Buried in Babel

June 24, 2011
UNTIL THE HEALTHCARE industry can coalesce around a standardized clinical language or languages, clinical decision support applications cannot
UNTIL THE HEALTHCARE industry can coalesce around a standardized clinical language or languages, clinical decision support applications cannot support critical business objectives.

Standardized clinical vocabularies exist. Unfortunately, until recently, clinicians have had few incentives or little opportunity to use them in their practices. Maturation of major clinical vocabularies and increased dependence on clinical decision support systems highlights the deficiency. Efforts to resolve the problem through technology will not be not enough (See "Coming to Terms," March, page 70).

What remains difficult is how to express healthcare information, declares Gunther Schadow, MD, a member of the HL7 technical committee, Berlin, Germany. "No interface engine will solve this problem. We must do it," he says.

As early as 1986, the National Library of Medicine’s (NLM) Unified Medical Language System (UMLS) project began wrestling with the issues. Its Metathesaurus, built for system developers, serves as intellectual middleware combining and cross-referencing national and international medical standard vocabularies. If there were no Metathesaurus, says Betsy Humphreys, NLM assistant director of health services research information, Bethesda, Md., there might be a stronger case for one single system. "As it is," she says, "I think that we are in a position where more people think that a workable and practical strategy is to standardize on some set of complementary, non-overlapping vocabularies that, together, cover the whole spectrum necessary for patient records."

System development and sales

It makes perfect sense to standardize on a prescribed clinical dictionary--either a single vocabulary or a set of vocabularies. But logic doesn’t necessarily drive development. Standard practice by most software vendors has been to deliver a starter vocabulary set when the system is installed. Individuals at the sites then customize the system by adding their own vocabulary. This may have helped get buy-in at the site level, but this strategy has resulted in banks of data so customized that even same-vendor systems can’t talk to each other.

According to Stanley Huff, MD, senior medical informaticist for Intermountain Health Care, Salt Lake City, and co-chair of the of the Health Level 7 (HL7) vocabulary special interest group (SIG), although all the large system vendors actively participate in standardization efforts such as those by HL7, he hasn’t seen much change in company behaviors. When it comes to really making the move to clinical language standardization, most vendors suffer from a "multiple personality disorder." One side of the personality wants the interoperability possible with a common vocabulary; the other side is comfortable with the status quo. That side sees system sales still rolling along--why change?

The growing demand for decision support systems may change all this. If the vocabulary is different in every system, software vendors can’t deploy a common decision support system even within the same software environment.

Adding value

Value-added is the goal for the HL7 standards organization. "We do not want to re-create vocabulary," Huff says. "We want to define the set of allowed meanings that can be in a message by reference to the UMLS Metathesaurus. We want to make very specific relationships that show how that set of terms is used in HL7 messages."

The HL7 SIG has undertaken two projects. One tackles the problem of drug names. The second deals with more general vocabulary issues. Despite the breadth of the projects, Huff is optimistic and hopes to have useable definitions by year-end.

Charlene Mariettiis senior technology writer at Healthcare Informatics.