The following editorial can be found in the May 2024 issue of the American Journal of Bioethics.
In “Revive and Refuse: Capacity, Autonomy, and Refusal of Care After Opioid Overdose”, Marshall, Derse, Weiner, and Joseph contend that patients who may appear to satisfy the standard criteria for decision-making capacity could nevertheless be making non-autonomous refusals of care in the aftermath of naloxone revival following opioid overdose. They point to the effects of chronic opioid use, overdose, and naloxone resuscitation coalescing in such a way that genuinely autonomous decision-making is frustrated while the patient declining observation may still appear to have decisional capacity.
Here, we briefly share parallel findings from our research with heart failure patients who refuse a left ventricular heart device (LVAD), a form of mechanical circulatory support which is the primary alternative to transplant. We argue that what both sets of cases have in common is that clinical or research observers might be inclined to classify them as instances of non-autonomous (or autonomy-impaired) choices in key ways, even though clinicians not-unreasonably take these patients to have decision-making capacity. In short, we have found many instances of what Marshall et al. would characterize as plausibly capacitated but non-autonomous decision-making.
What Drives Non-autonomous but Plausibly Capacitated Decisions?
Marshall et al. draw our attention to the high proportion of ED patients who choose to forego an examination period following their resuscitation with naloxone. They underscore that, depending on the circumstances of the reversal, these patients may be experiencing either a residual intoxication, or acute withdrawal, from opioids, both of which impact cognition and higher-level decision-making. Furthermore, they highlight how many such patients suffer from opioid use disorder (OUD), a chronic condition that “involves substantial long-term changes in cognition, risk evaluation, preference-ordering, and decision-making” (14). Substance use disorders, including OUD, are often marked by impulsivity or compulsive behavior that can overpower more long-standing goals or desires. For this reason, many patients with chronic substance use disorders describe their choices in the throes of addiction as “inauthentic”, and not preferences they would, all things being equal, reflectively endorse.
The authors explain how OUD undermines the patient’s reflective evaluation of the choice at hand (to stay or go) in several ways: risk evaluation, delay discounting, and impaired strategic planning. These patients typically make riskier choices than those without OUD and prefer riskier choices with larger rewards (risk evaluation). They have a marked tendency for valuing immediate rewards over larger future rewards beyond that of those without OUD (delay discounting). Finally, they also tend to demonstrate impaired strategic planning (i.e., executing tasks with multiple steps) that suggests impaired executive function.
Study: Personalized Risk and Decision-making When Considering LVAD Therapy
Our research team is presently collecting and analyzing data as part of a multi-site trial investigating the benefits of using a personalized risk calculator (www.lvaddecisionaid.com) in patient decision-making about LVAD therapy (grant #R01 HS027784).
A novel risk calculator incorporating patient lab values has been designed and used to generate a personalized risk score for each patient, which is incorporated into the LVAD education sessions with their clinicians. Each personalized risk score provides mortality estimates at 1 and 2 years both with and without the LVAD. Follow-up interviews are being conducted with patient participants and their caregivers about their recall of the personalized mortality estimates provided to them and other considerations factoring into their decision-making process. While analyzing these interviews, we are discovering apparent “autonomy-impairing” parallels between LVAD candidates and the types of overdose revival patients Marshall et al. focus on.
Case Vignettes: Seemingly Capacitated, but with Impaired Autonomy?
Cognitive biases and reasoning heuristics are often features of both our everyday reasoning and our most complex, weighty medical decisions. In their classic 1974 work on heuristics and biases, Kahneman and Tversky identify many ways biases can impact our beliefs in more or less reasonable ways. But just because someone’s decision-making relies on decisional heuristics does not necessarily mean that they lack the capacity to make medical decisions. Nevertheless, these biases can impair autonomous decision-making even if they do not spoil decision-making capacity. Below, we identify several themes that emerged in our qualitative research, and suggest that they represent similar “autonomy impairments” to those proposed by Marshall et al.
“Everything’s a Risk” – Impaired Risk Perception
Our first vignette concerns impaired risk evaluation (which according to Marshall et al. impairs autonomy). One LVAD candidate participating in our study was given a personalized morality estimate of having a 60% chance of surviving a single year without an LVAD. During his interview with our research team, he accurately recalled his risk of death without an LVAD, but remarked:
I can be changing my tire on the side of the freeway and get hit by a semi. That’s a risk too. Everything’s a risk.
That this participant had accurate recall but a skewed perception of risk comparisons (e.g. equating dying from heart failure with unforeseen, accidental death) suggests an impairment in risk evaluation. This participant was also comparatively pessimistic about their chances of meaningful longevity with an LVAD. When asked about his own prediction of longevity if he chose to get an LVAD, he said “I feel like I’m going sooner [than the personalized calculator predicted].”
“I Didn’t Really Want to Know” – Exaggerated Delayed Discounting
Our second vignette provides an example of delayed discounting, another autonomy-impairing condition according to Marshall et al. In this example, the participant had an active lifestyle and only recently received their heart failure diagnosis. She accurately recalled that her 1-year mortality risk without the LVAD was 50%, saying “I glossed over hearing those numbers because I didn’t really want to know.” While capable patients have a right not to receive medical information, this participant repeatedly referenced her contemporary feelings of good health and her healthy lifestyle when discussing her thoughts on whether to receive LVAD treatment. This patient evinced difficulty in seeing past the immediate burdens of LVAD surgery in comparison to the longer-term benefits of increased longevity and function, opting for the short-term decision of avoiding surgery.
“That’s Not Me”: Other Cognitive Biases and Heuristics
We have found other cognitive biases and heuristics used by this population which we believe are also potentially autonomy impairing but not incapacitating. One example is affective forecasting or impact bias. This bias in decision-making occurs when someone underestimates their ability to adapt to a change in circumstances. For example, some participants expressed difficulty in reconciling the prospects of living with an LVAD with their conception of themselves. One patient who had a mere 40% chance of surviving 2 years without an LVAD said, “That’s not me,” citing the need for having tubes and wires ‘sticking out’ and regular charging of the device’s batteries. Another participant (who had a 60% estimate of surviving 2 years without LVAD) cited personal vanity and that “having tubes out is not me.” While there are risks and burdens associated with LVAD therapy, many who choose to get an LVAD acclimate to living with the device—a point that may be underappreciated in a way that impairs truly autonomous choice.
Another cognitive bias we have encountered is the focusing effect, where a person reasons about a decision from a particular fixed point even if it is not the most relevant consideration to the decision at hand. For example, several patients cited their fear of a driveline infection (or keeping the site clean) as especially influential on their decision, with one indicating that the prospect of such an infection would “haunt me.” While drive line infections are possible, and wound care is important for avoiding infection, overly focusing on potential negative outcomes that are minor compared to the risks (earlier death) demonstrates a likely impairment of autonomy which would not be captured in decisional capacity assessment.
Autonomy Beyond Capacity
While an extended discussion of decisional capacity and its connection to autonomy falls beyond the scope of this brief editorial, it is worth concluding with the remark that some of the apparent tension Marshall et al. identify between decisional capacity and nonautonomous decision-making may be a tension more in appearance than in reality. As they note, that a patient appears to meet criteria for decisional capacity does not mean that they, in fact, do, and there are particular features relevant to this patient population that make performing an accurate capacity assessment particularly difficult (e.g., time constraints in the ED, the “epistemic gap” between stating reasons and exhibiting the corresponding motivations).
Most significantly, these two patient populations raise important questions about where nonautonomous decision-making may give us reasons to be concerned about a patient’s decisional capacity. Indeed, decisions that appear to truly “not be the patient’s own” are strong prima facie evidence against their having DMC in the first place, at which point questions of involving surrogates and involuntary treatment become important. However, we also ought to be wary about extrapolating too much from cases where the conditions for DMC are merely superficially or apparently met, as a more thorough analysis could reveal that capacity was lacking all along.
Funding
This project was supported by the Agency for Healthcare Research and Quality (AHRQ) through grant number R01HS027784. The content is solely the responsibility of the authors and does not necessarily represent the official views of the Agency for Healthcare Research and Quality.
Jared N. Smith, Joanna Smolenski, Ben H. Lang, and Jennifer Blumenthal-Barby