Can ICD-10 codes save lives?

There is a lot of potential in ICD-10 coding. The U.S. version, ICD-10-CM, has 68,000 diagnosis codes. All those codes represent data points that can be mined for rich medical research. We should be able to take that research and greatly improve patient care.

In other words, save lives.

But Carl C. Jaekel of Santa Rosa Consulting concludes that ICD-10 isn't living up to that potential and saving lives.

Jaekel looked at several studies based upon mortality data using ICD-10 codes from around the globe. He found no correlation between more specific data and mortality rates.

"Thus, the current state-of-ICD-10 in directly reducing mortality has little empirical evidence to support a relationship between ICD-10 codes and statistically valid improvement in mortality, as of mid-2012. However, as a tool to improve care delivery and reduce ambiguity of diagnoses, ICD-10 offers tremendous potential…Once ICD-10 becomes commonly utilized within the United States, we anticipate that a range of studies will demonstrate that ICD-10 is a valid tool in preemptively reducing mortality. However, this will necessarily wait upon broad adoption of ICD-10 and a period of elapsed time between ICD-9 cessation and ICD-10 utilization, such that statistically valid results can be drawn."


Jaekel focuses on studies from outside the United States because we have not implemented the ICD-10 code sets yet. But we have been using the World Health Organization's (WHO) version since 1999 for mortality reporting. That's 13 years of data.

The National Center for Biotechnology Information has an index of ICD-10 based studies if you're looking for some more poolside reading. But there don't seem to be a lot of examples of those studies leading to anything.

Maybe because ICD-10 only has about 14,000 potential diagnosis codes. Yes, that's twice as many as the almost 7,000 ICD-9 diagnosis codes. (Note that I'm using the unmodified WHO versions here.) Do we need to expand that to 68,000 ICD-10-CM codes to get actionable research?

Or the ICD-10 code set could lack fundamental improvements over the ICD-9 code set. That doesn't make the argument that we should leapfrog right into ICD-11 coding. Talk about potential. But we don't even know what ICD-11 will look like by the time it is ready for implementation.

I'm thinking that this is a very long-term process. Look at what needs to be done:

  1. Wait until there are enough cases to provide a statistically significant sample size that shows a problem.
  2. Collect the data, study it and write a paper.
  3. Distribute and let society debate its merits.
  4. Make changes to improve patient care and save lives.
  5. Wait until there are enough cases to provide a statistically significant sample size that shows if the problem is solved.


So we're at Step 3. It's going to be a while before anyone decides to improve anything. And it will be even longer before we know it's working.

In the meantime, visit ICD10Watch regularly to keep up with the developments of ICD-10 implementation.

Add new comment