Changes in how medical diagnoses are coded under ICD-10 may complicate the financial analysis, research projects and training programs that depend on look-back comparisons of healthcare data, according to researchers at the University of Illinois at Chicago (UIC). The report, a collaboration of researchers at UIC and at the University of Arizona, is also online in the Journal of the American Medical Informatics Association (JAMIA).
To this end, Andrew Boyd, M.D., assistant professor of biomedical and health information sciences at UIC and first author of the paper, and his health information science colleagues, have been looking at issues that could come up as physicians and hospitals change from one system to the other. Previously they found that some ICD-9 codes map well to ICD-10, but many more have highly convoluted mappings, and some don’t map at all. This forward-mapping is needed for continuing payments of ongoing medical conditions, Boyd says.
Boyd has been leading the charge in tackling this mapping issue for the last few years. In 2013, UIC researchers found that 60 percent of the ICD-9 codes are “non-entangled motifs,” which would map without discontinuity and should be immediately interpretable. However, it’s those remaining convoluted codes, which accounted for 36 percent of the codes the researchers looked at, which could cause problems. In one study, where the researchers looked at 24,008 clinical encounters in 217 emergency departments, 27 percent of the costs were associated with convoluted diagnoses.
And last year, Boyd and other UIC researchers looked at the coding ambiguity for hematology-oncology diagnoses to anticipate the challenges all providers may face during the transition from ICD-9 to ICD-10. The researchers used a web-based tool, developed in house, to input the ICD-9 codes and translated them into ICD-10 codes. They looked at whether the translation made sense; whether a loss of clinical information occurred; and whether a loss of information had financial implications.
“Now, we are taking the same methodology and looking backward,” Boyd says. Reverse-mapping from ICD-10 back to ICD-9 will be important for all sorts of retrospective analyses, he says, “because we have 30 years of data that we want. We don’t want to lose all this information.” Clinical researchers and analysts conducting studies across datasets—and hospital administrators who manage growth and watch trends for strategic planning—will need to pull data under both the new and the old codes, he notes.
Boyd says while there is a huge educational burden on the industry in preparing for ICD-10, memorizing codes, and understanding what documentations are necessary for the new codes, his focus is on is what ICD-9 codes are currently used for in healthcare. “There are plenty of consultants and other companies for that other stuff. We have tried to focus on what reports can you run in ICD-10, and after you code, can you map backwards and run old reports in ICD-9 until you have enough ICD-9 data to make clinical decisions?” he says. As such, Boyd says that organizations might not be able to run all of their reports meaningfully in ICD-10 until 2018. “Our focus has thus been on using the science of the network, the mathematical theories designed to help connecting networks to help find the hard parts or find the areas where the reports might not make sense. And you have to engage the clinicians to figure it out,” he says.
The aforementioned web portal tool and translation tables were created to provide guidance on ambiguous and complex translations and to reveal where analyses may be challenging or impossible. The tool lists all ICD-9-CM diagnosis codes related to the input of ICD-10-CM codes and classifies their level of complexity, which can be: one-to-one “identity,” or reciprocal, the simplest (28 percent of ICD-9 codes fall under this category); class-to-subclass (12 percent); subclass-to-class (22 percent); “convoluted” (36 percent); or “no mapping” (1 percent). “The healthcare system runs on data,” Boyd says. “We are fundamentally changing the way we record the data.” Although the new system will improve the way the data is sorted and recorded, he says integrating it with the last 30 years of information will be difficult.
The alternative to forward mapping, Boyd says, is to dually code in ICD-9 and ICD-10, a process that he says would double the cost of professional coding and double the time of physicians. “We have a $3 trillion dollar healthcare system, so not even all the big organizations will be able to code everything,” Boyd says. “Some will for internal purposes, but the cost is so huge to have everyone in the country do that.” As such, in the sense of mapping forward, Boyd says that it’s easier because you can map to the same general concepts. “Right now, for example, we say ‘ear infection’ in ICD-9,” he says. “You’ll have to specify right, left, or unspecified in the future. And if you map backwards you’re losing data. If you map forward, it maps forward to ‘unspecified ear infection,’ so at least you get the idea,” Boyd explains.
Boyd says that everything his team has done in this arena has been published online and is available for free use. They have created a tool where the user can create either the organization’s top 25 codes or 100 codes used in practice, and then the tool will give them a graphical output so the user can see how the codes are interrelated.
“Besides that, we also generate an online table so you can take that an incorporate it into your own reports,” Boyd says. “We also label the list of your own codes that you provide us into one of those five complexity categories. We have additionally created a separate online tool for when you’re in ICD-10; we provide the network backwards and we indentify the same categories,” he says. “All of this helps you understand the robust network in a comprehensive manner. We’re all in this together. The idea is to reduce the costs and burden for everyone.”