It's Starting to Add Up - A Few Observations from AMIA 2010

Nov 18, 2010, 11:16 AM, Posted by Steve Downs

I’ve just finished three days at AMIA’s Annual Symposium – the geekfest gathering of informaticians (or informaticists, if you prefer).  It’s a big conference, with many themes and tracks, so it’s hard to draw general conclusions as any observations are largely functions of which sessions one chooses to attend.  So I’ll try not to generalize (too much) but offer a few thoughts on what I saw and heard.

Meaningful Use policy is a really delicate business.  As provider organizations are starting to translate Meaningful Use requirements into operational plans, the details are getting really tricky.  I saw a panel representing some real EHR pioneers (e.g. Intermountain Health Care, Marshfield Clinic) that showed how even for them, who’ve been using EHRs rather meaningfully for years, they will have to make substantial changes to their systems in order to qualify for the subsidies.  Meaningful Use is a great concept because it uses large scale EHR implementation as a vehicle to shift practice in the direction of better quality (good) and more prevention (good again), but doing so without being over-prescriptive is much harder than it looks.  Finding that balance of rewarding the principles while maintaining a fair degree of flexibility on the implementation details seems to be the key (and, to be fair, easier said than done).

Health IT raises interesting questions about roles.  Dan Masys, in his perennially fascinating review of the past year’s research findings and key developments, pointed out that now study after study about clinical decision support shows the same finding:  that decision alerts always have more impact if they are sent to anyone on the care team but the physician.  He also noted that the teachable moment of an order entry system is at the moment of entry (when, for example, a drug-drug interaction or contraindication alert is triggered) and that physicians don’t learn from these opportunities if they delegate the order entry task, as is common.  Dryly, he observed that studies also show that decision support is most effective when it’s not ignored.  On a serious note, he commented that if the trend of these findings continues, disintermediation of physicians will become inevitable.  Ouch.

Rapid Learning is really starting to happen.  I attended a terrific presentation by Nirav Shah, a researcher based at NYU and Geisinger.  He’s been mining Geisinger’s EHR data, which goes back many years now, and doing just the kind of studies one would hope to see when we can start to tap large clinical datasets.  After a preliminary rundown of the pros and cons of mining EHR data vs. running prospective clinical trials, he showed how he was able to compare the effectiveness (not a sanitized clinical trial of efficacy, but actual effectiveness in terms of real-world experience) of a next-generation brand-name drug against a generic (no difference except in cost).  He also showed that he could predict, in 50 percent of cases, which patients would develop congestive heart failure six to 24 months ahead of time.  Wow.  While it would be great to see the 50 percent number climb, that kind of result opens the door for preventive interventions that could save many lives.

Power to the Patients!  One panel brought together some of the true pioneers of patient engagement.  (Okay, I’m a bit biased as RWJF/Pioneer funds two of them currently) Warner Slack, who’s a great-grandmentor of some of the students in the audience (he trained the students that trained the current generation of faculty), gave a hilarious talk about just how strange his colleagues thought him when he was advocating for greater patient participation in clinical decisions 40 years ago.  Tom Delbanco discussed the OpenNotes project , and Patti Brennan discussed Project HealthDesign and the value of tracking observations of daily living (“ODLs,” also a frequent topic on this blog).  Patti gave a very clear explanation of the value of each: OpenNotes, she said, gives a window into what the clinician is thinking; ODLs give the clinician a window into the day-to-day life that a patient is living.  Both, she said, are necessary to improving care.  Charlie Safran made some important observations about the need to drop often stereotypic assumptions about how groups of patients will behave and e-Patient Dave was both passionate and entertaining as he spoke about the value of engaging patients in reviewing their records (not only will they be better informed, they will inevitably fins mistakes – some serious). Dave’s comments raised for me a key point, which is that clinicians should make sure that patients know what’s in their records – not for any moral or philosophical reasons – but for the simple practical reason that it can improve care and safety. 

Those were a few things I noticed.  You can check out the Twitter stream for many perspectives.   I’d love to hear from others who were there and get their take.  And for those who weren’t, are these observations consistent with what you’re seeing?