Looping in Firas Wehbe, who is down deep with our data for his thoughts. Clinically at Vanderbilt, we are using a large SNOMED CT subset for our clinical problem lists, built with a term expansion to aid in search and a user interface that includes some features to make search more efficient (e.g., autocomplete, smart search results sequencing, NLP-based recommendations) and we use ICD9 for billing. Right now, we don't use mappings from SNOMED to ICD to assist with billing, so they are two separate steps. However, the systems we have in place allow ICD preferences based on patient and provider, so it can be efficient. Via this route, we have gone to a high proportion of SNOMED coded problems on problem lists over the past couple years.
________________________________ From: Klann, Jeffrey G. [[email protected]] Sent: Friday, November 28, 2014 5:43 PM To: Campbell, James R; Russ Waitman; Charles Borromeo Cc: Visweswaran, Shyam; Rosenbloom, Trent; Phillips, Lori C.; Murphy, Shawn N.; [email protected] Subject: Re: SNOMED in i2b2 Hi Chuck, Lori Philips has imported SNOMED clinical findings into an i2b2 ontology. A version is available here: http://i2b2.bioontology.org/ We considered distributing this with the SCILHS ontology, but the size of it made it unmanageable and our sites have not needed it for the CDMv1 domains. I suspect, like Russ, in CDMv2 we will use LOINC for labs and RxNorm for meds and we will not venture in to SNOMED at this time. But I’m definitely curious about Jim’s work with transitive closures in i2b2 queries. I am not familiar with this. Happy Thanksgiving, Jeff K. Jeffrey Klann, PhD Instructor of Medicine, Harvard Medical School Assistant in Computer Science, Massachusetts General Hospital PhD in Research Information Systems and Computing, Partners Healthcare ofc: 617-643-5879 [email protected] From: <Campbell>, James R <[email protected]<mailto:[email protected]>> Date: Thursday, November 27, 2014 at 10:12 AM To: Russ Waitman <[email protected]<mailto:[email protected]>>, Charles Borromeo <[email protected]<mailto:[email protected]>> Cc: Jeffrey Klann <[email protected]<mailto:[email protected]>>, "Visweswaran, Shyam" <[email protected]<mailto:[email protected]>>, "[email protected]<mailto:[email protected]>" <[email protected]<mailto:[email protected]>>, "Phillips, Lori C." <[email protected]<mailto:[email protected]>>, Shawn Murphy <[email protected]<mailto:[email protected]>>, "[email protected]<mailto:[email protected]>" <[email protected]<mailto:[email protected]>> Subject: RE: SNOMED in i2b2 Hi Charles! We have rolled SNOMED CT out in Nebraska and are deploying Clinical findings/Events/Situations for the problem list, leaving the encounter and billing diagnoses to ICD* as is mandated. Practically speaking, there is heavy cross-walk in content between the two domains, but when PCORI CDMV1 specified encounter diagnosis, the split seemed logical and appropriate given the cumbersome deployment in i2b2 of dual ontologies for a single set of data. We will be deploying surgical history in SNOMED CT procedures and will use other components of the SNOMED ontology as we go forward. There is a technical reason we have been slow since the combinatorial explosion of the SNOMED CT ontology for Clinical Findings using i2b2 standard approach to ontology construction led me to trim the hierarchical views and employ a different procedure for aggregation queries using transitive closure tables as first suggested by Kentucky. We are testing the speed and efficiency of the query currently. We will be distributing this to anyone who wants it as soon as I am sure it is technically sound. I have arranged with Olivier Bodenreider to get an RDF release of RxNORM/NDF-RT that we can access on a regular maintenance cycle and we will be testing that for deployment of presciptions and med orders. As Russ says, LOINC supports all lab, although we are testing our deployment model for molecular biology and microbiology. We have been rolling out LOINC/SNOMED CT for clinical observations and I am working to develop a reasonable ontology for clinical measurements and vital signs. ICD-O3 is aligned perfectly with SNOMED CT body structure/neoplasms and we will be using that for ontology layer there with the mapped ICD-O3 code. Jim ________________________________ From: [email protected]<mailto:[email protected]> [[email protected]<mailto:[email protected]>] on behalf of Russ Waitman [[email protected]<mailto:[email protected]>] Sent: Thursday, November 27, 2014 8:55 AM To: Charles Borromeo Cc: Klann, Jeffrey G.; Visweswaran, Shyam; [email protected]<mailto:[email protected]>; Lori C. <LCPHILLIPS@PARTNERS. ORG> Phillips; Shawn N. Murphy; [email protected]<mailto:[email protected]> Subject: Re: SNOMED in i2b2 Hi Chuck, Happy Thanksgiving! >From my experience here and looking at most i2b2 installs on babel (though I >defer to Jim Campbell): 1. Practically we’re largely ICD9 centric for diagnoses (and park IMO terms below the ICD9 folders for) because that’s how most of the data was recorded. I think once you’ve got basic concept terms like DX-ID in Epic that roll up using IMO to either SNOMED or ICD, you can build folders to hold them. I see actually more SNOMED used within Epic “smart data elements” that could be contained in notes (but are difficult to extract from Clarity; we want to tackle that once we get Chronicles access sorted out). My sense is most clinicians/researcher are familiar with ICD9 like folders so informatics teams have built from the UMLS or pulled from Harvard’s starter i2b2 demo data ontologies. “Most” people clinically don’t think in terms of SNOMED as an interface terminology which is probably why you don’t see it as much and until lots of raw data is coded in SNOMED it may not have as much visibility. Now Jim Campbell is the SNOMED/LOINC leader for us and I think we’ve got pathways for how we manage these conceptual linkages but practically much on purely diagnoses sits in ICD9 like flavors. Trent at Vandy might also comment. I think as we work past phase 1 and get the fundamental observations interoperable, people can prototype different i2b2 folders that use different interface “folders”/terminologies for researchers. But, for now, ICD9 like folders will cover the immediate needs we will see. I think because i2b2 allows you to search by text, browser folders and see counts, it helps the end user compensate for precise mapping deficiencies to some degree. 2. Now for cancer registries, ICDO-2/3 is used and Dan built a reasonable ontology for NAACCR files. 3. For microbiology results, Cerner nicely maps organisms to SNOMED which we’d like to replicate as well for our Epic data. That’s one place I’ve seen SNOMED used in an i2b2 install. See under CMH Microbiology. 4. For meds, it’s RXnorm. Depending on your EMR, you can either crosswalk off the vendor product (FDB/Multim) based on something like a gcnseq_no, off a representative NDC for the formulary file, or do NLP with something like MedEX. Not sure SNOMED would ever play a role there. 5. Labs are LOINC. 6. For clinical observables, look at some of the GPC clinical measurements folder on babel. Jim and the Nebraska team have mapped those to LOINC. Lots of other great things like cardiovascular results that will be great to see us map to LOINC as well for interoperability. Lori Phillips is probably best equipped to comment on i2b2 philosophy and Jeff Klann on where SCILHS is headed. Russ On Nov 26, 2014, at 2:01 PM, Borromeo, Charles <[email protected]<mailto:[email protected]>> wrote: Hi Russ, We are trying to extend our i2b2 ontology to cover more of the medications and diagnoses (our initial ontology only covered the elements required by our rare disease). I have not seen any evidence of the GPC or the Harvard CDRN importing the SNOMED ontology into i2b2. Is there a technical reason for this? I have not tried importing SNOMED yet and I was curious if there were some issues I needed to understand. Thanks, Chuck Borromeo PaTH Network Russ Waitman, PhD Director of Medical Informatics Assistant Vice Chancellor for Enterprise Analytics Associate Professor, Department of Internal Medicine University of Kansas Medical Center, Kansas City, Kansas 913-945-7087 (office) [email protected]<mailto:[email protected]> http://www.kumc.edu/ea-mi/ http://informatics.kumc.edu<http://informatics.kumc.edu/> http://informatics.gpcnetwork.org – a PCORNet collaborative The information in this e-mail may be privileged and confidential, intended only for the use of the addressee(s) above. Any unauthorized use or disclosure of this information is prohibited. If you have received this e-mail by mistake, please delete it and immediately contact the sender. The information in this e-mail is intended only for the person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Partners Compliance HelpLine at http://www.partners.org/complianceline . If the e-mail was sent to you in error but does not contain patient information, please contact the sender and properly dispose of the e-mail.
_______________________________________________ Gpc-dev mailing list [email protected] http://listserv.kumc.edu/mailman/listinfo/gpc-dev
