Since Daniel Kahneman’s magnificent book, Thinking Fast and Slow, made us think differently about ourselves—as Darwin and Freud had done before—we have become familiar with the ideas of flawed thinking, behavioural economics, and nudges to make us behave better or worse, but I hadn’t until this week understood how these ideas can improve patient care and possibly the lives of doctors. At a seminar organised by the Institute of Global Health Innovation, Emily Larson and Alix Brazier from The Behavioural Insights Team described what they can and might do.
The team, which started in the Cabinet Office, works to “generate and apply behavioural insights to inform policy, improve public services, and deliver positive results for people and communities.”
Larson started the meeting by illustrating our flawed thinking. She put up a list of about 10 related words like “bed, rest, tired, wake” and asked us to remember as many as we could. Along with 40-50% of any group I “remembered” the word “sleep” even though it didn’t appear in the list. As most of the words related in some way to sleep I thought that it must be there.
Kahneman and his colleague Adam Tversky described how we have two ways of thinking: the fast way that we use most of the time and is shot through with biases; and the slow, analytic way that requires much more effort—and still has biases.
The “availability heuristic” means that we resort to thoughts or explanations that are most available: a doctor might diagnose a rare condition in one patient and then think it possible in the next patient simply because it comes to his or her mind. With the “representative heuristic” we might when told about a surgeon assume that the surgeon is male because most are. With “present bias” a patient will put the immediate reward of eating cake now before the long-term benefit of not developing diabetes. “Friction costs” mean that even small things that we have to overcome in order to do what we are reluctant to do may stop us doing it. (I thought of going to a meeting, being told that I wasn’t on the invitation list and that the receptionist would have to contact the organisers, and taking that as enough to miss the meeting and do something more amusing.)
The Behavioural Insights Team has developed the EAST (Easy, Attractive, Social, Timely) framework as a tool to help use behavioural insights to make improvements. Making things easy includes removing friction costs, simplifying messages, and moving—as for example, with organ donation—from “opting in to opting out.” Increasing attractiveness might mean personalising messages (Dear Richard rather than Dear Sir) or adding rewards and incentives; adding to NHS appointment letters the message that missing an appointment cost the NHS £180 reduced non-attendances by 25%. Increasing timeliness means encouraging people to act now not in the future and to have a specific plan if acting now is not possible.
Increasing the social means recognising that people are social creatures and inclined to do what others do. Brazier described an experiment in which actors in a lift all faced the back of the lift rather than the door, which is what people normally do; people who were not actors also tended to face the back when they got into the lift and saw everybody facing the back. Having an influential messenger, perhaps the prime minister, makes people more likely to follow a message as does reciprocation: people are more likely to donate organs when reminded that they or their families might one day need organs replaced.
Unlike traditional economics, which depends mostly on analysis and modelling of data, behavioural economics prefers randomised trials, and Brazier described two trials of interventions derived from the ideas of behavioural economics.
One trial (published in BMJ Open) conducted in Imperial College Healthcare NHS Trust attempted to improve the prescribing of antibiotics by simplifying and redesigning the chart used for prescribing. Importantly the researchers worked with the doctors to study how they used the existing forms and then with them improve the design of the form through several iterations. Twenty-nine junior doctors were then randomised to use the two forms in a simulation of two patients: 15 used the old form and 14 the new. The new form resulted in considerable improvements: the signature was legible in 96% of the new and none of the old forms; correct information on the duration of treatment increased from 52% to 9%, on indication from 59% to 100%, and on the number to bleep if there were problems from 31% to 86%.
Another trial published in The Lancet tried to improve the prescribing of antibiotics among the 20% of highest prescribing GPs. Having identified the high prescribers the researchers randomised them (1581 practices) to receive a letter signed by the chief medical officer pointing out that they were in the top 20% of prescribers in their region and recommending three simple interventions—like encouraging self-care—to reduce prescribing. Brazier asked the audience what we thought the effect would be of one letter. My thought echoed that of a man who said it might have a short term effect. In fact it led to a reduction antibiotic prescribing of 3.3% in the intervention arm maintained over six months. The cost of the exercise was around £4000 and the saving £92 000. As somebody in the audience pointed out this was a substantial saving achieved at low cost, whereas an advertising campaign would be expensive and probably ineffective.
The team has now turned its attention to trying to prevent burnout among doctors. It is using it TESTS tool (Target, Explore, Solution, Trial, Scale) and has begun, using data from the General Medical Council, to identify a target, a high-risk group. They have identified trainees in anaesthetics and are working with the Royal College of Anaesthetists. They have explored burnout by interviewing trainees, shadowing them, and doing desk research. They have identified characteristics that may lead to burnout like being reluctant to ask for help, the NHS seeing you as “a bum on a seat,” not being given rotas well in advance, and not receiving gratitude from patients (as they don’t see them when they wake or at follow up) as well as workload, exams, and commuting distance.
Nudges have to be small, cheap, and scalable, so, although they have fed back to NHS England and the college on improvements that could make a considerable difference, the team has opted for a solution that consists of a series of text messages that provide links to stories from grateful patients and advice from the BMA on preventing burnout. They are about to conduct a trial with some 1500 trainees with outcomes being standardised of burnout and wellbeing.
It’s hard to believe that such a small, cheap, easily-scaled intervention will work (and the team accepts that it may not), but that’s the point of nudges: small interventions that can have a major impact not by making a big difference with individuals, although they may, but because they can be easily scaled.
Richard Smith was the editor of The BMJ until 2004.
Conflict of interest: RS is an unpaid professor in the Institute of Global Health Innovation.