The use of defaults are all around us, whether we recognize them or not. Consider Netflix and YouTube who have changed their default settings to automatically play the next episode in a TV series or similar content to encourage binge watching. Ecommerce giant Amazon displays additional, complementary items alongside your purchase to encourage purchase bundling. The same techniques being utilized by tech giants around the world, borrowed from the classic 401k defaults study by Benartzi and Thaler, are being deployed in the healthcare sector, and it’s working.
So applying behavioral interventions to problems within healthcare is just as important as the advancements we make in other aspects of healthcare like digitization, the use of machine learning, and wearables. Clinicians operating within healthcare face many challenges, and as the cost of healthcare increases without equivalent improvement in the quality of care (see our case study on Preventative Healthcare written by Dr. Leigh Christopher), inexpensive interventions that advance patients’ interests and minimize physician burnout are desperately needed.
Physicians mean well but the amount of tasks they must juggle throughout the workday is substantial, made worse by the introduction of EHRs according to this study. Here, the number of tasks physicians engaged in per minute increased after the EHR was implemented, indicating that physicians switched tasks more frequently when using an EHR. Frequent task switching raises patient safety concerns, which is the opposite of the aim of the intervention. Behavioral audits and insights are needed to fulfill the promise of EHRs in improving physician efficiency and patient healthcare outcomes.
Dr. Mitesh Patel and his team at the Penn Medical Nudge Unit were seeking answers on how to reduce healthcare costs. One way to help prescribers do this is to reduce prescription drug costs. Patel and his team developed an experiment to integrate a default towards automatically prescribing a generic version of a drug to the EHR. When applied, this intervention increased prescribing rates for generics over brand name drugs from 75% to 98%. This intervention both reduced the cognitive burden on physicians of manually changing the prescription from brand to generic as well as lowering prescription drug costs.
Within the same vein, patients admitted to cardiology after experiencing a myocardial infarction (heart attack) do much better when referred to a cardiac rehabilitation program (CRP). Physicians have normally had to refer patients to a CRP, and a lack of patient referrals is a key reason why these programs are underutilized. After redesigning this intervention as an opt-out option (actively choosing not to refer a patient) rather than an opt-in option (actively choosing to refer), the referral rate increased from fewer than 25% to more than 80%.
Kristin Kostick and her team were able to increase the implementation of a validated decision aid by physicians and relevant staff, in order to help patients choose a left ventricular assistance device (LVAD) by way of a default. LVADs are offered to those with advanced heart failure and have been shown to improve 1-year mortality outcomes when compared to pharmacotherapies alone by 30-55%. Kostick and her team augmented a pre-interventional 9.8% aid application rate to a post-interventional 70% aid application rate for eligible patients within six months. Increased aid adoption meant more eligible patients were educated on and provided informed consent for this life saving device.
We’re also seeing the use of BE and big data come together in healthcare. One such study sought to improve end of life care for critically ill oncology patients. Not every patient will respond to their treatment. Physicians and oncologists need to quickly identify those patients and conduct those difficult conversations about end of life care. With rapidly changing circumstances and a multitude of tasks at hand, oncologists and other clinicians sometimes miss these patients and do not issue a serious-illness conversation (SIC) in a timely fashion.
Using a machine learning algorithm to identify patients whose metrics indicate a ≥10% risk of mortality within 180 days (henceforth called ‘high risk’), practitioners were sent emails alerting them to which patients were high risk. Would receiving these emails be enough to prompt oncologists to make these appointments or would a default be necessary? Compared to practitioners that just received high risk patient emails (the control group), those defaulted into appointments with high risk patients were almost 5 times more likely to engage in SICs as compared to those practitioners in the control group. The use of big data in combination with behavioral economics’ interventions led to significant improvements in end of life care communications, at a time when time itself is of the essence.
In sum, these interventions show that some of the same behavioral insights and nudges adopted by our favorite companies, I.e., Netflix and Amazon, are also being applied to change physician behaviors in healthcare. Defaults are an incredibly powerful technique to use at a time when people are already juggling so much. In the case of overworked physicians, defaults can help them make better choices for their patients. In case of the binge-watcher, well, that is up to you to decide whether your default settings are hurting or helping you.