Get started
Work

What is a good medical choice?

Year: 2021
Type: letter
Abstract: Filson et al1 report a post hoc analysis of treatment patterns among men with localized prostate cancer enrolled in a randomized trial comparing a decision support tool—the Personal Patient Profile–Prostate—with usual care. Randomized trials have clearly shown that decision aids increase patient knowledge and improve several process outcomes, including patient involvement, value-choice congruence, indecision, and decisional conflict.2 However, there is a dearth of studies measuring preference-linked health outcomes, that is, whether patients exposed to decision aids subsequently experience the health outcomes that they said they would prefer and avoid the ones to which they were averse.2 Critically, downstream effects of decision aids on disease-specific health outcomes, treatment adherence, costs, and resource utilization remain largely understudied. We applaud Filson et al1 for performing this important study and posing the crucial question whether patients exposed to a decision aid went on to pursue guideline-concordant care based on disease risk. We are particularly interested in the use of active surveillance for low-risk disease because curative treatment is known to be associated with important reductions in quality of life due to urinary, erectile, and bowel dysfunction without relevant benefits in prostate cancer survival. Use of the Personal Patient Profile–Prostate decision support tool did not improve guideline-concordant care (adjusted odds ratio of receiving active surveillance for patients with low-risk prostate cancer, 0.89; 95% confidence interval, 0.36-2.20).1 However, the authors maintain that “this is not what the decision support tool was designed to do” and reemphasize the results for the primary end point, a reduction in decisional conflict in comparison with usual care,3 and they claim that this would improve psychosocial outcomes.1 However, this line of reasoning would, if taken to its logical conclusion, imply a fundamental realignment of how we think about medical decision-making. The conventional position is that a good medical decision is one that optimizes the benefit-to-harm ratio, with the important proviso that patients may vary as to the value that they place on particular benefits and harms. The authors instead appear to argue that we should judge a medical decision by how patients feel about their choice (value-choice congruence) and whether they worry about it afterwards (decisional conflict). Taken to its extreme, this would suggest that it would be acceptable for physicians to subject a patient to a painful, invasive procedure with important risks and no benefit as long as the patient felt sure that this was the right approach. By way of analogy, the authors’ argument would have us recommend an educational tool about tobacco use that had no effect on cessation rates as long as patients did not feel conflicted about their decision to keep smoking. We have previously argued that decisional conflict is a highly problematic end point for several reasons, most notably that it is not at all clear that “feeling sure” about a difficult medical decision is a desirable state.4 There are numerous scenarios where a good decision-making process actually increases decisional conflict (eg, a patient who believes that all cancer needs surgery eventually decides on active surveillance after a thorough discussion with a sensitive physician but does so with some qualms). Similarly, a problematic decision-making process may be associated with low decisional conflict (eg, a patient who says “I have cancer, I want it out, and there’ll be no more discussion”). Decisional conflict is also susceptible to cognitive biases such as cognitive dissonance. Endpoints such as survival and urinary incontinence are far less problematic. Yes, patients vary in the degree to which they value additional survival or would find urinary dysfunction to be problematic, but no reasonable patient would accept a risk of incontinence for zero benefit in survival. Active surveillance is recommended for low-risk prostate cancer in guidelines such as the American Urological Association guidelines5 not because of a paternalistic position of “the doctor knows best” but because, except in rare cases, management of low-risk disease is not preference-sensitive. A decision aid that does not improve uptake of active surveillance among patients with low-risk prostate cancer is severely limited. Furthermore, it is highly problematic to redefine the purpose of medical decision-making away from optimizing benefits and harms toward merely making patients feel confident about their decisions. We thank Filson et al for conducting this important study and for posing the critical research question of whether patient decision aids ultimately improve patient care. We hope that this study will spur discussions regarding suitable outcome measures of shared decision-making as well as research regarding what components of decision aids may dissuade patients from decisions that may be in their best interest. The authors’ work was supported in part by funding from the National Institutes of Health/National Cancer Institute (P30-CA008748). Sigrid Carlsson was further supported by a career development award grant from the National Institutes of Health/National Cancer Institute (K22-CA234400). Behfar Ehdaie reports consultant fees from Myriad Genetics and an honorarium from Koelis outside the submitted work. Andrew Vickers is named on a patent for a statistical method to detect prostate cancer; this method has been commercialized by OPKO Health. He also receives royalties from sales of the test and has stock options in OPKO Health. The other author made no disclosures.
Source: Cancer
DOI: 10.1002/cncr.33447
Language: English
Open Access status: green
OpenAlex ID: W3127857201
Cites: 6
Cited by:
Related to: 10