April 2015. Fifteen years after going live, the electronic Child Health Network has come to play an indispensable role in the health care of Ontario's children. The medical data in its repositories are now shared among 107 hospital sites and 250 other health facilities, plus more than 1,000 physicians' offices. Over two-and-a-half million young patients are represented in a database which includes nearly 50 million clinical reports, with almost half a million added every month. Approximately 10,000 health care providers have active eCHN user-accounts. So seamless and intuitive is the service it provides and so compelling the logic behind an integrated electronic health record that it's only too easy to overlook the mammoth challenges involved in actually integrating data from so many separate facilities -- challenges that must be overcome on a daily basis.
The most obvious problem, the numerous and disparate computer systems used by hospitals in very site-specific ways to store data, has been addressed to some extent by the development of syntactic standards, such as HL7, which format and define the transmission of messages among different systems. The language in the message, however, is another matter -- no universally accepted standard for terminology exists. A veritable Tower of Babel reigns among the various health facilities. Each new site arrives with its own local jargon from the era before electronic records and computer-to-computer talk -- and with it comes a myriad of unique translation challenges.
The problem extends across all the domains of health care data but nowhere is it more marked than in the domain of lab results and digital imaging. Labs can code and report test results in many different ways. Test names may vary widely; for example, the test for ionized calcium can be found to have sixteen different term variants. Many lab orders and results make use of proprietary coding systems or implement standard codes idiosyncratically. There is also a remarkable variety in the way units of measure, such as volume or time intervals are expressed across facilities.
What makes all this particularly consequential is that an estimated 80% of clinical decisions are based on just such results. If there is no generally agreed-upon meaning for the terminology being used, it becomes problematic for health care providers at different locations to interpret and make full use of shared data. They can't make accurate comparisons or trend lab results such as Haemoglobin A1C for diabetes patients to monitor changes over time. The quality of health care suffers and even patient safety may be compromised by ambiguous data.
From its inception, the people who built eCHN were keenly aware of the problem of language in integrating sites and made the creation of standards for data quality a top priority. The driving force behind eCHN's prioritizing of standards has been internationally known biochemist, Dr. Gilbert Hill. Retiring from his job after thirty years as SickKids Clinical Biochemist-in-Chief, Dr. Hill turned to consulting with eCHN in its formative years. He had already been actively involved via key provincial, national and international organizations in the development and implementation of standards for electronic transmission of clinical laboratory data. Dr. Hill's unique training in multiple fields -- medicine and chemical engineering, in addition to biochemistry -- rendered him an astute and vastly knowledgeable guide to the new frontier of integrated health records.
With Dr. Hill's expertise and leadership, eCHN has become Canada's pre-eminent exemplar of the complex art of "cleaning" site-specific medical data to make them meaningfully shareable among health care providers. This translation process is referred to as "standardization" or "normalization". (Although there is a subtle difference between the two, they are often used interchangeably.) An entire nine-member team of experts at eCHN called the MED team is dedicated to the task full-time. Composed largely of subject-matter experts in lab and digital imaging technology, the team is named after the Medical Entities Dictionary, a consolidated reference dictionary of medical and enterprise specific codes and terms that supports the standardization process.
Salve Achacoso, Project Manager of the MED team came to work for eCHN six months before it went live in 2000. She explained the intricate process of bringing a new member site on board the network. It starts with eCHN subject matter experts (SME's) conducting a day-long site survey to gather detailed information about the site's internal business processes, infrastructure and outbound messaging capabilities. Interviews are done with the site's equivalent SME's and detailed questions asked about everyday procedures such as how they update test results, change or cancel incorrect reports, how many reference labs they use and so on. The eCHN SME's write up a comprehensive report of their findings and recommendations on the site's capacity for interoperability and readiness to pursue integration. This detailed report is sent to the new member site for review and editing, if required. When the site gives its agreement on the report, the eCHN teams proceed to make the necessary transformations for the integration process.
On average, it's a six-month-long process. The MED team must familiarize itself with the facility's local catalogue of system codes, test names and specifications. "This is key," said Achacoso. "We need to understand what they do before we can normalize it." The team must translate the local catalogue item-by-item to the enterprise standards defined in the MED using LOINC, the international coding system for lab tests, observations, panels and assessment. It's called "mapping" and it's a long, painstaking job that requires in-depth expert knowledge of lab procedures and much back-and-forth communication with the facility's domain experts. There are many nuances involved in getting the codes right -- some test procedures, for example, may involve multiple timed samples. The MED team ensures that all test procedure complexities are properly represented. In the process, the site benefits in its internal operations, too, by being informed of any inaccuracies, errors and gaps in its local lab information system.
The Data team, meanwhile, operates on the messaging side, mapping the transformations needed in various messaging protocols to bring them into conformity with eCHN requirements. The two teams must work closely together as the elements they are transforming need to be in perfect sync.
Dr. Hill, who has been a prolific contributor to the LOINC database himself over the years, explained that it lists 75,000 tests and expands by about 2,000 every year. eCHN uses about 3,000 of them. "The challenge," he said," is to unequivocally identify a particular test. According to LOINC you need to know six pieces of information about that test in order to distinguish it from somewhat similar tests with somewhat similar characteristics."
Standardizing units of measure for lab tests is an even tougher job -- in fact it's the biggest issue for the MED team. Achacoso explained that generally, the MED team's primary objective is to report results using units that conform to the International System of Units (SI, Système International d'Unités), an international metric system of units used in most of the world. However, there are tests whose results cannot be reported using SI units -- or are not done so -- in which case the objective is to normalize the units. Normalization means the unit is still being regularized among all facilities but is not defined by a recognized standard.
When the process of mapping is done and the functionality of the new dictionary tested and certified by eCHN, the sign-off from the new member-site involves final validation and data acceptance. "Trust is key for us in our integration", said Achacoso. "We never go live with a facility unless we have its complete agreement. They sign a document saying that 'Yes, we've tested it together with the eCHN team -- we agree that everything is there and our data is represented properly with eCHN.'"
After the site goes live on the network, there are various maintenance methodologies in place to support and ensure that data integrity is preserved and mismatches between the site's dictionary and eCHN's MED are immediately identified. This is a crucial component as a test result can only be posted on eCHN and shared among health care providers if the test name exists in eCHN's enterprise dictionary. There is constant double-checking. Every night a daily extract is sent from the facility to provide eCHN, as Achacoso puts it, with the site's "latest and greatest site dictionary". eCHN analysts also receive daily reports which they monitor for message errors. These processes enable eCHN to be swift in responding to facility changes such as the addition of new or clinic-specific tests. As well, since the LOINC vocabulary is constantly changing, eCHN has a reliable and efficient method for quickly integrating each new release into its dictionary.
The benefits from normalization are vast. As patients move from hospital to hospital, each with diverse reporting systems, eCHN is able to provide an integrated, seamless and comprehensive display of their lab and digital information in the consolidated form of a single medical chart. Physicians and other health care providers need learn only one standard set of lab test names, rather than the multiple sets they would require to understand test orders and results from diverse systems at multiple facilities. It allows for the efficient display of cross-facility data and enables health care providers exchanging data to make accurate like-to-like comparisons. It also makes possible the graphical trending of lab results which is so crucial for monitoring changes in patient health over time.
With all its benefits, however, standardization seems to be a surprisingly tough sell. Dr. Hill conveys the distinct impression that bringing standards to lab information systems has been an uphill battle and that eCHN is an "outlier" in Canada for making it such a high priority. "To the best of my knowledge, no one else in Canada is attempting to encourage the adoption of standard units. We're unique in Canada and probably in North America, as well."
Ultimately, the most powerful reason for standardization, in Dr. Hill's view is the need to achieve interoperability. There are so many potential interfaces among facilities and health care providers that depend on a common language for meaningful connection. "Without it", said Dr. Hill, "the full potential of the electronic health record will never be realized." The higher the degree of standardization, he argues, the more interoperability there'll be -- and with it, the greater the quality and efficiency of our health care.