Raw Burgers and Future of Clinical Decision Making: A Ramble

The inspiration for this post was the final episode of "The Next Iron Chef". I was definitely psyched to see Jose Garces win. There was a moment in the show when the other finalist, Chef Mehta, served the judges a dish featuring a pork burger, and they commented that the pork was undercooked or even raw. This got me thinking - here are highly trained professionals, experienced practitioners of the art of cooking, and they are making such basic mistakes? In this day and age of advanced technologies, can't we come up with some gizmo to stick inside that burger patty, take its temperature and figure out - definitively - its state of doneness? Why rely on these rules of thumb (or rather the thenar eminence) - pun intended - by touching the meat to gauge whether it will still moo when I cut into it? Why am I talking about this? Well, think intravenous lines placement by highly trained professionals, experienced practitioners of the art of medicine. How are these lines placed? Essentially, by blindly poking around until you find the right blood vessel! Why not use the fancy technology? There are portable ultrasound thingies that help to visualize the needle and the vessel. Yes, they are far from perfect, but nothing new will be developed unless there is a demand for it, and there will be no demand until we stop thinking that relying on technology somehow makes doctors less cool.

Of course, same goes for information technologies. In his blog entry "Health IT: What’s the Future?" Steve Downs of Robert Wood Johnson Foundation recaps themes from a recent "Discovery and Innovation in Health IT" workshop. One presenter focused "on the need for cognitive support, showing a hockey stick graph of the number of facts that will be relevant to a given clinical decision over time (this theme reappeared several times over the two days). The number is expected to reach 1000 by 2020, while the number of facts that a human can contemplate while making a decision remains stuck at um, five." Healthcare practitioners need clinical decision support (CDS). I am surprised that such a statement could still be considered controversial...

Just like those ultrasound vein visualizing gizmos, CDS technologies are far from perfect. Thus far the CDS efforts are targeted at individual systems. The challenges are to figure out how to get/represent/manage/update/share clinical care guidelines logic that drives CDS recommendations, how to surface these recommendations to clinicians at the right place and time in a workflow-aware fashion so that they do not dismiss them outright, how to make these recommendations "actionable" to facilitate carrying out an order should a clinician decide to follow a recommendation. Necessarily, these efforts are hard - if not impossible - to generalize. These systems tend to be tightly bound to their initial implementation environments and are therefore non-interoperable in any meaningful way.

Ken Mandl and Zak Kohane proposed an idea of a plugin-friendly platform instead of a typical monolithic EHR. For this to work, the underlying clinical data must be handled in a way that abstracts it from individual applications. Alternatively, leave your monolithic applications alone and pool your data into a near-real-time repository that is application-agnostic. Ether way, if data can be separated from applications (how's that for a radical idea?), there is hope for the kind of interoperability that would result in scalable CDS. Automated processes are needed to abstract and represent domain knowledge encoded in clinical guidelines so that it can be machine-processable (consider baby-steps like HQMF); and it will probably take a miracle to figure out how to deliver CDS recommendations to practicing clinicians in such a way that they are useful. But despite these challenges, if the right incentives are in place to ensure a healthy demand for CDS technologies, we will see progress. The end-goal is the realization of the potential of healthcare IT - safe, appropriate, timely, high-quality care.