Related Links

In recent years, a lot of computer-assisted coding products have hit the health care market as organizations gear up for the ICD-10 coding system and implement population health management and data analytics programs to succeed in evolving care delivery and reimbursement methods.

Figuring out how to choose a CAC is the focus of an educational session at HIMSS13 in New Orleans. For instance, all CACs use natural language processing engines, which also may be called natural language understanding engines, to recognize and parse a word in clinical documentation, assign the word as a data element, categorize it (such as diabetes), and determine if the element is related to a diagnosis or procedure. Then the engine maps the data element to the correct terminology, such as ICD-10, SNOMED CT or LOINC. But how does an organization determine how well one vendor’s NLP engine works to do all this, compared with other engines?

That’s one of the topics of distinction that Deborah Kohn, principal at Dak Systems Consulting, will tackle during a roundtable session at HIMSS13. The NLP engines powering computer-assisted coding systems vary in their accuracy and speed among the vendors, as do their features. Some engines are able to automatically generate or suggest a code, which is a nice feature but there is a lot of workflow processes and integration necessary to do this.

And here’s the wild card: Medical care today is being coded in ICD-9 and CPT; there are very few users of ICD-10, SNOMED CT and LOINC, Kohn says. With ICD-10 on the near horizon, and SNOMED CT and LOINC to be used at least to some degree in future stages of EHR meaningful use, shoppers of computer-assisted coding software need to test a variety of engines with all kinds of terminologies to figure out which ones best support them, she adds.

There are a multiple of other questions that need answers before choosing a CAC that Kohn will address. Does the workflow work for both coders and clinicians? What changes could be made to clinical documentation systems? For instance, the EHR should have clinical decision support with an alerting system, so a physician documenting foot care is prompted if he or she does not specify the left or right foot.

The lasting impression Kohn hopes to leave is summed up in a simple warning: “There are a plethora of these products and it’s very hard to choose.” The session, “A Cacophony of CAC Solutions: How’s an Organization to Choose?” is scheduled at 9:45 a.m. on March 4.