Human error or Design error?
Recently, a healthcare provider received 5 times the dose of the Covid-19 vaccine in Singapore*. The article cites an investigation that showed it was due to human error resulting from a communication lapse. I’m glad to hear the staff is doing well and at low risk of a significant adverse outcome. Yet it doesn’t sit right with me that the cause is “human error”. It’s almost always “human error” in many instances yet, was so simple as someone who didn’t do their job? Or was there another pernicious factor at play?
In my opinion, “human error” is unfair to the staff who administered the vaccination. Having read the amazing “Design of Everyday Things” by Don Norman, I only agree this is human error insofar as it is human error in designing a multi-dose vaccine vial. Don Norman is an expert in usability engineering, cognitive science, and design. He works in the intersection of these disciplines and can honestly be said to have written the book on advocating good, usable design.
The Design of Everyday Things is a fantastic read not only for UX professionals but for anyone who has any stake in using or creating products. In particular, Norman writes of “user-centered design” part of which is designing for error. Here’s my favourite phrase from this part of the book:
“If the system lets you make the error, it is badly designed. And if the system induces you to make the error, then it is really badly designed.”
Chain of unfortunate events
On to “errors”. In the “Swiss Cheese” model of error causation, there were many “holes’’ of errors that led to this unfortunate circumstance; including design, communication, human action. I’ve done a quick summary in the diagram above. You can see it was a series of errors culminating in the incorrect dose being administered.
Additionally, to make matters worse, there is time pressure because the prepared vaccine needs to be used within 6 hours or it must be discarded*. But, right at the start, you can see the first hole in the cheese. The vial. The vial that allows incorrect doses of the vaccine to be drawn. To ignore this design, is to ignore a root cause in the chain of unfortunate events.
Many vs. One
The multidose vial is a design that increases the risk of wrong dosing. The steps are fairly complicated and require a high degree of precision. Firstly, the multidose vial must be thawed before it is diluted. The dilution process involves mixing the content with 1.8mL(!) of saline solution. Thereafter, the vial will contain 2.25mL and each dose is 0.3mL*.
Forcing someone to draw out multiple doses of an odd number (0.3mL) from each vial is not the best design for safety. Additionally, there is an increased risk of contamination and time pressure to expend the contents once removed from cold storage and diluted. Surely isn’t there a better design?
I think so. In fact, reconstituted vaccines have been mostly single-dose products. Compare this to a single dose system where the vaccine is already premixed and rests in a syringe. Thus, all that is required is to add on a needle (if needed) and administer the dose. In reconstituted vaccine systems which require dilution, the single-dose design requires the entire contents of a provided diluent to be mixed within the vial and the entire contents then administered. As you can imagine, this minimises the risk of drawing the wrong dose or even the use of an incorrect diluent.
There are also other benefits to a single-dose system. There is much less time pressure to cycle through patients and risks of contamination are minimised with a single draw compared to having to puncture the lid.
Hence, to restate my point, there was an error made by a human who drew and administered the incorrect dose. But equally, there was an earlier error in the design of such a system which permitted such an error to be made in the first place.
When I look at digital product design, I have learned to envisage critical errors users may make. From this, measures to prevent such an error would be conceptualised and designed. The goal is primarily to prevent the error from even being a possibility and if that was not possible, means to alert the user to the error and how to recover from it.
An example of error elimination would be the mandatory time out of an account after a set time period to prevent accidental or unauthorised access by another party. There are numerous examples of the latter 2 designs for errors; to ensure all fields in a form are filled, a user may be prompted when trying to progress (or click the next button). That is an example of an alert. If the error message contained instructions on which fields needed to be filled, that would help the user recover from the error.
In healthcare applications, this is particularly important, so that the wrong outcome is prevented. In terms of severity, this could match an incorrect dose of a vaccine. Yes, it may be of similar severity. For example, there are now apps that help patients determine their diet and dosing of insulin. An incorrect user input may trigger incorrect advice. An incorrect dose of insulin may have just as bad effects as with a vaccine, if not worse if the dose is too large and causes extremely low blood sugar levels.
Good design in healthcare
This is a timely reminder of the importance of good design in healthcare, where errors may have adverse outcomes for all end-users, from patients to healthcare professionals.
**For the Covid Vaccine SmPC-PL based on EU documentation dated 21-DEC-2020
I’m loving the use of subheadings, and images. When the piece is done, also consider adding in quotes from the piece to break up the text. But this is already leaps and bounds ahead of the first version.
Please excuse the images. I sketched them out real quick.