Thursday, May 06, 2010

HL7 and the Electronic Health Record

It is understandable that HL7 is devoting increasing attention to the EHR topic. One result is the ANSI-approved "Electronic Health Record System Functional Model Normative Standard", which provides a reference list of 132 functions that is designed to enable the  description and common understanding of the EHR functions needed or available in different settings. Version 1 of this standard was released in 2007. The revised version 1.1 is, I am told, still in development.

The original Electronic Health Record System Functional Model Normative Standard of 2007 is said to have received "an unprecedented amount of feedback from hundreds of reviewers from the standards community, the provider community, the international community, and other industry stakeholders". It is therefore troubling to see what has been provided by HL7 in the way of documentation for this standard.

Consider, for example, the Glossary made available on the EHR-S Functional Model website (select "EHR-S FM 2007" from the list of downloadable files on the right here). This is a list of more than 100 terms, together with definitions, some of them drawn from a potpourri of external sources, some of them created by the glossary's authors.

What is troubling is that not one of the hundreds of reviewers from the standards community, including those persons involved in the ANSI standards vetting process, seems to have noticed the quite peculiar mixture of ways in which the definitions in this glossary fall short not only of standard best practices but also of achieving HL7's own goals.

For some definitions:
Text =def. Computer functions that return a single value.
Uniquely identify =def. A standard that lets you specify a unique label to the set of element names.
it is difficult to conceive of how they could have survived even minimal scrutiny.

Other definitions, for instance:
 Clinician =def. An expert clinical physician and teacher.
(from Dorland's Medical Dictionary for Health Care Consumers) seem not to correspond to the normal meaning of the term in question.

Others contain non-trivial typographical errors. For instance:
Health condition =def. An observable finding about or state of health that persists over time and tends to require intervention or management, and, therefore, distinguished from an Observation made at a point in time; may exist before an Observation of the Condition is made or after interventions to manage the Condition are undertaken. Examples: wellness, impairment, chronic illness.
where the error on line 1, when corrected (by replacing 'or' with 'a'), yields crucial problems for HL7. This is because 'Finding' is listed elsewhere in this Glossary as a subclass of 'Clinical Data/Information', and findings so defined cannot require management, and cannot exist before an observation of the condition is made.

Other definitions are mere lists of synonyms, for example as here (taken from
Repudiate =def. To refuse to recognize or acknowledge: deny, disacknowledge, disavow, disclaim, disown, reject, renounce.
Others are circular (which means that they cannot address the primary need of glossary users, which is to be informed of the meanings of terms they do not already understand). Some are worse than circular, the logical equivalent of 'an apple = def. the eating of an apple' as for example:
Resource utilization =def. Measurement of the effectiveness of resource usage.
Yet others provide two or more logically conflicting definitions of a single term, as for example:
Result =def. The conclusion or end to which any course or condition of things leads, or which is obtained by any process or operation; an outcome. The act or process of applying general principles or formulae to the explanation of the results obtained in special cases.
Peculiarly, not one of the definitions in the glossary draws on HL7's own definitions and glossaries provided elsewhere. Indeed they seem to betray almost no awareness of the wider HL7 context within which this glossary was created, and some of the definitions are indeed inconsistent with definitions normatively required by HL7, most egregiously:
Entity = def. Something that has separate and distinct existence and objective or conceptual reality. Something that exists as a particular and discrete unit. An organization (as a business or governmental unit) that has an identity separate from those of its members (from Merriam-Webster),
where HL7 defines its RIM backbone class 'Entity' as: 'A physical thing, group of physical things or an organization capable of participating in Acts, while in a role.' This makes somewhat confusing HL7's assertion in the text of the EHR S FM 2007 glossary to the effect that terms in this glossary "will be submitted for inclusion in the HL7 Version 3.0 Edition 2006 Glossary".

Given that HL7 is being promoted as "the glue that will hold together the pieces of the electronic health record (EHR) of tomorrow", the question becomes all the more urgent as to what will serve as the glue that will hold together the pieces of HL7 itself.