Getting Help with Coding
Health Data Management Magazine, October 1, 2009
Throughout history, clear-cut turning points have triggered surges in demand for technology. The introduction of personal computers, for example, was a watershed I.T. moment. In recent months, the federal economic stimulus program has been in the spotlight, with experts debating whether it could be a strong catalyst for a surge in adoption of electronic health records. But a much lower-profile health care industry development - the shift from ICD-9 to ICD-10 diagnosis and procedure codes for claims - also could prove to be a powerful technology catalyst, giving a boost to demand for computer-assisted coding systems.
Despite years of development, the use of automated coding systems is still far from commonplace. The technology is much more prevalent among physician group practices than hospitals, primarily because outpatient claims are so much simpler to code.
Advertisement
The looming shift to more complex ICD-10 codes in 2013 could stir up interest in the applications because all providers will use the new codes when filing claims with all payers. The greater level of precision in the codes likely will improve the performance of computer-assisted coding applications and make them easier to use, says Sue Bowman, director of coding policy and compliance at the American Health Information Management Association, Chicago. "We're going to see much better and more sophisticated inpatient computer-assisted coding once ICD-10 is implemented," she predicts.
For now, group practices and hospitals are using a wide variety of approaches to auto-coding. These range from coding functions embedded in electronic health records to freestanding systems that simply automate code checklists formerly found on paper "super-bills."
As these systems become more sophisticated, they could have a huge impact on the role of coders. Many now spend the bulk of their time manually selecting codes after reviewing paper or electronic records. "Our coders are going to be more like auditors, checking every claim before it goes out to a carrier," predicts Deborah Grider, vice president of strategic development at the American Academy of Professional Coders, Salt Lake City, Utah.
"Computer-assisted coding ultimately will remove the mundane, routine coding tasks," AHIMA's Bowen says. "Coders will become editors, reviewers and auditors who make sure the codes are accurate and the documentation is complete."
The shift to ICD-10 will mean big changes for physicians as well, Grider argues. Hospitals and clinics will have to offer extensive training to make sure doctors include all necessary information in patient records-whether paper or electronic-to support specific codes, she notes. Payers also will require more detailed documentation to support pay-for-performance projects that measure the quality of care, she adds.
But Bowman cautions that providers must be careful when shopping for "computer-assisted coding" software because vendors use the term loosely to describe a broad variety of applications. "Computer-assisted coding in its truest form is linked to an electronic health record system, and it guides you to the correct code by linking to documentation in the record," she contends. But relatively few organizations use such systems, which rely on a technology called natural-language processing. The technology translates readable information stored in EHR databases into related codes.
An Important Test
University of Washington Medical Center recently began tests of such a system. The three-hospital medical center in Seattle is helping Cerner Corp., Kansas City, Mo., integrate the firm's recently acquired coding software with its inpatient electronic health records system.
In the tests, physicians are using the coding software, formerly known as ß LingoLogix and now called Discern nCode, to automatically generate E&M codes for outpatient treatment of bone marrow transplant patients, explains Thomas Payne, M.D., chief medical information officer. E&M (or Evaluation and Management) codes are modifiers to CPT (procedure) codes that indicate the complexity of the case on a scale of 1 to 5.
The coding software uses natural language processing to "read" documentation in the Cerner PowerChart electronic record and then suggest an appropriate E&M code. But the software goes one step further, offering a "detailed markup of the document" that describes how the code was derived, Payne notes. "That's an incredibly powerful educational tool for physicians to bolster their understanding of how coding works," he adds.
In the months ahead, the hospital also expects to test broader applications of the coding software, both inpatient and outpatient, such as for generating diagnosis and procedure codes.
The coding software applies the standard clinical vocabulary known as SNOMED CT to the content of electronic records, easing the conversion to other coding methods, Payne explains. In addition to applying SNOMED CT, the hospital can create other rules within the coding engine for interpreting phrases or abbreviations with "special biomedical meaning" specific to the organization, he says.
Beyond improving coding, the experiment also could lead the hospital's physicians to improve the structuring of documentation in patient records so it can be used to support meaningful outcomes research, Payne adds.
Variety of Approaches
While computer-assisted coding is still relatively rare at hospitals, many physician group practices are using a wide variety of applications to assist with coding. Some of these applications, such as one used at Radiological Associates of Sacramento (Calif.), also use natural language processing to automatically suggest codes based on the content of records.
The 100-physician practice uses coding software from CodeRyte Inc., Bethesda, Md. At the end of each day, the practice sends a print image of its radiology reports to CodeRyte's server, where a coding engine applies criteria developed by the practice to generate diagnosis and procedure codes. The next morning, coders review the suggestions on a portal and make final decisions before submitting claims, explains Michael Gonzales, billing operations manager.
Before acquiring the software four years ago, the practice would print out radiologists' reports for painstaking manual coding, Gonzales says. Today, the practice is handling coding for 30 more physicians but has one less coder, thanks to auto-coding, he adds. That's because coders now serve as editors and analyzers rather than code-chasers.
Page 1 of 2.
