Controversy surrounds the Centers for Medicare and Medicaid Services’ (CMS) decision to delay the date on which physicians and hospitals must transition to the ICD-10 (International Classification of Diseases) code-set. Some, like the American Health Information Management Association (AHIMA), disagree with this decision vehemently. Others, say it was a needed move.
Christopher Chute, M.D., head of the Rochester-based Mayo Clinic’s bioinformatics division, and chair of the World Health Organization’s ICD-11 Revision Steering Group, is among those that are in the latter group. Recently, Dr. Chute expressed his reasons in a paper alongside several other notable names in the industry, John D. Halamka, M.D., Stanley M. Huff, M.D., James A. Ferguson, M.D.and James M. Walker, M.D.
The group’s piece, which appeared in a recent edition of Health Affairs, involves many of the typical, yet pressing ICD-10 talking points that have arisen in this debate: cost, effectiveness and other federal regulatory acts, specifically meaningful use under Stages 1 and 2 of the American Recovery and Reinvestment Act/HealthInformation Technology for Economic and Clinical Health Act (ARRA/HITECH). Chute says the recent unveiling of the proposed rule for Stage 2 of meaningful use has reinforced his concerns.
“The meaningful use requirements, particularly now that Stage 2 requires SNOMED implementation, put
s an enormous burden on healthcare organizations,” says Chute, who despite his beliefs, is aware of the consequences the delay will have on those who were going to be ready for the original October 2013 compliance, including his organization. “In a sense, we’re in an awkward situation where those who have been diligent could conceivably be punished by such a delay, because of the costs associated with it.”
Christopher Chute, M.D.
Even with this awkwardness being inevitable, Chute and his colleagues, in the paper, attempt to dispel the notion that the improved effectiveness of ICD-10, in comparison to the currently used ICD-9, is significant enough to outweigh the cost factors that are associated with the transition. The authors conducted a study in which they compared the ICD-9 and ICD-10 systems and how effective they were in processing clinical data, and aimed to present a case for their argument.
For the study, which was an updated revisit of earlier, similar comparative research, the authors had fifty clinical [documentation] notes from four academic medical centers “exhaustively annotated,” according to Chute. The researchers identified 3,061 clinical concepts to be used for comparison. The researchers had the concepts recoded to ensure they were updated with the current versions of ICD-9 and ICD-10. It was then reviewed by the National Center for Health Statistics for accuracy.
Focusing on diagnostic concepts, the researchers scored the codes, in both ICD-9 and ICD-10, on the “goodness of fit of each element of that coded information.” Scoring was either fairly good, not good at all, or it got the idea. “It was literally a 0-1-2 scale, so it was subjective,” Chute says. “You either got it, didn’t get it, or you sort of got it.” The scores were then averaged out. What the researchers found was both ICD-9 and ICD-10 performed similarly. ICD-9 scored an average of 1.62. ICD-10 was a mere 0.2 points above that at 1.64.
While many will argue that ICD-10 contains a lot more code sets than ICD-9, 68,000 to 13,000, Chute notes that more than half of those are focused on injuries and external causes. The actual diagnostic code sets are numbered at approximately 25,000, which according to Chute, is slightly inflated by the fact many of the code-sets account for both left and right side diagnoses. “You permute these things out,” he says. “It’s fairly easy to get an inflated number of codes based on that expansion.”
No way to “skip over to” ICD-11 seen
Ever the careful practitioner, who was trained to weigh both sides of an opinion and make an informed decision, Chute does not offer the extreme belief that switching from ICD-9 to ICD-10 is completely useless in this regard. He doesn’t want to create the impression that there is no material difference. However, he says, the ways ICD-10 is more effective than ICD-9 are exhibited in a smaller percentage of cases.
Chute is also adamant that there is no possible reason or possibility that the U.S. could just skip over ICD-10 right into ICD-11. Even with his ties to ICD-11, Chute says there it’s not realistic, nor is it plausible, to have seven-to-nine more years of ICD-9 codes, while the medical industry waits for the World Health Organization to finish drafting ICD-11 and then waits for the U.S. to adapt it for its own use.
In terms of how ICD-11 differs, along with SNOMED integration and genomic content, Chute says it includes a “semantic core, what that is at its heart, a large graph of concepts. It’s a network of terms and concepts that are related by semantic web connections, using semantic web technologies.” This infrastructure, he suggests, should be used for ICD-10. He says future ICD-10 updates could be done through the “ICD-11 machinery,” and over time, such a process would lead naturally to an evolution into the latter code-set. This, he says, would make for a much easier transition from ICD-10 to ICD-11 than there has been for ICD-9 to ICD-10.
“I think there’s no way the country is going to have any appetite for yet another ICD conversion, hot on the heels of the ICD-10 conversion,” Chute says.
Delay Seen As Necessary
There are two ways to deal with the delay, according to Chute. One would involve CMS providing a grace period in which the compliance would not be enforced or have a dual-use period instead of a single switch-over date. Chute doesn’t advocate for those, indicating there would be chaos with duplicative coding.
The other way would involve a uniform delay of one-to-three years for the switch-over date. This is what Chute says would be best, but he does not have a specific delay time period in mind. First, he says, CMS must weigh the implications of meaningful use Stage 2 before it sets a new compliance date.