Designing a train ticket website? If the usability is poor, I might buy the wrong ticket. Designing an online booking system? Bad usability could lead to a lot of confusion and unnecessary phone calls. Designing a medical device? Bad usability could kill someone.
Last night’s UX Oxford talk by researcher Chris Vincent was about the specific challenges of usability in the design of medical devices. Dr Vincent’s work for CHI+MED links up industry, academia and clinical settings. He talked about the need for better communication between and within those three areas; without communication, he explained, you can end up in a situation where everybody believes they have no power to improve anything. To complicate the picture, you also have to think about the regulatory frameworks for the parts of the world where the devices are intended to be used.
CHI+MED’s work focuses on infusion pumps, which are very widely used in healthcare to deliver fluids, medication and so on into the patient’s body. I’m interested in them because they’re used to deliver chemotherapy drugs, and my work for EONS means I’m always interested in improvements to the chemo experience.
Dr Vincent explained that when you’re designing this kind of equipment, you don’t know who’s going to be using it and what level of training they have. You don’t even know if the person using it is going to have any medical training at all – there is an increasing trend for medical equipment to be used at home by patients or their carers. But you do know that in some cases, a mistake when setting up the dosage can make the patient very ill, or even kill them. That’s why there’s extensive research into seemingly trivial issues like where the decimal point should go on a numerical keypad.
User personas do play a part in CHI+MED’s work, but Dr Vincent says they use them differently from the likes of Microsoft: “The more detail you add to a persona, the less likely you are to find someone actually fitting that persona.” While some organisations use coffee mugs or posters to make user personas feel like real people, at CHI+MED they are just tools, with descriptions kept quite broad and subject to review.
One slide made me smile with recognition: the one about how users behave around alarms. What do you do when your neighbour’s car alarm goes off for the third time in a week? Assume there’s a crime in progress and spring into action? Thought not. Well, it’s exactly the same thing when you’re a nurse working on a ward full of infusion devices that seem to be going off all the time: false reports of occlusion, unexplained beeping and so on.
One nurse reported her perception that the alarms were always beeping. Dr Vincent didn’t say so explicitly, but when alarms on a device create this perception in the user, they are training the user not to react to alarms. They also have the effect of stressing out the patient – we’re all trained to think that when you’re in hospital with a device attached to you, sudden beeping means something has gone wrong. Another unexpected use: some chemotherapy nurses were deliberately setting the alarms to go off early so they would have time to prepare a new batch of drugs before the current one ran out. It’s an understandable user hack.
It’s an interesting time for the designers of medical devices, because the boundaries between home use and hospital use are becoming blurred – let alone the boundaries between what counts as an app and what counts as a device. Unfortunately the FDA has a very long and rigorous validation process for medical software, not just when it’s being rolled out but also when any changes are made. This means that seemingly trivial updates or bug-fixes can take a long time to get approved, with safety consequences that are presumably the opposite of what was intended.
In response to a question from Pete Aylward, Dr Vincent agreed that the move towards touchscreens is good because it’s easier to update how a device is used without solid-state buttons. That should have positive consequences for the longevity of such devices. He emphasised that when you design a device you don’t know where it will end up. You might design it for hospitals, only to find it being sold on eBay when it gets a bit dated and ending up in a nursing home. You might design it for home use, and find that people store it out of sight because they don’t want a visual reminder of their ongoing need for medical treatment.
You can’t predict what users will do, but you can learn a lot from what they’re already doing. That’s why this kind of research is so important.