Introduction

Psychological research has long identified explicit and implicit biases that tend to arise when humans make judgments or decisions. For example, the halo effect is when a general impression of a person, such as attractiveness, positively biases judgments on unrelated characteristics such as intellect or moral character [1].

Bias in health care is a systematic pattern of inaccuracy in social perception that can negatively influence medical decision-making [2, 3]. In 2003, the Institute of Medicine concluded that unrecognized biases against certain minority groups can affect the care offered and health outcomes for these individuals [4]. These biases are believed to be one cause of health disparities [3,4,5,6,7].

Most research on health disparities has involved racial and ethnic minorities; however, persons with disabilities are also a population with health care disparities [8]. As with racial minorities, persons with disabilities face experiences of prejudice, discrimination, and oppression [9]. The literature on health care bias with other minority groups can be instructive for investigating biases related to patients with disabilities. Unchecked biases in any population can undermine ethical caregiving.

Like other minorities, people with spinal cord injuries (SCI), an identifiable sub-group of patients with physical disabilities, have visible features (e.g., presenting in a wheelchair) that others may use in the formation of bias against them. Bias can be expressed directly (e.g., a clinician noting that “People with severe physical disabilities make me uncomfortable”), or more indirectly (the clinician makes eye contact with the caregiver instead of the patient presenting in a wheelchair). Implicit biases are typically not consciously recognized and occur despite best intentions to the contrary [6, 10]. It is the clinician’s role to accurately assess the health status of a patient and treat them effectively and with dignity.

The therapeutic relationship greatly influences clinical outcomes [11]. This relationship is undermined when bias is present, and especially if the clinician denies the possibility of bias being present. Nonetheless, the patient–clinician relationship can be restored if the clinician acknowledges the bias and attempts to correct it [7].

A core part of the patient–clinician relationship is an accurate assessment of a person’s risk of self-harm (i.e., self-injury, self-neglect) or suicide. Biases that impair the therapeutic relationship or the decision-making process would result in inaccurate perceptions by the clinician [6] that either overestimate or underestimate risk for self-harm or suicide, which likely alienates patients and harms the therapeutic alliance. Discussing uncomfortable topics can increase the chance for bias [2], therefore, it is important to look at how clinicians and patients interact in these circumstances and then determine what contributes to negative outcomes. The existential discomfort of discussing a patient’s interest in living and dying provides a rich opportunity to study occurrences of unconscious bias [2] and the potential reactions from clinicians assessing them.

Another bias found in the perception of people with SCI is that clinicians attribute lower quality of life to their paralyzed patients than the patients report for themselves [12]. Reasons for these biases are complex—and importantly, not deliberate [3, 13]. Bias can manifest during uncomfortable conversations involving death wishes or explicit suicidal thoughts with patients with SCI. Recognizing bias in the context of assessing risk for self-harm, where accuracy existentially matters, is especially important for people with disabilities, since suicidality is 3.5 times higher among this group as compared with nondisabled adults [14]. Bias can lead to an inaccurate assessment that in turn leads to faulty decision-making, damaging the therapeutic relationship, and causing further harm to people with SCI.

Case presentation

Three biases and their effects on risk assessment

There are three biases that likely impact a clinician’s decision-making regarding a patient with a disability [15] and these biases also can apply when assessing the patient’s risk for self-harm: (1) ineffectual bias, (2) fragile friendliness bias, and (3) catastrophe bias. These preconceptions are derived from a mix of paternalism, projection, low expectations, pity, and infantilization [15]. Each bias impacts clinical decision-making and harms the therapeutic alliance. We explain how each bias can affect clinical decision-making regarding diagnosis, treatment, prognosis, and prevention for patients with SCI within a case scenario: a patient with SCI is hospitalized for poor nutrition and severe pressure wounds. They spend their days sitting in a wheelchair, drinking whiskey, and watching television.

(1) Ineffectual bias

Persons with disabilities are often perceived as having low agency and diminished competence [15, 16]. Here the perceiver makes inferences from limited information, such as no longer being capable of ambulating, and applies those inferences to the person’s entire mental state. People with disabilities report lived experiences of feeling devalued by others in this specific manner related to the under-perception of agency [9]. For instance, one person with quadriplegia noted, “I am either a spectacle or invisible [in public]” (Personal Communication, July 5, 2018). Ineffectual bias in the perception of patients with SCI primarily manifests in clinical paternalism and a diminished therapeutic alliance.

(a) Diagnosis: clinicians are more likely to discount and place less emphasis on the experiences of a patient with SCI in diagnostic formulations, presuming they are not fully competent. Consequently, the quality of the diagnosis is likely reduced. Symptoms of self-destructive intentions may be discounted (e.g., biased perception is “Of course they’d drink and watch TV all day! We cannot expect them to do much of anything else.”). Unfortunately, this patient is exercising passive suicidality through self-neglect, which may be overlooked.

(b) Treatment: more conservative treatment options that “they can handle” are more likely to be made. Treatment for patients with SCIs is likely based less on patient preferences and via shared decision-making (e.g., a clinician does not discuss ways for the patient to stop drinking because those strategies are not considered “realistic” suggestions for the patient.).

(c) Prevention: the clinician’s efforts are more likely oriented toward external interventions such as the environment, caregivers, and social supports (e.g., fewer efforts to self-motivate the patient for better self-care and healthier life activities. “Maybe we can suggest to the spouse not to buy as much alcohol and perhaps work fewer hours to help them do other things than drink alcohol and watch television?”). Focus is less likely on patients with SCI to improve their own health as partners in intervention planning.

(d) Prognosis: skewed toward the pessimistic. Diminished hope from the clinician adds drain on the therapeutic relationship and the patient’s self-efficacy (e.g., “This is the best we can expect. Maybe drinking is their only source of pleasure?”) (Table 1).

Table 1 Ineffectual bias and impact for clinical decision-making of self-harm risk.
Full size table

(2) Fragile friendliness bias

There is an increased perception of the person with a disability, such as SCI, to be friendly and fragile [15]. Compared with nondisabled patients, patients with SCI are deemed to be more warm, trustworthy, and prosocial [15]. This fragility bias and saintly bias can result in either over- or underestimating risk for self-harm or suicide.

Clinical fragility bias

With the over-attribution of fragility, clinicians likely amplify the degree of suffering perceived in patients with SCI, as compared with the same symptoms reported by a person without that disability [15].

(a) Diagnosis: clinicians are likely to overestimate the severity of symptoms; overdiagnose and test (e.g., the patient states that their life is like groundhog’s day, with boring redundancy and not much to look forward to, declaring drinking as a rare source of limited pleasure in life. The physician insists that the person speak to the on-call psychiatrist and recommends a full workup on mood disorders and suicidality). This likely results in the patient not wanting to share their experiences with clinicians.

(b) Treatment: the perception of heightened fragility likely leads to more conservative treatment decisions by clinicians, because unpleasant or uncomfortable assessments or interventions will be avoided (e.g., clinicians avoid suggesting interventions with negative side effects; or avoid having direct conversations about the patient’s disinterest in living for fear of upsetting the patient further).

(c) Prevention: amidst a perception of high fragility, it is more likely that extreme or unnecessary forms of prevention are enacted (e.g., an order is placed for the patient to have 1:1 supervision throughout this admission even if their suicidality is passive without any intent or plan to harm themselves). The patient likely learns distrust for psychological or psychiatric services and may be less likely to disclose their honest thoughts in the future.

(d) Prognosis: clinicians are more likely to underestimate resilience in the patient with SCI, often with suicidal projections of hopelessness onto the patient when none are present (e.g., the clinician may unconsciously think, “This is a permanent condition with no chance of getting better. I would also feel bored and dissatisfied and think about dying too.”). With a diminished prognosis for the patient with SCI, the clinician’s decision-making is truncated in its hopefulness for possible interventions and biased toward end of life planning.

Clinical saintly bias

With the over perception of friendliness and prosociality in patients with SCI, clinicians likely underestimate the darker or dysfunctional sides of patients with SCI, including self-harm, anger problems, sexually transmitted diseases, and marital infidelity. When this quintessentially human side of life minimized, the patient–clinician interaction will be distorted, and the patient will be less likely to feel that they are “allowed to be a regular person,” as expressed by Olkin [9].

(a) Diagnosis: not wavering from the saintly script, clinicians are more likely to overlook self-destructive behaviors in their diagnostic formulations (e.g., the rehabilitation team pleasantly provides full treatments without questioning if recent behaviors were an attempt for passive suicidality).

(b)/(c) Treatment/prevention: interventions will be less likely for the same presenting problem if it violates the saintly script (e.g., patient’s thoughts regarding deliberate self-harm are more likely to result in clinician thinking, “They are such an inspiration to others and would not hurt themselves. No interventions are needed at this time.”).

(d) Prognosis: the saintly view of prognosis more often includes a happy ending, and so is likely biased toward being overly optimistic (e.g., the patient is returned to medical baseline and discharged back home with providers expecting positive health behaviors going forward. How to prevent the patient from becoming medically compromised by sitting in one spot all day drinking whiskey [passive suicide] was not addressed) (Table 2).

Table 2 Fragile friendliness bias and impact for clinical decision-making of self-harm risk.
Full size table

(3) Catastrophe bias

Clinicians likely project onto the patient with SCI the clinician’s own perceptions of catastrophe for being in the state that the patient is in, and this biases therapeutic interactions toward more perceived suffering and less hope [9]. Quality of life is likely perceived as being much lower than it really is for the patient with SCI. Suffering (hopelessness, wanting to die) is likely overestimated relative to what the person actually experiences. Catastrophe bias likely leads to clinical pessimism, more nocebo (inverse of placebo) effects, and enhanced disconnection between the patient and the clinician.

(a) Diagnosis: more severe diagnoses will likely be given for the same presenting symptoms (e.g., the clinician diagnoses substance use disorder (SUD) without a comprehensive clinical interview when a proper assessment would have ruled out SUD).

(b)/(c) Treatment/prevention: clinicians will more likely assume that the patient with SCI is in mourning or highly distressed over their disablement, and treatment/preventive efforts will be more aggressive than without the catastrophe bias. However, when the catastrophe bias goes beyond a threshold (i.e., the patient is presumed to want to die), treatment will then be less likely since the clinician will embody hopelessness in their treatment planning and “give up” on the patient sooner than they would have otherwise.

(d) Prognosis: diminished expectations for outcomes. Resiliency is underestimated and so is the chance for adaptation or growth (e.g., clinicians expect the same SUD behaviors and an unchanged evaluation on the next visit) (Table 3).

Table 3 Catastrophe bias and impact for clinical decision-making of self-harm risk.
Full size table

Discussion

All humans are susceptible to biases, including well-meaning and morally principled clinicians. Unrecognized biases toward patients with SCI can undermine ethical and effective health care delivery. Explicit and implicit biases can unintentionally affect bidirectional communication between patient and clinician and significantly hinder the therapeutic alliance. Consequences can be detrimental within a medical context when such interactions impact the clinician’s ability to accurately assess a patient’s risk of self-harm, such as their view on wanting to live or die, to have or to forgo treatments or preventative interventions. For patients with SCI, poor interpersonal experiences with their clinicians can undermine trust in their care, leading to less engagement, perhaps increased apathy about having a disinterest in living, and less adherence to healthy treatment options.

Research on biases in health care toward the physically disabled, such as patients with SCI, is scarce despite the behavioral patterns of avoidance seen in the presence of persons with physical disabilities [17]. We, therefore, call for more research attention on this topic, both for persons with SCI, and patients with disabilities more broadly.

Other interventions should target social psychological biases that impair the care for persons with SCI and the accurate assessment of their risk for self-harm. Implicit biases occur without conscious awareness and are often not in line with one’s personal beliefs [13]. Therefore, it may be helpful to conceptualize implicit bias as a “habit of mind” [10] that can be ameliorated through habit-breaking interventions.

Such interventions ought to be both educational and clinical. Education and raising consciousness are paramount [15]. As with all successful changes in behavior, educators, and clinicians must first understand their susceptibility [18]. Faculty who develop and conduct training around bias should prepare for personal and other challenges that may affect implementation in their institutions [7, 19, 20]. For example, intergroup differences such as race and gender, as well as the amount of exposure the clinician has with a certain population, can influence bias [20]. Interventions to raise awareness and improve skills for limiting bias are best when trainees are directly exposed and cooperatively interact with people who have disabilities [18], particularly when patients use wheelchairs [21, 22]. Case studies should be conducted involving actual stakeholders rather than vignettes—for example, patients with SCI like those we treat in our clinics.

All members of the disciplinary team should be involved in efforts to reduce bias because false assumptions about a person’s quality of life can change how we present treatment options and negatively influence outcomes [23]. Clinical formulations in rounds should include discussions about biases while at the same time individualizing recommendations for the specific patient [24]. This can be done by a commitment in education and practice to finding out and attending to the patient’s actual experience and mental state, rather than deriving perspectives from the heuristic of an identified category or group. Rather than attending to what clinical determinations and interventions do people with SCI require as a generic category, clinicians should ask what factors are this particular person struggling with, and what can be done to increase their health outcomes and resilience? Education efforts should begin at the start of clinical training and include all clinicians who care for individuals with SCI (doctors, physician assistants, physical therapists, occupational therapists, nurses, psychologists, technicians, etc.). It could also be potentially helpful to increase the number of clinicians with SCI and other discernable disabilities. In addition to allowing the clinical staff to better reflect the population at large and the patients they care for, such individuals exhibit significantly less bias against people with disabilities, and their presence increases opportunities to see people with SCI or other disabilities in less stereotyped and more empowering roles [25]. Health care professionals, especially those working in areas other than medical rehabilitation, would be more likely to realize that disability is a part of the human condition, and any member of the human family can acquire a disability, sometimes instantly. And undoing disability bias takes much longer than an instant.

Table 4 shows patients from our clinic who read and responded to the biases as described in this.

Table 4 Current patients with SCI read the paper and shared their thoughts about the described biases.
Full size table