Lab Interoperability Plays Catch Up
The rise of the EHR increases expectations to access laboratory results in digital format. That requires a laboratory information system (LIS), hospital information system (HIS) and EHR to communicate between various departments, hospitals and possibly health information exchanges (HIEs).
While lab results often contain a variety of fields—both text and numerical—that are complicated to report electronically, perhaps standards themselves are the most challenging aspect of laboratory interoperability. A recognized lack of standardized nomenclature has proved tedious for clinicians trying to translate information. Nevertheless, organizations are finding ways to meet the goal.
Walter Henricks, MD, director of the Center for Pathology Informatics at Cleveland Clinic, speaks to the changing expectations of physicians. "In the past, physician offices would have been satisfied with getting results on paper records, or maybe through the use of an ad hoc laboratory portal. In the future, physicians and practices are no longer going to be satisfied with getting a paper result, or even going to a separate lab portal system. They're going to expect that laboratory results are going into a new EHR system."
Logical Observation Identifier Names and Codes (LOINC) are getting more attention as a result of the push for interoperability, and it is a standard recognized by the Office of the National Coordinator of Health IT (ONC), according to Henricks. The LOINC dataset of universal identifiers purports standardized test codes, but to date, has had limited applicability because it is difficult to use and there hasn't been much incentive. Additionally, it only facilitates interoperability if both sides of the interface use it.
"There are different ways LOINC can be implemented, creating its own challenges," says Henricks. "It's a workable standard, but not easily implemented in a widespread manner."
Ulysses G.J. Balis, MD, director of the division of pathology informatics at University of Michigan Health System (UMHS) in Ann Arbor, anticipates the day when standards allow for plug-and-play interoperability. Like Henricks, Balis cites deficiencies with the standards and nomenclature.
"The hardest part is the standards for defining a test," says Balis. "LOINC did an amazing disservice to the laboratory community, not by virtue of establishing it which is a great idea, but by the way it's adjudicated. There is no universal lexicon within LOINC."
The standard suffers from synonyms, and picking terms that can be read on both sides of an interface can be challenging. "In the absence of having a centrally adjudicated universal list, there's an added burden of validation, which is really quite vexing," says Balis.
While implementing a new EHR at UMHS, its vendor's version of LOINC had only about 80 percent overlap with the version the system was using. Because the two systems don't use the same version of LOINC, physicians have to manually translate tests.
"Standards are supposed to solve problems," says Balis. "In this case, the standard itself is the problem."
"The current versions that most people use are versions 2.3 to 2.5. There is a better XML-based version, which no one uses because vendors that have designed their own interfaces can't support it," says Carter. "Vendors have a lot of difficulty keeping up with the pace of technological change because it is a huge expense. In addition, when it comes to interoperability, all vendors have to agree to update to the better standard over the same timeframe. These are huge challenges."
Other issues associated with interoperability include finding a place for certain types of information within the HL7 message itself, and there are unresolved questions as to how much data the HL7 message should carry. In some cases, there may not be a place for the data within the LIS, and this limits the ability to transfer such information by HL7 messages. For example, some new tests performed at Emory—including genetic and molecular testing—require vast amounts of data space.
"Our laboratory information systems aren't even close to being able to handle the data," says Carter. "It's in part because the data are large and complex, but also the technology is still really new."
"We were able to share our laboratory results across native vendor and LIS platforms without going through a detailed, time-consuming individual interface development exercise," says Balis. "Once they have the interface, they're compatible with the LIDDEx cloud."
The challenge was less about the technical requirements than it was about getting a vendor community to work with academia to accomplish exchanges of identified patient results using a cloud service. They found that vendors and users both benefited through the collaboration.
Meanwhile, as hospitals attempt to increase their interoperability, there are things that both hospitals and the HL7 community should take into consideration.
Recognizing the issue is critical to solving it, says Henricks. "It is now time for laboratories to examine their systems and adapt them to LOINC, or define LOINC codes within their system," he says.
"People should insist that the HL7 community wake up, and serve their needs," adds Balis. "The HL7 community itself needs to recognize these requirements are urgent. We don't have decades to wait."
While lab results often contain a variety of fields—both text and numerical—that are complicated to report electronically, perhaps standards themselves are the most challenging aspect of laboratory interoperability. A recognized lack of standardized nomenclature has proved tedious for clinicians trying to translate information. Nevertheless, organizations are finding ways to meet the goal.
Walter Henricks, MD, director of the Center for Pathology Informatics at Cleveland Clinic, speaks to the changing expectations of physicians. "In the past, physician offices would have been satisfied with getting results on paper records, or maybe through the use of an ad hoc laboratory portal. In the future, physicians and practices are no longer going to be satisfied with getting a paper result, or even going to a separate lab portal system. They're going to expect that laboratory results are going into a new EHR system."
LOINC & HL7
Laboratory interoperability is far from "plug and play," and one of the most fundamental issues is aligning test vocabularies. Standards exist, but they lack specificity that would foster manageable interoperability. "Maybe a glucose test has a certain code or abbreviation within the EHR, but a different code in the LIS," says Henricks. "There has to be a reconciliation or translation for interoperability to occur."Logical Observation Identifier Names and Codes (LOINC) are getting more attention as a result of the push for interoperability, and it is a standard recognized by the Office of the National Coordinator of Health IT (ONC), according to Henricks. The LOINC dataset of universal identifiers purports standardized test codes, but to date, has had limited applicability because it is difficult to use and there hasn't been much incentive. Additionally, it only facilitates interoperability if both sides of the interface use it.
"There are different ways LOINC can be implemented, creating its own challenges," says Henricks. "It's a workable standard, but not easily implemented in a widespread manner."
Ulysses G.J. Balis, MD, director of the division of pathology informatics at University of Michigan Health System (UMHS) in Ann Arbor, anticipates the day when standards allow for plug-and-play interoperability. Like Henricks, Balis cites deficiencies with the standards and nomenclature.
"The hardest part is the standards for defining a test," says Balis. "LOINC did an amazing disservice to the laboratory community, not by virtue of establishing it which is a great idea, but by the way it's adjudicated. There is no universal lexicon within LOINC."
The standard suffers from synonyms, and picking terms that can be read on both sides of an interface can be challenging. "In the absence of having a centrally adjudicated universal list, there's an added burden of validation, which is really quite vexing," says Balis.
While implementing a new EHR at UMHS, its vendor's version of LOINC had only about 80 percent overlap with the version the system was using. Because the two systems don't use the same version of LOINC, physicians have to manually translate tests.
"Standards are supposed to solve problems," says Balis. "In this case, the standard itself is the problem."
Outdated standards, too much data
Laboratories are facing other issues with interoperability as well. Health Level Seven (HL7), the authority on standards for interoperability, governs the messaging format used to communicate between different information systems. The HL7 format has limitations that make it difficult to interface with different entities, according to Alexis B. Carter, MD, director of pathology informatics at Emory University School of Medicine in Atlanta. Currently, many hospitals are using an outdated version of the standard because vendors have yet to design their interfaces to use the newer version."The current versions that most people use are versions 2.3 to 2.5. There is a better XML-based version, which no one uses because vendors that have designed their own interfaces can't support it," says Carter. "Vendors have a lot of difficulty keeping up with the pace of technological change because it is a huge expense. In addition, when it comes to interoperability, all vendors have to agree to update to the better standard over the same timeframe. These are huge challenges."
Other issues associated with interoperability include finding a place for certain types of information within the HL7 message itself, and there are unresolved questions as to how much data the HL7 message should carry. In some cases, there may not be a place for the data within the LIS, and this limits the ability to transfer such information by HL7 messages. For example, some new tests performed at Emory—including genetic and molecular testing—require vast amounts of data space.
"Our laboratory information systems aren't even close to being able to handle the data," says Carter. "It's in part because the data are large and complex, but also the technology is still really new."
Creating a solution
Many within the laboratory community are anxiously awaiting new HL7 standards, but physicians within UMHS created the Laboratory Information Digital Data Exchange (LIDDEx) interoperability project. While not in clinical use, developers including Balis have unveiled the project publicly on multiple occasions to demonstrate the possibility of interoperability between organizations."We were able to share our laboratory results across native vendor and LIS platforms without going through a detailed, time-consuming individual interface development exercise," says Balis. "Once they have the interface, they're compatible with the LIDDEx cloud."
The challenge was less about the technical requirements than it was about getting a vendor community to work with academia to accomplish exchanges of identified patient results using a cloud service. They found that vendors and users both benefited through the collaboration.
Preparing for interoperability
Within five years, the UMHS' goal is to have an enterprise data bus—a subsystem that transfers data between computers—which will facilitate the interoperability of their laboratory, according to Balis. "When new systems are brought online, they are by design implemented in a way that lives on the bus, either through a translation layer or natively, if possible," he says. "Additionally, we have a large number of data silos, data marks and repositories. It's not practical to swap out modern techniques all at once. We will write wrappers on those and put them on the bus. Eventually, legacy technology, contemporary technology and everything else will live on the bus, at which point the entire health system's data will become available."Meanwhile, as hospitals attempt to increase their interoperability, there are things that both hospitals and the HL7 community should take into consideration.
Recognizing the issue is critical to solving it, says Henricks. "It is now time for laboratories to examine their systems and adapt them to LOINC, or define LOINC codes within their system," he says.
"People should insist that the HL7 community wake up, and serve their needs," adds Balis. "The HL7 community itself needs to recognize these requirements are urgent. We don't have decades to wait."