OHDSI Home | Forums | Wiki | Github

Ask for help from a newcomer

(Rita Rb-Silva) #1

Hi everyone,

I have some doubts regarding the use of ATHENA. Is there anyone with some availability to help me?

Specifically, I am looking at the exercises of “Chapter 5 - Standardized Vocabularies” of the OHDSI book. The first (5.1) is very easy-peasy, but the second (5.2) is not (for me)… I keep getting different possible answers and I don´t understand why the suggested answers do not include all the possibilities.

I tried these strategies among others:

  1. Searching for “Gastrointestinal hemorrhage” as keyword and filtering “ICD10CM” or “ICD9CM” in Vocabulary;
  2. Searching for “Gastrointestinal hemorrhage” as keyword and filtering Standard concept = “Standard”; selecting Concept_ID=192671 and looking at the Term connections…

I’m sorry to post this naive question, but I having a hard time trying to figure this out by myself.

(Katy Sadowski) #2

I believe the second strategy you suggest should be sufficient to locate all of the codes - however, I also do not see ICD9CM code 578 “Gastrointestinal hemorrhage” on the ATHENA page for the standard Gastrointestinal hemorrhage concept.

I think this might actually be an issue in ATHENA, because in ATHENA, this ICD9CM code only maps to the standard concept for Hematemisis (https://athena.ohdsi.org/search-terms/terms/44830170), which would seem to be incorrect. But when I check on my local ATLAS instance, it maps to the standard concept for Gastrointestinal hemorrhage. :thinking:

(Don Torok) #3

When I followed your second strategy using Athena and looking at Term Connections I see Hemorrhage of gastrointestinal tract, unspecified 44819812 ICD9CM under the section for Standard to Non-standard map (OMOP). And concept id 44819812 is for ICD9CM code 578.9 [Hemorrhage of gastrointestinal tract, unspecified]

(Rita Rb-Silva) #4

The suggested answer for exercise 5.2 is:
a) ICD-10CM codes:
* K29.91 “Gastroduodenitis, unspecified, with bleeding”
* K92.2 “Gastrointestinal hemorrhage, unspecified”
b) ICD-9CM codes:
* 578 “Gastrointestinal hemorrhage”
* 578.9 “Hemorrhage of gastrointestinal tract, unspecified”

My second strategy gives the ICD-10CM codes K29.91 and K92.2, but for ICD-9M we only get 578.9. However, if we click on “Hemorrhage of gastrointestinal tract, unspecified”, we can see that this concept has two term connections:

  • “is a”, “Gastrointestinal hemorrhage”, concept_id = 44830170, which has the concept_code = 578
  • “Non-standard to Standard map (OMOP)”, “Gastrointestinal hemorrhage”, concept_id = 192671.

My doubts are:

#1: Why (“Gastrointestinal hemorrhage”, concept_id = 44830170, concept_code = 578, vocabulary = ICD9CM), does not appear in the list of Term Connections of (“Gastrointestinal hemorrhage”, concept_id = 192671, concept_code = 74474003, vocabulary = SNOMED)?

#2: If (“Gastrointestinal hemorrhage”, concept_id = 44830170, concept_code = 578, vocabulary = ICD9CM) has 3 subsumes, shouldn’t we consider the three?
I understand that “Gastrointestinal hemorrhage” is “site-unspecific” and “equivalent” to Hemorrhage of gastrointestinal tract, unspecified, while Hematemesis and Blood in stool are “site-specific”. But Hematemesis and Blood in stool are types of Gastrointestinal hemorrhage, so I think they should be included in the “ICD-9CM codes that map to this Standard Concept”.

#3: Why Blood in stool has a term connection to Hematochezia but not to Melena?

(Katy Sadowski) #5

#1 - See my post above. I think this is an issue with ATHENA.

#2 - The ‘Maps to’ relationship only covers direct relationships of non-standard to standard concepts.
Parent/child relationships are covered via ‘Is a’ and ‘Subsumes’ in the standard concept hierarchy. So those “site-specific” ICD9 codes you list will each map just to the respective standard concepts for those site-specific conditions. The relationship of those child concepts to the parent Gastrointestinal hemorrhage concept will be represented in the standard concept hierarchy. (Note that the ICD9 hierarchy is separately represented using ‘Is a’/‘Subsumes’, which are the subsumes you’re seeing on the ICD9CM 578 page)

#3 - It looks like the standard concept for Hematochezia has a synonym of Blood in stool, so these terms are considered equivalent in the hierarchy. The standard concept for Melena, however, is on its own, separate branch of the tree.

(Yevhenii Bezeha) #6

Hi, @rita.rbsilva , @katy-sadowski ,
Thank you for your interest in OHDSI vocabulary. Your feedback helps us to improve it.
I will add the information to the previous answer.

  1. Here is an issue in Athena. Gastrointestinal hemorrhage ICD9CM concept is mapped with a mistake. You are right. We will change the mapping to 192671 in the next release.

  2. Here I would like to add.
    Gastrointestinal hemorrhage ICD9CM concept has 3 subsumes, you are right, according to http://www.icd9data.com/2012/Volume1/520-579/570-579/578/default.htm

Hematemesis ICD9CM concept now has mapping to 26727 concept_id, Hematemesis concept_name, SNOMED vocabulary_id and it is correct because Hematemesis SNOMED concept has as a parent concept 192671 concept_id, Gastrointestinal hemorrhage concept_name, SNOMED vocabulary_id. Please take a look.
http://ontoserver.csiro.au/shrimp/?concept=8765009&system=http://snomed.info/sct&versionId=http://snomed.info/sct/900000000000207008/version/20190731. We won’t map one concept to a parent and a child concept at the same time in this case. And we map it only to a child concept. So the mapping is correct now.
In other information @katy-sadowski is right. Thanks.

  1. Blood in stool ICD9CM concept has a term connection to Hematochezia SNOMED but it will be remapped to Melena SNOMED as you said because it is better to map like that. And it will be fixed in the next release.
    Also we will add here mapping to 192671 concept_id, Gastrointestinal hemorrhage concept_name, SNOMED vocabulary_id because Melena SNOMED doesn’t have this concept as a parent one, it is on its own, separate branch of the tree.
    I hope I answered your questions. If you have any other questions do not hesitate to ask.

(Rita Rb-Silva) #7

Shrimp seems a great tool for someone who is learning SNOMED and OMOP CDM from scratch!
I hope that my contribute gets more significant as OHDSI Vocabulary becomes more familiar to me.
Thank you @katy-sadowski, @DTorok and @Yevhenii_Bezeha