We used the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) checklist (Page et al., 2021). One author (NB) searched the following citation and abstract databases from inception until August 5th, 2021: PubMed/MEDLINE (US National Library of Medicine), Philpapers, and PsycInfo (American Psychological Association). The search strategy used keywords and controlled vocabulary terms (MeSH and Thesaurus of Psychological index Terms) for the topic of interest. The searches were limited to English language and peer-reviewed publications only where possible. The full search strategy (Supplementary Materials) was reviewed and validated by an independent librarian from the National Institutes of Health Library, and the search was updated to February 15th, 2022. Finally, we used the snowball method and our own experience to add publications that were not detected in the three databases. All search results were exported to EndNote X9 and duplicate citations were identified and removed.

We included an article if:

We excluded purely descriptive publications that do not make an argument for either position. We did not require that publications focus exclusively or predominantly on the issue of RS-DMC, only that they make an argument for or against RS-DMC. Additionally, because some of the initial discussion about RS-DMC originated in books, we eventually included five books that appear in the debates or were known to us in the literature.

To minimize bias, two readers (NB and a research assistant) independently performed the title/abstract screening and the full-text screening following the predefined inclusion and exclusion criteria. Discrepancies were resolved through discussion until agreement was reached, and when needed, a third reader (SK) assisted in this process. We identified and extracted frameworks and reasons as follows. Both authors (NB, SK) read 50% of the eligible manuscripts and independently created coding schemas. After comparison and discussion, a preliminary schema was agreed upon and adjusted as necessary while coding the eligible manuscripts. The resulting coding scheme had two parts: frameworks, which refer to structural or content features of a publication's conception of concepts related to RS-DMC, and specific reasons used in a publication. Both authors identified passages in each publication that constituted a “reason” or “framework.” Reasons were categorized into three groups: (1) Arguments for RS-DMC, (2) Arguments against RS-DMC, and (3) Counterarguments defending RS-DMC². This categorization naturally emerged from the progression of the debate in the literature, as the first publications discussing RS-DMC defended it, and were then critiqued, opening the door for counterarguments. Thus, our categories reflect the progression of the dialogue characteristic of the debate. In addition, we tracked the year of publication, type of journal, and author background.

Dates of publication ranged from 1984 and 2021, though most (32/41) were published between 1984 and 2006. Most articles were published in Bioethics/Philosophy journals (20/36), followed by Clinical journals (6/36) and Law/Policy journals (5/36). Nonclinical (philosophy, bioethics, law, policy), was the most common author background (25/41), followed by mixed background (10/41) and clinical (medicine, psychiatry, etc.) background (6/41).

Three main stances emerged: (1) the substantive view, according to which the threshold required for competence itself should vary with risk, (2) the epistemic view, according to which the amount of evidence or certainty required for a finding of competence should vary with risk, and (3) neither (Table 1). A slight majority (22/41) supported the substantive view of RS-DMC, while fewer supported the epistemic view only (12/41). Most publications in clinical journals (5/6) or books (4/5) supported the substantive view. Notably, most articles or books published before 1990 held the substantive view (6/8 publications). All epistemic view articles were published in 1990 or later.

Some publications endorsed both the substantive and epistemic views. These publications were categorized under substantive view, as most of the debate about RS-DMC focuses on whether the substantive view of RS-DMC is appropriate. For instance, articles critical of RS-DMC always targeted the substantive view, never the epistemic. In many cases, articles critical of substantive RS-DMC endorsed the epistemic view. More specifically, the epistemic view is often used to allow for the inclusion of risk in DMC assessment, thus satisfying the common intuition that this is appropriate, while avoiding the supposed flaws of the substantive view. Thus, the debate largely is between those who support the substantive view and those who do not, regardless of whether they support the epistemic view. Accordingly, hereafter we will use ‘RS-DMC' to refer to the substantive view of RS-DMC, unless otherwise specified.

Further, there is debate in the literature about the distinction between substantive and epistemic RS-DMC. Some authors question whether these views differ in practice (Wilks, 1999), while others maintain that there is a significant distinction (Wicclair, 1999; Parker, 2004).

We identified 26 primary reasons in the literature that were used in more than one publication (Tables 2–4). We classified 8 reasons as “Pros” (Table 2), 10 reasons as “Cons” (Table 3), and 8 reasons as “Counterarguments Defending RS-DMC” (Table 4). This reflects the progression of the debate over time.

The most common reasons in favor of RS-DMC rely on the intuitive appeal of RS-DMC and its practical merits (Table 2). For example, many publications argue that RS-DMC allows for a balance between autonomy and welfare (P2) and the balancing of potential errors (P3), but generally do not explain this balancing in detail, although there are exceptions (Schopp, 1994; Buchanan, 2004). Similarly, “Coheres with Current Practice” (P1), “Rejecting fixed level of competence safeguards against broader paternalism” (P4), “Avoids unnecessarily burdening the system” (P5), and “The opposing view needs to articulate a natural ‘adequate level' of decision-making abilities” (P7) focus on the practical difficulties that would result from rejecting RS-DMC. Finally, though cited infrequently, “Respecting patient's wishes has value” (P6) and “Tailored DMC assessment” (P8) do not fall neatly into either intuitive appeal or practical merits.

Most of the reasons against RS-DMC reflect concerns about paternalism (Table 3, arguments C2 through C9). Some arguments are direct charges of paternalism (“RS-DMC is paternalistic” [C2]) while others imply this concern. For example, “Form of outcome-based DMC” (C3) is the argument that when risk is incorporated into DMC assessment, a patient may be deemed incompetent and have their decisions ignored based on merely the decision they make. Similarly, “Imports assessor's values” (C6), “Introduction of values into value-neutral assessment” (C8), and “Falsely finds incompetent persons competent” (C9) are rooted in concerns that DMC assessment will be driven by physicians' values rather than any objective measure, and patients will be judged on whether their decisions/values differ from those of the evaluating physician. Finally, “Conflation of DMC and Decisional Authority” (C4) and “RS-DMC is tautological” (C5) argue that RS-DMC is conceptually flawed and only nominally protects patient autonomy, as a physician could seemingly deem a patient “incompetent” as long as he deemed the risk high enough, rather than assess competence with no consideration of risk and independently decide whether the patient's decision should be respected.

“Standards vary with complexity, not risk” (C7), “Coherence is not sufficient reason” (C10), and “It is unclear where to set the threshold” (C11) do not fit as neatly into this umbrella of concern over paternalism. C10 denies that coherence with current practice is sufficient reason for RS-DMC. C11 highlights a practical concern about RS-DMC implementation. C7 is a disagreement over what explains the intuitive appeal of raising the threshold when decisions are high-risk, but it is disputed by even some critics of RS-DMC.

The asymmetry between consent and refusal (C1) is the most discussed argument regarding RS-DMC. It says that RS-DMC implies the conceptually incoherent view that a person can be competent to consent but not to refuse. Some argue that even if this is not conceptually incoherent, it leads to ethically unacceptable evaluations involving bad faith or deception (Lawlor, 2016).

Most of the counterarguments defend RS-DMC against charges of paternalism (Table 4). Some do so by directly refuting the criticisms (“RS-DMC is not paternalistic,” [CA2], “RS-DMC is not tautological,” [CA5]), but others defend RS-DMC by clarifying how the model functions in practice or by disputing the conceptual premises of the critics. For example, “Outcome alone does not determine DMC” (CA3) clarifies that RS-DMC may include risk or outcome as part of the assessment, but does not rely on these factors alone; therefore, outcome alone does not determine DMC as is suggested by critics. Similarly, “Consistent with the reasonable person standard” (CA6) points out how consideration of risk is consistent with the commonly used reasonable person standard, and thus does not import values inappropriately. “Not a conflation” (CA4) argues that the disagreement is due to differences in the conceptual frameworks of decision-making capacity used. Similarly, “Inherently normative” (CA8) asserts the position that DMC assessment cannot be value-neutral.

The counterarguments against the asymmetry argument (CA1), however, are quite varied. Some simply acknowledge that the asymmetry seems odd but embrace it as part of RS-DMC (Buchanan and Brock, 1986; Howe, 2010), or consider consenting and refusing two separate decisions³ (Brock, 1991; Wilks, 1997). Others show that asymmetry appears acceptable when determining whether to maintain or rebut the presumption of capacity in a given situation (Checkland, 2001; Brudney and Siegler, 2015). Other supporters of RS-DMC argue further that aside from the context of evaluating the presumption of capacity, RS-DMC not only does not need but should not include asymmetry of consent and refusal (Bolt and van Summeren, 2014; den Hartogh, 2016; Graber, 2021).

Authors varied on how competence and related concepts are defined or understood. We use “frameworks” to refer to structural or content features of the author's conception of competence and related concepts (Table 5). We categorized conservatively: a publication was a given category only if it explicitly endorsed or clearly made use of a particular framework or definition; thus, it is possible an author in fact holds a certain framework but we could not code a publication as such.

The first framework category of “Externalism” vs. “Internalism” is a distinction noted by some authors (Wilks, 1997; Berghmans, 2001). Internalism holds that competence is solely a function of a person's internal abilities relevant to decision-making; externalism holds that competence is determined by both internal abilities and other contextual or relational factors external to the person's decisional abilities, such as risk. Of those publications codable on this issue, most were externalist (17/26).

A closely related framework category—“One-step” vs. “Two-step”—addresses how decisional authority should be determined (Buchanan and Brock, 1986; Culver and Gert, 1990). The two options are either a single step that incorporates information about a patient's abilities in addition to relevant contextual factors such as risk or taking two steps, first determining competence, then separately determining whether the person should have decisional authority. Of the 19 publications codable on this issue, most held the one-step view (12/19).

Similarly, the third framework category asks, “Does having DMC imply having decisional authority?” Some hold that a finding of DMC grants an individual decisional authority, so their decisions must be respected, while others hold that a finding of DMC only means that an individual has the abilities to make decisions, not that they should automatically have decisional authority. The majority (17/24) endorsed the view that having DMC implies having decisional authority. The first three framework categories seem closely related conceptually.

The fourth framework category is the predominant conception of well-being that a publication relies on in its model of DMC. There were three coding options: (1) predominantly objective, when publications primarily use an objective or shared understanding of welfare, (2) predominantly subjective, when publications primarily use an individual's own subjective understanding of their welfare, and (3) both objective and subjective used, with neither clearly favored over the other. Among codable documents on this issue, half (15/30) had a predominantly objective conception, while 9/30 relied on both, and the remaining 6/30 relied primarily on a subjective perspective.

Finally, publications were categorized according to their view on the scope of competence, that is, whether competence is about a specific, particular decision or about a type of decision. Those publications that hold the first view argue that the specific decision a person makes is relevant to competence assessment, while those that hold the second view argue that an individuals' competence should instead depend on an individual's ability to make the relevant type of decision. 17/24 codable publications on this issue supported the “specific decision” view.

The framework categories identified in Table 5 are highly associated with an author's stance on RS-DMC.

For example, 16 of 17 publications that hold an externalist view endorse the substantive view of RS-DMC, while all 9 publications that hold an internalist view do not (Table 6). Similarly, all 12 publications that support a one-step determination of decisional authority endorse the substantive view, 12 out of 15 publications that use an objective conception of well-being support the substantive view, and 14 out of 17 publications that use “competence of specific decision” support the substantive view.

The frameworks are sometimes also associated with reasons for or against RS-DMC. This was most obvious when there was a logical connection between the frameworks and the reasons. For example, the two closely inter-related framework elements of “two step vs. one step” view of decisional authority and whether DMC implies having decisional authority were highly associated with two arguments against RS-DMC, namely, whether an author criticized RS-DMC as conflating DMC with decisional authority (C4) and as providing only tautological prohibition of paternalism (C5). For example, 5/7 publications that hold that having DMC does not imply having decisional authority argue that RS-DMC involves a conflation of DMC and decisional authority (C4), whereas authors who view DMC as implying decisional authority understandably do not see a conflation (0 among 12 codable papers). Similarly, 6/7 publications that hold a two-step view of DMC assessment and 6/7 publications that hold that having DMC does not imply having decisional authority argue that RS-DMC prohibits paternalism by definition only (“RS-DMC is tautological” [C5]). These findings suggest that at least some areas of debate over RS-DMC arise due to differing underlying premises.

There is a lack of uniformity in vocabulary and definitions in the debate. The publications were in journals from a variety of disciplines, by authors with diverse backgrounds and across jurisdictions, which may explain some of the differences in vocabulary. But authors often understand the concepts of DMC and decisional authority differently, as captured by the five framework categories that we tracked in the literature (Table 5).

These differences can make it unclear whether disagreement reflects misunderstanding or substantive ethical disagreement. It appears at least some of the disagreements about RS-DMC may be due to differences in frameworks and premises⁴. The frameworks endorsed also sometimes relate to the reasons each publication used. For example, “Conflation of DMC and decisional authority” (C4) or the tautology argument (C5) require the view that having DMC does not imply having decisional authority, a minority view in the literature. The significance of frameworks has been relatively neglected in the debate. Future debates on RS DMC may benefit from explicit attention to this issue.

Though there are many distinct reasons for and against RS-DMC in the literature, a few broad patterns became apparent. First, the pro RS-DMC arguments tend to rely on the intuitive and practical appeal of RS-DMC. Second, arguments against RS-DMC mostly have to do with two concerns: one, concerns about paternalism (although sometimes this is only implicit) and, two, concern about the coherence of asymmetry of consent and refusal that is said to be part of RS-DMC. Finally, the most notable feature of the counterarguments defending RS-DMC (aside from defending against the variety of charges of paternalism) is that there were a variety of responses to the asymmetry argument with differing views among defenders of RS-DMC. Given that the RS-DMC debate has perhaps focused more on the asymmetry argument than any other issue, this is an interesting finding and suggests that further research is needed.

In addition to the issue of asymmetry, future work should focus more on the distinction between the substantive and epistemic views of RS-DMC. It is curious that those critical of RS-DMC often permit the incorporation of risk into DMC assessment through the epistemic view, rather than arguing risk should be entirely irrelevant. Thus, the intuitive appeal of risk-sensitivity carries a significant weight even among interlocutors who disagree with the substantive view of RS-DMC.

Additionally, it is notable that no publications critical of risk-sensitive DMC assessment targeted the epistemic view. However, whether there is a practical difference between the epistemic vs. substantive view of RS-DMC is disputed (Wicclair, 1999; Wilks, 1999; Parker, 2004). Given how differently each view is treated in the debate, clarification of the precise differences between the views would be valuable.

Both construction and application of codes require judgment and interpretation. For example, some codes have significant conceptual overlap—e.g., “Does having DMC imply having decisional authority” and “Two-step vs. One-step”—but we felt it was important to track both codes separately to capture their nuances, particularly because, in order to minimize bias, we coded conservatively and only coded a reason when it was explicit. Our inclusion of books in addition to our systematic search of articles is a further limitation, as a systematic search of books was not possible. However, it seems unlikely that this led to our missing any major arguments or reasons for or against RS-DMC. Only publications written in English were included in our search, and most of the publications included are from the US or UK. Literature in other languages or published in other countries may provide different perspectives on RS-DMC. For example, the emphasis on autonomy may vary among different cultures (Lepping and Raveesh, 2014), and this could also affect views regarding RS-DMC.