Fantasy baseball for healthcare informatics
March 31, 2009 7:39 PM   Subscribe

If you were the IT advisor to the Obama Administration, how would you implement the national medical informatics system that is proposed as part of the stimulus package?

I'm doing some work for a small healthcare policy foundation, and one of the things we think about is how informatics can be used (and potentially misused) in the American medical system. The hive mind is an amazing source of IT know-how, so please pretend you're the IT Guru & Healthcare Czar.

How would you deal with issues such as:

*Patient privacy/access to medical records

*Interoperability, given the various hardware and software in use across the country

*Accuracy and error reduction

*Data archiving/backup - medical records have to be maintained for many years, but physical media degrade and software changes

*Data fraud or tampering

*Open-source vs. commercial software (database, text documents, image acquisition and processing, etc)

*Deployment - can it be staged, and who goes first?

*Is existing technology enough for the job, or do we need some new inventions?

*Other ...?

Go nuts, but please explain in terms that non-IT laymen can understand.

Thank you for your expertise and insights!
posted by Quietgal to Computers & Internet (13 answers total) 9 users marked this as a favorite
My understanding is that the VA has solved a number of these problems and their software is public domain, but I think establishing document format standards is a critical layer.
posted by idb at 7:48 PM on March 31, 2009

Existing technologies/privacy: Google Health is great, and I'll never touch it out of concern that the info will somehow be used to deny me coverage for pre-existing conditions. The same goes for practically any medical records scheme, I have no confidence that it won't end up being used to deny coverage wholesale, at least until the insurance companies can start denying coverage based on genes.
posted by mullingitover at 7:55 PM on March 31, 2009

I'm in bioinformatics, but I stay far away from health informatics, because the system is, as you well know, a colossal clusterfuck. I know data and big systems, but don't have much inside experience in health records - the doctors among us can probably give more insight.

- I'm a big believer in open-source software, but this seems like a venture where open source doesn't matter as much as having open and standardized formats. I don't care who makes the end-user software packages, as long as they all produce data that everyone can read and write.

- Privacy/accessibility: I'd like to see strong encryption play a role. That way, everyone's medical data can live in some cloud that's accessible from everywhere. Authentication could be based on a number of different protocols, possibly including biometrics.

- Storage/Backup: Even if stuff is being stored locally, and there isn't a big central repository for this data (as would be ideal), backup isn't a big problem. Each local space signs up for something like amazon S3 - a big, online storage provider. These places keep something like triplicate backups of all data in different locations around the globe. It's about as safe as you can get these days. Use decent encryption on the data before uploading it, and there no way anyone unintended can read it.

- Is existing technology enough for the job.
Absolutely. It's just a matter of putting money and manpower towards the unsexy problems of data management. Unless all the major hospitals band together to demand better (and they haven't yet, to my knowledge), I don't see this happening without some sort of government mandate.
posted by chrisamiller at 8:12 PM on March 31, 2009 [1 favorite]

I read one real-life doctor's opinion (who has many years' experience with EMR systems) on this very topic a few days ago:

Musings of a Distractible Mind: Doing It Right
posted by ThyroidBob at 8:14 PM on March 31, 2009

Sounds like the NEJM just ran some articles on electronic records. They may be useful.
posted by chrisamiller at 8:25 PM on March 31, 2009

Privacy and security: Practically applicable technologies and existing practices do not meet the requirements at this time. That's an opinion derived from first-hand experience. Healthcare privacy is simply too important an issue for any 3-nines solution.

Interoperability: Top down requirement specifications. Web services using XML namespaces built upon existing data standards (e.g., HL7, ICD9/10, SNOMED, etc.) and transformation services (e.g., UMLS).

QA/QC: A provider issue.

As I see it, the security issue isn't a reason not to do this but it is a reason to start with a reasonably limited approach to what information is included in the system. NYC Electronic Health Records initiative may give you some ideas to think about.
posted by McGuillicuddy at 8:35 PM on March 31, 2009

I love hearing from the programmers/IT folks here.

As a physician, I would ask, when you consider Healthcare Informatics - what do you want from medicine? What would you want from your doctors? The best of what, say, an agency like the DMV might have to offer? Keeping records straight, making sure everyone's vaccinated? OK, that's easy, relatively.

if we're talking about public health, then the challenge is to aggregate good data from individuals for the benefit of the community, however widely that is defined for a particular problem, while maintaining the autonomy and right to privacy of each individual. That too seems easy compared with the challenge that I face in my work as a physician.

Because if we're talking about individual health records, the doctor's chart say - or the sum of a number of doctors' charts, since each of us may be under the care of more than one physician (at a time, or over a lifespan) - I can offer this from the point of view of a practicing primary care physician:

Electronic medical records are a disaster for the physician as artist.

Just as live art, a dance performance, a concert, or even something as static as a painting - cannot be captured electronically without fatal degradation (though one can make art digitally, that I don't dispute) - so the art of medicine defies capture by digital means. The real practice of medicine in the trenches quickly transcends the order imposed by the static quality of every electronic medical record I've seen.

A physician worth their salt starts from the place of the artist, open to experience, allowing data to fall into place for sure, and noticing patterns as they emerge, but more important in the case of disease - leaves themselves open to possibility - for discovery of something new - a new variant of the problem/disease, a new solution/therapy, say. Yet this quality is missing from any software I've encountered. Electronic medical record software, despite what might be promised, abandons the physician at the doorway of what Keats called "negative capability," or what in Buddhist philosophy is referred to so sweetly as "not knowing."

Yet this is the most important point in the work. It is where we as patients want to be with our doctors, in communion with "not knowing." Yes, we want the doctor to know. We want the doctor to heal. But we want the doctor to use all their powers to understand (rather than pigeonhole), and that understanding can happen only in a place that allows for every possibility, even chaos. Because this is the place of discovery. This is the painter's canvas, the chemist's lab bench, the dusty stage floor.

So, you folks who are involved in healthcare informatics, that would be my challenge to you. How do you create and invigorate a system for record-keeping that doesn't destroy what is most precious in what we as physicians do in the face of human frailty? How can computers help us help you as individual patients?

What do you want from your physician? If it's immediacy, how does a computer help us hold your hand? How does a computer help us love?
posted by pammo at 10:05 PM on March 31, 2009

Mostly what Chrisamiller and McGuillicuddy said.

- Is existing technology enough for the job.

Yes. This could have been done in 1994 if there was a will.

- Accuracy and error reduction

I think the biggest benefit will be from eliminating the silo mentality that many doctors have. The vast majority of doctors are in small practices without peer review. Universally accessible records will aid in error reduction simply by making it easy for a second set of eyes to review the same data.

*Deployment - can it be staged, and who goes first?

Sure, any place that is big enough to have a real IT staff should go first. Small medical practices should go last and just get satellite nodes from larger local groups who have worked out the bugs and have support staff.

Pammo: What do you want from your physician? If it's immediacy, how does a computer help us hold your hand? How does a computer help us love?

Personally, I don't care if you hold my hand. Just don't kill me or amputate the wrong leg while I'm in your care. I've gone to medical groups with fairly good electronic health record systems - having the doctor type a summary of his/her diagnosis and recommendations into a database, and then print out a copy makes me a lot more comfortable than a hand on my knee. I think you are under the delusion that intuition is incompatible with good record keeping, much like saying that a real artist shouldn't keep their accounting in order.
posted by benzenedream at 12:48 AM on April 1, 2009 [1 favorite]

Ross Anderson, who's professor of security engineering at Cambridge University has written pretty extensively about security of healthcare informatics. Survey page here. This paper is a good introduction, and sets out a sensible security policy for medical records.
posted by crocomancer at 4:24 AM on April 1, 2009

If you're interested in public health informatics, you might want to look at the public health informatics institute: or this CDC site: (I fail at HTML today)
posted by pointystick at 4:53 AM on April 1, 2009

Consider reading up on the implementation or failure to implement various systems in the UK with its NHS records.

You learn from mistakes and assumptions made. Look at the medical records and systems issues in various healthcare systems and home in on those.
posted by jadepearl at 6:57 AM on April 1, 2009

I do this for a living (albeit in Canada) and to be quite honest, halfhearted attempts at this on a national scale will be 10x worse than no efforts. The absolute myriad of insurances, state, local and national laws would be mind boggling. That being said, a policy based upon patient safety would be paramount i.e addressing the 7,000 deaths a year in US hospitals due to medication errors would be a hell of a start.

All (?) Health Care Information Systems use the same HL-7 messaging standard that allows the interface of different system to view or transfer information between systems.

I could go on but PM me if you need more resources...

posted by Country Dick Montana at 7:34 PM on April 1, 2009 [1 favorite]

Response by poster: Great answers, hive mind! I'd have to mark a whole slew as "best answer", so I'll just offer my thanks and appreciation here.
posted by Quietgal at 9:10 PM on April 2, 2009

« Older How much extra child support does he need to give?   |   So I've slept on it, now what? Newer »
This thread is closed to new comments.