ICD-10 is certainly no quick fix for EHR vendors' interoperability woes, but it is a step in the right direction.
One of most exciting aspects of the HITECH Act is that it's supposed to encourage portability of medical information across EHR software, hospitals, clinics and even state lines. Unfortunately, we’re not there yet.
Interoperability of software and other systems has been — and will likely always be — an issue faced since the advent of language. If a Greek farmer sent a shipping manifest written in Greek to a Roman merchant, either the Roman merchant would need to read Greek or have someone translate the document into Latin. Just as in ancient shipping, when two software systems speak different languages, someone or something has to translate in order for them to be able speak to one another.
Translating from one software’s language to another is difficult, but not impossible. One must simply know both languages. However, imagine if the software didn't consistently use the same language. Or, imagine if the software manufacturer refused to reveal anything about the language or structure of the software. Now you’ve got an idea why interoperability among EHR software systems is so difficult; many EHR vendors guard their code as if it contained a nuclear launch sequence.
To further complicate things, physicians often don't even use the same terminology when discussing identical procedures or diagnoses. For example, a heart attack could be referred to in a procedural note as a heart attack, a myocardial infarction or simply MI. Depending upon how a hospital or clinic operates, these terms could be used interchangeably; a human knows that all three terms mean the same thing, but a machine would record those as three separate diagnoses, making for either a corrupt database file or painstaking reconciliation of data and elimination of duplicates. Natural language processing software is making great strides in textual sentiment analysis, but the consequences of a mistake in interpretation of nuance in a tweet are far less severe than the consequences of a mistake in the interpretation of a physician’s order, especially if the analytics software either doesn’t have learning capabilities, or has yet to see sufficient data input to have learned the subtleties and patterns of orders in a particular hospital or department.
According to HealthIT.gov, there are four areas in which standards must be developed and adhered to for semantic interoperability of EHR software to become a reality:
- How applications interact with users (such as e-prescribing)
- How systems communicate with each other (such as messaging standards)
- How information is processed and managed (such as health information exchange)
- How consumer devices integrate with other systems and applications (such as tablets)
Moving to the ICD-10 code set as a mandated medical standard, eliminating CPT, ICD-9 and other competing terminologies, will allow consistency in output of medical data. Sure, the myocardial infarction we used as an example earlier could be one of 14 different codes, but, no matter what software a physician uses, I21.01 will always be a STEMI involving the left main coronary artery. This addresses three of the four bullet points above: ICD-10 standardizes the terminology used by a provider to interact with an EHR, the terminology used to communicate a clinical diagnosis fromone software to another, and to a lesser extent, the terminology used to process and manage data on health information exchanges.
Hospitals usually have an advantage over clinics in that they operate at scale. A smart hospital CIO will select EHR software for the entire hospital system that integrates with legacy software. They’ll also encourage or require physician and specialty groups operating within the system’s hospitals to implement compatible software. Even smarter CIOs make sure their hospital’s software will integrate with area health information exchanges and even competing hospital systems. Kathleen Myers, MD, a practicing physician in the Pacific Northwest, and CMO of Essia Health, says providers operating in Portland, Ore.-area hospitals were lucky, as "all the major hospital systems are on Epic." This makes the transfer of charts, along with other encounter information, much easier across multiple facilities.
ICD-10 is certainly no quick fix for EHR vendors' interoperability woes, but it is a step in the right direction. ICD-10’s biggest impact will be felt in billing outputs, population health tracking and Medicare/Medicaid and insurance reimbursements. If SNOMED inputs and ICD-10 billing outputs and reporting can be standardized and automatically cross-referenced, as some EHR vendors are attempting to do, the amount of truly granular, structured data sources available for predictive analytics, treatment and outcome analysis and population health modeling should enable great strides in all measurable areas. And once again, lest we forget, all of this data will serve to create the pay-per-performance architecture as physicians and hospitals shift away from a fee-for-service model and into one based on health outcomes and managed care.
Drew Settles is a content writer at TechnologyAdvice. He covers Healthcare IT news, with a focus on federal regulations and emerging technologies. You can connect with him on Google+.