Why T2T is Too Risky for Some Patients Save

“Humans often make irrational choices.” As you read that statement, you probably agreed with it. You even might have privately smirked and thought, “That’s probably true of most people, but it’s less true of doctors. Or me.”

I recently finished a book on behavioral economics, Thinking Fast and Slow by Nobel prize-winner Daniel Kahneman, and was intrigued by the possibility that, like the general public, rheumatologists sometimes do not make rational choices.

A few months later, during an interlude of a meeting of 12 U.S. rheumatologists brought together to review some clinical trial data, my colleagues indulged me in a small experiment. I set before them $100 in cash and offered each of them a choice: take the $100, or flip a coin for a 50% chance to win a certain amount of money that they could specify. They would tell me the minimum amount of money I needed to offer in order to prefer the coin flip over taking the $100.

The expected value of a 50% chance to win $500 is (0.50 x $500) = $250. Therefore, a “rational” person would prefer the coin flip if what I was offering them was $201 or greater. To my surprise, the median amount of money I needed to offer was $500. Meaning, it was only after I offered them a chance of winning $500 or more (someone requested $1,000) that most rheumatologists preferred the coin flip over the sure thing. The principle underlying this behavior is that humans are typically risk-avoiding when it comes to a “gain” or a benefit. There is a premium they are willing to forgo to be guaranteed a “sure thing”. In this case, this average premium for the “sure thing” was (the expected value of a 50% chance at $500 = $250) – $100 (the sure thing) = $150.

Although we didn’t have time to test it that day, the opposite is also true. Humans are typically risk-seeking when it comes to avoiding a loss and will take extra chances to avoid experiencing a potentially avoidable harm. And humans usually fear a loss much more than they are pleased by a gain of equivalent magnitude. Applied to money, the pain of a financial loss is about double that of the equivalent financial gain. The same paradigm applied to RA disease activity is shown (Figure).

Figure - For any given change in disease activity (on the x-axis), there is less symptomatic benefit (on the y-axis) for patients than the amount of harm incurred by the same magnitude of worsening. Visually, even though they are equidistant from the origin, the dotted green bar (representing the amount of benefit associated with improvement) is shorter than the height of the dashed red bar (representing the amount of harm associated with worsening).

What does this have to do with treat-to-target? Two recent studies from Europe and the U.S. presented at ACR 2015 showed that treat-to-target was not implemented at more than one-third of the visits that it should have been, despite a study protocol that “mandated” treatment acceleration [ACR abstracts 3184 and 3185]. In both studies, “patients refusing to escalate care” was either the first or second most important reason. This issue may resonate in your practice: you have patients whom you urge to change treatments to attain remission or low disease activity, but they refuse. Maybe they began in high disease activity and after starting methotrexate and a biologic, they are now in moderate disease activity. Now at four months, they tell you that moderate disease activity is “good enough” and are unwilling to change therapy.

Sound familiar?

Let’s now apply a behavioral economics framework to this problem. When you tell patients there is a chance that new drug (e.g. a different biologic) might not work as well as their current biologic and they could worsen, they perceive this as risking a loss to their present level of RA disease activity. When you tell them about a heightened risk of potential side effects (e.g., GI perforation, herpes zoster) related to a new drug’s mechanism of action, they perceive this as risking a loss to their present health state.  For many patients, these risks outweigh the potential upside of lower RA disease activity. And moreover, you have no “sure thing” to offer them in terms of the future benefit of a new drug. In RA, predictive modeling using clinical or biomarker data is not up to the task. From a decision-making perspective, this is the worst possible decision scenario – we ask them to risk a loss, and have no offsetting certainty of a gain.

So is there a fix?

Perhaps we need to shift the tenor of the conversation and talk about how undertreated RA is likely to cause “losses” over time. Accruing data suggests that ongoing moderate disease activity (not just high disease activity) has an unfavorable effect on cardiovascular event risk, serious infections, and other safety events. Long-term co-therapy with glucocorticoids (e.g. prednisone 10mg/day) has toxicities that rheumatologists love to hate yet sometimes use for years, without attempting to try and taper. And we need long-term outcome data from registries for Patient Reported Outcomes (PROs) that can show undertreated RA leads to worsening of PROs over time. The NIH’s PROMIS system (www.nihpromis.org) may help because it encompasses life domains that patients may care about in a way that they will never care about joint counts, a CDAI, or the DAS. These and other negative consequences of undertreated RA can be framed as a loss to your patient’s future health, counterbalancing the perceived risks of changing to a new therapy.  

The “loss aversion” described above is not the only relevant principle from behavioral economics. “Status quo bias” describes a tendency to prefer to maintain the status quo, or select the “default” option, because it is perceived as less risky. Changing the default from an opt-in to an opt-out has been shown to change behaviors, including dramatically increasing national rates of voluntary organ donation in some countries, or prompting more individuals to contribute to a matched 401(k).

Consider a recently diagnosed RA patient who has been treated for 4 months with methotrexate and 10mg prednisone. She’s better, but isn’t near disease activity goals. Absent information from you to the contrary, the status quo is for her to maintain her current treatment and change nothing. In contrast, if you set the expectation at her initial visit that standard RA treatment in your practice is to use combination therapy starting at 4 months for everyone not in low disease activity, you have shifted the default option, and she will be more likely to be willing to escalate care.

Like it or not, doctors are frequently in the position of persuading patients to make a specific decision. Using behavioral economic principles applied to medical decision-making in no way diminishes patient choice nor retards shared decision-making. Rather, behavioral economics sometimes invoke the notion of “libertarian paternalism”, the idea that an agent (e.g. a rheumatologist) is attempting to influence their patients’ choices in a way that will make the patient better off, as judged by the patient themselves.

Much of what we say to patients about their treatment options reflects the influences of our own preferences, values and goals, and we owe it to our patients to be aware of our own biases. Effective clinicians recognize that the art of communication is more than just providing data, it’s framing the information in a way that is concordant with patient’s own goals and values. Behavioral economics may offer you an improved framework to communicate with your patients the next time you have one of these conversations that seem to go amiss.