Could Clinicians Create Better HIE Tools?

The following is a guest blog post by Andy Oram.His post reminds me of when I asked “Is Full Healthcare Interoperability a Pipe Dream?“

A tense and flustered discussion took place on Monday, August 11 during a routine meeting of the HIT Standards Committee Implementation Workgroup, a subcommittee set up by the Office of the National Coordinator (ONC), which takes responsibility for U.S. government efforts to support new IT initiatives in the health care field. The subject of their uncomfortable phone call was the interoperability of electronic health records (EHRs), the leading issue of health IT. A number of “user experience” reports from the field revealed that the situation is not good.

We have to look at the depth of the problem before hoping to shed light on a solution.

An interoperability showcase literally takes the center of the major health IT conference each year, HIMSS. When I have attended, they physically arranged their sessions around a large pavilion filled with booths and computer screens. But the material on display at the showcase is not the whiz-bang features and glossy displays found at most IT coventions (those appear on the exhibition floor at HIMSS), but just demonstrations of document exchange among EHR vendors.

The hoopla over interoperability at HIMSS suggests its importance to the health care industry. The ability to share coordination of care documents is the focus of current government incentives (Meaningful Use), anchoring Stage 2 and destined to be even more important (if Meaningful Use lasts) in Stage 3.

And for good reason: every time we see a specialist, or our parent moves from a hospital to a rehab facility, or our doctor even moves to another practice (an event that recently threw my wife’s medical records into exasperating limbo), we need record exchange. If we ever expect to track epidemics better or run analytics that can lower health case costs, interoperability will matter even more.

But take a look at extensive testing done by a team for the Journal of the American Medical Informatics Association, recently summarized in a posting by health IT expert Brian Ahier. When they dug into the documents being exchanged, researchers found that many vendors inserted the wrong codes for diagnoses or drugs, placed results in the wrong fields (leaving them inaccessible to recipients), and failed to include relevant data. You don’t have to be an XML programmer or standards expert to get the gist from a list of sample errors included with the study.

And that list covers only the problems found in the 19 organizations who showed enough politeness and concern for the public interest to submit samples–what about the many who ignored the researchers’ request?

A slightly different list of complaints came up at the HIT Standards Committee Implementation Workgroup meeting, although along similar lines. The participants in the call were concerned with errors, but also pointed out the woeful inadequacy of the EHR implementations in representing the complexities and variety of patient care. Some called for changes I find of questionable ethics (such as the ability to exclude certain information from the data exchange while leaving it in the doctor’s records) and complained that the documents exchanged were not easy for patients to read, a goal that was not part of the original requirements.

However, it’s worth pointing out that documents exchange would fall far short of true coordinated care, even if everything worked as the standards called for. Continuity of care documents, the most common format in current health information exchange, have only a superficial sliver of diagnoses, treatments, and other immediate concerns, but do not have space for patient histories. Data that patients can now collect, either through fitness devices or self-reporting, has no place to be recorded. This is why many health reformers call for adopting an entire new standard, FHIR, a suggestion recognized by the ONC as valid but postponed indefinitely because it’s such a big change. The failure to adopt current formats seems to become the justification for keeping on the same path.

Let’s take a step back. After all those standards, all those certifications, all those interoperability showcases, why does document exchange still fail?

The JAMIA article indicated that failure can be widely spread around. There are rarely villains in health care, only people pursuing business as usual when that is insufficient. Thus:

• The Consolidated CDA standard itself could have been more precisely defined, indicating what to do for instance when values are missing from the record.

• Certification tests can look deeper into documents, testing for instance that codes are recorded correctly. Although I don’t know why the interoperability showcase results don’t translate into real-world success, I would find it quite believable that vendors might focus on superficial goals (such as using the Direct protocols to exchange data) without determining whether that data is actually usable.

• Meaningful Use requirements (already hundreds of pages long) could specify more details. One caller in the HIT Standards Committee session mentioned medication reconciliation as one such area.

The HIT Standards Committee agonized over whether to pursue broad goals, necessarily at a slow pace, or to seek a few achievable improvements in the process right away. In either case, what we have to look forward to is more meetings of committees, longer and more mind-numbing documents, heavier and heavier tests–infrastructure galore.

Meanwhile, the structure facilitating all this bureaucracy is crumbling. Many criticisms of Meaningful Use Stage 2 have been publicly aired–some during the HIT Standards Committee call–and Stage 3 now looks like a faint hope. Some journalists predict a doctor’s revolt. Instead of continuing on a path hated by everybody, including the people laying it out, maybe we need a new approach.

Software developers over the past couple decades have adopted a range of ways to involve the users of software in its design. Sometimes called agile or lean methodologies, these strategies roll out prototypes and even production systems for realistic testing. The strategies call for a whole retooling of the software development process, a change that would not come easily to slow-moving proprietary companies such as those dominating the EHR industry. But how would agile programming look in health care?

Instead of bringing a doctor in from time to time to explain what a clinical workflow looks like or to approve the screens put up by a product, clinicians would be actively designing the screens and the transitions between them as they work. They would discover what needs to be in front of a resident’s eyes as she enters the intensive care ward and what needs to be conveyed to the nurses’ station when an alarm goes off sixty feet away.

Clinicians can ensure that the information transferred is complete and holds value. They would not tolerate, as the products tested by the JAMIA team do, a document that reports a medication without including its dose, timing, and route of administration.

Not being software experts (for the most part), doctors can’t be expected to anticipate all problems, such as changes of data versions. They still need to work closely with standards experts and programmers.

It also should be mentioned that agile methods include rigorous testing, sometimes to the extent that programmers write tests before writing the code they are testing. So the process is by no means lax about programming errors and patient safety.

Finally, modern software teams maintain databases–often open to the users and even the general public–of reported errors. The health care field needs this kind of transparency. Clinicians need to be warned of possible problems with a software module.

What we’re talking about here is a design that creates a product intimately congruent with each site’s needs and workflow. The software is not imported into a clinical environment–much less imposed on one–but grows organically from it, as early developers of the VistA software at the Veterans Administration claimed to have done. Problems with document exchange would be caught immediately during such a process, and the programmers would work out a common format cooperatively–because that’s what the clinicians want them to do.