Disrupting Improving Access to Psychological Therapies: I just want to not feel shit

IAPT is top down and exists because a government said so. How might digital make IAPT something that suits people?

Beep boop! A giant robot menaces the national IAPT leads (A still from Devil Girl from Mars show a giant boxy robot, a devil girl from Mars and some scared Scots on a moor)

The following is the text of a speech delivered by Mark Brown at Disrupting IAPT: how do we digitise the pathways to therapies? conference at De Vere West One, London on 10th December 2018

I’ve been asked to speak to you this morning about how The Digital Mental Health consumer might strike back and what that might mean for the disruption of current models of delivering the national Improving Access to Psychological Therapy programme in England, something the New York Times called in wonderment in 2017 “England’s Mental Health Experiment: No-Cost Talk Therapy” .

The first thing to bear in mind is that IAPT is a disruption in itself. It was launched specifically to change the way that people with mental health difficulties such as depression and anxiety accessed the NICE guideline recommended low intensity psychological therapies.

The two hallmarks of IAPT, derived from its initial impetus as a health solution argued for in economic terms, are evidence based psychological therapies and routine outcome monitoring. IAPT is a machine for delivering evidence based psychological therapies at scale and for collecting the data to know how and if they are working and for whom. Also integral to IAPT is regular and outcomes focused supervision, making it almost the franchise model of psychological therapies, a kind of Londis or Nisa or Spar of Budgens but for the mind. Fidelity to the model but with local variations for taste.

Digital disruption

Digital disruption, on the other hand is something else instead. Digital disruption is, broadly speaking, what happens when a digital entrant into a market changes the conditions of that market to such an extent that the operation of all other actors in that market must change. Whether you support your local bookshop or make sure you always buy your sexy undies from your local Ann Summers, the availability of a far wider range of good, often at lower prices, made possible by internet shopping has changed the conditions under which such high street shops operate. Digital disruption is experienced by existing services as a reduction in their revenues and a reduction in their customers.

Digital disruption is consumer-led. Consumers move to digital services for reasons like price, convenience, portability, speed, flexibility or because digital services can offer a very different user experience or offer a wider range of goods or services overall or a wider range of good or services within a niche. An online bookshop can offer millions of titles where your local bookshop might offer lots of great ones but not the specific one you’re seeking. An online shop selling sexy undies can be preferable because you don’t want to run into your old headteacher while surveying a stand of dildos and strap-ons. Email can be preferable to letters because you can get a reply in minutes, not in six days, no matter how friendly your postie is.

The point is, real digital disruption happens when the bulk of people move to new ways of doing a thing. It’s not that everyone moves to exactly the same online service, it’s that people move to a variety of online or digital services that impact upon the bricks and mortar services overall.

So, I want you to hold in your minds the idea that digital disruption is always consumer-led. If you have no customers, you have no business. The world is littered with the baffling husks of failed online shops and failed online services. Not everything you can offer online or digitally will actually be something that people want online and digitally. A digital disruption happens when enough customers move to online because online suits them better. Many people feel some things are better done in person. The successful digital service is one that serves the needs of people better than its offline competitor so that they choose to use it.

IAPT and disruption

So, you may say, ‘We thought you were going to tell us about how we could put the digital mental health user in the driving seat and then how we could make even more gains for IAPT.’ I am going to do that, but I thought it was important to hammer home a few points first.

What you’ve heard should tell you a few things about IAPT and disruption.

The first is that IAPT exists because a government decided it should. While many people might agree that not getting help and support around depression and anxiety is awful; not all of them would immediately say that IAPT was the answer they would choose. Indeed, IAPT was a policy idea that was put into action; a kind of massive experiment. The objective was to get people to use IAPT; the objective wasn’t to modify IAPT until people did.

The second is that IAPT delivers a limited range of goods. If bookshops sell books; IAPT sells IAPT. IAPT can’t become Fortnite or Minecraft and it can’t open a new homewear department online. It exists to deliver evidence based psychological therapies to a particular range of people. At its core, IAPT cannot be something else and still remain IAPT.

The third thing to note is that IAPT provides services to a huge amount of people each year, but only roughly half of them are considered to be ‘moving toward recovery’. According to the Nuffield Trust: “Successfully finishing a course of treatment is only one outcome for those referred to IAPT. Other outcomes include: declining or dropping out of treatment, being referred to another service, not being suitable for IAPT treatment or dying before they reach an outcome. In Q1 2018/18, 147,229 people finished a course of IAPT treatment, which is slightly less than it was two years ago. This observation is worrying, since the number of people starting treatment has gone up but those finishing treatment has remained the same. It could indicate an overstretched system in which people start a course of IAPT treatment but are unable to schedule regular appointments due to a lack of capacity.”

The fourth and final thing is that, to date, the driver for digital products and services within IAPT has been from an NHS perspective of saving resources.

IAPT didn’t come from what people wanted, it came from others ideas of what their need was and how to answer it. It can only deliver a limited amount of approved things. Lots of people drop out and don’t complete but we aren’t really sure why and we’re pinning our hopes on reaching far more people with digital as that seems to be the only thing we can afford or the only thing we’re not already trying.

What that last couple of sentences are is not a pessimistic overview of the possibilities of using digital to really change people’s experience of IAPT. What it actually should be taken as is a very good statement of our initial problem to which digital could be applied.

It’s not true that there are grounds for pessimism. Digital delivery of psychologically informed apps and services is booming. A visit to the itunes store of the Play store for Android will show you thousands of depression and anxiety apps. Headspace, an app that helps users to carry out mindfulness based meditation and breathing, passed 1 million users worldwide in June 2018. Of course, overall, these apps are not bound by the kinds of rigorous reporting that IAPT is, so it’s hard to judge how many active users they have and many will be absolutely tit-awful and some will be actively dangerous. Ones that eventually want to be provided as part of healthcare or insurance based systems will, however, try to collect the kinds of data that would allow them to be provided within evidence-based services.

What matters, though, is that people search for them and download them because they are looking for something to help them with the kinds of things that would also mean they reached the threshold for IAPT. People are looking for digital answers to their problems and, at present, IAPT is not supplying them in the places they are looking.

Design thinking

Starting from problems is absolutely where thinking about digital solutions should begin. The things that I’m going to talk about next are all in one way or another derived from Design Thinking. Design thinking is about understanding people. Digital developments in mental health, as with many digital developments in public services are often dead in the water before they begin because they do not begin with the user. Developing digital products is a very particular form of engineering that included the requirement for a project to understand some stuff that is, by outsiders to the process a bit kind of vague and touchy feely.

In my experience, NHS service designers and academics often find it hard to properly get their head around what design thinking is doing, because it sometimes looks like research, sometimes looks like art, sometimes looks like farting about and sometimes looks like setting off half-cocked. Designing thinking deals with what you might call ‘provisional knowledge’ — what we know so far — and builds in stages and cycles the development process to refine that ‘what we know so far’ until it is certain enough to move to the next stage of the process. It’s prime mover is the question ‘have we understood the problem?’ and its prime duty is to the users of an app or service. If you want to understand the various stages and tools that can be deployed in Design Thinking I’d advise going and checking out The Government Digital Service as a starting point.

Design thinking also asks ‘what do people want to happen and how would they know when it has?’ The objective is understand what you’re trying to do from the perspective of all of the people who want something to happen and then to build something first to test if you’ve understood correctly. Design thinking is about answering the problem in the best way, for the people that matter, within the budget available and so it works in the context it will be found.

Ideally you want to have the kinds of multidisciplinary design teams within the NHS, but at present they mostly don’t exist except for Leeds’ excellent mhabitat who can work with NHS organisations on building digital products. Design thinking is how you put the user in the driving seat. And you do that by arranging around their needs teams that can understand, test, build and research.

It would be really cool, maybe, to invent a rotating fridge or a drone powered teapot, if your problem was ‘can i make a spinning fridge?’ or ‘do I believe a brew can fly?’ but they’re solutions in search of a problem. No gets up in the morning and wishes their fridge was rotating or that they had to bash their morning cuppa down from the ceiling with a broom.

A drone powered teapot (apologies to GONG)

Potential users don’t want IAPT, they want to not feel shit

In contrast to flying teapots, spinning fridges and other fripperies; in relation to what IAPT is intended to do, millions of people get up in the morning and think ‘why do I feel so shit and why can’t I feel better?’ So, the use case for IAPT is not ‘how do we extend IAPT?’ but ‘how do we make digital things that work with IAPT that answer that early morning feeling of hopelessness or anxiety?’

Your potential digital IAPT users don’t know they want digital IAPT. To be honest, a lot of the time we don’t want IAPT at all but are so used to settling for anything rather than nothing we’ll take any crumb of possible help we can find. What we know is we don’t want to feel awful anymore. That’s why we ask for referrals or refer ourselves. We all have different preferences, different experiences, different expectations and different desires. To do anything successful to disrupt IAPT digitally will involved understanding as many of those different things as possible then thinking through what that means for a digital intervention.

This will become even more pressing when IAPT is extended to providing interventions for people with other long term health conditions. Each of those conditions make life different. If you have ME and anxiety your life is very different from someone with chronic backpain and depression. If you are visually impaired and feel so low you worry you might lose your job you are very different from someone with Parkinson’s and a terror of being unable to make your own cup of coffee in two years time. If you add this to the variables of age, digital enthusiasm, experience of poverty, sexuality, gender, where you grew up or the injustices and prejudice you might face in day-to-day life; how a digital intervention looks and feels and whether it feel like its for someone like you is of massive importance.

To answer that question from the perspective of users, or more importantly, perspective of prospective users, the only answer is to spend a lot more time them.

The five or six stages of design

There are lots of ways of setting out the stages of design thinking approaches, but all roughly follow the pattern. It’s not a magic spell or anything. What’s important is that the stages do what needs to be done to make something that’s really, really close to what someone using it will actually use because it solves their problem.

Initiation

The first stage is the Initiation phase. This is the time for kicking of the the process of finding stuff out. Finding stuff out can involved desk research, talking to people, visiting places, looking at what’s already out there. You can sometimes call this the Discovery Phase, the phase of working out what the problems actually are that you might like to try to solve and which might get you closer to the world you and your users want to see. Discovery phases can last for ages or can be really quick; depending on how much time you spend building the insights that point you toward a good problem to solve. There is one methodology that calls this stage the ‘Empathise’ stage, as in the stage where you really feel what’s going on for people. In digital IAPT terms, you would at the very least need to be speaking to past IAPT users, current IAPT users, people who didn’t complete IAPT, the kinds of people you would hope would use IAPT, the staff, managers and commissioners of IAPT, tech developers who have tried to sell IAPT -ish solutions to the NHS, academics who understand implementation and anyone else who seems interesting.

Definition

The second stage is the Definition phase, where you really dive into what you’ve found out so far and try to work out what it means and what kinds of problems it suggests would be good to answer. It includes also trying to balance competing problems. A big one is the problem that IAPT needs to collect data and the reality that most people find it boring and arduous to provide it. I’ve worked on a project to turn various measurement scales into an app and the reality of it is that everyone finds questionnaires awkward in different ways, staff included. But during this stage you keep developing your hunch of what looks like a good problem to solves and then checking whether you’ve understood it correctly. Once you’re certain of what problems what you develop has to solve you move to the next stage…

Design

Which is the design phase, where you do everything you can to come up with something that will answer the problems you’ve identified, testing and checking and prototyping as you go along until you are sure that what you actually end up building has the best chance possible of meeting people’s needs. One methodology calls this the ‘ideate’ stage. It about sorting out exactly what the thing to be engineered is and how it show look, feel, function and fit in to people’s lives.

A digital CBT app intended to be used alone will be eff all use to a target audience with limited dexterity if the buttons are too small. An therapeutic app that intends to support people with anxiety might not be very useful if it send two confusing push updates an hour saying ‘OPEN THE APP NOW OR ELSE’.

This stage is a great stage for codesign, building a team of people you can ask for help who really understand what it’s like to be the kind of people who will use the app. If the app includes elements for professionals, you need to ask for their codesign help on the bits that they will use. *whisper* don’t always ask for their help in all of the bits that will face the end user because they get a bit excited and think they’re app developers and talk over your real consumer users and say ‘it has to be calming colours’.

By the end of the design phase your team should know exactly what they need to build to test whether something has a chance of working for people.

Development

In the fourth phase, Development, something is built. And then shown and tested with people. Then rebuilt. Then shown and tested. Then tweaked, then shown and tested. Either until it is perfect or, more likely, until it does something well enough and safely enough to actual let people try it and use it. Budget should always be left to keep this process of revising and redeveloping going until the thing is the best it can be for the people who use it.

Implementation

In the real word, the fifth phase, sometimes called testing or Implementation is when the thing that is built is implemented in the real world. This is one of the stages where digital development and the hangover of medical trials sort of snarl each other up, when they shouldn’t really. Both are focused on understanding what happens because of an intervention, but medical trials are often focused on the outcomes and digital implementation is about understanding in much more detail how the thing is working in practice and how it might be changed and made to work better. One is about capturing a snapshot of an app or intervention at a particular time, the other about capturing a sense of how people are using throughout the time it is being tested. Too often, this stage is used to either bin a project or to decide to use it at scale without looking at what it being live in the field can tell developers about further development and tweaks.

Follow up

Not all design thinking includes a sixth phase, but I think it’s pretty vital for IAPT related applications or services. This would be the Follow-up phase. Getting an application to launch is may be the end of the work for an outside development team, but it’s the beginning of the story for the people and services in which the product is deployed. It’s like building the thing is giving it the best possible start in life; but the important thing is what happens as it grows and meets new people and new challenges. Each new site or new service or new use case into which the app is deployed should be documented, understood, used to inform future iteration or potentially the creation of something new. Each new event adds to the ‘provisional knowledge’ created by the process of it existing and being used.

The amazing thing about digital is that if you get the core of something correct, it’s possible to tweak that something into all manner of different products for slightly different use cases. People often misunderstand this element of digital. It’s often sold as you can roll the same thing about to huge numbers of people with little additional cost — in fact seldom true as something being available isn’t the same as people using it — when the real advantage is you can roll out something more specifically tailored to large numbers of people if you know what it is that they specifically need.

Instead of trying to make one implementation of an intervention you know works that is intended to work for everyone; it’s better to make twelve different implementations of the effective element in slightly different forms that better suit specific groups of people. As yet though, we aren’t clear in research terms how to isolate the effective element of interventions, tending to treat them as if they were medications where the effects cannot be separated out. The wonder of digital is that you can test specific elements at different times, often with the same trial participants. Research and evaluation hasn’t quite caught up with that, although any research with the name of David Mohr on it probably has.

Here’s my problems and why they need solving

I mentioned earlier that I was slightly pessimistic about the chances of IAPT delivering really successful user-facing apps and interventions. The reason for that are informed both by experience and by the research of others.

The pessimism lies in four main areas:

Teams
Implementation
Interfaces
Funds

My pessimism about teams is that I don’t think think many NHS organisations or commissioners have as much will as they might do to create the kinds of multidisciplinary teams required for a design thinking approach to digital within IAPT. I don’t trust that the funds will be made available for the kind of lengthy discovery phases required to really bottom out problems and really understand them.

The NHS is bad at commissioning teams and processes, and there’s a huge danger that something is commissioned that looks on paper like a design thinking approach but ends up in practice becoming a specification/commission approach missing out the need to keep a consistent and clear vision of the user need at the heart of every single meeting and every single decision.

I think some of that could be solved by both creating project manager roles but also product manager roles. A product manager is the person who keeps the vision and purpose alive and is always the champion of the user in the process. A product manager can say ‘no, this isn’t what users said would meet their needs and preferences’. A product manager says ‘no, we can’t skip this stage’ or ‘we need to know more about this before we can go forward’. Without them, projects focus on being delivered without thinking what they’re delivering. As I said, no one wants a drone powered teapot. A product manager should guarantee that the voice of future users is the loudest in any conversation. It’s not about listening or consulting, it’s about making and testing and conversing and checking.

Last week the Lancet Psychiatry published “Factors affecting implementation of digital health interventions for people with psychosis or bipolar disorder, and their family and friends: a systematic review.Reviewing it for The Mental Elf, i found that it described many of the problems that are common across the implementation of digital mental health interventions. The paper found there were very few papers about implementation in general, and that few really described the actual intervention they were implementing. It also found “The majority of factors for effective implementation of digital interventions were centred at the level of the individual or the intervention. Digital health intervention users were more likely to complete an intervention if facilitated by staff or peer support, and if the intervention had been proposed by a staff member who found it useful.

In the studies the review did identify, there was no consistent definition of what engagement with a digital intervention meant and no data in the studies reported on the representativeness of the study samples, meaning it is difficult to generalise across studies. The authors acknowledge that the diversity of the interventions themselves varied, as did the setting of their implementation, again suggesting that it is difficult to draw meaningful comparisons.”

Overall we are not terribly good at understanding, from a research perspective what it is we’re implementing and what factors really influence. I’m not sure we’re going to get better tomorrow; but I think IAPT, with its focus on data collection, evidence and a degree of standardisation might actually make an amazing test bed if anyone were able to actually get into that bed and test it.

I think there are significant challenges for the NHS in engaging with commercial organisations beyond those I addressed above. Many developers would love to develop a product that could eventually be rolled out at scale within IAPT services. Often the NHS is terrible at setting out what the costs and constraints of doing that would be and at stating what their red lines are in terms of interfaces with existing data collecting; in terms of passing evaluation and on standards of evidence and sharing the costs of trials. Many commissioners trying to commission digital products don’t know their arse from their elbow and don’t have a team who can help them. Other commissioners are incredibly reticent to work with outside developers. The result of this is a relationship that doesn’t have a strong set of guidance on ethics on either side and which makes developers just go straight to consumer, with all of the associated dangers of that. Often developers don’t have enough knowledge of the problems that people really have and launch products that just wouldn’t work for people who need them most.

The other extreme is the NHS ending up in very difficult and challenging waters; such as the continuing fallout from the deal with Google Deepmind or in the case of Babylon Health and GP at Hand the weird spectacle of a current secretary of state for health and social care appearing in an advertorial for a commercial product in an un-bylined interview in a newspaper.

I think something like IAPT, with its kind of franchise model could iron out a lot of these challenges; making funds available for partnerships to work upon innovations in digital IAPT that could potentially be rolled out nationally, making sure that user need was placed at the heart of everything that is done by stipulating some version of a design thinking approach.

If I were an entrepreneur what I would be focusing on is finding ways of meeting with people who didn’t complete their IAPT course or intervention and developing a product that suited them and worked for them; approaching the NHS and saying ‘we developed something that has reached the parts of people that you didn’t.’ But even then, I’d be looking to markets overseas or going direct to the consumer because I wouldn’t gamble on the NHS even answering my calls or finding a means to commission me. I know any investors I’d to support that long process of exploration and development wouldn’t want me to put all of my eggs in that basket, no matter how much my politics and ethics wanted me to.

I think overall, if we are looking to disrupt IAPT through digital means, the only route to that is knowing what people actually want to happen. We need to know what brought them to IAPT and what they hoped to have by the time they left. Efficacy and being ‘on the road to recovery’ happens in people’s lives, not in your IAPT services.

The objective is not to solve the problem of digital IAPT. The objective is to come up with a way that keeps solving the problems through digital means of the kinds of people who might access IAPT. It’s not about getting one idea right. It’s about moving to a properly resourced, properly integrated, properly respected, process of keeping getting things more right than wrong and then doing better the next time.

You’d disrupt IAPT via digital, and make things better for people, by making things that are better for them that what is currently there to choose from.

When people don’t find what they need, they go elsewhere or give up.

IAPT can be an amazing machine if it works for you, but if it’s the wrong machine, or the right thing delivered in the wrong way, it’s just something that gets their hopes up and then lets them down. Digital might be a way to let less people down. You provide IAPT, but what people are looking for is feeling less shit.

@markoneinfour