Wednesday, May 13, 2009

The BIDMC Problem List Vocabulary

In several blog entries, I've been discussing the adoption of SNOMED-CT as the vocabulary to support BIDMC problem lists and our Google Health interface.

The National Library of medicine has mapped 93% of BIDMC problem list terms to SNOMED. The International Health Terminology Standards Development Organization (IHTSDO) has given me permission to share the BIDMC problem list vocabulary and the SNOMED-CT codes on my blog via the following terms

"The SNOMED CT® identifiers are posted with permission from the International Health Terminology Standards Development Organisation (IHTSDO). SNOMED CT is available for use under the terms of the IHTSDO Affiliate License Agreement, which is also Appendix 2 of the License Agreement for Use of the UMLS Metathesaurus. Use of SNOMED CT is free in IHTSDO Member countries (12 countries as of May 2009, including the US), in an additional 49 countries characterized as low-income economies by the World Bank, and for qualifying research projects worldwide. The National Library of Medicine (NLM) is the US Member of the IHTSDO. For more information, those in the US should contact NLM. Those in other countries should consult the IHTSDO for appropriate contact information."

Our plan is to load these terms into our electronic health record and code all our matching historical problem list data into SNOMED-CT. Our Google and Microsoft interfaces will be changed to send SNOMED-CT so that all problem list data can be mapped to disease monographs and consumer decision support tools.

In June, the NLM will forward us their consolidated list of the 6000 SNOMED-CT coded problems from the leading early adopters of SNOMED in the US. If we find any gaps in this problem list, we'll work with NLM and IHTSDO to fill them.

I hope the BIDMC problem list is helpful to your applications and institutions!

14 comments:

Claudio Luís Vera (@modulist) said...

Even if SNOMED-CT is the product of a non-profit standards body, it's proprietary intellectual property.

It should send up red flags to everyone that you need to secure permission to reprint the condition codes. It's even more troublesome that there's a future potential of having to pay royalties for public standards.

That's so wrong on so many levels. It reminds me of the court battle in the 90s over who owned basketball game scores. The NBA prevailed, and every news organization ended up having to pay royalties to the NBA. Tweet the score of a game, and technically you're violating copyright.

To me, using proprietary IP as a public standard is a pact with the devil, even if the devil is an international NGO. It may seem like a marriage of convenience at first, but a conflict of interest is inevitable.

IHTSDO has so, so much to learn from the open source community. An open standard, should be well, open. IHTSDO should be a consortium like the W3C with a standard that is free to all -- not a GATT-like circle of most favored countries.

e-Older American said...

I emphatically agree with "modulist"!

GreenLeaves said...

It's good to hear that for a change the US is working with an international body on preparing definitions in the healthcare domain.

What kind of time frame exists, for resolving problems e.g those from BIDMC within the structure of IHTSDO?

Alan said...

I couldn't agree more. People really need to wake up and smell the coffee.

It seems that payers, providers, and the government could rally around a REAL open approach. They all have so much to gain.

Instead it is setup to keep the mainstay health vendors in place and allow them to charge more than is necessary. This also causes "vendor lock". Government, Payers, Providers, and citizens beware! As @modulist, said we are being sold down the proprietary river.

This means high prices for health providers seeking to procure for health IT systems. If you've been asking why is health IT is so expensive, here is exactly the reason why. It costs a lot but it doesn't need to be quite so expensive. Demand OPEN formats for the sake of the US Health care system and our economy.

Read my related Article:

Should Health Care Standards be Open Source?

http://bit.ly/2mqpkK

-Alan

AL said...

SNOMED costs many millions of dollars to develop and maintain. The editors are mostly physician and nurses who are paid for their time and expertise. In the past, inexpert modelers caused major corruption problems in SNOMED. This is a full-time job for a team of medical professionals, not a fly-by-night operation. Medical care cannot tolerate GIGO done by inexperienced part timers. The completely unpaid model is untenable, unless some country or organization steps up to the plate and funds it. Making it open source would ensure that nobody would pay. No money means no SNOMED.

e-Patient Dave said...

Fwiw, here's the e-patients.net post about various data vocabularies. (I don't know enough to express an opinion on any of them.)

Also fwiw, people I know who feel strongly about clinically useful vocabularies are pleased with John's (or whoever's) choice of SNOMED. As the post above says, CPT is only available by license from the AMA.

For other considerations on usability and fitness for purpose please see that post.

David said...

John, thanks for the post and, as always, for sharing your experiences (free of charge!). BTW, since HITSP selected the VA/KP subset (apprx 17K terms) of SNOMED-CT for coding of problems, how does the BIDMC subset (approx 800 terms) relate to that? Is BIDMC a fully contained subset within VA/KP? Mm=ight the existence of multiple subsets of different sizes cause some confusion?

Robin Shorrock said...

I agree with most of the comments here, although lean towards the @modulist point of view.

As @ePatientDave implies, we have to use what's available, given their relative fitnesses for purpose.

I do believe however that there should be concerns about standards that are controlled by commercial organisations with licensing restrictions that may not be in the public interest (e.g. restricting use). Funding for development is an important issue but doesn't have to come from the users of the standards.

Pressure should be applied to ensure critical health standards are not controlled by those who may exploit our dependency on them to our overall detriment, so transparency around the various licenses is very important at this stage. Also key is the discussion around implications of each of the license offered.

Unknown said...

John, while most of your problem list is findings and disorders with two situations (history of), there are about 20 procedures (e.g. appendectomy) and one event (motor vehicle accident).

Do you actually use those procedures and MVA as problems, or as reasons for presenting?

gemstest said...

I wonder which issues with problem lists coding is intended to resolve?

John Halamka said...

The BIDMC problem list vocabulary contains the most common free text terms entered by our clinicians over the past 15 years.

Rather than using a locally invented vocabulary, we'd like to use a more standard vocabulary that is a subset of SNOMED-CT. Using such a vocabulary will improve our interoperability with other EHRs, PHRs, and Quality measurement entities.

Per the NLM, one such subset, the VA problem list vocabulary, contains about 17,000 SNOMED concepts. The BIDMC list has 821. Between the two lists, 683 (83%) concepts are shared.

In June, the NLM will send me their "best practices" list of 6000-7000 terms gleaned from several institutions including Regenstrief, the VA, and Kaiser Permanente.

krowwolf said...

DavidH has a very good point - just because a SNOMED concept has a description that sounds like what you want doesn't necessarily mean you can use it. The specific hierarchy it comes from is significant and that needs to be taken into account when used in a record or you end up with data that uses SNOMED codes but has no valid interpretation (after all, the whole point is not the storing of information but being able to query/retrieve it).

Betsy Humphreys said...

As the US Member of the General Assembly of the IHTSDO, I have been following this thread with interest. As AL said, the ongoing development, maintenance, and distribution of SNOMED CT costs millions of dollars each year. The IHTSDO’s business model is based on governments contributing to the maintenance of a global public good, with an end goal of making it available worldwide for all uses and users. Annual Membership fees paid by the National Library of Medicine (NLM) on behalf of the US and by other current IHTSDO Members (on behalf of Australia, Canada, Cyprus, Denmark, Lithuania, New Zealand, Netherlands, Singapore, Spain, Sweden, United Kingdom) support the ongoing development of SNOMED CT. This financial support enables free use, which is obviously key to promoting adoption and meaningful use, in Member countries and also in low-income countries, as noted in John’s blog entry.

The current IHTSDO funding levels are estimated to be about two-thirds of what would really put the ongoing maintenance of SNOMED CT (with its new technical platform that supports distributed input) on a firm footing to support safe and effective electronic health records worldwide. The use fees that now apply in non-IHTSDO countries with adequate resources were established as an inducement for more of these countries to join the IHTSDO. Once that occurs it should be possible to make a well-maintained and highly functional SNOMED CT freely available everywhere.

When that Great Day arrives, open license terms will still be needed to deter the creation of divergent versions that are the enemy of interoperability. In the meantime, the license terms that govern free use in the US and other Member countries provide a guarantee against the sudden appearance of future royalty payments.

Claudio Luís Vera (@modulist) said...

I think it's great that Betsy from IHTSDO would weigh in. By no means do I want to be critical of SNOMED-CT as a standard. Instead, what I want to point out is that public industry standards should be open source or public domain -- by definition. Otherwise, they're a monopoly in disguise.

Unfortunately, the current arrangement between NLM and IHTSDO provides neither the advantages of licensed software (profits to IHTSDO) nor the advantages of open source software.

In the open source world, IHTSDO would have an army of self-organizing contributors, maintainers, and reviewers contributing to SNOMED-CT code and vetting it for accuracy. Provided that the reviewers are doing their job, SNOMED-CT would be a quality product. Developers would be free to embed SNOMED-CT codes in the software for their EHRs and PHRs -- as long as they provided proper attribution and licensing details.

IHTSDO on the other hand is maintained by country-level agreements (read: subsidies) to fund maintenance and updates of its code base by a paid staff of medical experts. That's a business model that's far less sustainable than open source.

By not opening up SNOMED-CT to a public license IHTSDO is creating a barrier to innovation, particularly to the small practices and rural hospitals that would most benefit from healthcare reform. Things could be far worse, too, as IHTSDO could be as overpriced as the AMA's CPT codes.

There are many very successful precedents for public standards in the open source community such as LInux, Apache, and the Internet itself. it sounds like IHTSDO is taking its first baby steps in this direction with looking at distributing the responsibility of maintaining SNOMED-CT.