150 UX iterations with key users optimized mobile app design
150 UX iterations with key users optimized mobile app design
In this podcast series we ask experienced Appreneurs for one success story and one fail. In this episode of the Mobile Engagement Podcast:
Roger Killen is CEO of VitalPac, an App that is saving lives in hospitals. I’m always interested to interview developers of B2B apps because I truly believe that its the second and more valuable wave in mobile App evolution.
Vitalpac is an example of this value: not just purely on a monetary level but value to the community. The interview had a few audio challenges (Roger was on a cell) but I think the content is extremely interesting and uncovers a lot about the healthcare system and how old systems are ripe for disruptive improvement.
We score all out interview anecdotes based on 3 axes of: acquisition, UX and retention – I’ve listed the actionable points below in TL;DR but the main takeaway was Roger’s focus on embedding themselves with nurses and doctors is primarily UX with a focus on embedding the tool into processes thus driving retention – so Roger’s learnings scores are 0,10,5.
Let us know on our Twitter account how you score it.
- The key message here is how mobile App design is spending time with your core userbase and learning the roadblocks in the flow, so:
- Stay close to your key users.
- By iterating you learn the key workflow moments.
- and…by doing this you will learn how to subtract UX steps.
- Mobile app design in the context of medical processes can really help catch where the “ball could be dropped” – stressful environments benefit from tools that remind users to gather data.
- “So actually high value information shared in the right place in workflow is – you get quick feedback on”
- The aggregation of data results in more objective measures of quality care and safety.
- Capturing the data accelerates early detection of situations such as infection before it spreads to other patients.
- Extremely valuable solutions exist in solving real problems and real issues. Its an easy argument to make that solutions for real problems (like hospitals) are more net positive than “Clash of Clans” (go ahead – shoot me!)
DAVID: Hi, this is David from Street Hawk. I’m home this evening. I’ve poured myself a red wine and I’m talking to a man who’s probably just finished his bacon and eggs, I’m not quite sure, he’s in the UK. We’re going to be talking about an application that’s being used in the medical industry. Tell us about your business.
ROGER: Yeah. Well, there were some big problems that I became aware of in hospital care. I used to do at my first founding company, we did the league tables in hospitals, and it really shocked me that the number suggested that if everybody was as good as the best hospital in England that something like 40,000 fewer deaths would happen per year in English hospitals. You know, these are massive numbers. Two and a half thousand people die on the roads a year and perhaps 40,000 not necessarily in hospital. Why was that? Because clearly the doctors and nurses were doing their very best to look after these patients.
So I’d come to the conclusion there’s something about the system that supported the caregivers, and I wanted to explore that more deeply. So I started working in clinical teams with doctors and nurses who’s at the front line of care to see just what was it, what information did they need, what interpretation of information was needed at point of care to promote the best possible care outcomes and progress for their patients?
It quickly comes clear when you do that, I mean, that it’s doctors and nurses that adds to value in healthcare. Generally doctors plan care and nurses deliver care. The doctors will decide what prescription you have, but the nurses will be give you drugs every few hours as they provide the treatment. So if you really want to see what’s happening to the patient, you want to be in that gap between the nurse and the patient, not just deliver healthcare.
So I ended up thinking, “Okay, that’s a really interesting point of healthcare, and I’d like to know more about that. But if I want to know more about that, I have to capture electronically everything the nurse does to the patient. Now if we’re going to get the nurse—and there are a lot of nurses, hundreds of thousands—if we’re going to get the nurses to capture this electronically, you ought to remember that it’s paper that they use in those in the past on the medical record, then we’re going to have to be really convenient, we’re going to have to be very thoughtful about their workflow, so that hopefully in implementing we can make the administration associated with their job less than it is when they’re using traditional means. If we could do that, and noticeably reduce the burden of admin, then I think they’d give their data willingly electronically, and if we capture data electronically, we can work out how valuable it is after that.
DAVID: So how does this work in the context of, you know, if I walk into a hospital? Typically what I see is a clipboard on the end of the bed and the nurses will come in while I’m visiting a patient and they’ll look at the clipboard and the doctor will look at that too. You’re saying that you kind of inserted yourself at the clipboard level or you’re inserting yourself as a tool that the nurse carries around with them.
ROGER: Yeah. We replaced the clipboard in hospitals that implement our system. The product is called VitalPAC. “PAC” stands for Personal Assistant for Clinician. The idea is we give every nurse—in practice these days, it’s pretty much most of the time it’s an iPod touch, occasionally an iPad mini—every nurse gets their personal device and that has a list of their patients in their ward areas, and they can document care with those patients as they need to.
The chart at the foot of the bed used to retain the blood pressure, the heart rate, the temperature of the patient, and it’s that collection of data that effectively it’s just monitoring “how is the patient?” And “how is the patient” is an important concept in healthcare. If they are not well, there are certain times when a specialist should come and see them, and depending on just how unwell and also the urgency of how quickly they should be seen.
DAVID: Right, okay. So basically they’re carrying the iPod, iPod touch, or iPad mini around with them, doctor comes in for his rounds. How does he actually look at “Bill Blogs” patient’s chart?
ROGER: The doctor would either be carrying iPod or iPhone or an iPad mini or an iPad. Our application changes how it presents, depending on the size of the tablet that they’re using. So it displays differently on an iPad than it does on an iPhone, although the code underneath is the same.
ROGER: So when they come in to see the patient, they can single-click on the patient’s name and we show them the chart on their iPad. Generally, if at all, they would have been looking at it whilst walking towards the patient. They’ll not have to pick the chart up from the foot of the bed.
DAVID: Yeah, they’re already pretty informed.
ROGER: So the information is with them all the time, yeah. So really, one of the keys to improvements here was making sure that information was shared in as timely a fashion to the key [piece] of added value in healthcare wherever they were.
DAVID: This is interesting because, you know, my father was in hospital quite a lot in the last six months and I saw the process where information would fall between the cracks because, you know, you would tell somebody something and then that would just kind of not be handed off to somebody else. I’m not saying that’s a bad institution; I’m just saying there’s kind of like human error type thing there.
ROGER: It’s exactly that kind of building reliability in so that you minimize the chance of that happening.
ROGER: A good thing about electronic data generally is it’s got memory. It’s retained and it can be shared, and it can be shared multiply with different care providers involved in the care of the patient. That’s hard to do through socializing data or through database system. So what we do find is that the right people involved in the care of the patient do get to know the key information a long time in advance than how the old previous “Chinese Whispers” paper based system [laughter] and they got the information through to them. Actually we published about six months ago in one of the British medical journals and, you know, a peer review journal here in the UK.
But what was the impact after we rolled the app out? In our first year in hospitals, which were both major hospitals, Portsmouth and Coventry, it showed that in the 12 months following rollout in Portsmouth measured at the order of 390 fewer deaths a year, and in Coventry at the order of 370 fewer deaths a year. So the actual number of patients dying in the hospital fell was very marked. I mean that was about 15% of the fewer patients who died per year.
Another key finding is just by looking at patterns in the data collected at point of care by the nurses and, again, we just submitted to one of the Lancet Journals. In the last four years, Portsmouth hospitals, which would have about 30 outbreaks of infections a year, where people wouldn’t notice the patient had infection and then they bring patients who start to catch the infection. And the staff actually, they’ve reduced it from about 30 a year to 1 a year, just by using information in a timely fashion and sharing it with the infection control specialist so that when a patient does have signs and symptoms indicating infection, they are quickly identified and isolated so that they don’t transmit to other patients.
The system forces the nurses to ask questions every time they observe the patient, so a collection of questions that we would force them to ask every time they observe the patient: Do you feel nauseous? Have you vomited recently? Have you had a stool [maintenance]? Actually we have this little what’s called Bristol Stool Chart, which is a picture of different stools.
DAVID: Oh, no! [Laughs]
ROGER: When the nurses show it to the patient actually, the patient can identify what the movement was.
DAVID: Okay. That should change my App Store rating. [Laughter]
ROGER: It’s kind of funny but that has made such a difference to reduction of infection. If the approach is as successful as it is Portsmouth and replicate the drop, then there’d be 20,000 fewer patients catching an infection in hospital a year. You know, just a simple way of identifying but doing it routinely and reliably is really different.
ROGER: The way they used to do it, if they have any thought that the patient might have an infection, they would have to find a stool sample and they usually have to wait until the next stool movement, so there’d be a time lag. Then when they’ve got a sample and they send it to a lab, then these tests of the sample typically take more than 24 hours, so then you respond after they got a positive test result, which might be one to two days later.
In the meantime, the neighboring patients have already caught it (infection) in the hospital wards start to get close, and so there’s change from a controlled environment to a real prevention approach. It’s been usually impactful. In fact, an interesting associated statistic is the number of infection control specialists in that hospital have almost halved, because they spend so much time, if you like, clearing the mess of the outbreak, and now that they prevent outbreak, it requires far fewer people. So high-quality care is efficient. It’s cheaper to the taxpayer as well as that of the patient.
DAVID: So in the context of this podcast where we ask about what were kind of game-changing or “aha” moments in regards to the adoption, I mean I can hear there in the usage of this thing, by enforcing a particular protocol of asking particular questions and covering things, that sounds like a great thing in terms of making the system work. But, you know, ultimately, do the nurses resent having to enter the information there every time they see the patient? Is that a negative impact?
ROGER: I wondered if that would be the case when we started, and actually the response…they are a very operational group and the response that I get from nurses, generally speaking, is “I like it. I like that it reminds me to do everything, ‘coz I know I’m supposed to, but without it, it was very easy to forget something.” Because there may be 10 or12 pieces of data that they have to gather each time they see a patient, and if they just try and do that through memory, when you’re busy and you’ve got lots of different patients—“this one needs to have their diabetic condition monitored, and this one needs to have their breathing rate monitored because they have a respiratory problem—when all of that is reminded to you in the context of the personalisation of the individual patient, the nurses know they’ve done their job right.
ROGER: There have been instances over time where the system has shown that perhaps the nurse didn’t do their job right and there’d been a problem. That can cause a bit of a pushback, but if well handled, we found that that’s often been used as leverage for a really improved practice. The fact that if a Big Brother system and it monitors – it’s very double-edged Big Brother aspect. If you actually do all the jobs you’re asked to do, you will look good in every scrutiny and the system demonstrates that you’ve done what you’re asked to do, and the payoff for the nurse, if you like, is that we can show them the pictures. But when you do what you’re supposed to do, the outcomes for the patients are much better.
DAVID: Yeah. I think probably the type of people that are attracted to nursing are the people who are, you know, their very nature is actually the willingness to do good, and so therefore, it’s not like they’re going to be trying to slack off in the first place anyway.
ROGER: Yeah, it’s interesting actually in working in this environment, the difference in the sort of philosophical position of the nurse versus the doctor. You know, the doctor is used to planning treatment, so forcing direction on doctors, from an etiquette approach view, is much more difficult. Doctors react in a different way to that kind of workflow management. Nurses like it, generally speaking, and doctors can be irritated by it, so you got to be quite careful in your app development that you tailor to the etiquette of the profession.
DAVID: You know, I’ve had a little bit of experience with technology and doctors in general. They kind of fall into two classes: You’ve got the guys who are very intelligent and therefore being a bit geeky is part of their nature. Then there’s the other doctors who actually think that technology’s a waste of time or a nuisance and so on. So you got those; at least those are the personalities I noticed. Has there been a situation where [on the whole] the doctors have embraced it as well too?
ROGER: Yeah. I mean, there’s a spectrum. You know, we’ve been going a few years now and in that period of time, technology has become so much more – I mean, even though [die-in-the-wood] doctors, you know, who’s not got an iPhone in their pocket or a smartphone in the pocket, are very few now. So adoption of technology, sort of generally in a social use, is making people much more open to embracing, I have found, in a professional use.
DAVID: Do you find that those doctors will take up the system a lot more because they can actually use it on their own device, the device they’re carrying in their pocket anyway?
ROGER: There are not many hospitals that have adopted what’s called the “bring your own device”, because there are also a lot of issues around it, so generally they’ve taken the view that we will supply the tool for the job.
DAVID: Right, okay.
ROGER: So it’s a device with which they’re familiar but not necessarily their own. There are the odd ones that are doing some investigative work with us, and whether or not that will take off and grow, I actually think the professional device, the tool for work, is a helpful thing.
DAVID: Right. So what’s an example of a big “aha” moment you had in terms of – I mean, obviously you’re not out there marketing to masses of users, so user acquisition is probably more a very high-touch type process—I can’t imagine the sale processes you’ve been through [this run] to get through the NHS or the hospital system or whatever—but in terms of where you’ve been with the app and so on, what’s really worked for you in terms of you’ve been struggling with something but then suddenly the game changes?
ROGER: Sometimes when you understand the problem, when you get to understand the problem really thoroughly, it can often lead to [gentrifications] in designing. When we’ve done that, when we’ve just made something simpler and actually it’s really “cut to the nub” of what the problem was, there’d definitely be moments there when I’ve looked and I thought, “Oh, that just makes it so much quicker for that nurse documenting.” It is, and then we may measure that. I remember in the early days of adoption, when we published in peer review papers that it was 40% quicker for an untrained nurse to use the handheld than it was for that same nurse who’s used to using the paper in their ward, to document on the paper on their ward, and that’s because with being able effectively by using electronics cleverly, to take a lot of jobs off them and do those jobs best by using a computer. That was a bit of a breakthrough.
DAVID: Right, yeah. In fact, a previous interview I did with a company called Handshake.com, the CEO refers to a negative speed limit, which is that the way you understand whether your product is really working is that if it falls below the speed of efficiency of doing it on paper, you know you’re in trouble.
ROGER: I agree with that. And that’s, you know, simple, when you get and ask the nurse what are the things that they say that makes the job easier, and making the job easier for a [busy] person is like a welcome embrace.
DAVID: You’re saying that you had an “aha” moment where you went from something that was very procedural and you got down to the way that this really wanted to communicate it.
ROGER: In the early days, I would spend two to three days on the ward with my tech team [behind] and we would do a new release almost every day. I think over the first 20 months, we did 150 software releases. It was very intuitive and the nurses used to say, “Your system doesn’t work very well, but every time you come they get a bit better, so we’re gonna stick with it.” We sort of moulded it gradually into how they worked. I can think of a few moments that I thought, “Oh, that’s a breakthrough.”
ROGER: Some is a step forward and a step back. You know, I think we went through phases when I’d be asked by some of the senior doctors, “Is this approach getting in the way of the nurses’ thinking about their patient or they’d too much concentration on the screen and not enough concentration on the patient?” We had to put things into the application to try and address some features which I felt were appropriate raised concerns and how do we use information presentation to try and make sure that there’s no such thinking about the patient’s condition.
It’s interesting when you look at the data analysis in this area. The mathematical models are good but they are far from perfect. So most patients will end up [trickling], what we call triggering. In other words, the system will say they need to see a doctor urgently, in a period before they have a serious clinical incident or die. But still 25% won’t but they were clearly very sick because of what subsequently happened to them, and they will show other kinds of signs and symptoms like anxiety or unusual behaviour or pallidness or clamminess or other softer signs and symptoms. And the nurses need to be awake, in other words, in observation to more than just the data.
So the relationship, this personal assistant for clinician, you know, the title of the product, was always important to me because it says, “This is your PA but you’re the boss.” This is going to do all it can to help you deliver your job, but do not forget that you’re clinically responsible for this patient. You have more trained capability than this little device ever is going to have, but what it is really good at is organizing your appointments, bringing all the right data and making sure you don’t miss stuff.
DAVID: So does that mean that the design included a way for somebody to kind of break out of the process and jump somewhere else to where they really needed to be?
ROGER: Sometimes it is. Sometimes it’s about putting out a trigger in the data or a data capture. An example would be measuring the pulse rate. When nurses measure pulse these days, they take the oximeter, the medical device, up to the patient and they’ll click something on your finger and it will come up on the screen that the pulse is, you know, 95. But the nurses used to count that by feeling the pulse. When you feel the pulse, you discover more than just the count. You can feel clamminess in the wrist. Sometimes you can feel threadiness in the flow of the blood, and you can measure, let’s say, the regularity of the pulse, which won’t be indicated on the medical device.
So there’s more information to be gotten by feeling the patient, so it’s generally a good thing to feel the patient. It puts you in a closer contact, facial contact, to discover just how they really are in themselves. So we put some features into the data cap saying, “mandate, you have to tell us whether it’s an irregular or regular pulse.” They can’t do that from the machine. The machine will tell you what the pulse rate is, but you’re going to have to feel the pulse to get the regular or irregular. So we’re actually putting workflow in to drive you closer to the patient.
ROGER: And then with some other stuff, by putting things like the charting. Trends are important. It’s not just [now] but how quickly are they changing over time. When we first did it, we didn’t run trends in the workflow but we decided that we would from some of the feedbacks from the nurses who were really thinking about the patient. When I put it in, you know, you’re quite nervous about doing that because it puts that trend screen, mandated into every time you’re going to document, [I’m] going to insist that you look at the trend. But actually the nurses just responded well to it and said, “That’s helpful, you know.” So actually high value information shared in the right place in workflow is you get quick feedback on.
DAVID: Right. That’s really good.
ROGER: Nurses are good at telling you what’s right and what’s not right. [Laughter] When we make a mistake, they’re equally good at telling you, “That doesn’t work, why are you making me do that?”
DAVID: I can absolutely see that happening. [Laughs]
DAVID: So out of that process, is there one big fail that you experienced that really stands out to you as well?
ROGER: I think the thing that was the hardest and most challenging is that as we grew, we worked in many different hospitals, and each hospital would have personal nuances, “Oh, we just like to ask that question this way or that way,” “We use this language instead of that language.” You look at these, what we might call customisations, and think, “Oh, gosh, I can’t believe that makes any difference in the quality outcome for this patient.” But somebody has invented that wording inside that hospital and has a deep ownership for it.
Well, I think in the early part of our program, we were perhaps more accommodating than we should have been to customising to different, let’s say, nuance in language, in the way that documentation was managed across different sites. I think we were too flexible, and actually the weight in flexibility was to build evidence base, that one approach actually you’re able to demonstrate was better than another approach. So we actually used two different risk models in all hospitals, but now we’ve converted them, almost all, to a single risk model. In fact, our risk model have been taken on by the Royal College of Physicians and recommended as a national risk model across the UK for all hospitals.
ROGER: Because there’s an evidence base to it, it just works a bit better to collect the big data base. So we were able to get them all back on to the same risk model. Those approaches, when you’re for the scaling of the business, those customisation nuances, they start to trip you up because when you try and redeliver your software and update to new functionality, you’ve got to take account of all these twists and turns on each side. Suddenly you haven’t got one product; you’ve got 20 products. So we’ve done a lot of work in the last couple of years in reversing out of some of that into more standard, what we call, standard builds, so that we’re able to keep everybody marching forward. But I think in terms of what we did wrong was being over-flexible and not perhaps questioning the value of some of those nuances. We came to learn the cost later really.
DAVID: Yes, certainly. I’ve experienced that myself. You know, we had a company where we basically had 40 different custom build, so you want to do a new release to the core product and to roll that out across the 40 custom builds and maintain all those differences is a killer sort of thing. There was a famous quote that said, “How did God manage to create the earth in seven days? Well, because he didn’t have to worry about backward compatibility.” [Laughter]
ROGER: That has been effort in the last two years. I think perhaps by the end of this year, all of our clients will be on the same code build and that is speeding the whole company up again.
DAVID: Yeah. And it gives you that uniformity of data that you can actually use in your risk models and things like that, which is far more hugely valuable type capability, the first time it’s probably ever been measured in a consistent way.
So that’s great. Thank you very much for all of that. That’s really appreciated. Can you tell me a little bit about where to next for the Learning Clinic in VitalPAC?
ROGER: Well, we definitely – I mean, we’re building some additional products, so we’re just launching one called VitalFlow, which actually captures the data from the clinicians at point of care but uses it to manage the flow of the patients to the hospital, and we have some good early results in shortening the stay—the amount of time the patient has to stay in hospital—in the first hospital we’ve done it by 20%, which again means they can get far more patients through and they can do it with fewer doctors and nurses. And generally speaking, patients don’t want to stay in hospital longer than they need to. So it’s good all round: there’s lower cost of care and better outcome, so some sort of associated operational products, if you like, to our clinical products, VitalPac is very clinical.
Then also, we have just started really to look internationally. We’re in conversations in America, in Australasia, in South Africa, the Middle East, and Ireland. So, over the next three or four years, we’d like to be more international than we are now. We’re very domestically based at the moment, but I think the product has capability and stretch to be relevant. Anyway, you know, patients’ physiologies are the same in Australia as they are in the UK.
DAVID: Yes, absolutely. So the question I’ve got there on that is when I did speak to Handshake and Happy Inspector, two other companies that do B2B type applications or very business-focused type things, they’re still in the App Store. Are you in the App Store?
ROGER: No. We sell through the enterprise. At the moment, we get our clients to sign up to the advance price agreement, so they manage their own sort of professional app store, if you like, related to their [medical] organisation.
DAVID: Yes. It’s just interesting. Those two companies I mentioned, the App Store still ends up being kind of like a distribution/evaluation system for those people so you can get it in the hands of one person who then starts to evangelize it inside their own organisation.
ROGER: Yeah. We’ve talked about this today. I can see some advantages to that, but that has always been very “professionally led”. It’s always, you know, it is very B2B. It is bought by the hospital trust on behalf of their clinicians. We quite like the idea of getting maybe a demo app so that people can say, “Look, if we had this, this would be great for me, and get some to the doctors to give a ground flow, and the nurses to give a ground flow to the management of the hospital to go and procure.” It’s always been a bit of a tradeoff about showing too much intellectual property too much but too early.
DAVID: Yes. I can understand in your situation. All right, very good. So, Roger, thank you very much for your time. I appreciate getting you at the start of the day. It’s always hard to sort of do these things early, so I appreciate that.
ROGER: No. It’s a pleasure – it’s a beautiful morning in London. I see the sun shining on the Thames.
DAVID: My goodness, that’s a shock!
ROGER: Yeah, in February!
DAVID: All right, thank you so much for your time, Roger.
ROGER: It’s a pleasure. Bye.
DAVID: Take care.