Hello,
I am working on an OMOP based project for my Master’s. As part of it, I am performing ETL from a FHIR like resource into OMOP CDM. Some of the conditions in the source system are coming from DSMIVTR and DSM5CA vocabulary. I could not find a mapping for these in my database or in Athena. I have searched the forums but could not find any information regarding my issue. Has someone run into this? If so, how did you go about addressing this requirement? As a newbie to the OMOP world, looking for any information/recommendations that could help me get over this challenge. Thank you!
Hi @Krishna , excuse my ignorance, but is DSM5CA the DSM-5 version used in Canada? But DSMIVTR is probably not the Turkish version of DSM-4, is it?
In any case: it seems that you are facing what almost everyone faces when doing an ETL: while the almighty Athena has a very comprehensive set of vocabularies and concepts, it is not covering each and every terminology and concept in the world that is. Athena covers lots of sources and mappings to standard target concepts but for the ones that are not covered, you have to do the heavy lifting yourself. How many concepts are we talking about? Do you have frequencies / counts of occurrence in the source data alongside with? You probably know that already, but the process described in the Book of OHDSI including the use of the Rabbit family of tools should already take you a long way. I am quite confident that you can identify good targets for example in SNOMED. If you however can describe a valid use case, find more support in the OHDSI community and maybe even some funding and hands to help, maybe at least the DSM-5 could be introduced to Athena as a new license restricted vocabulary (given that the APA permits).
Hi @mik,
Thank you for responding. I should have been little bit clearer - my project is for a Canadian institute. So, they are specifically DSM-5-CA and DSM-IV-TR (DSM-IV-TR - an overview | ScienceDirect Topics). I don’t think TR is related to Turkey, but I could be wrong.
There are 318,869 DSM-5-CA concepts and 49,087 DSM-IV-TR concepts in addition to 29160 ICD-10-CA concepts that I need to map to the standard vocabulary.
I have come across a mapping from DSM-5 to ICD-10-CM and ICD-9-CM here
but I don’t think it helps me as I need DSM-5-CA and DSM-IV-TR.
I have tried both the Rabbit tools but they were not much help with the Vocabulary challenges. I was going to try the Usagi tool as the last resort after confirming here.
In regards to providing more details about my use case, let me follow up with my superiors and if I get their permission, I will definitely share more.
Hi @Krishna ,
your link provided the answer: " DSM-IV-TR, a text revision of DSM-IV ". DSM-4 is a fairly old terminology, so there is less drive to be expected by other community members to adopt this as a vocabulary.
If you can get hold of reference sets (like the one you linked), this is a good first step. The Rabbit tools are accompanied by Usagi (which is also a “Rabbit” tool, only in Japanese) and their use actually prepares you a little for creating mapping input. I would also assume there is a lot of overlap between the Canadian and the US version. You might want to create a delta file and reserve that for special processing. Same goes for the ICD-10-CA. As you can see, in order to create your mappings, a multi angle approach is normally employed: some use of reference sets and existing vocabularies and for what is left over, try Usagi (which is helpful as you can filter what Usagi does the string matching against, e.g. only conditions and particular target vocabularies). The numbers that you mention seem to be your overall number of records. Now you need to sort them by frequency so that you have a first criterion for prioritization for your mapping efforts. Another criterion will be your research question and special interest in particular areas or even individual codes. Go for it! It will require quite a bit of manual work and domain knowledge but you will get there.