I am a fan of the ISMP acute care newsletter, and was pleased to see them tackle what they are referring to as their "Targeted Best Practices" list.
Number 1 on that last involved entering orders on the wrong patient.
While the precautions and solutions all made sense, the discussion focused my thinking on a subject that came up a couple of years ago at the Informatics Institute at the ASHP Summer meeting where the speakers talked about making certain that an automated replacement for a manual process address all the needs of all the stakeholders in that process. If you were there, you may recall that the example was an Emergency Department white board that was replaced by a display, where the replacement focused entirely on replicating specific patient details, and failed to include the semaphores that drove other processes (like cleaning and restocking of the cubicles between patients, and ensuring that the attending physicians had checked the patient out before discharge). The result as reported by the presenters was that many of the processes that ensured a consistent and high-quality experience in the ED came to a grinding halt.
What brought me back to that discussion was thinking about why, in my 12 years of practice, I cannot recall ever seeing orders written on the wrong patient. I will be the first to admit my memory might be selective, but, if it happened, it was certainly pretty rare.
As I thought about the process I observed during those years, a lot of the prevention of this kind of error was physical. It wasn't that physicians didn't interrupt one ordering session to handle something urgent, but the process involved physically obtaining, opening, managing, and then closing the chart and moving it to a location where nursing knew it had new orders. That physical process had become engrained in the muscle memory of the medical staff. It physically maintained a certain level of context within their minds. They could get multiple patients deep in the process, but, by closing out and removing each chart from the pile of charts, they could keep their focus on the patient at hand, close and move each chart as it was completed, and then refocus on the next patient at hand.
Without a doubt, that process had its problems:
- If a physician put the chart back into the main chart rack, nobody knew there were orders to be managed there.
- The orders were still often incomplete and fundamentally illegible
- There was no decision support available during the process
- There was a definite lag time between the completion of those orders and the time that they were completely implemented
- There was work involved for both nursing and pharmacy in ensuring that medication document was up to date
Many of these problems have been addressed in electronic medical records. But in the process, we seem to have lost those physical clues that we need to re-orient our thinking to the next patient. This condition has a name: e-iatrogenesis.
How did we miss this? I, of course, have an opinion. This is simply my opinion, so take it for what it may be worth.
When we invent something new and wonderful, we tend to focus on the problems with the old system that the new system is intended to solve at the expense of what is (was) good and right about the old system.
In the days where unit-dose drug distribution was the "new thing", we argued extensively over the patient-safety aspects of the packaging while ignoring the workflow and sociological impacts of removing multiple-days' supply from patient care areas and forcing nursing to become dependent on our ability to service them (an ability that sometimes failed). Not that there weren't problems with the floor stock system; studies showed that it had an atrocious error rate in medication selection and dose calculation. But there were benefits as well; benefits that ultimately sold much of our profession on cartless medication distribution.
One of my old mentors (Don McDonald) used to force our pharmacy residents through the exercise of coming up with five reasons to keep the floor stock system. It was an excellent thought experiment. Not that Don wanted us to abandon unit dose; he just didn't want us to lose track of the impact centralization of drug distribution on the people at the sharp end of the medication management spear: the nurses. The original unit dose systems did solve a bunch of medication accuracy problems, and reduced the amount of discarded meds, but it also increased substantially the amount of time nurses had to spend ensuring that they had the meds they needed to care for their patients. I often wondered if the legendary "nursing/pharmacy conflict" arose from that centralization.
Back to the point: in addressing the bullet list above, which CPOE generally does, somehow we seem to have lost the physical signals that helped ordering caregivers keep the patients straight. It wasn't a problem we saw in the old system, so we didn't address it in the user interface on the new system. Or if we did, the solution seems to have been less than adequate.
The ISMP suggestions to address this are interesting. Pretty clearly the notion of maintaining some kind of screen presentation that clearly and always displays the patient information is one solution.
I have become aware of another that I thinks deserves attention to our thinking, which is to enhance our computerized decision support systems (CDSS) with the notion of patient context. Current systems are rule-based and doggedly apply the rules they have been given. They can therefore be only as good as the rules they have been given. A CDSS that is contextually aware of the current patient should be able to recognize things that may be inappropriate that do not fail drug-drug interaction, drug-dose, or allergy checking, such as ordering birth control pills for a male patient, or ordering oncolytics on a patient with no documented cancer. Indeed, knowing the problems we have with getting reliable allergy information, wouldn't it be good to know that a patient with a documented Codeine allergy had been taking Acetaminophen with Codeine without problems preoperatively?
For those of us in the informatics community, the lesson is, I believe, that we have to thoroughly understand the current process both in terms of what is good and what is bad, and to articulate all the problems that have to be solved (some of which may be well-solved in the old system) before we design and implement our automated solutions. And that was the lesson taught at the Summer Meeting Informatics Institute at its seminal session.
I just thought it was a lesson worth repeating.
What do you think?
Dennis A. Tribble, Pharm. D., FASHP
Ormond Beach, FL
DATdoc@aol.com
The opinions in this blog are my own, and not necessarily those of my employer or of ASHP.