Assessing autonomy in psychiatric practice: a narrative literature review

Abstract

This article’s “Key Question” is: Does enforcing treatment restrict a person’s autonomy, or is their mental illness their source of autonomy restriction with treatment seeking its restoration? While autonomy assessments occur in manifold contexts, distinguishing autonomy-undermining from autonomy-promoting influences can be difficult. This narrative literature review aims to conceptualise autonomy in a clinically useful way. Its construction is guided by the Scale for the Assessment of Narrative Review Articles and the Narrative Overview Rating Scale. 126 journal articles, 11 peer-reviewed encyclopaedia articles, and 2 classic bioethical book chapters are reviewed. Autonomy is defined as “self-governance”. Clinical and ethical utility lies in assessing autonomy’s dimensions of “self” and conditions of “governance”. These are: personal identity, liberty, relatedness, agency, decisional capacity, coherence, desire-orders, authenticity and temporality, further classifiable as procedural or substantive accounts of autonomy. Paternalism may not always conflict with autonomy. The Key Question is an ethical and clinical prism through which these various refractions of autonomy are examined and rendered usable. Autonomy ambiguities and conflicts are identified, along with contexts and patient presentations which may recommend that their autonomy assessment and Key Question answer lean towards one direction or the other, depending on the patient’s unique evolving formulation. Through synthesising the broad interdisciplinary autonomy literature and its Key Question applications, this twelve-factor framework is offered to assess and enhance patient autonomy. Ultimately, it is hoped that in applying this framework to their own lived experience, people with mental illness might be empowered to expand awareness of their self and self-sovereignty.

Introduction

Objectives

Does enforcing treatment restrict a person’s autonomy, or is their mental illness their source of autonomy restriction with treatment seeking its restoration? This question is fundamental to psychiatry and is this review’s “Key Question”. It can be encountered by clinicians considering using a Mental Health Act to enforce treatment, tribunals reviewing cases, and people with lived experience of mental illness. Indeed, enforced treatment is a common occurrence in clinical medical practice, particularly psychiatry, and its ethics remain one of the most controversial issues [1–3]. Yet, autonomy assessments frequently lack consensus [1–3].

People with mental illness do not always refuse treatment, nor is treatment always necessary. Autonomy is not the only ethical principle involved [1–4]. Alongside this, the Key Question recognises that autonomy-promoting and autonomy-reducing influences can be difficult to discern when a person with mental illness refuses treatment. However, this ambiguity is often ignored, with literature reviews demonstrating that many articles regard autonomy as a singular absolute criterion which is always antithetical to enforced treatment [1, 2]. In contrast, other research recognises that “the notion of autonomy is complex and diversely understood” [1], “the relationship between autonomy and coercion is complex and challenging” and “the boundaries between coercion and autonomy might be blurred. The mental illness may itself limit one” [3]. This is typified in reviews reflecting that enforced treatment may not violate and indeed may promote or recover autonomy [1–3], and that respecting the patient’s autonomy may reduce the need for enforced treatment [3]. This literature review will illustrate how these conflicting perspectives of autonomy may arise, explicitly delineate autonomy ambiguities, and elaborate potential nexuses between autonomy and enforced treatment. “Rather than seeking to reach a single answer, ethics requires a posture of constant questioning, where multiple perspectives and ambiguities are embraced” [2].

Despite the potential ambiguity in assessing it, “respect for autonomy is no mere ideal in healthcare; it is a professional obligation” [4, 5]. There are a number of relatively recent reviews addressing ethical challenges when considering enforced treatment in mental healthcare [1–3]. Yet, while all recognise the importance of autonomy, none offer how to assess it [1–3]. Many of the dimensions which this literature review will reveal are mentioned; however rarely is their relationship to autonomy defined or discussed, and nowhere are they systematised [1–3]. At the same time, their use of these concepts reveals their clinical relevance [1–3]. Notably, other autonomy concepts which are well established in the applied bioethics literature are absent in mental health research regarding enforced treatment [1–3]. Hence, the range of these concepts is poorly understood by many for whom they are most relevant. Articles call for further research to clarify key concepts, provide unifying terminology, ethical analysis, and applicable evaluation tools [1], along with making implicit ethics explicit [3], to improve the existing clinical context of confusion and misinterpretation [1, 3]. This review answers these calls and attempts to methodically address these gaps.

Research demonstrates that some mental health staff do not experience enforced treatment as ethically problematic and wonders if this may indicate a deficit of ethical vocabulary [3]. Other clinicians considering or enacting enforced treatment may experience moral doubt or distress [2, 3]. Moral doubt is uncertainty regarding the ethical choice [2]. In moral distress, one feels pain, anguish or discomfort in an ethical dilemma where the moral choice is also perceived as wrongdoing [3]. For clinicians to justify enforced treatment “it is crucial to be able to identify all relevant ethical elements needed for the assessment” [1]. “Prior knowledge of existing arguments, conceptual nuances, and controversies” may create space for enhanced ethical capability [3]. Expanding clinician awareness and ethical discernment may be applied to moral dilemmas [2, 3]. “When we ignore fundamental elements in the evaluation, we can expect strong moral disagreements between those involved (patients, caregivers, relatives), as well as risks of abuse of power and errors in weighing the benefits and risks of the measure” [1]. Many studies discuss the central importance of dialogue with patients, human contact, and involving them in all stages of treatment, in influencing mutual (staff and patient) perception of enforced treatment and patient incentive to engage in treatment voluntarily [2, 3]. This review will collate autonomy nomenclature, providing a refined vocabulary to facilitate these conversations and deliberations.

Ultimately, this review aims to conceptualise autonomy in a clinically useful way. Specifically, its intent is to clarify autonomy ontologically, theoretically, linguistically, and semiotically, to assist its translation into psychiatric practice and the lived experience of patients. At the outset, it was hypothesised that autonomy can be assessed as a multi dimensional construct. These dimensions were sought to be identified through a narrative literature review of key academic materials. This review extracts and applies these concepts to the Key Question. The aim is not to offer an exhaustive interpretation but to explore their application to psychiatric disorders and lived experience. The Key Question is used as an ethical and clinical prism through which the various refractions of autonomy may be examined and rendered usable. Rather than concretely answering the Key Question one way or the other, this article purposely approaches with a neutral framing. This is to invite readers to deeply explore their own perspectives untethered from potential author bias. It is also to assist in identification of any potential variables or contexts in which one might consider that the patient’s autonomy assessment and Key Question answer may balance in a different direction. Hence, both sides of the debate and spectrum will be represented if applicable. The hope is to offer clinicians and patients tools to carry out their own independent and mutual assessment of the patient’s autonomy by creating awareness of the range of possibilities and sources of potential conflict. This review will synthesise all this material in its Conclusion, to offer a framework to understand autonomy, and consider patient presentations which may recommend that their autonomy assessment and Key Question answer appropriately lean towards a specific direction, depending on the patient’s unique evolving formulation. This may guide autonomy assessments of greater clarity, nuance and rigour.

The separate issue of autonomy and restraint use warrants a review in its own right, such as Al-Maraira and Hayajneh have published [6], and it will not be specifically addressed here. This review focuses on determining autonomy with respect to the broad issues of mental illness and enforced treatment. As far as the authors are aware, no reviews have been published directly addressing this central issue.

While this is an academic resource that invites revisitation, it is “systematic reflection on real cases [that] promote ethical awareness” [3]. Readers are encouraged to apply the autonomy dimensions to their own experiences, to encountered patients who bring the concepts alive, and to begin familiarisation with its everyday use. This will pave the pathway towards conscious and unconscious competence [7], and ethical and emotional resonance.

Background

“Autonomy” derives from the Greek autos meaning “self” and nomos meaning “law” or “governance” [4]. Initially referring to city-states, autonomy also denotes the self-governance of individuals [4].

Autonomous personhood is a global condition, distinguishable from being locally autonomous relative to a desire, decision, or action [8]. Basic autonomy is the minimal status of having the responsibility and ability to choose for oneself [8]. All adults are presumed to have basic autonomy unless challenged by constricting conditions or pathologies [8–10].

Respect for autonomy in biomedical ethics recognises the patient’s self-sovereignty and self-awareness of what is best for themselves, whereas their doctor may only be aware of what is best for their health [2, 11]. With compromised patient autonomy, clinicians are tasked with its restoration, or turning to a surrogate decision-maker [4, 12].

Self-governance in autonomy concerns two vantage points: that of the governing authority and the governed [13]. It is contentious whether mental illness may usurp one’s governing authority such that one is governed, not by one’s elected authority, but rather by a coup dictatorship. With one’s psyche hosting all the relevant powers, it can be challenging to distinguish any tyrannising powers from one’s own [13].

Methods

Narrative literature review guidelines

Two rating scales guided the construction of this narrative literature review – the Scale for the Assessment of Narrative Review Articles, i.e. SANRA [14], and the Narrative Overview Rating Scale, i.e. NORS [15].

Search strategy

Multiple forms of academic materials were searched to provide a depth and breadth of concepts and empirical information pertinent to the Key Question [14, 15]. Journal articles were sourced from the PhilPapers database [16]. A search was conducted in August 2022 in the PhilPapers section of Autonomy in Applied Ethics for:

• Autonomy AND;

• Psychiatry OR Psychology OR Mental (incorporating mental health and illness).

Inclusion and exclusion criteria are described in Table 1.

Table 1.

Inclusion and exclusion criteria

Inclusion criteria	Exclusion criteria
A. Peer-reviewed English-language journal article	1. The format was not a journal article
B. Autonomy was discussed	2. Autonomy was not a central issue
C. Dimensions or conditions of autonomy are discussed	3. The content was outside the scope of this narrative literature review
D. Autonomy concepts apply to the Key Question	4. Duplicates formerly retrieved were excluded

Academic compendiums were then reviewed, including the classic textbook of biomedical ethics by Beauchamp and Childress and two peer-reviewed encyclopaedias of philosophy:

• Principles of Biomedical Ethics [4, 17]

• Stanford Encyclopaedia of Philosophy [8, 9, 13]

• Internet Encyclopaedia of Philosophy [5]

Themes of autonomy identified were then reviewed in both encyclopaedias: Personal identity [18, 19], Liberty [20], Agency [21], Decisional capacity [22], Authenticity [23], and Paternalism [24].

Screening and data collection

Titles and abstracts from the PhilPapers searches were read. The journal article was retrieved and reviewed when Table 1 inclusion criteria A and either B or C were met, and criterion D was suggested as possible within the article body.

Text selection

Along with the encyclopaedia and bioethical textbook chapters, journal articles were included in the Discussion when all inclusion criteria were fulfilled.

Data synthesis

Autonomy dimensions were extracted as thematic headings, informing a description of critical concepts and subdimensions [15, 25, 26]. They were then narratively integrated through application to the Key Question [15, 26].

Results

The methodology results regarding journal articles are summarised in Fig. 1. 412 journals articles were screened. A total of 126 journal articles (of which 72 were discussed), 11 peer-reviewed encyclopaedia articles, and 2 bioethical book chapters were reviewed.

Fig. 1.

Literature search, screening and selection strategy flow chart

Search results which did not meet the format criteria for retrieval included books, theses and dissertations. Health-related database studies excluded for being inapplicable to the Key Question featured autonomy related to: euthanasia, physical illness, organ donation, genetic testing, intellectual disability, children, reproductive health, research participation, and neurosurgery.

Discussion: autonomy dimensions

Procedural vs. substantive

A prominent distinction is between value-neutral procedural and value-laden substantive accounts of autonomy (Fig. 2) [5].

Fig. 2.

Procedural vs. substantive conditions of autonomy [5, 9]

Procedural autonomy is the capacity to critically consider one’s motivations and beliefs and regulate one’s preferences accordingly [9]. As long as the specific self-reflective process is achieved, the agent is considered autonomous regardless of the values they employ [9]. Procedural conceptions are criticised for disregarding the impacts of manipulation or limited options on agents’ motivations, or for spuriously conceiving value-neutrality [9, 12, 27–29].

Substantive autonomy is value-laden, subject to normative constraints suggesting that impermissible preferences render an agent non-autonomous [9, 30]. A substantive account might claim that one cannot autonomously choose enslavement [9, 28]. Substantive conceptions are criticised for collapsing “self-rule” into “right-rule”, as though others can know what is right for oneself [9].

This introduction of procedural and substantive autonomy prefigures that each subsequent autonomy dimension which this literature review discusses may be interpreted through either of these lenses.

Key question application

Procedural and substantive autonomy may conflict. Privileging either form may be deemed paternalistic. In proceduralist accounts, a person with mental illness would be assessed as autonomous if demonstrating deliberation using the procedure of critical endorsement, in which case enforcing treatment would restrict their autonomy [9, 31–33]. People with addiction or anorexia nervosa (AN) could demonstrate procedural autonomy in preferencing substances or thinness over other values [34–37]. Whereas if mental illness impaired their procedural autonomy, enforced treatment could restore it.

Others suggest that mental illness is value-laden, creating pathological values – in AN, malnourishing thinness, in major depressive disorder (MDD), suicide, and in addiction, substance misuse – displacing formerly salient values and undermining the bearer’s substantive autonomy [12, 29, 38–42]. However, critics of this approach observe that people can value harmful things, regardless of mental illness [22]. “Pathological values” are further contested as a misnomer, as while harmful to an agent along one metric like health, these values may be positively perceived along other metrics [41]. In AN, thinness may be associated with beliefs of self-control, aesthetic ideals, and admirability [12, 34]. It is possible that a person autonomously prefers mental illness, which partially serves as a maladaptive coping mechanism, e.g. disorders of eating, addiction, and obsessive-compulsion, with treatment enforcement undermining their autonomy [33, 34, 43, 44].

Personal identity

The “self” of self-governance might be clarified as a “person” and furthermore as one’s “personal identity”. Questions of personal identity include Characterisation and Persistence [18].

The Characterisation Question asks, “Who am I?” [18]. Personal identity refers to properties one feels a special attachment to and ownership of, defining oneself in some essential sense [18]. It is contingent and changeable [18]. Two people might possess the same properties and yet identify with them differently [18].

The Persistence Question asks, “What is required for a person to persist over time?” [18]. If someone’s identity changes, the Characterisation Question responds that they are not the same (kind of) person [18]. However, they still exist, such that the Persistence Question recognises them as the same person, persisting through time [18]. The Persistence Question has been particularly applied to dementia and vegetative states [11, 18, 45].

Key question application

It can be stated that the person with mental illness, and their healthier version, is the same person existing in different states on a continuum. Quite often, a person’s identity resonates as their present state, whether that is one of illness or health. The Key Question could privilege either version. Assessing their mental illness as autonomy-undermining and treatment as autonomy-promoting privileges their health identity. Whereas assessing enforced treatment as autonomy-undermining respects their ill-health identity. This might particularly occur with treatment-resistance when the health identity cannot be restored.

Identifications with ill-health, i.e. “egosyntonicity”, have been described by people with many conditions, particularly those featuring identity disturbance, where illness can provide a sense of identity [34, 46–48]. Relinquishing illness identity may risk turmoil, e.g. Erik Erikson’s “role confusion” [34]. In a personal identity study, AN was “both experienced as a distinct, alien entity” and “a part of themselves”, concluding that from an identity vantage having AN was not obviously worse than being free of it [34].

In contrast, a study of people with Bipolar Affective Disorder (BPAD) quoted one person stating: “When we have an episode of high or low mood, we are not ourselves” [48]. Similarly, Weiner described their schizoaffective disorder: "We experience this as coming, yes, from within us, but not ultimately of us" [49]. In case law, courts have debated if a person with schizophrenia could competently prefer homelessness, even when clearly distinct from their premorbid high-functioning preferences [42]. It is argued that mental illness, through pathological biopsychosocial mechanisms, could “sever” one’s “character” and supplant identity-defining features [42].

Liberty

Beauchamp and Childress regarded liberty, analogous to voluntariness, as an essential condition of autonomy [4]. They defined liberty as independence from controlling influences, from sources external or internal to the person [4].

Isaiah Berlin distinguished negative liberty as “freedom from” constraint by others, and positive liberty as “freedom to” be or become (Fig. 3) [20, 50].

Fig. 3.

Liberty subdimensions of autonomy, conceived by Isaiah Berlin [20, 50]

Negative liberty reflects an absence (e.g. of interference), while positive liberty requires a presence (e.g. self-mastery) [20, 50]. Positive liberty holds the sources and nature of an individual’s motivations and beliefs as pertinent in establishing their freedom, while negative liberty regards such questions as violating the person’s “dignity or integrity” [20, 31].

Positive liberty holds a paradoxical danger of authoritarianism [20, 50]. For example, a conception of positive liberty might regard freedom as being one’s “higher rational self” rather than a “slave” to one’s passions [20, 50]. If some individuals consider themselves more rational than others, they might contemplate “liberating” people by forcing them to realise their rational “true” self [20, 50]. Positive liberty defends against this criticism by proposing that freedom necessitates “capabilities” and their lack serves as a constraint [20, 27].

Gerald MacCallum clarified freedom as the agent, being free from something, to be or become something else (Fig. 4) [20, 51]. Negative and positive liberty extend these individual variables differently [20, 51].

Fig. 4.

An application of Gerald MacCallum’s concept of freedom [20, 51]

Key question application

Enforcing treatment would restrict a person’s autonomy by constraining their negative liberty to refuse treatment and be mentally ill. This is encapsulated in: “Even when I’m mad, I’m still a human and have the right to make decisions, even if they are bad ones” [48].

Whereas, assessing the person’s mental illness as constraining their positive liberty to be healthy with all the potential that entails, might regard enforcing treatment as autonomy-promoting. Where the person’s mental illness influences them to be suicidal or refuse life-sustaining treatment, it constrains their positive liberty to live, as seen possible with MDD [39, 42, 52] and AN [39, 53]. Likewise, Weiner felt “oppressed from within” by their schizoaffective disorder [49].

The complex disorder of addiction has been regarded as a contextually irresistible compulsion, while others deny that it constrains voluntariness [36, 37, 39, 54, 55].

Relational

“Relational autonomy” distinguishes from atomistic conceptions of “self” (Fig. 5a) by recognising one’s interpersonal embedding (Fig. 5b) [4, 5, 9, 28, 56, 57]. It claims that autonomy does not require self-sufficiency [4, 5, 9, 28, 56, 58–61].

Fig. 5.

Relational autonomy (B) distinguished from atomisation (A)

Relational autonomy includes three forms of interdependence: developmental, contextual and compositional (Fig. 5b) [62]. Developmental nurturance and attachment security impact the formation of self-capacities [62]. One’s societal context either enriches or oppresses autonomy [28, 30, 62, 63]. Culture provides compositional materials to formulate oneself as “I” e.g. language, ideas, and values [62].

Feminist literature debates whether specific responses to oppression illustrate diminished autonomy [9, 27, 28, 30]. Rational choice theory discusses “adaptive preference formation”, where preferences are unconsciously adjusted to oppressive circumstances to avoid the cognitive dissonance of desiring what one cannot have [9, 27]. “Deformed desire” occurs when the oppressed person desires that which oppresses them [9]. Agents may also consciously adopt oppression that harms them as a societal requirement, e.g. genital cutting [9]. “Preference-deformation” in constraining conditions that remove “de facto power and authority over choices and actions” is considered non-autonomous [9]. Others argue that agents may autonomously accept oppressive conditions by prioritising counterfactual values [9, 64].

Key question application

Enforcing treatment reflects restricted relational autonomy in a person subjected to societal, legal and healthcare (relational) contexts which privilege their healthier state, when the person prefers their mentally ill state. Yet, enforced treatment may also enhance their relational autonomy if the person’s mental illness impairs their relationships and roles, as seen possible with schizophrenia, AN and addiction [34, 39, 42, 53].

AN and body dysmorphic disorder with muscle dysmorphia, where the person is respectively preoccupied with excessive thinness or musculature, can be associated with perceived societal standards of beauty. There is debate whether internalisation of intersubjective norms, such that appearance becomes tied to self-worth, reflects adaptive autonomy or subjugation [9, 28, 30].

Agency

Beauchamp and Childress regarded agency as another condition that is essential for autonomy [4]. The standard conception of an agent is a being with the capacity for intentional action [21]. Agency denotes exercising this capacity [21]. Unintentional actions and undesired outcomes can derive from intentional actions, thus still constituting agency [4, 21].

Standard or central agency is distinguished from more advanced agencies, e.g. “self-controlled, autonomous, and free agency” (Fig. 6) [21]. It is also distinguished from more basic agencies that do not ascribe “representational mental states” e.g. intuitive action without conscious thought [21].

Fig. 6.

Kinds of agency [21]

Research demonstrates that two mental processing systems underly agency: one is “automatic, effortless, and heuristics-based” and the other is “conscious, deliberate, and rule-based” [21]. Specifying these two types of agency as pre-reflective and reflective, Schlimme phenomenologically described “lived autonomy” (Table 2) [65].

Table 2.

Types of “lived autonomy” conceived by Jann Schlimme [65]

Lived autonomy	Experienced through
Pre-reflective	Everyday actions
Reflective	Producing well-founded reasoning for one’s actions or intentions

Waller proposed that psychological requirements of autonomy are an internal locus of control and a sense of self-efficacy [66, 67]. On a continuum, actions have a perceived locus of causality that is internal or external [56]. Agents with an internal locus of control believe their existence is self-directed, while those with an external locus believe that events are not under their influence [67].

Key question application

An agent with mental illness has agency in intentionally refusing treatment, even if that unintentionally maintains their mental illness. Enforcing treatment would undermine their autonomy by constraining their standard agency and internal locus of control. However, if mental illness contributes to the agent experiencing unintended outcomes, reducing their sense of self-efficacy and free agency, enforcing treatment might be autonomy-promoting. An internal locus of control is generally salutary for patients, but that effect can be reversed without a sense of self-efficacy, or with a loss of situational control [67]. Healthcare environments are recommended to encourage autonomy in as many actions as possible, addressing what the agent can control, and fostering self-efficacy [56, 67, 68].

Enforced treatment might also be autonomy-promoting if mental illness restricts the person’s agency by pathologically limiting their intention formation and range. The “experiential workspace” of depressed persons can be perceived as having a restricted scope of actions, pre-reflectively valuing actions as onerous while devaluing their abilities and motivation [65]. Their “lived autonomy” might consequently be impaired though “immediately pre-reflectively or retrospectively reflectively” perceiving themselves as helpless or unworthy [65].

Decisional capacity

There is consensus that autonomy requires decisional capacity, for how could one self-govern without legitimately deciding? [12]. While their definitions are disputed, “decisional capacity” and “competency” are terms used interchangeably in medical practice [4, 22]. All adults are presumed to be competent to consent to or refuse treatment unless contrary evidence arises [4, 12, 69]. “Decision-relativity” refers to capacity being decision, time and situation specific [4, 22, 70]. Numerous instruments assess treatment competency, with the MacCAT-T being most widely used [12, 22, 32, 71]. Figure 7 demonstrates Appelbaum and Grisso’s established four-abilities capacity criteria [32] which are normatively used in medical practice and additional proposed decisional capacity criteria [22, 34, 44, 71, 72].

Fig. 7.

Appelbaum and Grisso’s [32] most influential treatment decisional capacity criteria (top) with other proposed criteria in the wider literature (bottom) [17, 29, 39, 66, 67]

The four-abilities capacity assessment is established as procedural [22, 32, 73]. However, others maintain that rationality is substantive [22, 40, 74, 75]. For example, delusions are considered a failure of appreciation within the four-abilities capacity assessment [4, 22, 32]. It has been countered though, that patients meet procedural rationality with delusional beliefs, given that their reasoning is based on them and is as internally consistent as non-delusional people [12, 29]. Epistemic constraints, e.g. prohibiting “reality distortion”, are argued to reflect a substantive rationality standard, with the beliefs and values themselves distorting the decision-making process rather than the person’s procedural thinking [12, 27, 40, 76].

Key question application

In a study investigating the views of 115 people who received enforced treatment upon losing competency, one month post-treatment, 83% of individuals with restored capacity gave retrospective approval for their enforced treatment, compared with 41% of people who provided retrospective approval from those who did not regain capacity [77]. This suggests that if a patient’s decisional capacity is restored with enforced treatment, they are most likely to retrospectively consent to that enforced treatment, however a small proportion of these patients will not do so. It also suggests that even if the patient’s decisional capacity is not restored with enforced treatment, a proportion (41% in this study) will retrospectively consent to treatment [77]. While the consent of an incapacitated patient may not be ethically utilised, it may nonetheless be important for their felt sense of autonomy.

If a person retains the standard four-abilities of decisional capacity, they may be assessed as autonomous to refuse treatment. Whereas, if mental illness impairs their decisional capacity, enforcing treatment might be autonomy-restoring [78]. The latter argument has been debated regarding covert treatment to improve insight in severe schizophrenia [79, 80].

It might be argued that the person’s mental illness creates pathologically evaluated, affectively distorted, or identity-deviating decisions, such that enforced treatment might be autonomy-promoting. Critics have suggested that the standard four-abilities do not sufficiently consider these impacts upon decision-making [9, 12, 22, 34, 35, 52, 69, 74]. Some authors have argued that despite patients with MDD and AN being procedurally competent, their treatment refusal due to substantive factors including indifference to recovery or valuing illness features, reflects premorbid departure, which should not be regarded as competency [22, 35, 52, 71]. Grisso and Appelbaum responded that appreciation testing captures values based on reality distortion and that these patients would not have met procedural competency [76].

Coherence

Coherence is an identity-based autonomy condition [9]. In Coherentist autonomy, an agent governs their action only if their motivation coheres with some ultimate mental state [13]. The mental state varies with accounts, e.g. highest-order desires, evaluative judgement, long-term plans, cares, character traits or most integrated psychological states [13, 33, 52, 54, 75]. These mental states can be autonomously changed; however, an action is considered unattributable to an agent if, even while performing it, they occupy some ultimate vantage repudiating their action [13].

Coherentist conceptions are criticised for failing to account for a person being compelled to endorse their supposedly ultimate motivation [13]. Yet one may autonomously be moved by cares “that are not only irresistible but so integral to one’s identity that one could not possibly will to resist them” [13, 54]. Describing this experience of volitional necessity, Luther famously declared, “Here I stand, I can do no other” [54]. Distinguishable from an alien compulsion, “possession” by caring can render one “liberated” [54].

Key question application

Coherentist autonomy conceptions might question the person’s treatment refusal because it does not cohere with their highest-order desires, values, long-term plans, cares or character; thus, assessing that it reflects, not the person’s choice, but one compelled by mental illness. However, the person could deny impaired Coherentist autonomy, arguing that they have autonomously changed their preferences with regards to these motivation sources. It has been proposed that in mood disorders, when behaviour and evaluative commitments do not cohere with a person’s past range, objectivity regarding their autonomous preferences may be sought through obtaining collateral history or through enforced restitution of their mental health [40, 52].

Sometimes, it may not be possible to enforce treatment for a sufficient duration such that a person attains or sustains mental health, e.g. the person with schizophrenia who ceases their clozapine. There may come a point where enforcing treatment only restricts their Coherentist autonomy because they “could not possibly will to resist” their mental illness [54].

However, ethical clarity of the Coherentist framing of autonomy may at times be questioned [13]. Take for example, the person with AN who develops their illness while young prior to the formation of values separable from their illness [22, 34, 35]. They may have internalised controlled dietary consumption as a “volitional necessity”, and as a reflection of their own cares, plans and character, rather than as an alien force. Indeed, these Coherentist ultimate motivations may remain consistent. It is debatable whether this reflects their Coherentist autonomy which enforced treatment would restrict, or whether enforced treatment which achieves illness remission, may allow the person to develop wholly different ultimate motivations which illness restricts [13].

Desire

Harry Frankfurt’s “hierarchy of desires” is an influential Coherentist autonomy account which is generally established as procedural (Fig. 8) [5, 12, 28, 33]. It involves the capacity to control one’s “first-order” desires through “second-order” desires [4, 5, 8, 12, 33]. Distinguishing “between first-order and second-order desires is wanting something and wanting to want it” [12, 29, 33].

Fig. 8.

Hierarchy of desires, conceived by Harry Frankfurt [33]

The hierarchy of desires model was criticised for failing to account for manipulation of an agent’s higher-order endorsement [8, 9, 12, 28, 30, 31]. Critics also propose that one can be autonomously conflicted and ambivalent [5, 9, 23, 30, 62].

Marilyn Friedman argued against restricting autonomy to a misconstrued “true self” at the hierarchical desire apex, proposing instead an integrated model of bidirectional desire-order influence (Fig. 9) [5, 30, 81]. Painful emotions could alert one to an oppressive higher-order superego, with appropriate “bottom-up” regulation [30, 81]. For example, shame stemming from punishing self-criticism could alert one to consider being gentle with oneself rather than further striving towards perfectionism.

Fig. 9.

Models of desires, where arrows indicate the direction of autonomous regulation. A Hierarchical (unidirectional) model [33] B Friedman’s bidirectional integration model [81]

Key question application

The person who refuses treatment but wishes that they could desire otherwise can be regarded as having diminished autonomy, e.g. the person with AN or addiction who wants weight loss or heroin, but wishes that they did not [12, 39]. Contrastingly, if they endorsed losing weight as their “raison d’être” or being a “willing addict” they might be considered autonomous [33, 36, 37]. However, refutations claim that this reflects their severity of illness and non-autonomy [39, 42]. Others have alternatively proposed conflicted second-order desires in addiction or AN, as not negating autonomous first-order preferencing [36, 37], as demonstrating non-autonomy [33] or as indeterminate regarding autonomy [35]. These latter interpretations suggest that desire-orders may be reconceptualised as substantive autonomy accounts by some.

Authenticity

It is debated whether autonomy represents an “authentic” self and if an authentic self exists [5, 82]. “Authentic” refers to being “of undisputed origin” or an “accurate representation” [23]. Yet what is it to truly be oneself or represent one’s self, when being oneself is inescapable? [23] The authenticity ethic proposes that there are characteristics and capacities central to one’s identity and liberty, such that their displacement threatens self-alienation [23, 30, 31].

Psychoanalyst Karen Horney traced the authentic self to the personality’s “active centre”, creative potential and talents [83]. This tradition regards authentic individuals as integrated through consciously interrelating their life’s dimensions and possible selves [83]. Contrastingly, Friedman warned that “true self” conceptions privileging an aspect of a person or what they “identified” with were mistaken in appropriating autonomy, when autonomy rightfully applies to the whole person, including their unconscious aspects [81].

Key question application

Changes in the person’s identity or decisional style arising with psychopathology can “alienate” their authentic self, as people have described experiencing with many different mental illnesses [22, 34, 42–44, 48, 49, 55, 84]. This considers the person’s healthy self as authentic and restorable with treatment [22, 72, 77, 85]. Contrastingly, in authentic treatment refusal, the person’s perspectives remain consistent. Examples include people with personality disorders, e.g. the dependent or borderline patient prolonging the sick role to elicit nurture, or the avoidant or schizoid patient finding healthcare intimacy intolerable [43, 86].

Others have raised doubt about determining authentic desires [85, 87, 88]. Clarifying the person’s usual views might be unhelpful if they developed their illness while young, prior to identity formation separable from illness-related values, as often occurs with AN [22, 34, 35]. Furthermore, there is a danger of confusing authentic changes with pathological changes [22, 72, 88]. Experiencing illness, one may make decisions that one never anticipated before becoming ill, reflecting learning rather than pathology [22, 72, 89].

An alternative stance denies a true self. The potential future self, arising out of enforced treatment, and the mentally ill self can both be considered authentic [72]. The person can be defined through their narrative self-evolution [23]. Rather than “restitution” or “chaos”, their health-illness journey can be a “quest” [44]. This stance might regard enforcing treatment as autonomy-undermining. It may also ask the person to consider that whether treatment is enforced or not, and whether they are ill or healthy, all these are part of their experiences and may be owned as authentic and autonomous.

Temporality

John Christman’s procedural historical account considers an individual’s intent or desire as autonomous only if they did not resist its formation, and would not have resisted were they aware of its arising [5, 8, 9, 28, 31]. Temporally, autonomy is governance of and by one’s synchronic (existing at the present point in time) self, and the diachronic (existing over a period of time) unity of one’s later and earlier selves [8, 9, 13, 90]. One has diachronic autonomy in exerting one’s will across time [90]. Yet, an agent may autonomously compromise their future autonomy, ranking it lower against counterfactual values [9].

Substitute decision-making requires caretakers to fulfil the patient’s prior autonomous decisions regarding treatment, i.e. their precedent autonomy [4]. However, disputes can arise about the validity and applicability of Advance Directives [4, 11, 91].

Pre-commitment or mental health Advance Directives (Fig. 10) can enable autonomous self-control and self-direction in people with episodic mental illness [48, 92–94]. Patients “take responsibility for their disorder and give form to its regulation” [92]. “Ulysses contracts” which are a type of Advance Directive classified in Fig. 10, derive from Ulysses’ story: tying himself to his ship and ordering his men not to heed his commands for release towards the Siren’s bewitching song [92, 95].

Fig. 10.

Types of pre-commitment directives stipulating conditions at the time of effect [88]

van Willigenburg and Delaere regard self-binding pre-commitments (Fig. 10) as autonomous self-control, and Ulysses contracts as the opposite, i.e. acknowledging that one may become non-autonomous with control presided over by others, pressured by one’s former self [92]. While someone might appreciatively regard it as astute management, others might experience compulsory pre-committal execution as humiliating or traumatising [92]. They suggest that Ulysses contracts offer, not self-control, but restoration of and guidance by one’s authentic identity [92].

Key question application

There are four potential temporal authorities regarding the patient’s autonomy (Table 3) [72, 93]:

• Their current self (synchronic autonomy). Yet, if the patient is currently incapacitated, what they “truly” want may be regarded as unknown [22, 72].

• Their former healthy self (precedent autonomy). In a study investigating 565 patients with BPAD, 82% endorsed Ulysses contracts, 7% were ambivalent, and 12% were rejecting [48]. Ulysses contracts may enhance precedent autonomy in most patients with episodic mental illness who regard it as empowering [48]. However, this cannot be presumed, as others fear outdated Ulysses contract-based treatment [48].

  Wrigley argued that competent persons do not have authority over decisions taking effect when they are incompetent [45]. Whereas autonomy confers revisability of choices, this is precluded in Advance Directives [45]. It is proposed that epistemically, no one can provide informed consent for circumstances beyond one’s experience, e.g. the future [11, 45]. Studies demonstrate that healthy people are poor at forecasting their preferences upon becoming ill [22, 72, 89]. However, this approach requires more than basic autonomy. Possibly, the patient recollecting episodic mental illness can provide substantially informed consent.

  A historical approach would explore how the person arrived at refusing treatment [31]. If they would have resisted mental illness development, enforcing treatment might be autonomy-promoting. If they would not have resisted becoming ill if given the chance, their treatment refusal might be autonomous. Paul Benson disagreed, arguing that it is unreasonable to expect individual resistance to internalising oppressive socialisation without societal norms changing e.g. the valuing of thinness in AN [9, 30, 34, 39].

Table 3.

Temporal autonomy [72, 93]

• Their future self (prospective and diachronic autonomy). Davis argued for parity Ulysses contracts (Fig. 10), stating that healthcare providers could justifiably enforce Ulysses contracts over competent refusal [93]. The case for keeping Ulysses to their pre-committal is that they will likely later not want the ramifications of being released, enabling Ulysses to shape their life over time [93].

  Large-scale multi-hospital studies of patients’ experiences of enforced treatment show variable results [68, 77, 85, 96–99]. Although a substantial proportion of patients retrospectively approve enforced treatment as necessary, many other patients disapprove [68, 77, 85, 96–99].

• A hypothetical self. Questioning what the patient would want now, if witnessing the events free of illness, would require evidence of reflecting the patient’s autonomy rather than paternalistic imagination.

Paternalism

Beauchamp and Childress delineated paternalism as a conflict between respect for autonomy versus beneficence or non-maleficence [17]. They rejected autonomy-based paternalism justifications as conflating beneficence with autonomy, e.g. children are controlled, not because they will retrospectively consent, but so they will have better, safer lives [17]. In contrast, Edwards proposed that paternalism could only be justified by autonomy, i.e. restoring the person’s identity which mental illness usurped, because it can be supposed that is what that person most wants, rather than for any beneficent justifications [42]. There are several classifications of paternalism (Table 4).

Table 4.

Paternalism [17, 24, 31, 72, 93]

Soft paternalism intervenes in someone’s life based on beneficence or non-maleficence when they are non-autonomous, or to determine their autonomy status [17, 24, 72]. Hard paternalism intervenes to protect or benefit an autonomous person, e.g. preventing autonomous suicide [17].

Weak paternalism interferes with the means agents select to fulfil their goals if those means will thwart those goals [24, 31]. Strong paternalism considers that agents may have erroneous goals and justifies inhibiting their attainment, e.g. enforcing the wearing of seatbelts for safety regardless of opposing preferences [24].

In Libertarian paternalism, a “nudge” changes the presentation of choices so that the healthiest option is more likely to be chosen [24]. It is “Libertarian” in preservation of the choice set and freedom of choice [24]. Its degree of paternalism depends on the influence’s transparency.

Temporal paternalism justifies that the person’s long-term autonomy is furthered by limiting their short-term autonomy [24, 93].

Key question application

Forcing a person with mental illness to have treatment is traditionally regarded as paternalism justified by non-maleficence or beneficence despite impinging on the person’s autonomy to refuse. However, several stances of paternalism presuppose not to discord with autonomy:

• Soft paternalism, where enforcing treatment is justified, as the person is non-autonomous and treatment restores their autonomy, e.g. when mental illness has impaired their competency [24, 40, 77].

• Weak paternalism, where enforcing treatment is justified as the person’s treatment refusal discords with their goals, e.g. when a person with AN refuses life-sustaining nutrition which is discordant with their aim to live [24, 53].

• Libertarian paternalism, where the patient is nudged to accept treatment by framing, e.g. offering the patient an antipsychotic medication, not for their delusions which they think are reality-based, but to feel less stressed [24].

• Temporal paternalism, where enforcing treatment is justified as a short-term restriction in service of the person’s long-term autonomy [24]. Mental illness may restrict their self-governance by shortening their lifespan or choices. Temporal paternalism presumes the person would prefer choices that earlier mental illness precluded [93]. This stance is criticised for presuming the person’s future perspective, and given the person may prioritise counterfactual values over their future autonomy [9, 17, 72].

Conclusion

Autonomy may be defined as “self-governance”. Clinical and ethical utility lies in assessing autonomy’s dimensions of “self” and conditions and capacities of “governance” (Fig. 11). These are: personal identity, liberty, relatedness, agency, decisional capacity, coherence, desire-orders, authenticity and temporality, which are further classifiable as procedural or substantive accounts of autonomy. Paternalism may not always conflict with autonomy (Table 4). This literature review’s methodology effectively revealed the dimensions of this twelve-factor framework, and that they may overlap and conflict. A person might be autonomous in some dimensions and not others, and on balance, autonomous by degrees.

Fig. 11.

This narrative literature review’s framework for assessing autonomy

In each autonomy dimension, there may be a subdimension assessing enforcing treatment as undermining or promoting autonomy, or assessing the person’s mental illness as autonomy-undermining or constitutive of the person’s self and authority to self-govern. These alternatives need not be mutually exclusive and may exist in a dialectic. Each autonomy dimension will be constitutively and contextually unique for each person. Individuals will perceive and prioritise each autonomy dimension, in isolation and compositely, differently from other people, and likely differently from themselves in different contexts and across time. This affirms that each case of considering enforced treatment necessitates a new assessment factoring in all relevant moral elements and their granted relative weights [1].

Enforcing treatment might be assessed as predominantly autonomy-promoting, if there are indications that enforced treatment reflects the person’s higher-order, precedent (former healthy) or diachronic (over time) autonomous desires. These may be suggested through their conflicted or apathetic state, their pre-commitment directives, or collateral history. They might reveal the person’s identifications with being healthy, decision-capable, or self-efficacious (having free agency), which mental illness restricts. They may also express the person wanting to be offered the potential of their positive liberty, relationships, existential coherence (values, character or most integrated psychological states), and their full range of intention formation (agency), which mental illness impairs. In these cases, it may be assessed that mental illness restricts their authentic autonomy and enforced treatment offers its restoration. The corollary is that in the absence of such indications, it may not be possible to establish this assessment. Discovering this lack may iteratively inform the creation of a pre-commitment directive to guide the fulfilment of autonomous treatment in potential future states of mental illness.

On the other hand, a person’s treatment refusal might be assessed as autonomous, if they are unconflicted and decision-capable (although the empirical validity of their capacity testing may be questioned). Their refusal might also be considered predominantly autonomous if impinging upon their negative liberty and internal locus of control (agency) is likely to be traumatic, or if it reflects their temporal and identity-based autonomy. Perhaps they did not and would not have resisted becoming mentally ill if given a chance, e.g. if mental illness serves as a preferred coping mechanism, or is interwoven with counterfactual values, culminating in volitional necessity (coherence). Their treatment refusal may cohere with their personality dynamics or predominant treatment resistance. An anti-psychiatry perspective may consider all these rationales immoral or unnecessary, by regarding the person’s synchronic (current) treatment refusal as their autonomous choice, with their mental ill-health reflecting an authentic state of their “self”. In all these cases, it may be assessed that enforced treatment would restrict the person's autonomy more than any contribution from mental illness or indeed because of the extent of their illness. However, the person's autonomy may be particularly difficult to determine when they develop mental illness while young, such that they do not experience their illness as alien, and without treatment may not develop an identity separable from it.

That either enforcing treatment or sanctioning the patient’s treatment refusal might be classed as predominantly respecting their autonomy, depending on the patient’s presentation affecting the construction of each autonomy dimension, illustrates that autonomy is not a unitary phenomenon but rather an alive principle that needs an assessment. It may be difficult to know what the person’s perspectives are, were, or are likely to be with regards to each autonomy dimension, or how to weigh conflicting dimensions. This complexity with “proper balancing” may also be found with other ethical principles considered with enforced treatment [1]. Determining and supporting the person’s autonomy will require iterative dialogic exploration throughout their mental health-illness continuum. The interim query around enforced treatment may need to be guided by other ethical principles and elements [1, 3, 4].

A potential criticism of this multi dimensional assessment of autonomy is that clinicians may have predetermined perspectives and personal preferences with regards to evaluating each autonomy dimension. Indeed, research notes that the evaluation of enforced treatment may depend on moral values prioritised by clinicians, local legislation, institutional culture, and peer opinions [1, 2]. The argument could be made that rather than clarifying the patient’s autonomy, this twelve-factor framework merely gives clinicians the language to cement their preformed answer to the Key Question. It may be observed that certain clinicians, health departments or regions incline towards the same dimensional or composite autonomy assessments. Alternatively, it may be detected that they assess one way with certain patients or mental health conditions and differently for others, or that the assessment is influenced by the patient's stage of illness, or the assessment's conducted time or season. These revelations may serve as personal, professional and systemic knowledge worth having and reflecting on further individually, with peers and as a discipline. It is also possible that in having a richer shared language of autonomy with patients, one may find one’s professional or intuitive answer to the Key Question, being modified with certain autonomy dimensions and on balance. Perhaps through dialogue, a more therapeutic consensus on autonomy may be possible [2, 3].

Authors tend to address subdimensions of autonomy while neglecting others. They may also refer to autonomy as a unitary concept when their article content reveals that they mean a specific autonomy dimension. Alternatively, authors often discuss autonomy whilst presuming readership knowledge of erudite concepts. This review synthesises the broad interdisciplinary autonomy literature, offering a multi dimensional framework that can be used to clinically assess and enhance autonomy. It provides a twelve-factor lexicon by which professional and emerging (student) mental health workers, policymakers, families and patients may collaboratively explore and support the latter’s autonomy.

It is hoped that in applying this framework to their own lived experience, people with mental illness might be empowered to expand awareness of their self and self-sovereignty. This is through giving patients language to explicitly consider their potentially polysemous and multiplicitous intrapersonal and interpersonal worlds. This may facilitate a more integrated self-knowledge, self-expression and therapeutic alliance towards co-scaffolding their autonomy. Clinicians might be experienced as companions and allies of their self-sovereignty amidst mental illness, rather than adversaries. This mutual and methodical autonomy dimension weighing has the potential to tip the balance in favour of voluntary treatment or sanctioned non-treatment. Finally, research demonstrates that patients’ perspectives of enforced treatment is contingent upon whether they “believe staff acted towards them in good faith” [3]. Even if there is disagreement and enforced treatment, an understanding facilitated through shared language, of how their treating team has tried to honour their autonomy, may seed its felt sense.

Limitations and research recommendations

This narrative review was limited by its restriction to English-language academic sources. However, the literature reviewed was without regard to geography, rendering it internationally informed and applicable. The search strategy was weighted towards finding relevant concepts in Applied Bioethics. It may be built upon through Biomedical and Psychodynamic database searching. Nevertheless, this manuscript addresses a significant gap in the literature and provides a scoping review foundation necessary to enable this further research. This study did not specifically address autonomy concerning restraint use. Further research could explore any influences of patient, illness, clinician, systemic, regional, environmental and climate factors on multi dimensional autonomy assessments.

Author contributions

TD conceived researching the Key Question, conducted this narrative literature review, read the referenced research and academic materials, and wrote the main manuscript text. EP created the tables and figures, assisted with referencing and provided constructive feedback. SG assisted with planning the literature review and structuring the manuscript, provided constructive feedback, and undertook a similarity check using Turnitin software available through the University of Adelaide. SG also provided linkage with peers for research advice, recommended journal articles regarding writing peer-reviewed narrative literature reviews, and provided advice on publishing. All authors contributed to writing the manuscript, revised it critically for intellectual content, and reviewed the manuscript.

Funding

No funding or grants were received for conducting this study.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

The corresponding author confirms reading the policies of Discover Mental Health and submitting this manuscript in accordance with those policies. All authors agree to the Discover Mental Health publication policies. The text, images and figures were created by the authors, who consent for them to be published.

Competing interests

The authors declare no competing interests.