3 ms·
The closed systems and lack of support for interoperability between EHRs upsets me the most because it leads to patient harm. As a patient here's what gets me:
by mindcrime 9y ago
The closed systems and lack of support for interoperability between EHRs upsets me the most because it leads to patient harm.
As a patient here's what gets me: UNC has one EMR system, my GP has another, and Duke has yet another. I've been a patient at all 3 at various times. I also use Strava, Fitbit, and the like. Now you would think that it would be trivially easy to use an API and export and aggregate all of my lab results, for things like blood pressure, cholesterol, etc., so I can graph, say, my blood pressure and my Strava activity together. But nooooo... all of these EMR systems either have no export functionality / API, or if they do, it's something byzantine and just-short-of-impossible-to-use.
Of course not every patient is a data geek who's going to use machine learning, statistical analysis and visualizations on their own data. But for those of us who want to, the roadblocks to doing so are infuriating.
I just had a patient transferred from an outside hospital for abdominal pain and somehow their CT scan was lost in transit. Because it was an emergency we had to CT scan them again which doubled their radiation exposure and their risk for a kidney injury from the iv contrast. It was midnight so it would have taken until business hours to obtain the scans from the hospital.
The doubly sad part is that this is almost 100% a policy / business issue and not a technical one. We've known how to share data for a long time. Heck, IIRC, a scenario much like the one listed above was used as an illustrative example for justifying the WS-Federation[1] protocol way back when.
Edit: Yep, this document[2] explaining WS-Federation actually uses an emergency room scenario to justify the need for WS-Federation. Not the exact same scenario as above, but the point stands. We've had protocols and technologies for doing this stuff for a long time.
[1]: https://en.wikipedia.org/wiki/WS-Federation https://en.wikipedia.org/wiki/WS-Federation
[2]: http://download.boulder.ibm.com/ibmdl/pub/software/dw/specs/ws-fed/WS-FederationSpec05282007.pdf?S_TACT=105AGX04&S_CMP=LP http://download.boulder.ibm.com/ibmdl/pub/software/dw/specs/...
- ch4s3 9y agoWell, they are all required by law to support import/export of a CCDA, and Duke's EHR does. The problem is that it's poorly supported in the workflow, and may require using something like a HIE. Also, Fitbit data is medically useless beyond establishing that you do or don't exercise.
- mindcrime 9y agoI use a Fit bit brand scale, and I would argue that my weight data is useful.
- ch4s3 9y agoThey weigh you at any doctor's office, on a more accurate scale. Weight trends are important, but at a granularity of a handful of measurements per year. Unless a patient some specific chronic illness like diabetes or asthma, most patient recorded data isn't medically useful.
- mindcrime 9y agomost patient recorded data isn't medically useful. If you're talking about diagnosing a particular disease, you're probably right. That's not what I'm talking about. I'm talking about having an overall holistic view of my health and fitness related data, and being able to combine it so I can look for interesting correlations / patterns to suit my own interests. They weigh you at any doctor's office, on a more accurate scale. Weight trends are important, but at a granularity of a handful of measurements per year. The "more accurate scale" part is irrelevant given that your bodyweight can fluctuate 3 or 4 pounds from one day to the next, based on hydration levels, undigested food in your stomach, etc. Trends over time, even if measured on slightly less accurate scales, are really more meaningful. Anyway, I'm not really sure why we're even having this discussion. Are you arguing that EMR systems shouldn't make their data easily available to the end user? Because otherwise, this whole thing about Fitbit data is a red-herring. What I do with the data, what other data I mash it up with, etc. is really not what's at issue here.