Q and A With Dr. Ed Hammond

PDF File: 

HCT3_wp_hammond.pdf

Duke University

Professor Emeritus at Duke University explains the historical evolution oftechnical standards.

HCT: From a technology standpoint, what’s required for EHR implementation?

Ed Hammond: I think it’s important to discuss, briefly, what an EHR (electronic health record) is. It seems to me that historically, almost every organization that starts to do something with the EHR attempts to define what an EHR is. And, invariably, that is always an open issue. There is no universal definition of that term. Clearly an EHR serves multiple roles. Its major role is documenting the care of the patient. In addition to that, increasingly in the future if not so much in the past, it’s been the source of information for research and analysis. It certainly performs a legal service in the sense of documenting what happened to the patient. It’s the source of decisions related to reimbursement. So there’s a multiplicity of purposes.

The EHR also is used in many different sites; most specifically the difference between the EHR used in the inpatient environment may be very different from the EHR in the outpatient environment. In the inpatient environment, largely it is a process of dealing with an acute event. The data that’s collected is stored in the EHR and has a more immediate value. You do a test, analyze the data and do a treatment. So you’re trying to solve an acute care problem where your changes are rather rapid. And the long-term value of that, I think, is much more limited than it is in the outpatient world. In the outpatient world, the data tends to change more slowly. Much of that data is focused on either chronic problems or in screening problems and trying to determine the health status of an individual, preventative care and that type of thing, and thus changes occur at a slower rate and therefore the data has more persistent value than it does in the inpatient care setting. I think those are two distinctions that need to be combined in the EHR. Another factor that impacts the EHR is that most patients go to more than one site for care.

A recent study suggested that patients see from three to five different providers in different sites of care. To get a complete picture of the patient, we need to be able to bring all of that data together in a single EHR to understand what’s really going on with the patient. So the combination of data from a primary care setting, a secondary care setting, a specialist’s setting and an inpatient setting is necessary to provide a complete view of the patient. Women, for example, typically will have a primary care physician as well as a gynecologist; at a minimum that scenario gives you two sources of data that need to be combined into a single record. So when we talk about EHR implementation, I think that’s the kind of thing that we’re talking about. There is another perspective of the EHR that will impact implementation as well. That perspective is the different views of the EHR. I think the first view is the one that I’ve really been talking about, and that’s the view that I call the provider view.

This view is an institutional view, either inpatient or outpatient, and is the documentation of events occurring in the organization delivering the care. It’s used for credentialing of physicians; it’s used for the reimbursement process; it’s used for accreditation; it’s used for patient care; it’s used for auditing quality and patient safety; it’s used for research; and it’s used for education. Two other views of the EHR are being developed; one is a population view that may be identical to the summary record at which there is an aggregation of data about a patient from multiple sites of care. The population view also serves for health surveillance, where you’re looking for outbreaks of disease or analyzing similar episodes of care in a regional area or in a particular population group. In most cases, that population data functions as the public health documentation and involves multiple sites of care. And more recently, there is increased sophistication and awareness of individuals — consumers, if you will, of healthcare — into their own healthcare, so they’re a much more knowledgeable group and much more interested in being involved in the decisions made about their healthcare. They want to be informed sufficiently so that they can make decisions and trade-offs between risks of death and quality of life. And again, much of that results from an EHR.

So a personal health record is another view of the EHR that most groups believe is important for involving the patient in his or her healthcare. I think part of this is that the changing model of healthcare has raised questions about who the individual is , an identifiable person, which accepts the responsibility for an individual’s care. Who worries to make sure that the chronic disease, whether it’s hypertension or diabetes or asthma, is under control? Who worries about whether the patient does all the right things to make that happen? Who worries about whether the levels of the medication are adequate to control the disease and things of that nature? Increasingly the model doesn’t seem to identify that individual. The role of a primary care provider as the gatekeeper seems to have disappeared, and again that responsibility appears to be falling back to the patient, to the citizen, to the consumer. And to do that effectively requires that data. So I think that again is a requirement.

Now in terms of the implementation, that’s an open-ended question. To some it would involve defining what the architectural structure of the EHR is, whether this is a series of data points stored in a relational database or, perhaps, in others’ view, it’s a series of documents that are stored, and some of those documents constitute the EHR. And some of the standards work that’s being done supports that view. So there’s not a single agreed-upon answer to the question of what an implementation is, and I don’t think that we will see a single answer to that question. Certainly in the next 10 years, I think you’ll see a variety of things.

HCT: What’s the historical evolution of technical standards?

EH: My assumption is that we’re talking about health data standards, rather than healthcare standards. And to my knowledge, there were probably three or four different groups that approached the standards that would be used, health data standards, for different purposes.

ANSI created a standards group called ASC X12 for the purpose of creating transaction standards. In other words, this group created the standards that were responsible for carrying the transaction data that is used for the reimbursement process. The HIPAA legislation in 1996 identified the X12 standards as the transaction standards that would be used for reimbursement purposes. Another group, the National Council of Prescription Drug Programs (NCPDP), started because of the need for pharmacy-based systems for PBMs to reimburse for prescription drugs. The NCPDP created its first transactions standard for the reimbursement of clinical drugs back in the late ’80s and early ’90s, and this is the dominant standard for that purpose.

DICOM came originally from the American College of Radiology and the National Electronics Manufacturers Association, which jointly created a standard that was used in PACS systems. Basically these are the imaging standards that are used primarily for radiology and are now expanded to almost any images involved in healthcare.

HL7 (Health Level 7) got its start when a group of people were trying to create a hospital information system from what’s called the best of breed. In other words, you look at the best ADT system, you look at the best laboratory order entry system, you look at the best materials management system, you look at the best pharmacy system and things of that nature and you need to put those together.

HL7 began as a clinical messaging standard that was used, initially, primarily in single settings, namely hospitals, to carry the data from one department to the other to create a consolidated look at the patient. Initially HL7 focused largely on that inpatient environment, and most of the electronic systems were in the inpatient world because that’s where most of the money was. But over time, that change began to move into the outpatient world. So HL7 continued to support clinical data in both the inpatient and the outpatient arena.

As the needs of what we were doing required more standards, HL7 changed its scope to cover some of those needs. Now HL7 is really addressing the broad set of standards that are required for EHR. Implementation of that includes definitions of EHR functionality; it includes the clinical data architecture; it includes clinical statements, clinical templates, data types, a reference information model and so forth. So HL7 now covers a very broad horizontal scope of the things that are associated with the implementation of the EHR and the movement of data from various sites to be able to consolidate that.

The RSNA, HIMSS, and DICOM, in beginning to try to connect PACs systems and radiology information systems with the rest of hospital information systems, began to broaden its scope a little bit. This was later in the world of standards, so it recognized the existence of standards and started looking beyond standards for specific and single purposes and at the broader aspects of putting standards together. So the IHE (the integration of the healthcare enterprise) activity started, and the kinds of things that are required to make an enterprise system that uses standards came out of that.

There are two other groups that I would talk about in terms of evolution of standards. One of these is the ASTM E31 Committee. Dr. Clem McDonald led the effort; I was a major part of that. The first kind of standards that we created in ASTM were similar to the standards that HL7 created, and those first standards were focused on moving laboratory data, primarily from the commercial clinical laboratories to provider institutions. ASTM has had typically smaller groups working on topics, but they’ve been fairly broad and included such things as dealing with the EHR in terms of content, in dealing with transcriptions, in dealing with standards related to security, privacy and things of that nature.

Most recently the ASTM’s major contribution is the continuity of care document standard that’s been used to move electronic data from one place to the other. MedBiquitous is a very recent ANSIapproved standard organization that is trying to create standards that would support distance learning and education and patient simulation and activities of that nature. The American Dental Association has had standards focused entirely on the dental community for a number of years. Veterinarians have created standards — more in other groups than independent efforts — but they have been identifiable as veterinary standards.

So basically in this country we’ve ended up with six or seven groups that have been creating standards that, when they first started, had a clear-cut and understandable difference in purpose and being, but in pursuit of those standards and in meeting the needs — the broad needs of the people that use those standards — have begun to increase their scope to the point where almost all the standards activities have overlaps in redundancy with other standards efforts. And a specific example that recently occurred is that the NCPDP standard for capturing and reporting prescription drug data was used in an overwhelming percentage of transactions involving prescription drugs, and as such, these databases, whose contents have been acquired through the process of reimbursement, are the most complete record of all the prescription drugs the patients are taking.

As e-prescribing has come into vogue, the Medicare Modernization Act is going to require electronic prescribing of drugs over the next three years. The NCPDP standards appear to be the ones most likely used for this, taking advantage of that database. But the problem is when you begin to look at the broader requirements of e-prescribing, you begin to need clinical data. At a minimum, you need some physical parameters such as the height or the weight of an individual to be able to calculate the correct dosages; you need to have something about the diseases a person may have that would impact the metabolism of the drugs; you begin to need some social kind of information, the ability of the patients to take complex drugs. So that variation has now expanded the requirements beyond just the drugs and how they were administered.

The NCPDP, for example, is working on a medical data standard that will increase that capture. So you begin to see the overlap between ASTM and the CCR, HL7 and its messaging standards, the NCPDP, and through claims attachment you begin to see the overlap among the standards groups. In fact, the IHE effort in moving clinical documents begins to overlap with this as well. So you’ve got all of these groups, including the ADA, that has a little bit of everything that had this overall redundancy. So it will be interesting to see how we solve that problem, because probably worse than no standards is too many standards There’s not a clear-cut way of finding which standard you want to invest your money in using, and consequently you don’t use any standards. So that’s sort of where we are and how we got there.