DIY ICD-10 conversion – Part 13

May 30th, 2014 / By Ron Mills, PhD

In Part 12 we looked up unused ICD-9 codes in your policy in the 9-to-10 GEMs cluster table. If we found a code in there, it would lead us to one or more translation alternatives, each of which consists of two or more ICD-10 codes which have to appear on the patient’s record together in order to convey the same meaning as the ICD-9 code.

Here again is one of the examples we looked at:

806.00 Closed fracture of C1-C4 level with unspecified spinal cord injury

for which the GEMs provides four alternative translates, all clusters. Here is the first one:

S14.101A Unspecified injury at C1 level of cervical spinal cord, initial encounter

With one of

S12.000A Unspecified displaced fracture of first cervical vertebra, initial encounter for closed fracture

S12.001A Unspecified nondisplaced fracture of first cervical vertebra, initial encounter for closed fracture

(The other three are similar clusters for the C2, C3 and C4 vertebrae.) The 806.00 is a combination code – it specifies both a closed fracture and a spinal cord injury. To convey the same meaning, it takes two codes in ICD-10-CM: one for the closed fracture, one for the spinal cord injury.

If you had 806.00 in your ICD-9 policy, would you want this cluster in your ICD-10 policy? To answer this question you have to consider what your policy list is for.

  • Is it a list of codes for finding all patients with a spinal cord injury? If so, you will want the S14 codes on your ICD-10 list, but not the S12 codes.
  • Is it a list of codes for finding all patients with a closed fracture of cervical vertebra (or perhaps of any vertebra, or perhaps of any bone)? If so, you will want the S12 codes on your ICD-10 list but not the S14 code.
  • Or is it a list for finding all patients with both cervical vertebra fractures and spinal cord injuries? If so, your policy needs to preserve the cluster as a cluster, and you may have a problem – the problem we alluded to back in Part 10. Can your ICD-10 system handle clusters?

If you find yourself in the third category (needing the cluster – in this example, looking for patients with both a spinal cord injury and a cervical fracture), then you will want to consider how the policy is implemented. If it is “just” a written specification in a policy document, then you can carefully spell out that you are looking for patients who have both the S14 and the S12 code on their record, and hope that the human brains that read the document will understand what you mean.

If the implementation is for a computer program (for example, creating a report on which cervical fractures with spinal cord injuries have their own row) then you may have to do some re-programming. For each scenario you include in your policy, the computer program will have to look for any of the codes with choices 1 on the record, then if it finds such a code, look for any code with choices 2, and so on. The order of choices in the GEMs does not matter. Nor do the codes in a cluster have to occur in the record adjacent to each other. S14.101A may be before or after S12.000A or S12.001A and there may be other codes between them.

When dealing with 9-to-10 clusters, you should also look to see if any of the ICD-10 codes in the cluster are already on your list, thanks to the 10-to-9 singles translation you performed in the first phase of this process. If they are, then the cluster is what we call a redundant cluster and you can ignore it.

To continue with our example, suppose you had ICD-9 code 952.00, Spinal cord injury without evidence of spinal bone injury, C1-C4 level with unspecified spinal cord injury, in your ICD-9 policy list. (That right away tells you that the policy is about spinal cord injuries and not about vertebral fractures.) The 10-to-9 singles phase of the translation has already put S14.101A on the ICD-10 version of the policy. So when you translate 806.00 and get a cluster that includes S14.101A, you can safely ignore the cluster. Why? Because you have already determined that if S14.101A is on the patient’s record, then it is a patient you want. Checking to see if the record also has one of the S12 codes listed above (which is what cluster processing requires) buys you nothing. You are already including the patient in the policy.

Uncork a bottle of champagne, crack open a beer, pour a glass of chocolate milk – your celebratory drink of choice – for we have completed GEMs-based translation of ICD-9 policies!

To summarize:

Phase 1: 10-to-9 singles GEM with reverse lookup

Phase 2: 9-to-10 singles GEM with clinical review to find additional ICD-10 codes that belong in your policy

Phase 3: 10-to-9 cluster GEM for ICD-10 combination codes that should also be in your policy

Phase 4: 9-to-10 cluster GEM, to find “necessary clusters” for your policy, if your system can handle clusters

In Part 14, we will move on to the second of five ways that ICD-9 codes may be used in your organization—analytics.

Ron Mills is a Software Developer for the Clinical & Economic Research department of 3M Health Information Systems.

You can find the complete DIY ICD-10 series here.