search

chb / ccdaScorecard

Automatic C-CDA scorecard to promote best practices
Other
31 stars 12 forks source link

ScoreCard 'Errors' #9

Open Dapostol58 opened 10 years ago

Dapostol58 commented 10 years ago

I am sorry if this is not the correct forum, but I have some questions regarding the actual scoring results. While checking our newly created CCD code against the API we've received several messages that don't seem to make 'sense', per se. In the 'General' category:

  1. Best Practice: Codes that claim to be from SNOMED CT, LOINC, and RxNorm should be present in UMLS 2012AB. - The codes that are identified are in UMLS 2014AA, can we presume to 'ignore' this warning?
  2. Best Practice: Codes in a C-CDA should assign a valid @displayName that reflects the meaning of the underlying concept. A best practice is to use preferred labels from UMLS.

Your Results: 1 of 31 codes didn't match their displayName Your code Preferred term What now? SNOMED CT:173747005 "Roux-en-Y gastrojejunostomy" "Roux-en-Y gastrojejunostomy" See http://purl.bioontology.org/ontology/SNOMEDCT/173747005 - since our term matches theirs (exactly) we're presuming this is a 'bug'?

In the Lab Results category:

  1. Best Practice: Lab results should be coded using LOINC. In pratice LOINC is huge, but 2000 codes cover 98% of real-world usage. This means that most results should be covered by the 2000+ most common LOINC codes published by Regenstrief.

Your Results: 1 of 1 lab result codes weren't in the recommended value set Note: This may be normal, if this C-CDA document includes unusual labs for which no common LOINC code exists. But look through the un-matched codes below to make sure you don't have a mapping error in your export pipeline.

Recommended value set:

LOINC Top 2000 Codes Your code The issue What now? LOINC:26380-6 "Hand-L XR 3V" Not in value set See http://purl.bioontology.org/ontology/LNC/26380-6 - again, is this a bug? This is a valid LOINC code

In the Vitals category:

  1. This is more of an observation than anything else: Best Practice: Vitals in C-CDA should be coded with LOINC. Specifically, with codes from the HITSP Vital Sign Result value set.

Your Results: 5 of 47 vital sign codes weren't in the recommended value set Note: This may be normal, if this C-CDA document includes unusual vitals for which no appropriate LOINC code exists. But look through the un-matched codes below to make sure you don't have a mapping error in your export pipeline.

Recommended value set:

HITSP Vital Sign Result Your code The issue What now? "Pain severity" Not in value set See http://purl.bioontology.org/ontology/LNC/38214-3 "Pain severity" Not in value set See http://purl.bioontology.org/ontology/LNC/38214-3 "Pain severity" Not in value set See http://purl.bioontology.org/ontology/LNC/38214-3 "Pain severity" Not in value set See http://purl.bioontology.org/ontology/LNC/38214-3 "Pain severity" Not in value set See http://purl.bioontology.org/ontology/LNC/38214-3

MOST healthcare facilities now consider Pain as the 'sixth vital sign' and are requesting that it be recorded in the same section. Are there any initiatives to do so?

Thank you,

Debra S. Apostol

jmandel commented 10 years ago

Thanks for the detailed report + questions! I look forward to digging in in greater detail - but first, do you have a document that you can share so I can check out these issues directly? You can post for example in a GitHub gist or just email me (joshua.mandel@childrens.harvard.edu).

Thanks,

Josh

On Fri, May 16, 2014 at 3:06 PM, Dapostol58 notifications@github.comwrote:

I am sorry if this is not the correct forum, but I have some questions regarding the actual scoring results. While checking our newly created CCD code against the API we've received several messages that don't seem to make 'sense', per se. In the 'General' category:

  1. Best Practice: Codes that claim to be from SNOMED CT, LOINC, and RxNorm should be present in UMLS 2012AB. - The codes that are identified are in UMLS 2014AA, can we presume to 'ignore' this warning?
  2. Best Practice: Codes in a C-CDA should assign a valid @displayNamehttps://github.com/displayNamethat reflects the meaning of the underlying concept. A best practice is to use preferred labels from UMLS.

Your Results: 1 of 31 codes didn't match their displayName Your code Preferred term What now? SNOMED CT:173747005 "Roux-en-Y gastrojejunostomy" "Roux-en-Y gastrojejunostomy" See http://purl.bioontology.org/ontology/SNOMEDCT/173747005 - since our term matches theirs (exactly) we're presuming this is a 'bug'?

In the Lab Results category:

  1. Best Practice: Lab results should be coded using LOINC. In pratice LOINC is huge, but 2000 codes cover 98% of real-world usage. This means that most results should be covered by the 2000+ most common LOINC codes published by Regenstrief.

Your Results: 1 of 1 lab result codes weren't in the recommended value set Note: This may be normal, if this C-CDA document includes unusual labs for which no common LOINC code exists. But look through the un-matched codes below to make sure you don't have a mapping error in your export pipeline.

Recommended value set:

LOINC Top 2000 Codes Your code The issue What now? LOINC:26380-6 "Hand-L XR 3V" Not in value set See http://purl.bioontology.org/ontology/LNC/26380-6 - again, is this a bug? This is a valid LOINC code

In the Vitals category:

  1. This is more of an observation than anything else: Best Practice: Vitals in C-CDA should be coded with LOINC. Specifically, with codes from the HITSP Vital Sign Result value set.

Your Results: 5 of 47 vital sign codes weren't in the recommended value set Note: This may be normal, if this C-CDA document includes unusual vitals for which no appropriate LOINC code exists. But look through the un-matched codes below to make sure you don't have a mapping error in your export pipeline.

Recommended value set:

HITSP Vital Sign Result Your code The issue What now? "Pain severity" Not in value set See http://purl.bioontology.org/ontology/LNC/38214-3 "Pain severity" Not in value set See http://purl.bioontology.org/ontology/LNC/38214-3 "Pain severity" Not in value set See http://purl.bioontology.org/ontology/LNC/38214-3 "Pain severity" Not in value set See http://purl.bioontology.org/ontology/LNC/38214-3 "Pain severity" Not in value set See http://purl.bioontology.org/ontology/LNC/38214-3

MOST healthcare facilities now consider Pain as the 'sixth vital sign' and are requesting that it be recorded in the same section. Are there any initiatives to do so?

Thank you,

Debra S. Apostol

Reply to this email directly or view it on GitHubhttps://github.com/chb/ccdaScorecard/issues/9 .

jmandel commented 10 years ago

Summary

Hi @Dapostol58,

I've had some time to investigate -- overall we have:

  • some good fixes for the Scorecard
  • some errors in your document
  • some typically hard-to-resolve stuff

Let me take your points in order...

Best Practice: Codes that claim to be from SNOMED CT, LOINC, and RxNorm should be present in UMLS 2012AB. - The codes that are identified are in UMLS 2014AA, can we presume to 'ignore' this warning?

I've take this opportunity to update the Scorecard to UMLS2014AA. But after investigating the file you shared with me, the problem wasn't actually with our UMLS version: the problem is your document had extraneous spaces in some code names. For example:

  • code="55561003 "
  • codeSystemName="LOINC" code="54094-8 "

(Note the extra space before the close quotes in your codes: those shouldn't be there. Apologies for the superlatively unhelpful error messages :-))

Best Practice: Codes in a C-CDA should assign a valid @displayName that reflects the meaning of the underlying concept. A best practice is to use preferred labels from UMLS.

SNOMED CT:173747005 "Roux-en-Y gastrojejunostomy" "Roux-en-Y gastrojejunostomy" See http://purl.bioontology.org/ontology/SNOMEDCT/173747005 - since our term matches theirs (exactly) we're presuming this is a 'bug'?

This was indeed a bug. Now fixed -- thanks!

LOINC Top 2000 Codes Your code The issue What now? LOINC:26380-6 "Hand-L XR 3V" Not in value set See http://purl.bioontology.org/ontology/LNC/26380-6 - again, is this a bug? This is a valid LOINC code

This is a valid LOINC code for sure -- and the Scorecard agrees with you on that point! It just happens not to be in the top 2000 most common LOINCs, so it gets flagged for review. (The reason for this behavior is we've seen many people use incorrect/obscure codes for common things like WBC, and we want to make sure that uncommon codes are chosen intentionally.)

This is more of an observation than anything else: "Pain severity" Not in value set See http://purl.bioontology.org/ontology/LNC/38214-3 MOST healthcare facilities now consider Pain as the 'sixth vital sign' and are requesting that it be recorded in the same section. Are there any initiatives to do so?

I couldn't agree more about the clinical importance of recording subjective pain scale observations! On this point though, the value set called for in C-CDA is maintained (or rather... isn't actively maintained) by HITSP. This would be a good issue to raise on HL7's Structured Documents Working group mailing list: http://www.hl7.org/myhl7/managelistservs.cfm

For now, my recommendation is to keep doing what you're doing and include pain scale observations with your vitals! You'll note that the Scorecard doesn't actually deduct any points for this -- it's just raising a flag to make sure you know what you're doing (which you do!).

jmandel commented 10 years ago

Two more notes:

Since your "Results" is in fact a diagnostic imaging report, you may want to look at the "Findings" section which C-CDA uses specifically to convey the body of such reports:

image

Also, there are lots of lab results embedded in free-text in your document. Ideally these would be coded as Results :-)

  • © Githubissues.
  • Githubissues is a development platform for aggregating issues.