Beyond Nudges to Behavioral Strategy

First, the case: This involved a chronic condition that has a history of being difficult to properly diagnose and treat. It is a serious condition, given to unpredictable periods of flares and remission, with a broad range of effects that can vary considerably from person to person. Symptoms can be debilitating, and broader long-term complications can emerge that have negative impacts on both lifespan and the patient’s quality of life. Underlying disease mechanisms remain mysterious, but there is a candidate set of treatments, cutting across different classes, that has been built up over decades from which physicians can select to successfully tackle flares, reduce symptoms, and stall disease progression, the trick being finding the one that will be right for the patient at hand.

It was in this context that the client’s treatment was situated, itself part of a newer, cutting-edge class of therapies that had already shown great benefit in this condition and in multiple others. In this case, the client’s treatment had the edge of data demonstrating not only that its efficacy could beat the standard of care, but that it could also be effective in mitigating one of the condition’s long-term complication risks as alluded to earlier. Yet, despite increasing clinical experience and comfort with this class of treatments, a nontrivial percentage of specialists were staying with older therapies, and not demonstrating openness to the client’s drug, when it came to treating the condition in question. As it turned out, many of these physicians were also expressing confidence that the long-term risk the drug could mitigate was rare, sometimes discounting epidemiological evidence of the risk’s true prevalence when it stood in contrast with the physicians’ own expectations.

The case seemed, in short, like a repeat of the late- and non-adopter problems that are ubiquitous in healthcare – albeit likely driven in this instance by dynamics that would have been unique to the combination of physician specialty, condition, and professional context that was central to the case at hand. To adequately address the core challenge, it was necessary to develop a better understanding of these dynamics so that tactics could be selected that would have the best chance of increasing physician receptivity to data, given the forces that were likely attenuating the risk perceptions and standing in the way of the client’s drug being utilized. And that meant pulling together as many facts as could be marshaled about the physicians and their world that could act as clues to the concerns, practices, and mental processes they were relying on as they treated the condition of interest.

What the facts seemed to suggest was this:

• These were hardly incurious or habit-bound physicians. On the contrary, they understood the gravity of the conditions they were tasked with treating, took their implications for patients seriously, and seemed to revel in the opportunity to unravel the puzzles that each patient presented them with, often playing an active, creative role in clinical judgment and decision-making in lieu of formal published guidelines to support them.
• They also thought in ways that were exceptionally concrete and experiential: They emphasized what made each patient unique, attended closely to the phenomenological manifestations of disease (i.e., symptoms) which they then made central to their case representations, and then made treatment decisions by recalling past cases, defined by similar symptom, sign, and lab value clusters, for whom a particular treatment had appeared to be effective. They had a well-developed sense of the set of treatment options available to them, and they relied on the if-then rules implicit in past case exemplars to flexibly select and tailor treatments based on the unique, granular constellation of case features that was presented to them in the moment.

The fact that the physicians relied on a curious, experience-based reasoning style to problem-solve made sense in a specialty area known for conditions with significant heterogeneity, little associated disease mechanism knowledge, and scant guidelines – circumstances to which the physicians would have needed to have been well-adapted if they were going to be effective. What it didn’t do, though, was explain why they would be reluctant to use the client’s treatment or would downplay the long-term risk that the treatment could address, particularly given their own recognition of the limitations of standard therapies.

Some other clues proved to be illuminating:

• The first had to do with the evidence used to assess long-term complication risk. As with diagnosis and treatment, the physicians extended their experience-based reasoning style to their risk perceptions, prioritizing personal clinical experience over what epidemiological data might tell them – at times, going so far as to actively disparage such data on various post-hoc grounds. Their experiences with the condition often included fluently recalled, disturbing memories of cases in which the risk came to pass – but it also included steps the physicians had taken with patients to vigilantly monitor for signs of disease progression, along with steps they had actively pursued, and clearly understood to take, once complications emerged. As the physicians based their risk estimates on their own local prevalence rate, they also now opened themselves to experiences in which the occurrences of the complication in question could have been distributed sparsely across time and cases – quite possible given the magnitude of the true aggregate rate and how it would be expected to play out locally based on purely statistical grounds.
• Then there was the nature of what the physicians focused on when they evaluated treatment pros and cons, which was frequently downside-heavy. In effect, they recognized the limitations of current treatment practices, but they worried about new treatment risks as much as they worried about the condition they were treating. And they did so in a context where the standard of care had conditioned them to treat reactively, forcing them to adopt a state of vigilance with a condition that, historically, had always been unpredictable and never curable.

With the facts assembled, a behavioral diagnosis was developed that emphasized the presence of three things:

• A bias toward underestimating the complication risk owing to the physicians’ high levels of personal involvement in monitoring and responding to it (effectively encouraging a feeling of personal control that is known to drive risk perceptions downward), combined with the statistical artifact of how the risk would likely manifest at a practice level – all potential counterweights to any tendency to overestimate the risk arising from fluent recall of its past occurrences;
• A chronic inclination toward a reasoning style, well-adapted to the nature of the domain in which the physicians are specialized, that prioritizes experiential data, pattern recognition, naturalistic decision-making, and intuition over theory-based, hypothetico-deductive reasoning – in this case, placed at the service of problem-solving conducted with high levels of engagement and curiosity; and
• A motivational orientation that sensitizes physicians to the negative consequences of the condition (including making them vigilant in risk detection, however unlikely they judged it), but also makes it likely for them overlook treatment potential over the possible downsides of change.

These components had a good chance of being synergistic, with the reasoning style reinforcing the emphasis on intuitive-based risk judgments over abstract data, the style’s focus on concrete, local facts being a good fit with the known cognitive preferences of an avoidance-like orientation, the orientation itself being amplified by the heightened levels of engagement, and even the active disparaging of data over experience being explained, in part, by the prevention-minded desire to avoid untoward influences on beliefs felt to have good standing. Together, they painted a portrait of the physicians approaching the focal condition less like bench scientists than like hunter gatherers who, without the benefit of clear guidelines or solid disease process theories, needed to adapt their thinking and motivational styles to feel their way through a clinical terrain that was heterogeneous, unpredictable, and threatening – all requiring curiosity, intuition, and fast-and-frugal thinking to keep up with the terrain while taking steps to protect against behavioral disruptions, missed signals, and mistakes that could harm the patient along the way. The portrait needed to be treated as provisional absent the opportunity for further testing, but it provided the most coherent, parsimonious fit to the available evidence relative to other alternatives that were considered. And it created a solid base from which to recommend evidence-based tactics, from data presentation formats and outcome framing methods to methods for building causal narratives, that could potentially deliver information with the right resonance, support more hypothetico-deductive thinking styles, and allow epidemiological data to land with greater force.