In my recent posts about the Limitations of Administrative Data and the Lessons Learned, I've concluded that clinical observations of symptoms and conditions coded in SNOMED-CT are the most relevant problem list data to share with patients and use for data analysis.
We have a 3 step approach to implementing SNOMED-CT in BIDMC clinical systems
1. Our existing problem list dictionary was developed in 1998 and hence it's called BI-98. We contributed it to the National Library of Medicine and it was incorporated into the Metathesaurus.
About 70% of the terms we used are easily mapable to SNOMED-CT Codes. The National Library of Medicine will send us a BI-98 to SNOMED-CT mapping in the next few days and we'll incorporate it into our existing dictionary, giving us a SNOMED-CT vocabulary for 70% of the existing problem list entries in our system.
2. The NLM will be giving us a compendium of the 6000 most commonly used terms in the local problem list vocabularies of large health care institutions, and their equivalent SNOMED CT codes. We'll incorporate that list into our systems and create a novel "Problem List Picker" using AJAX technologies that will assist doctors in choosing the best problem description associated with a SNOMED-CT term. This will give us a great framework for the terminology of newly entered problems.
3. I'm working with other organizations, such as Kaiser Permanente, to gather problem list "best practices". We'll leverage their experience to innovate at BIDMC and I'll share the experience broadly via my blog. I'll post their problem list dictionaries as I receive permission to do so.
I look forward to your participation and feedback as we work together to improve the usefulness of data in EHRs and PHRs nationwide.
Wednesday, April 22, 2009
Enhancing our Problem Lists
Posted by John Halamka at 3:00 AM
Subscribe to: Post Comments (Atom)
John, I agree that the only terminology system we have that makes sense for problem lists, diagnosis, and other clinical findings is SNOMED-CT.
That raises the problem, however, in that very few people (including some medical informaticists and nearly all EHR vendors) understand how to use SNOMED properly.
Even basic uses like negation (i.e. saying the patient doesn't have some disorder) require a level of training/experience which isn't common place.
It is going to be crucial that organization,particularly HL7, provide explicit guidance on how to use SNOMED-CT, and how to bring together all the related metadata (things like dates that a problem existed, who recorded the problem, and so on).
Look for this to be forthcoming from the HL7/ISO detailed clinical models efforts.
I was thinking more about EHR,
I came across an article in the latest issue (4/20/09) of eWeek called 'Mashups give Defense Department strategic edge'. A mash up, as the article defines it, is an application that combines information from multiple sources to create a single new piece of information. The article also says that U.S. military intelligence uses an off-the-shelf product from JackBe software to "bring together
strategic information from disparate sources (p 36)", in a mashup to help make better decisions. Perhaps doctors can use this type of idea too. Electronic Health Records from all hospitals for each patient could be combined to help make better decisions regarding the patient's health. Or this mash up software can be used to create EHRs for patients by combining information from each hospital that has a record for the patient. Another thing, if hospitals adopt that kind of system, military level of information security can be appropriate to keep patients records private.
mabramso [AT] gmail.com
Electronic Health Records from all hospitals for each patient could be combined to help make better decisions regarding the patient's health. Or this mash up software can be used to create EHRs for patients by combining information from each hospital that has a record for the patient.
Recep Deniz MD
Post a Comment