Margaret Dessypris Thomas is a Manager with CapTech Ventures in Richmond, VA. She has over 12 years of experience in the health care IT industry, working with an array of technologies in web development, service oriented architectures, and business intelligence. Her business focus includes process improvement, operations management, and organization change management. In these IT and business projects, she has served as lead business analyst, project manager, and scrum master. Margaret earned a B.S. in Information Technology from Virginia Tech and an M.S. in Information Systems from VCU. She holds PMP, CSP, and CBAP certifications.
Code Set Translations: Crosswalks – Are they problem solvers or problem generators?
Apr 10 2012
Within the health care industry, the ICD10 project lingers as the Centers for Medicare and Medicaid Services (CMS) has just set the October 2014 deadline for implementation. The International Classifications of Disease (ICD) codes are part of a large code set used within the health care industry to designate the symptoms or condition of the patient. Currently, everyone is using version 9, and CMS is mandating the use of version 10. When it comes to an upgrade, this is far from simple. One of the larger challenges within the ICD10 project is that it seeks to “crosswalk” the two different ICD code sets. A crosswalk can essentially be thought of as a translation from one code set to another. (See these links for an explanation of the ICD codes and the difference between ICD9 and ICD10.)
Additionally, the industry must also determine at which points within the system’s decision-making process are crosswalks required as there are a number of places where ICD codes are used within the system. They are used for patient administration and records, claims and billing, business intelligence, predictive analytics, profiling efforts. With a code set that contains approximately 13,000 ICD9 codes and another with approximately 65,000 ICD10 codes, translations must be performed in both directions. The direction and the cardinality of the relationships between new and old codes are critical for creating an efficient crosswalk within the values. In the direction of ICD9 to ICD10 – there’s at least a one-to-one relationship, and, in many cases, a one-to-many relationship. When translating from ICD10 to ICD9, there is additionally a one-to-zero case where there’s no previous code in existence. The translation is manageable. Several vendors have already provided loose translations, but they're not necessarily guaranteed. Understanding the system logic which enables the decision-making-process while using the new ICD10 codes is the basis of new business rules which determine the semantics and relevance to achieve the same or better results as the ICD9 codes. When a claim is received, the ICD10 code may need to translate back to an ICD9 code so that the claim can be processed according to old rules. When a hospital is trying to understand the history of patient utilization in different areas of medicine, the old ICD9 codes need to be translated to an ICD10 code in order to have all reports be consistent for purposes of comparison.
There’s a huge amount of complexity, not just relating to the upgrade itself, but also in making the switch without disrupting the current status-quo for payer and providers. So does evaluating the current logic and status-quo lead us to believe that a crosswalk isn’t appropriate? While determining the right ICD10 code and processing with new business/medical rules is the better path, a crosswalk is needed when dealing with the ICD9 codes that must make a step forward.
The health care industry recently went through a similar (but simpler) exercise in the case of the National Provider Identifier (NPI), a unique identifier for all the providers and hospital that all the payers must use. The use of a crosswalk, in this case, made sense because both the legacy set of identifiers and the tax identifiers remain in existence. Business rules were available to quickly resolve discrepancies, and the crosswalk still remains in place today even after several years since implementation. However, with ICD10 project, the intent is for the ICD9 codes to go away entirely. Creating and maintaining this crosswalk is more involved and expensive, has a shorter time line, and is significantly more complex.
So do crosswalks solve a problem or do they generate one? In my opinion, using a crosswalk for the ICD10 project creates more problems than it solves. For the NPI project the crosswalk was appropriate because the cost of overhaul wasn’t as tumultuous. This is all due to magnitude of complexity. With the NPI project, the crosswalk logic relied on the relationship between the provider and the payer. With the ICD10 project, everyone needs the same crosswalk or there will be severe discrepancies and disruptions for the patient records and claims processing with ramifications in downstream systems. Everyone needs to learn a new language in the health care industry. Operationally speaking, since some codes cannot be translated in a one-to-one manner, manual intervention will likely need to be made in a process that was once automatic. Many provider and payers will require extra time to perform actions they never needed to do before.
In order to get around the need for a crosswalk, the entire logic dependent upon the ICD9 code set must be overhauled in order to accept the ICD10 codes without breaking the process, and that is a considerable overhaul of the health care system’s IT systems as well. The cost involved to achieve a needed upgrade needs to be balanced with a feasible date, a roadmap and in-depth guidelines provided by CMS.
So how might one navigate through such a complex project? Part 2 will focus on the Analysis and Project Management that will be required.