BE360: nudging our white coats

The potential of technology to advance healthcare with AI-assisted healthcare and robotics, telemedicine and personalised medicine is the great hope.^[ 1 ] But there is also strong evidence from the behavioural sciences that tiny changes, with minimal costs, can also have a big impact on improving our quality of care. And we need every tool in the box to help our current fragile healthcare system.

This article – the ninth in our BE 360 series – takes a step back from consumer- or citizen-facing nudges and looks at how people – patients in this case – are getting better outcomes through the application of BE to guide and improve the decision-making of doctors and healthcare workers, ensuring more accurate diagnoses, safer and lower cost treatments and more transparent decision-making.

In an ideal world, doctors and clinicians would have unlimited time, boundless cognitive energy and resources which they could use to collate all relevant information for each patient to make a diagnosis or treatment decision. They would be able to draw on multiple resources such as scans and screening, as well as basic monitoring to assess patients, before finally applying cognitive energy to process the information. With all that precision and care, they’d be reasonably confident that the decision they made would be the right one.

We all know the ‘ideal world’ scenario is unrealistic. In the real world, they are often stretched for time, short staffed, short of beds, short of resources and restricted by budgetary constraints. In addition, they are often tired and stressed by their heavy workload and decisions are made under extreme pressure.

However, there is convincing evidence from the behavioural sciences that diagnoses and treatment decisions made using systematic mental shortcuts and evidence-based rules of thumb can perform just as well and, more often than not, better than more complex processes.

Research by behavioural scientists such as Gerd Gigerenzer have shown that more information does not always lead to the most accurate answer. Rather, application of a rule-of-thumb approach, using simple and transparent processes which ask only a few sequential yes/no questions and rely on a few key pieces of information, can be far more accurate. These processes are known as Fast-and-Frugal decision trees.

Improving diagnoses and treatment decisions with Fast-and-Frugal decision trees

Behavioural economists Julian Marewski and Gerd Gigerenzer tell the tale of Professor Complexicus and Doctor Heuristicus:

• Professor Complexicus is known for his scrutiny – he takes all information about a patient into account, including the most minute details. His philosophy is that all information is potentially relevant, and that considering as much information as possible benefits his decisions.
  
• Doctor Heuristicus, in contrast, relies only on a few pieces of information, perhaps those that she deems to be the most relevant ones.
  

Who do you think is the more effective physician? You might assume that it’s Professor Complexicus due to his thoroughness. Yet you’d be wrong. In test after test, in a wide range of diagnoses and treatment decisions – from diagnosing heart failure, treating children with pneumonia, HIV testing, cancer screening or diagnosing depression – less is more.^[ 2 ]

Take this fast-and-frugal decision tree developed and tested by Lee Green and David Mehr at the University of Michigan Medical School. In a rural hospital in Michigan, doctors were sending 90% of patients complaining of severe chest pain to the coronary care unit (CCU) with a suspected heart attack, preferring to err on the safe side in a climate of litigation. Yet this diagnosis was leading to too many false positives; the actual proportion of patients who had suffered heart attacks was only 25%. 

Not only was this over-diagnosis expensive, but it was also leading to an overcrowded coronary care unit, reducing the quality of care for those who really had a heart attack and putting the patients who did not need to be there at risk of hospital infections. Not to mention instilling unnecessary panic amongst numerous patients and their families.

They designed a simple fast-and-frugal decision-tree where doctors only needed to ask three crucial questions (see diagram). The first question looks for anomalies on the patient’s electrocardiogram. If anomalies are found, patients are sent straight to the CCU. If not, a second question asks if the patient’s primary complaint is chest pain and a final question checks if five other factors are present.

They compared their simple decision-tree with an existing, more complex decision-tool; the Heart Disease Predictive Instrument (HDPI) where doctors need to check for the presence, absence and combination of seven symptoms and match findings to a chart containing around 50 probabilities and then calculate a logistic regression using a calculator to determine if a patient should be admitted…sounds complicated doesn’t it?

No surprise then that many doctors didn’t like using it. But how accurate was it compared to Green and Mehr’s decision-tree?

The simple decision-tree was more accurate – 95% of diagnoses were correct compared to 70-80% using the more complex HDPI tool.^[ 3 ] Moreover, doctors also preferred using it; it’s easy to remember and simple to apply, and, significantly, it was better suited to the time-stretched, cognitively demanding context in which they worked.

Improving patient care by changing the default treatment

Another trial drawing on behavioural science has also succeeded in improving cardiac patient care. The Nudge Unit at Penn Medicine Center, University of Pennsylvania, increased rates of referral for cardiac rehabilitation – known to reduce mortality by as much as 30% in high-risk patients – from 15% to more than 80% simply by making referral the default for all patients.

Qualitative research in the form of interviews with cardiologists revealed that the existing referral process was manual, so they had to take action to initiate the referrals, opting patients in.  Redesigning the process as an opt-out system led to far better treatment and quality of care for patients, while retaining freedom for clinicians to opt out of the default if they did not think rehabilitation was necessary.^[ 4 ]

Reducing costs by making generics the default prescription

Healthcare institutions are continually striving to balance quality of patient care with cost efficiency. One area of cost saving comes from the prescription of generic drugs – chemically identical to branded or proprietary versions, just as effective and yet usually a fraction of the price. Given that doctors write hundreds of prescriptions each week from a wide range of treatment drugs, these costs can quickly add up.

In the NHS, generic prescribing has been rising since 1976 and stood at 84% in 2015. The Kings Fund estimates this has saved the NHS around £7.1 billion in total. However, it calculates there is still room for improvement, with potential for rates to rise to 90%, especially as there is variation between general practices.^[ 5 ]

In the US, generics prescription rates stand at 89% on average but again, could be higher. In a society where the payment often falls on the patient, generics are even more important as patients are nearly three times more likely to abandon a branded medication because of the high cost.^[ 6 ]   

To encourage more generics prescription, Mitesh Patel, director of The Nudge Unit at Penn Medicine Center^[ 7 ] recently trialled a tiny tweak to the prescription order system on the University of Pennsylvania Electronic Health Record system.

When doctors there select the drug they want to prescribe they click on a drop down menu. Previously, branded drugs were listed at the top of that menu and generics at the bottom. Patel flipped the order – it had an astounding effect. Before the trial, the generics prescribing rate at Penn Medicine was around 75.0%. Immediately after the change in the drop down order, the generic prescribing rate increased to 98.4% and remained there for the 10-month test period.^[ 8 ]

Why might this tiny change have had such a big impact? Sometimes the order in which items are listed can have subconscious effects on our decision-making. For instance, it may be that doctors presumed the medical community’s preferred choice of drug was the one listed first. Or perhaps, short of time and energy, they scrolled down to the first drug they saw that matched what they were looking for and looked no further.

While Patel admits that one of the biggest barriers to these kinds of interventions is that many clinicians are resistant to change and the concept of being ‘nudged’, he counters that “we often don't realise that we are already being nudged by the design and choice architecture of whatever electronic health record system we are using. It influences our choices every day, but often this is overlooked."^[ 9 ] 

The examples provide a snapshot of just a few of the simple, low-cost, yet astoundingly effective initiatives leveraging behavioural science to improve patient care by improving clinician decision-making.  Moreover, we may soon see more units like the Penn Nudge Unit, currently one of its kind, working to integrate behavioural science into medical processes and IT systems. While AI-assisted healthcare is taking the limelight, BE-assisted healthcare may be equally deserving of some attention.

Reference: