This question comes up on the HL7 Structured documents mailing list about this text in CCDA:
If the diagnosis is unknown or the SNOMED code is unknown, @nullFlavor SHOULD be "UNK". If the code is something other than SNOMED, @nullFlavor SHOULD be "OTH" and the other code SHOULD be placed in the translation element (CONF:10142).
The querant wants to know what that means, and provides a few examples (which I've used below).
If all you have is an ICD-9 (or ICD-10 or other code), and you don't know the SNOMED CT code or cannot map to SNOMED CT, then what you should do is in his example #1 below:
1. Just provide
the ICD-9 code in a translation
<value xsi:type="CD"
nullFlavor="OTH"
codingSystem="2.16.840.1.113883.6.96">
codingSystem="2.16.840.1.113883.6.96">
<translation
code="401.9"
displayName="Hypertension, essential"
codeSystem="2.16.840.1.113883.6.104"
codeSystemName="ICD-9-CM"/>
</value>
2. Provide the
(mapped) SNOMED CT code up-front, and ICD-9 in translation
<value
xsi:type="CD" code="59621000"
codeSystem="2.16.840.1.113883.6.96"
codeSystemName="SNOMED CT"
displayName="Essential hypertension"/>
<translation
code="401.9"
displayName="Hypertension, essential"
codeSystem="2.16.840.1.113883.6.104"
codeSystemName="ICD-9-CM"/>
</value>
This is exactly the same guidance that I provided in Stage 1 for using the HITSP C32.
Using "OTH" as the null flavor isn't ideal, but may be the best you can do in CDA R2. OTH says "this concept can't be expressed in the specified code system", when what you actually want to say is "I haven't tried to express this concept in this code system".
ReplyDeleteIn either case, datatypes R1 says that you can only use OTH if you declare the codeSystem attribute. (In R2, you can declare either the codeSystem or the valueSet.)
Thanks for that correction Lloyd. I knew that and forgot to check for it. I've updated Josh's example to reflect the correction in sample #1.
DeleteAnd perhaps better would be to say: nullFlavor="UNK" when you haven't tried to map, rather than cannot find a good mapping.
DeleteIn the current vocabulary, there's a specific code for this - UNC (unencoded) which is a specialization of INV (invalid) which itself specializes NI (no information). Given that neither UNC nor INV are available in CDA R2 (though they will be in R3 I expect), "NI" is probably the best flavor to use. UNK means the information isn't known - an obviously it is known if you can provide a translation. When you need to adopt a code that doesn't have the exact meaning desired, your best bet is to settle on the nearest ancestor available or something you're sure no-one will use for another purpose. I expect UNK and OTH will both be used for their "real" purposes, and thus would be reluctant to recommend their use for dealing with "unable to translate to the desired code system" situations.
DeleteLive link to the example that raised this question:
ReplyDeletehttps://github.com/chb/sample_ccdas/tree/master/Cerner%20Samples
Cerner contributed this sample; we'd love to collect lots more!
Is there a standard way of representing data that is not encoded in any system? We have an example for an allergy reaction being entered as free text and there is no element available for this case.
ReplyDeleteIs this kind of translation still valid for the CCDA requirements in MU stage 2? We have seen examples using this, but are now being informed that it is not sufficient to pass certification. We only use ICD-9.
ReplyDeleteThanks for all you do!
In the translation element (based on the examples above), we use "2.16.840.1.113883.6.104" as our code system for ICD-9-CM. What do we use as our codeSystem for ICD-10-CM? I'm so frustrated searching on google for the codeSystem for ICD-10-CM and came up with nothing! Thanks for your help!
ReplyDeleteIf some of the medications in CCDA are represented with @nullFlavor "UNK" or "OTH" and the rest of the medications are represented with the Standard RXNORM code is this CCDA MU compliant? or Does MU require all in a section represented with a standard code?
ReplyDeleteThank you,
Tara