Transcript

HAROLD THIMBLEBY: Let's go to the beach. And it's a lovely day, and there are people playing. Just suppose some of them got into difficulties. You'd want to call the coast guard to get the help you need. I've got a mobile phone. We tend to think our mobile phones are wonderful-- technology is wonderful.

Well, what's interesting is I choose this and I like it, but very often we have to use devices we can't choose. Well, let's try and call the coast guard. Maybe you don't have coverage on your mobile phone, so you're going to run to use the coast guard phone.

The coast guard phone-- you can see on the sign that it's saying call 999. 999 is the right number to call to get the coast guard. It says 1 2 3. How are we going to dial 999? That's guaranteed we're going to take longer. I'm confused. I don't know what to do. I can't dial 999. I was told to dial 999. What do I do?

Well, there's a really interesting thing going on here-- that somehow the Coast Guard has put a 999 phone on the beach for everybody to use, and you can't do that. Something's gone wrong with the design. Here it's really obvious that it's gone wrong.

Possibly the coast guard don't notice because, of course, the only people who ring them up have succeeded in using it, and they're not going to say, oh, there's a design issue here. They're going to say, come and rescue the kids who are drowning. So it's an obvious design problem, but nobody knows about it.

In hospitals, everybody makes slips. And obviously what's good about hospitals is everybody's professional and those slips gets sorted out. What worries me is people make slips and the technology isn't part of the team that stops those slips turning into catastrophes.

And the cash machine is a really good example of how the catastrophe of losing your cash card is a problem that's been fixed by redesigning the cash machines so that you can't leave your card behind. Because you don't leave it behind now. You want your money. That is a good illustration of how redesigning infusion pumps and calculators and stuff like that would make hospitals a much safer place.

Denise Melanson was a cancer patient who died from an overdose of fluorouracil. Fluorouracil is a chemotherapy drug, and you want to get the right amount of it, and it's really critical to get the right amount. If you have too little, it doesn't kill the cancer. And if you have too much, it kills the patient.

And unfortunately, a mistake was made, and I want to explore what the mistake was. I'll tell you what the story is. Denise Melanson was what's called an ambulatory patient. She was walking around with this infusion pump and a bag of fluorouracil. So this is really nice. It means you can walk around. You're not stuck tied to a bed.

Her bag of fluorouracil runs out, and she goes to the Alberta Cancer Care Centre to get some more drug. And the nurses go to the pharmacy. They get another bag of fluorouracil, and they have to calculate how much fluorouracil the pump is going to give her. And it should have been 1.2 millilitres per hour. They accidentally forgot to divide by the number of hours in a day, so they ended up with 28.8 mils per hour.

Denise Melanson should have come back four days later with an empty bag. Instead she came back four hours later. And basically, she'd be given an overdose of fluorouracil and that killed her.

What went wrong? There are several things that went wrong. One is the way the calculation was done. It's actually very difficult to do the drug dose calculation. And the best I can do is it takes 22 keystrokes on a calculator. If you make any slip while you're doing that calculation, you just get the wrong result.

If you're using a calculator, the whole point using a calculator is you don't know what the right result is. So if you make a slip and get the wrong result, you don't know. So here, they made a slip. They forgot the 24 hours in a day. The calculator had no idea what they were doing, and it just gave them the wrong answer.

The normal take on this is that the nurses should be trained to do the sum properly. My take on it is why are they using a calculator that doesn't know what they're doing? It's easy enough to set up the device. These infusion pumps can do that sort of sum. And in fact the bag of fluorouracil the nurses were given said 1.2 millilitres per hour on it. There's actually no need for them to do the calculation at all.

But the point is, if they make a slip-- and from time to time, everybody will make a slip-- the calculator doesn't help. This is the infusion pump that Denise Melanson was using. The root cause analysis, the study of why she died included a study of this pump. Five nurses from the ward where she was treated were taken for two hours to study how to go through the same scenario that ended up killing Denise Melanson.

Basically, none of the five nurses they had in that two hour study could use this pump effectively. They had problems with it. One of the obvious problems is the decimal point is the same as the up arrow key, and that confuses nurses.

The conclusion that the root cause analysis had is that the nurses need to be trained to use the pump. That seems like an obvious comment to make, like, you've got these things-- people ought to be able use them.

I'm interested in design. And the different issue is, why didn't the manufacturers do a two hour study with five nurses and find out that they're selling an infusion pump that people can't use? They could've redesigned it so that it was easier to use. And it only takes five nurses for two hours, and you discover all sorts of problems with it. Those problems should have been fixed before it was sold.

I'm arguing that design can make things safer. There are two sorts of infusion pump, and the question is which design is better? There's one sort like the pump I've just show you which has got a calculator keypad on it. If you want to enter a number like 12.3, you key in 1 2 point 3. But you key in the digit. The other sort is an incremental keypad, whereas if you want to 12.3, you press up and down, and you gradually increase the number to 10, and 12, 12.3, or you decrease it 100 down from 90, 80, 70 to 12.3.

So one sort of infusion has up/down keys to increase or decrease a number. The other sort has got a keypad on it with 10 digits and a decimal point, and you type in the number that you want. Which is better?

We've done some experiments, and it turns out that the up/down keypad is twice as good. And the reason for this is if you're using a numeric keypad, you want to enter 12.3 or whatever number, you just press the keys. Your eyes are looking at the keypad. You don't look at the display because you know what you're doing. You look at the keypad because you need to know where to put your fingers. If you make a slip, you don't notice.

With the up/down keypad, you hardly need to look at the up/down keys because you know where they are. You need to look at the number. And of course, you keep on adjusting the number until it's right. And unsurprisingly, when you think of it like that, you end up with being more accurate and more reliable if you use the up/down keypad.

So that's a simple example where an experiment tells you one sort of design is twice as good as the other sort of design in terms of error rate. The point is, if the technology is re-designed to detect errors, then everybody could do a much better job. At the moment, technology ignores errors and it does anything. We've done some peer reviewed papers that showed that if you detect errors in this process, you can halve the death rate.

I've given you a few examples of how bad design leads to problems from the coast guard phone on the beach to infusion pumps, calculators, up/down keypads. You can redesign things to make them safer and better support the tasks that people are doing.

In health care, you've got IT systems and gadgets that are letting nurses and patients down. They're killing people and nurses are getting sacked rather than solving the design problem. I'm on a mission to get this message out-- that everybody needs to be aware that the technology that we're using could be a lot better.

[SIRENS]