Mixing, matching and missing

It's no secret that early work on the proposed National Health Information Network (NHIN) has focused on standards. Without broad agreement on common ways of building electronic health records (EHRs) systems, that vision is an unrealizable dream.

The good news is that, for the most part, there's little need to build standards from scratch. The diverse health care industry has devised many ways to electronically represent the complex information used to describe a patient's care, drug and treatment regimen.

But that's also part of the problem. Dozens of elements make up an EHR, and each can describe information differently. Medications and allergies are two elements that will always be in a record, for example. But there are many standardized ways to describe one medication, and no recognized standards exist for allergies.

So far, most organizations that have set standards have operated in isolation. Most health records agreements occur between hospitals and clinics in the same geographical region; they have business reasons for building compatible systems to electronically swap data.

The formation of regional health information organizations extends the reach of those agreements to more medical facilities. But such standards, though broader, are still localized. An electronic record kept by a facility in one region would probably look like gobbledygook to the system a facility in another region uses.

It is no surprise then that standards were one of the top concerns that individuals and organizations identified in responses to a request for information about what is needed to build the NHIN.

Respondents overwhelmingly agreed that standards are crucial to achieve the interoperability needed for the NHIN, according a report summarizing the 512 responses released by the Office of the National Coordinator for Health Information Technology (ONCHIT) at the Department of Health and Human Services.

In addition to that consensus, the report states that "there was great variety of opinion regarding who should create the standards, how they should be created and maintained, and which standards should be considered to achieve interoperability."

Where to begin

Before members of the health care community can decide on standards, they must understand the role those standards would play, said Rod Piechowski, vice president of technology leadership at the National Alliance for Health IT, a partnership of organizations from all health care sectors.

The alliance "has identified some 1,200 health care standards that are in use today," he said. "So we need to decide on which of those can be applied to the NHIN. We're in the process now of defining a collective vision of what those standards need to do."

Health care organizations must decide not only what standards will apply to which elements of a patient's EHR and how the NHIN will transmit that data, but also how those standards will work together. Those organizations must agree on the most basic definitions, such as what interoperablity means in reference to the NHIN.

The alliance took a necessary first step earlier this year by producing one of the first widely supported interoperability definitions in the health care arena.

"In health care, interoperability is the ability of different [IT] systems and software applications to communicate, to exchange data accurately, effectively and consistently, and to use the information that has been exchanged," the alliance's definition states.

That definition is the goal of interoperability, Piechowski said, adding that it's also intended to serve as a guiding principle for technical specifications and standards development.

Some technical requirements for interoperability are fairly clear.

Michael Solomon, vice president of strategic planning at IDX Systems, a health care solutions provider, sees two main needs: the messaging and transport layer of the network — how the data physically gets from one point on the NHIN to another — and the data layer, which requires descriptions of the message that carries the data.

Transport layer standards are fairly easy to deal with because the consensus is that Web-based technologies will deliver medical records via the Internet. So the NHIN is likely to use already well-accepted IP-based and Web services standards, such as Extensible Markup Language and its variants.

But the data layer is another matter. Solomon said the needs fall into two categories: one for the message format and another for content.

"The format is the schema for what the message should look like," he added. "And while there are still some growing pains, for the most part, I'd say we also have the standards that would ensure interoperability."

Although not as entrenched as IP-based standards, a number of health messaging standards are already in use, including Health Level 7's standards for clinical and administrative data, the Digital Imaging Communications in Medicine standards for describing images and diagnostic information and the American National Standards Institute X.12 standards for defining various business and administrative functions.

But a standard for message formats does not help without standards for interpreting messages' content.

Randy Thomas, vice president of advisory services at Healthlink, an IBM company, compared it with semantic differences that occur when people use different languages and vocabularies to talk about a particular issue.

When people speak the same language and use the same vocabulary, the topic at hand is usually transparent, she said. But when a conversation requires a translator, then something is often missing, even if the terms are translated correctly.

"So we've all got to learn a new language, and these [interoperable health] systems need to be working with a single vocabulary, or at least a common set of vocabularies," she said.

Some standards are available in this area. The Systematized Nomenclature of Medicine Clinical Terms (SNOMED CT), for example, is an increasingly popular terminology standard for health care organizations. HHS announced in May last year that it would make the English edition of SNOMED CT freely available through the National Library of Medicine.

But health care organizations have been slow to adopt those standards because it's not easy to adapt long-established commercial products that use different terminologies, Thomas said.

"These vocabularies go to the heart of how commercially available systems are designed and built," she said. "They're not something that can be easily bolted on."

Janet Dillione, chief operating officer of Siemens Medical Solutions' health IT division, said the standards discussion relates to what needs to be included in the common patient dataset that medical facilities will share with one another via the NHIN.

To be useful, the patient data would need to include precise and clearly understood clinical terms about details such as recent scans, medicinal allergies and recent treatments. Such information is critical, particularly when the patient is unconscious or cannot provide that information.

"That is fundamental to dealing with patients in a facility separate from where they may have been treated earlier," Dillione said. "The problem is that those facilities may have different terms for things, even when they are talking about the same thing."

The state of content-related standards is spotty. In some areas, such as laboratory tests and procedures, the standards are sophisticated, which provides a solid foundation for building interoperability, Dillione said. But in other areas, such as tracking allergies, there is little to work with.

"Even with relatively complete standards such as SNOMED and [the Logical Observation Identifiers, Names and Codes,] there are still gaps," Dillione said. "Nothing is complete enough to handle the breadth of what's needed."

A total semantic understanding requires more than agreement on terms and vocabulary, said Thomas Jones, chief medical officer at Oracle. Different operations in an organization have different needs for the data.

"Emergency room doctors and nurses, for example, will use the information in these records differently from other people," he said, "So you also have to be able to transform the data to fit the many different views, the different virtual records, that people will need."

More to be done

Some standards are more than a matter of technical specifications.

A lot of work, for example, must be done to develop security standards and best practices specific to health care interoperability.

Also, in addition to questions of data and message format, policies and procedures are necessary to ensure proper data transfers. The United States, for example, does not assign national patient identification numbers, so a unique system must be developed for the NHIN. Even with that system in place, policies must dictate how and when physicians can access patients' records.

Such policies might apply nationwide, but others will have to be fitted to state and local regulations.

"We'll need policies that are technically configurable according to such things as geography, whether a patient can opt in or out of the process and in what circumstance, and that can also be applied facility by facility," said Dan Garrett, vice president and managing partner of Computer Sciences Corp.'s Global Health Solutions. "This is what people are beginning to understand is needed for [interoperability] in the real world. But it's not really being written or talked about yet."

However, health care standards may be close to reality — certainly faster than has happened in other fields — largely because of the federal government's intervention.

In June, the ONCHIT released four requests for proposals related to health care interoperability. One RFP calls for developing processes that would lead to unification and harmonization of industrywide health IT standards.

That RFP addresses standards that are similar or at least have significant overlaps in meaning and intent, creating a need to identify and merge them. There's also a need to understand how those various standards will operate with one another on the NHIN.

Congress has also been busy producing legislation to support health care IT, much of which also stresses standards development.

No one expects the imminent construction of a complete standards environment. But there's considerable optimism that a quick agreement could be reached on at least a core set of standards that will allow the first stages of the NHIN to begin.

"The next six or seven months will really be an interesting time," Garrett said.