Philosophies differ on optimal data exchange
When it comes to equipping itself for data sharing, the healthcare industry is facing a monumental task ahead. Even those wearing rose-colored glasses admit there is a tremendous amount of hard work ahead to make IT systems interoperable and accommodate electronic health records by October 2013.
Progress toward interoperability varies widely among provider organizations, as some have already laid the foundation pieces, while others are still contemplating how to proceed. One of the inhibitors is proving to be semantic interoperability, which proponents say is an important step in the connectivity process, while detractors see it as yet another starting gate hoop to jump through.
Semantic interoperability is a term used to describe the ability of computer systems to communicate information and have that information properly interpreted by the receiving system in the same sense as intended by the transmitting system. It requires that any two systems will derive the same inferences from the same information.
Arguing in favor of semantic interoperability, John Hatem considers it a vital thread in the fabric of connectivity.
“It is not good enough for diverse systems to simply exchange data,” said Hatem, director of healthcare product strategy for Redwood Shores, Calif.-based Oracle Health Sciences. “The use cases increasingly rely on semantic interoperability where the receiving system also ‘understands’ the data. This requirement increases the complexity on both the sending and receiving systems to prepare the data, and then to consume and interpret the data. The good news is that there is now a more widespread understanding of the challenges and consequently, we are seeing the incorporation of standards into national standards setting bodies.”
For example, in the United States there are standards for supporting patient summary documents such as the Continuity of Care Document known as the CCD-C32 specification, he said.
“In the past it was sufficient to exchange data without requiring semantic interoperability,” he continued. “Today, there are compelling business drivers which are helping to push the healthcare system towards interoperability. In the United States, providers are being obliged to adopt interoperable standards due to the HITECH Act of 2009. Other countries are at various levels of adoption of interoperable standards as well, often as a result of government mandate. Still, there may also be compelling business justifications for interoperability. Interoperable systems typically will drive efficiency to reduce costs, while helping to improve patient care.”
William Fera, MD, characterizes himself as “a family physician and not an informatics guy,” but the vice president of medical technologies and medical director of interoperability at the University of Pittsburgh Medical Center is a strong supporter of semantic interoperability.
“From my perspective it is absolutely necessary, and health information exchanges will fail without it,” he said. “One reason is that as a clinician when I look at an un-semantically organized list, it is like someone bringing in every single piece of paper in a medical file and dumping it on my desk.”
While efficiency is one advantage, Fera also believes semantic interoperability is a key to better accuracy and patient safety.
“Unless semantic interoperability is adopted, there will be a new era of medical errors based on assumptions,” he said. “Otherwise, when I see a lab report, I can’t be sure I’m seeing all the results because of the different vocabularies.”
Last year UPMC completed the initial phase of its interoperability initiative using architecture from Pittsburgh-based dbMotion. Through its Web-based viewer, the technology delivers through a single, integrated view of demographics, diagnoses, problems, allergies, medications, clinical notes, microbiology results, laboratory test results, documents and discharge summaries. Among the positive results to date is an 82 percent reduction in the time spent collecting preoperative patient information at one hospital while patient readiness for surgery improved 50 percent.
‘Just Get Started’
While not an outright opponent of semantic interoperability, James Walker, MD, believes the most important thing right now is to get the data exchanges started and that the function can be layered in once they are up and running. The chief information officer of Geisinger Health System in Danville, Pa., points to his own Keystone Health Information Exchange as an example.
Keystone is a fully functioning HIE that was developed before semantic interoperability was even a consideration.
“My advice is to not worry about it right now,” he said. “Get started and do it in a way that in five to eight years it will be according to industry standards and interoperable as it can be.” By “keeping the technology simple,” Keystone has grown in five years into a network of 13 independent provider organizations serving residents in 41 of Pennsylvania’s 67 counties. It has 2.5 mil- lion patients in its record locator service and 400,000 patients allow their information to be shared.
“Our region is largely rural, so we have the capability of sharing IT services that very few organizations in the region have,” Walker said. “Therefore, IT knits the region together.”
Waukesha, Wis.-based GE Healthcare provided the architecture for the Keystone project and in its new partnership with Wayne, Pa.-based InterComponentWare, is focusing on taking health data exchange to a new level, said Blair Butterfield, vice president of connectivity solutions for GE.
“Our partnership with ICW will strength- en the technology layer to more easily build in features such as quality reporting, clini- cal decision support and evidence-based medicine,” he said. “It goes beyond being confined to one hospital or organization and provides literally dozens of advanced services that depend on HIEs.”
ICD10: It's coming down
Fitting into the interoperability/health infor- mation exchange picture is the coding tran- sition from ICD-9 to ICD-10, which is the new “language” of connectivity. While the rest of the world has migrated to the new system already, the United States still faces significant hurdles to its adoption, said Tori Sullivan, manager for Capgemini Government Solutions in Washington.
“Because of the complexities, it has to be thought through to accommodate our payment structure,” she said. “The most complex change is from a reimbursement perspective, which is extremely complicated because of the algo- rithms and fundamental changes in technol- ogy. Many organizations are stalling because they don’t want to go through the process. In addition, other regulatory requirements have come into play that distract the industry from focusing on this implementation.”
The ICD-10 structure is much more spe- cific and detailed than ICD-9.
Where ICD-9 has 13,000 codes, 855 cat- egories and ranges from three to five alpha- numeric characters on the diagnosis side, ICD-10 by contrast has 120,000 codes, 2,033 categories and three to seven alphanumer- ic characters. On the procedural code side, ICD-9 has 4,000 codes with three to four characters, while on the ICD-10 side there are 200,000 procedural codes and seven characters.
“So there are many new categories and items that we were never able to code before that are now available and technologies need to accommodate them,” she said. “There are three fundamentals to moving toward bet- ter information – better data, more acces- sible information and information that is integrated. We can’t move forward without all three.”