- “Niche” Computer Systems
- Meaningful Use
- “Wrong Patient”
- Cognitive Friction
- Dialog-Box Rooms
- What’s in a word?
- Cost Disease
- Model T
- Signal-to-Noise Ratio
- Anti-Data Pixels
- Fitts’s Law
- Bad Apple
Updates, December 2014, October 2016: short addenda at end.
Speaking of “Bad Design Killing” a big part of the discussion at the ACEP Informatics Section meeting in San Francisco this month was about one particular usability problem with CPOE: entering orders on the wrong patient. I’ve done this myself – as far as I know I’ve always caught it before anything bad occurred. Many of my partners have had the same problem, and so have many of those at the Section meeting, with widely-different CPOE systems.
There are a variety of suggestions for why this is so, and what to do about it.
With the system I use, it seems to me that there are two main sources of the error. Orders are usually entered by right-clicking the patient on the spreadsheet-metaphor tracking system, and then entering orders. However, even though you think you’re right-clicking the patient, in the split-second after your brain sends the message to your finger to right-click, a new patient is entered or deleted by someone else above yours; your patient moves up or down a line, and you actually click on the patient above or below what you intended.
For the second cause of wrong-patient order entry, I have to explain the tracking system we use. There is the main spreadsheet window, which is usually maximized onscreen. Then there is a large popup window with other information; a great variety of information populates this popup window, including old records, a list of orders that have been entered, and even the order-entry screen. It’s possible to have one patient selected on the spreadsheet, and a different patient in the popup window. That is the second major cause of wrong-patient entry.
Regardless of the source of error, however, using the Swiss cheese error-prevention analogy (see the post on Color) we need a method to catch such errors (a second piece of Swiss cheese).
There are many suggestions for how to prevent such errors in the informatics literature. One suggestion is that we put a picture of the patient on the screen when one is ordering. This is used, for example, by the CPOE system used by the US Veteran’s Administration. However, for someone who has just met the patient for the first time, this may not be particularly useful. Actually, those in the ED tend to identify patients more by location and chief complaint: “the chest pain in room 5” and perhaps having these two things prominently shown would work better for ED systems.
It’s also true that we can only see a narrow field of “foveal vision” as described in the post on Mental Models, Input Modes and Cognitive Friction. So whatever we put on a screen as an error-prevention label (picture, “chest pain/Rm 5” or whatever) is within that six-degree field of view. Even if it’s within the realm of foveal vision, it may still be ignored, as was the gorilla that strolled through the ball-players in a famous psychology experiment: when we are tightly-focused on a task, we can ignore almost anything. In such experiments, some people even looked right at the gorilla (as seen by eye-movement tracking) but didn’t “see” it as they were so narrowly focused. There was one experiment reported at the Informatics Section meeting where placing the patient name in bold helped prevent the “wrong patient” errors, I would suspect that having the name (or “chest pain/rm5,” or a picture) closer to the task at hand would help.
I look forward to the results of more research into this issue, and will post new information here when I receive it.
Addendum, Christmas Day 2014:
An article in the Annals of Emergency Medicine at the very end of 2014, entitled Intercepting Wrong-Patient Orders in a Computerized Provider Order Entry System found a way to significantly reduce the incidence of wrong-patient orders.There was a mandatory 2.5 second popup to confirm that the order was on the right patient. Good news? Not really. If we do a cost-benefit analysis, we find that we have to waste 1.5 hours of each emergency physician’s time to prevent an error. And we know, as discussed in a previous post, that if a popup lasts for more than a second (and maybe a bit less in a busy ED) than the person sitting in front of the computer will simply switch from the computer to some other task at hand. Sounds like a great opportunity for someone to design a method of preventing wrong-patient errors that is more cost-effective. For example, the data in the above referenced-post would suggest that reducing the time for the mandatory pop-up to about 3/4 of a second would actually make the system work better, by preventing the user’s attention from straying.
Addendum, October, 2016:
I have documentary evidence, in the form of a cellphone-camera screenshot, of one of the two leading large HISs forcing me to enter orders on the wrong patient. Due to HIPPA concerns, I cannot post the actual picture here. What it shows is two windows, one in the background and one in the foreground, which according to the HIS’s design are supposed to be on the same patient. The background window is tagged at the top with the name “Abel Baker” and the foreground window tagged at the top with the name “Charlie Delta.” The clinical information in the foreground window, though, is not for Charlie Delta but for Abel Baker. And if you used the foreground window to enter orders, who would it be on?