systems that don’t join at the beginning will see the cost and difficulty of joining rise, rather than fall. On the other hand, using simple tools to raise the technical acumen once participants have joined a working network is in everyone’s interest; inclusion becomes a way to drive rising technical sophistication, but the corollary is not true.
Thus the question for broad participation in meaningful use is not: “What will the most complete system look like for the richest and most technically adept institutions?” Rather, it is: “What’s the simplest and most low cost way for a small vendor or new market entrant to get a small practice tied in?” This test won’t exclude the technically adept institutions from using more complex tools and standards, but it will keep the design of any set of national standards from locking out the bulk of the nation’s health care providers from the most basic ones.
The simplest set of standards will address both secure transport (the movement of arbitrary data from point A to point B) and content (the encoding of particular kinds of data, such as labs or meds.) These standards must also be co-developed with policies relating to patient privacy and constraints on data use. Clear policies governing both network and site security are necessary to reduce risks from both commercial and criminal misuse.
The transport standards will allow for secure, point-to-point transactions between two known entities. There may be more complex transactions that will use other communications patterns, but a system too complex to allow a basic point-to-point transaction will be de facto inaccessible to most potential users.
Here’s what a workable set of transport standards will not do: It will not assume to know what kind of applications any given network participant is running locally. Once the data are delivered, it should be usable by everything from the simplest to the most complex application, since the recipient of the data will have the best understanding of what works in their local context.
This ability to separate data from transport and applications from data is the essential pre-condition for innovation — a group that has a valuable new idea for presentation of data for clinical use should not also be forced to think about the data encoding or the way the data are transported. Groups working on new data encodings should not be tied to a pre-existing suite of potential applications, nor should they have to change anything in the transport layer to send the new data out, and so on.
The content standards should ratify those encodings already in wide enough use, such as DICOM and LOINC. Where there is broad agreement, the standards will document that agreement, and lower the adoption costs for all subsequent users. (And, of course, the transport and content standards must be kept separate, so that any content can travel over the transport layer, and any part of the transport layer can handle arbitrary content.)
And a workable set of content standards will not try to create standards that are yet to be tried in the real world, nor will they try to forestall competition between standards with large constituencies. Standards have two components — specification and adoption. Specification means getting the expressive content of the standard right. Specification is the theoretical part of standardization.
Adoption is the number of people or institutions who use a standard in a real-world setting. The more people using a standard, the more effort goes into documenting it, debugging it, integrating it into various hardware and software platforms, and so on. Other things being equal, the better adopted a standard becomes, the more widely and effectively it works, because the theoretical parts of the specification are worked out in practice.
If that were all there were to it, we might be able to imagine a simple pipeline, where fully detailed standards are designed in theory, shipped out to their eventual users, and then implemented. This pipeline model, however, doesn’t work, because the relationship between specification and adoption is more complex than that.
The essential asymmetry between specification and adoption is that better adoption drives tighter specification, as various implementing parties have to work together to get their work done, but tighter specification doesn’t drive better adoption. As a result, the model of incremental simplicity — specify the minimum to get participants in the system, observe the subsequent use, and specify the next degree of standards — is actually better for large-scale adoption than systems that target a high degree of complexity for a small number of actors.
Imagining a network simple enough that a small practice can participate in basic transactions has the downside of being less than perfect. It will not deliver a Big Bang, in which participants suddenly operate with complete technical flexibility and semantic clarity. In place of perfection, however, a minimally necessary approach to standards would have one significant upside: it might actually work.
Clay Shirky
Tags: FACA, Federal Advisory Committee Act, HIT Standards Committee, ONC
