I read a summary of a JAMA article that described a research study on the impact of restricting access within an EMR system to one patient chart at a time, the notion being to test whether such restrictions would reduce placing orders on the wrong patient. Interestingly, the restricted group had a slightly higher error rate (90.7 errors/100,000 orders vs. 88/100000). Further, it was discovered that over 66% of the order sessions for unrestricted users involved having only one patient record open at the time.
While that last statistic makes evaluating the impact harder, it does appear that, at least in this population of users, physicians had learned a habit of only being in one chart at a time. In my mind, that says a lot about their discipline.
I remember the “old days”, when physicians wrote orders in physical charts. Then it was pretty easy to keep things straight; to “open a new chart” you had to pull down a binder, or card file or whatever it was that contained the patient record. So the discipline was that you pulled open a chart, reviewed its contents, wrote orders on that chart, flagged it for attention, and put it back in the rack with the other charts. So orders on the wrong patient occurred primarily because you pulled the wrong chart out of the rack. It would be interesting to know if there are any baseline statistics from that era. I do not recall seeing any. My general experience was that these kinds of errors were pretty rare in the old physical world.
Pretty clearly, restriction isn’t the answer. Since the users who could only open one chart at a time still entered orders on the wrong patient as often as those who were unrestricted, my interpretation of the data is that the problem comes in selecting the right chart to begin with.
So, given that restricting behavior doesn’t seem to solve the problem of entering orders on the wrong chart, it appears to me we need more contextual clues that we picked the right patient to begin with. How might we do that?
- Might EMR products permit accessing the patient’s chart directly from a diagnostic report sent to the physician for review?
- Would putting up pictures of patients help?
- Would some kind of AI that looked at the orders and the patient and realized there was a disconnect help?
How would you go about reducing this error rate while not burdening the prescriber with all sorts of prompts and verifications?
Please let me hear from you!
Dennis A Tribble, PharmD, FASHP
Ormond Beach, FL 32174
DATdoc@aol.com
The opinions expressed herein are my own, and are not necessarily the opinions of ASHP or my employer, BD.