At the end of Part 9, we were translating a list of ICD-9 codes – a policy – into ICD-10. We used the 10-to-9 single GEMs with reverse lookup to find ICD-10 codes that should be in your ICD-10 version of the policy. We had some ICD-9 codes left over that no ICD-10 code translated to. You tried to look them up in the 9-to-10 single GEMs. You found some translated to ICD-10 codes already in your ICD-10 policy list, so you could feel assured that their meaning was taken care of. A few may have translated to single ICD-10 codes not already on your list. Those ICD-10 codes (“pink” in CTT) might be appropriate for your policy, but a clinical review of them was recommended.
Finding all the ICD-10 codes that might be on a patient’s record, and that might imply the patient satisfies the policy, is the objective of our process. Have we now found them all? Consider this case from Part 10:
One ICD-10-PCS procedure
037K34Z, Dilation of Right Internal Carotid Artery with Drug-eluting Intraluminal Device, Percutaneous Approach
is the result of translation from a cluster of four ICD-9 procedure codes
00.61, Percutaneous angioplasty of extracranial vessel(s)
00.40, Procedure on single vessel
00.55, Insertion of drug-eluting stent(s) of other peripheral vessel(s)
00.45, Insertion of one vascular stent
Suppose you have 00.55 in a policy designed to pay differently for procedures where a drug-eluting stent is inserted. The policy may also contain 00.60 and 36.07. When you try to translate 00.55, you do not find it in the 10-to-9 singles, and all you find in the 9-to-10 singles is NOPCS.
I picked 00.55 because it is an extreme (though not isolated) example. 00.55 is an ICD-9 adjunct code – a “code” that is never supposed to be coded on its own, but rather included to modify the meaning of another code (in this case the PTCA, 00.61). ICD-10-PCS does not use adjunct codes. Every meaningful combination of 00.61 and its adjunct codes is represented by one ICD-10-PCS code.
Adjunct procedure code 00.55 is just one example of an ICD-9 code that requires other ICD-9 codes to be on the record with it in order to specify the same patients as can be specified with just one ICD-10 combination code. This happens with diagnoses, too, such as:
I25.110, Atherosclerotic heart disease of native coronary artery with unstable angina pectoris
requiring two ICD-9 codes to convey the same meaning
414.01, Coronary atherosclerosis of native coronary artery
411.1, Intermediate coronary syndrome
Let us pursue this second example a little further. Suppose you had a policy for finding all patients with coronary atherosclerosis. 414.01 would be on it. When you translated 414.01 using the 10-to-9 singles GEM table, you would have found I25.10, Atherosclerotic heart disease of native coronary artery without angina pectoris. Good enough? What about those people who also had angina? Assuming you want them too, you would have wanted I25.110 (and several other similar codes) in your ICD-10 policy. But you would not have found them with the process as outlined so far.
So here is where we add the third phase to the process – look up each ICD-9 code in your policy in the 10-to-9 clusters GEM table.
When I recommended dividing the GEMs files into four tables back in Part 6, I suggested making a reverse index for the 10-to-9 singles GEM, but I forgot to suggest that you also do so for the 10-to-9 clusters GEM table. I apologize. With any ICD-9 code in your policy, you will want to see what ICD-9 clusters it is a part of, and that is only practical if you’ve resorted (or re-indexed) the 10-to-9 clusters GEM by ICD-9. CTT and suchlike tools do this for you.
What do you do when you find one of your policy’s ICD-9 codes in a 10-to-9 cluster, and the ICD-10 code you find with it isn’t already on your ICD-10 policy list? In theory, you do not have the whole story – you don’t know what other ICD-9 codes are expected to also be on the record to equal the precise meaning of the ICD-10 combination code you’ve found. We could go through all the rigmarole of using the scenario and choices columns of the table to find them (which we will unfortunately have to get to later) but you really don’t have to know them. (Another benefit of CTT and similar tools is that they will show all that to you with just a few clicks.) No, all you need to do is review the ICD-10 code to see if it deserves to be in your policy. Its code title (or failing that, its Tabular entry) should give you all you need to know.
Going back to our example, I25.110, Atherosclerotic heart disease of native coronary artery with unstable angina pectoris, tells you that the code selects patients not only with coronary atherosclerosis, but also unstable angina. What is the policy about? Everyone with coronary atherosclerosis? Then you want it. Only patients without angina? Then you don’t. You don’t have to know what ICD-9 codes the GEMs equated I25.110 to.
Now let’s look at the adjunct code example. When you looked up 00.55 in the 10-to-9 GEM with reverse index, you didn’t find anything. When you looked up 00.55 in the 9-to-10 singles GEM, you only found NOPCS. There is no PCS code that just means “insertion of drug-eluting stent.” But when you look up 00.55 in the 10-to-9 cluster GEM, you see 202 ICD-10-PCS codes that are expressed in ICD-9 as clusters, where the cluster contains 00.55. According to the GEMs, these are all ICD-10 codes that say, among other things, that a drug-eluting stent was inserted into a peripheral vessel. If your policy applies to all patients where a drug-eluting stent was inserted into a peripheral vessel, then you can safely add all 202 codes to your ICD-10 policy list. If not, you should look at all 202 and pick the ones that satisfy the intent of the policy.
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.