A case example and standard ethical analysis

Consider the following fictional case example:⁴

David, a 23-year-old Black man with psychosis, is treated in a German psychiatric hospital on a voluntary basis. During his stay, David has complained about feeling like he is being given less time and treated with less respect than other service users. Members of staff say they feel uncomfortable being alone with him and find him intimidating. One nurse states that he has difficulties connecting to him. One morning, after the dose of his antipsychotic medication has been increased, David refuses to take his medication. He says that he has not been informed of this change and asks to speak to a doctor. The present staff say that he has to wait until the next ward round. They offer him sedative medication in order to calm down. David starts to shout that he refuses to take any drugs until he has seen a doctor. The situation escalates. He jumps towards the people, trying to escape the room. The staff see a risk that he might harm them or others on leaving the room due to his anger and restrain him mechanically.

In Germany, ongoing mechanical restraint, involuntary commitment and forced medication are regulated by the mental health laws of the states and all require judicial approval. The coercive measure used most often in Germany is mechanical restraint, followed by seclusion.12 Continuous one on one care is legally required for the duration of mechanical restraint.13

We will here assume that coercion in cases of potential harm to others is sometimes ethically justified.14 The use of coercion in such cases is often justified by reference to Mill’s harm principle, which states that infringements on another person’s liberty may be permissible to prevent harm to others.14 15 Mental healthcare staff may sometimes even have a role-specific moral obligation—that is, a duty—to avert harm to third parties, for example, when one service user poses a threat to the well-being of other service users under their care.14 In such cases, the obligation to avert harm to others may override the obligation to respect service users’ autonomy.

It is generally assumed that coercion is only ethically justified when certain criteria are fulfilled.2 16 For example, Gather et al suggest that a coercive measure in mental healthcare is only justified when it is a ‘suitable means to remove the risk of harm’, meaning that it is effective at preventing that harm, when ‘there are no less restrictive alternatives’ to avert the potential harm, and when the harm caused by the intervention is ‘proportional in relation to the impact of the risk of harm’ posed by the affected person.13 These criteria are also embedded in the German legal regulation of coercive measures.17 For example, Article 5 (2) of the mental health law of the German state of Bavaria18 states that involuntary hospitalisation is only permissible ‘if the risk of harm cannot be averted by less drastic means’ (least restrictive alternative). It may also not ‘lead to a disadvantage that is recognisably disproportionate to the desired benefit’⁵ (proportionality).

How does the above case example fare with respect to these criteria? The staff believe David poses an imminent risk of physical harm to them or others. Mechanical restraint is a suitable, that is, effective, means to avert physical harm. It seems to be the least restrictive alternative to avert harm as well, especially since the alternative de-escalation strategy of offering sedative medication has already been attempted. The coercive intervention also meets the criterion of proportionality: as the staff perceive others’ physical safety to be at risk, temporary mechanical restraint appears to be a proportionate means to avert this harm.

According to this standard ethical framework, the coercive intervention thus appears justified. However, as we demonstrate in the following sections, we believe that this analysis is insufficient.

Proportionality criterion

The proportionality criterion assesses the probability and impact of the potential harm the coercive measure aims to prevent and compares it to the harm caused by the coercive intervention itself. A coercive measure is only regarded as justified if the latter harm is proportional to the former.13

In what follows, we will demonstrate that in the context of structural racism, this assessment is vulnerable to two biases: (1) an overestimation of the probability and impact of the potential harm and (2) an underestimation of the harm caused by the coercive measure to the service user.

Biases are explicit or implicit associations between members of social groups (eg, based on gender, race, ability) and certain traits (eg, intelligence, athleticism) or affectivity.19 Implicit biases are unintentional and automatic associations acquired during socialisation. Healthcare providers exhibit the same level of implicit bias as the general population, including implicit racial biases with positive attitudes towards White service users and negative attitudes towards Black people. These biases significantly affect patient–provider interactions, diagnostic processes and patient health outcomes.20 21

The case example contains several possible indicators that staff’s perception of David was influenced by implicit biases. The description of David as ‘intimidating’ and the nurse’s ‘difficulties to connect to him’ may suggest that David was perceived in line with gendered racial stereotypes.

Different biases in psychiatric settings may influence assessments of dangerousness.22 First, a German study found that men are perceived to be more aggressive than women by staff and experience coercive measures over a longer period of time. However, women were roughly as likely to harm third parties and committed graver physical assaults than male patients.23 Second, there is strong evidence for racist bias, for example, Black men are (falsely) strongly associated with being threatening and dangerous.24 25 Third, people with severe mental illness, especially with psychotic symptoms, are stereotyped as dangerous. Yet, various authors argue that it is not possible to reliably predict violent behaviour based on severe mental illness.26 27 Keating argues that the effects of these biases are exacerbated at the intersection of gendered racism with mental illness discrimination, leading to the ‘mad’ Black male subject being perceived as inherently dangerous.28

Studies have shown that the impact of bias is more likely to manifest under stressful conditions.29 In situations of possibly imminent danger, staff have to make fast decisions given the risk that non-intervention might endanger them and others. Such situations may thus be especially prone to the influence of implicit biases.

Therefore, it is likely that the described biases concerning gender, racism and mental illness led staff to perceive David as posing a greater danger than he actually did. This may have led to an overestimation of the probability and impact of the potential harm. In contrast, if it had been a White female university professor instead of a young Black man who started to shout that she wanted information about a change in her medication, jumping up and wanting to leave the room, it seems likely that staff would have interpreted the situation differently.

A second potential type of bias concerns the assessment of the coercive measure’s impact on the service user. Studies show that in the USA, Black people are still systematically undertreated for pain and that medical students and doctors have false beliefs about bodily differences between Black and White people regarding pain sensitivity.30 Even though this has not yet been empirically examined, it seems likely that providers might similarly underestimate the impact of coercive measures on Black service users.31 Additionally, the evaluation of coercion by staff depends on personal and professional proximity to service users.32 Hence, racist bias may lead to coercive measures being judged as less problematic in their case.

In fact, however, the impact of coercion might be even more severe for Black service users. Black communities have overproportionally experienced trauma and violence, which heightens the risk of retraumatisation.33 Furthermore, psychiatry has historically contributed to pathologising and oppressing Black communities,34 leading to low trust in psychiatric institutions and late engagement with services in situations of need.35 36 Therefore, experiencing coercion may lead to further loss of trust.

With respect to David, staff may have underestimated the severity of his experience of the coercive measure. The combination of both overestimating David’s dangerousness and underestimating the impact of coercion on David may have led to lowering the threshold for applying coercion. In this case, the proportionality criterion, which requires a high probability of significant harm for a coercive measure to be ethically justified, would not have been met. Therefore, the coercive intervention would not be ethically justified according to the standard framework.

We believe that in theory, the proportionality criterion is helpful for distinguishing cases of justifiable vs non-justifiable coercion. However, our analysis has demonstrated that this criterion is susceptible to bias, which is why coercion is more likely to be falsely regarded as ethically justified in the case of Black service users. To ensure that proportionality is assessed adequately in ethical analysis, we suggest supplementing the proportionality criterion by the following safeguard question: Considering the service user’s intersectional social identity, is it possible that dangerousness was overestimated OR that the impact of coercion on the service user was underestimated? If the answer is yes, the proportionality assessment is most probably problematically biased and should be revisited. The safeguard question considers the intersections of mental illness discrimination with other systems of discrimination, such as racism, cis-heteronormativity or ageism.⁶ In David’s case, intersectionality helps to identify biases at the intersection of mental illness discrimination and gendered racism. Thus, the proposed safeguard question explicitly considers the service user’s social position within structures of power. Structural competency37 or implicit bias training may enhance the competencies needed to assess the safeguard question, though more research on the effectiveness of such trainings is warranted.38 Such training may be beneficial for all those in charge of ethically evaluating coercion.

Insufficient consideration of context

While it is likely that the use of mechanical restraint in the case example did not meet the proportionality criterion, it is also possible that the providers’ judgement that David was dangerous and would cause physical harm to staff or other service users was correct, and that mechanical restraint was a proportional means of averting this harm. In that case, the coercive measure would be justified according to the standard ethical framework.

Let us now suppose, for the sake of the argument, that David was dangerous and all criteria for the ethical justification of coercion were met. Yet, even then, we believe that David may have been morally wronged by the coercive measure. This is because the existing ethical framework does not consider how restraining David is related to prior injustices he has experienced.

Empirical studies show that racialised service users report negative treatment experiences within mental healthcare, which they often trace back to their marginalised social identities.9 39–43 Users identify different discriminatory practices, such as derogatory language, dismissal of needs, not being listened to, and not being included in the treatment planning process. Furthermore, studies indicate that providers lack competencies and knowledge about racism, so that they fail to identify instances of discrimination, or trivialise and disbelieve instances of discrimination reported by service users. In the view of marginalised service users, these negative treatment experiences impact their relationship to providers and overall quality of care. This provokes negative emotional responses by service users, such as fear, anger and frustration. Discriminatory practices impede shared decision-making and individual treatment agreements, which are important for avoiding coercion.44

In our example, mechanical restraint was ultimately used because David was angry and aggressive. It seems that David’s anger was provoked by the staff’s refusal to respond to an understandable request: receiving information about a drug change. It also resulted from having been given less attention and respect than other service users during his in-patient stay. Given the empirical evidence on institutional racism within mental healthcare services, it is likely that this was due to racist discrimination David may have experienced during his stay, as is suggested by a nurse’s statement that he has trouble ‘connecting’ to him. Hence, his anger might have been an understandable reaction to the low quality of care and discriminatory treatment that he had been receiving so far. In the context of psychotic illness, anger might trigger violent behaviour,45 so that ultimately, if David was in fact dangerous, his dangerousness stands in a direct link to the experienced racist discrimination. It is likely that in the absence of racist discrimination, David would not have been angry and aggressive, and that therefore, the coercive measure would not have been applied. Racist discrimination has thus directly contributed to the situation in which David was mechanically restrained.

It is helpful, here, to refer to the concept of compounding injustice.46 According to Hellman, an actor compounds injustice if she (1) engages with a prior injustice by basing her actions on facts resulting from an injustice and (2) augments that injustice by causing additional harm to the person affected by the initial injustice.

Hellman proposes the example of a life insurance company that charges a woman subjected to intimate partner violence a higher rate because she is more likely to die in the following year. This elevated mortality risk is accurate, and it arguably does not seem morally wrong per se to charge people with a higher mortality risk higher life insurance rates. However, charging this specific woman a higher rate compounds injustice, because (1) it engages with a past injustice by using the fact that she experienced intimate partner violence as a reason to charge her a higher rate and (2) it augments the prior injustice by causing additional harm: not only has the woman suffered the consequences of the abuse and gender-based violence itself, but now the abuse will also lead to detrimental financial repercussions.

In David’s case, the mental healthcare providers (1) engage with the prior injustice by using the fact that David is angry and aggressive due to the injustice of having been treated with less care and attention as a reason for applying coercion. They (2) augment this prior injustice, since David has now not only suffered the injustice of having been provided inadequate care due to structural racism, but this inadequate care has also led to further harm caused by the coercive intervention. According to the standard ethical framework, the harm caused by mechanical restraint would be ethically justified if it is proportional to the danger posed by David. However, by being linked to the prior injustice of racist discrimination, the harm caused by the coercive intervention compounds this prior injustice and carries it into the present.

Hellman argues that we have a duty not to compound injustice, irrespective of who is responsible for the initial injustice. By engaging with and augmenting the prior injustice, the actor is partly responsible for worsening the harm caused by the initial injustice. This constitutes a significant moral reason against carrying out such actions.

Therefore, it seems that the providers in the case example have two conflicting moral obligations. On the one hand, they are morally obligated to avert harm to third parties. According to the harm principle and the discussed criteria, they are also prima facie morally permitted to do so by restraining David. On the other hand, the providers have a moral obligation not to compound injustice. If, as we have shown, restraining David compounds injustice, they have a moral obligation not to restrain David. Thus, it appears that the providers are facing a moral dilemma. These are situations in which ‘the agent is required to do each of two (or more) actions; the agent can do each of the actions; but the agent cannot do both (or all) of the actions. The agent thus seems condemned to moral failure’.47

One could respond that this is no genuine moral dilemma, because the moral obligation to avert harm to others overrides the obligation not to compound injustice.⁷ Hellman states that the duty not to compound injustice can be overridden by stronger obligations,48 and it might seem plausible that averting imminent physical harm to others, especially in the context of in-patient care, creates a strong moral obligation for mental health staff.

However, we think that the obligation not to compound injustice is similarly strong due to the significant impact of structural racism, which also includes systematic physical harm. In 2022, the average life expectancy at birth for Black people in the USA was 5 years shorter than for their White counterparts.49 Structural racism shapes health outcomes both by chronic exposure to direct discriminatory practices and by unequal distribution of opportunities, resources and risks.50 51 Racism is assumed to be a driver of psychotic, substance use and affective disorders.52 When comparing the obligation to avert harm to third parties with the obligation not to compound racism-based injustice, the structural harm caused by racism must be taken into account. Given the fatal impact of racism and considering that healthcare professionals have a special commitment towards promoting service users’ health, we think that their moral obligation not to compound injustice cannot easily be dismissed.⁸

This shows that the standard ethical criteria discussed above are insufficient for a comprehensive ethical analysis of the situation. The criteria are designed to ethically evaluate the use of coercion in narrowly defined interpersonal interactions, without taking into account how these interactions arise. Yet, as racism and other systems of oppression are structural and operate on the institutional and organisational level within mental healthcare institutions, it is insufficient to limit ethical analysis to individual actions.

This points to a more general shortcoming of the standard mental healthcare ethics approach. Rather than focusing on individual interactions, bioethics must examine how medical and social structures need to be set up to avoid discriminatory practices and to allow agents to fulfil their duty not to compound injustice. This has been pointed out by contributions from Black bioethics, which are concerned with the impact of structural systems of discrimination, such as racism, on bioethics53–56 and clinical ethics.57–60 However, this scholarship is not yet integrated into mainstream bioethics.61 In Germany, it is neither applied to mental health ethics nor incorporated into clinical guidelines on coercion.44 Our analysis may thus both raise awareness for racism within the German context, and broaden the international discourse by connecting Black bioethics and mental health ethics.