July 25, 2011

Making the Case for Translation from ICD-9 to ICD-10


If you want to see an ICD-10 program manager or business lead squirm, bring up the topic of translation from ICD-9 to ICD-10. The ICD-10 to ICD-9 translation topic doesn’t produce nearly the same level of angst due to the availability of the more one-to-one-laden reimbursement mappings.

Many of our clients are “pulling an ostrich” (planting their heads deeply in the sand and ignoring the world) when it comes to the concept of translation from ICD-9 to ICD-10, and hoping that they can avoid it entirely. Many payers and providers so far have chosen not to translate, opting for a “step back” approach to analytics (i.e., after Oct. 1, 2013, they will translate new ICD-10 claims/encounters to ICD-9 using the reimbursement mappings, and cross-correlate legacy data to the step-backed ICD-10 codes for trending).

This approach will work for many companies and for many types of analytics. However, the intent and approach to the reimbursement maps may not be clinically relevant for every purpose, and we believe that for some purposes (and for some companies looking to gain benefits from the increased specificity in ICD-10), translation from ICD-9 to ICD-10 will be a requirement, especially for diagnosis codes. Consider that for some regulatory, quality and trading partner analytics, trending in ICD-9 with stepped-back ICD-10 codes will not be acceptable.

But maybe translation from ICD-9 to ICD-10 is not so bad.

In our body of work providing analysis and translation services for clients, we’ve found that translation from ICD-9-CM to ICD-10-CM is achievable with very little manual intervention and a high degree of automation. The numbers might surprise you.

There are approximately 14,000 ICD-9-CM codes listed in the GEMS 9-to-10 maps that map out to one or more ICD-10-CM codes. Of these mapped codes:

  • Approximately 78 percent are compatible with “one-to-one” mappings (i.e. the GEMS maps contain one and only one correlating ICD-10-CM code to the ICD-9-CM code.) Of these, 25 percent are exact while 53 percent are approximate.
  • Approximately 17.5 percent translate as “one-to-many” approximate mappings (i.e. the GEMS maps contain more than one correlating ICD-10-CM code to the ICD-9-CM code, and GEMS specifies that the mappings are all approximate: any of the ICD-10-CM codes are a potential map).
  • Approximately 4.5 percent are “one-to-many” combination mappings (i.e. the GEMS maps contain multiple ICD-10-CM codes for the ICD-9-CM code, with GEMS specifying that the mapping is of the combination type. There can be one or more scenarios, with potentially multiple choice lists per scenario).

It’s the 22 percent of the codes that are one-to-many (approximate or in combination) that are problematic in that they cannot be translated automatically using the base GEMS maps. However, with some analysis a high percentage of these mappings can be simplified. My company has invested in a value-added mapping called the “HighPoint GEMS” to get to a higher percentage of mappings that can be translated automatically.

  • Of the “one-to-many” approximate mappings, for roughly 70 percent of the codes it is fairly straightforward to pick the most generic mappings (i.e. where there are some very specific codes and one that is much less specific, using “other” or “unspecified”) and arrive at a one-to-one approximate mapping.
  • Of the “one-to-many” combination mappings, for roughly 47 percent of these codes there is only one scenario and the number of ICD-10-CM codes is equal to the number of choices, making it fairly straightforward to translate.


The net of it is that approximately 92 percent of the mapped codes can be translated automatically from ICD-9-CM to ICD-10-CM.


That’s not so bad.


About the Author

John Wollman is the Executive Vice President of Healthcare for HighPoint Solutions, a Management and Information Technology consulting firm focused on Healthcare and Life Sciences.  John is responsible for HighPoint’s Healthcare industry group, catering to Payers and Providers.  John is a recognized expert in several healthcare business domains (Reform, HIPAA 5010, ICD-10, Platform Strategy) and technical domains (Master Data Management, Analytics).  Since graduating from Duke University, John has held executive level positions at consulting and technology companies over his 25 years in business.

Contact the Author

To comment on this article please go to

Read 9 times Updated on September 23, 2013