Consultation-Liaison Case Conference: Overcoming Bias in the Differential Diagnosis of Psychosis

Abstract

We present the case of a 34-year-old Black patient with no significant psychiatric history who presented with catatonia and psychotic symptoms following a recent severe acute respiratory syndrome coronavirus 2 infection, whose diagnosis of coronavirus disease 2019 encephalitis was delayed by premature attribution of his symptoms to a primary psychiatric etiology. Top experts in the consultation-liaison field provide guidance for this commonly encountered clinical case based on their experience and a review of the available literature. Key teaching topics include the diagnosis and management of coronavirus disease 2019 encephalitis, cognitive bias, and racial bias. Specifically, this case illustrates the role of the consultation-liaison psychiatrist in identifying medical conditions that may overlap with psychiatric presentations and in advocating for marginalized patients.

CASE PRESENTATION

Dr. Joshua Franklin and Dr. Bruce Leewiwatanakul

A 34-year-old Black male master’s student with no past psychiatric or medical history, and rare cannabis use, presented to the emergency department with his partner for altered mental status after 4 days of auditory hallucinations and abnormal behaviors.

About 2 weeks prior, he presented to a different emergency department with acute change in behavior, stuttering speech, and tinnitus; he had tested positive for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) 3 days earlier. At that time, he was admitted to the hospital where he had an episode of abnormal shaking. Neurology was consulted for a presumed seizure, and magnetic resonance imaging and electroencephalogram were performed, which were unremarkable. He started levetiracetam 500 mg twice daily and was discharged with outpatient neurology follow-up. He appeared relatively well for 6 days despite intermittent adherence with his levetiracetam but then began to behave unusually, cry intermittently, and experience auditory hallucinations. These symptoms worsened over 4 days leading up to his presentation to our emergency department. Table 1 summarizes his hospital course from this point forward.

TABLE 1.

Summary of Hospital Course

Day	Examination	Data and medications	Medicine team assessment and plan	Consultant assessment and plan
ED/0	Initial examination • Verbal, agitated, episodes of body shaking • BFCRS 17 Re-evaluation after medications • Calm, drowsy, minimal speech • Unable to follow 1-step commands, poor coordination, perseveration • BFCRS 12	Data • ESR 33mm/h (elevated) • CRP 0.80mg/dL (WNL) • Vitamin B12 356pg/mL (WNL) • UDS: cannabis Medications • Haloperidol 5mg and lorazepam 2 mg IV once • Lorazepam 1 mg IV once • Lorazepam 1 mg PO q8h standing	Assessment • Most likely primary psychiatric, due to life stressors • Low likelihood of encephalitis given no infectious symptoms Plan • Start lorazepam per psychiatry • Defer neurology consult unless new infectious signs present	Psychiatry Assessment • Differential diagnosis includes COVID-19 encephalitis or other autoimmune encephalitis due to acute onset, severity, and tachycardia • Primary psychiatric disorder less likely Psychiatry Plan • Start lorazepam • Admit to medicine • Neurology consultation
HD 1	Psychiatry • BFCRS 2 • Marked cognitive slowing Neurology • Hallucinations elicited	Medications • Lorazepam 1 mg PO every 8 hours	Assessment • Patient at baseline Plan • Psychiatry to evaluate for disposition	Psychiatry Assessment and Plan • Patient not yet at baseline • High concern for neuroinflammatory etiology Neurology Assessment and Plan • First episode of psychosis: differential includes schizophrenia, structural brain lesion, encephalitis • LP
HD 2	BFCRS 0 • Cognitive slowing • Intermittent hallucinations	Medications • Lorazepam 1 mg PO every 8 hours	Assessment • Primary psychiatric disorder Plan • LP to rule out neurologic cause		Psychiatry Assessment • Likely COVID-19-related resolving encephalitis • Unlikely to be primary psychiatric
HD 3	Cognitive slowing • Intermittent hallucinations • Mild paranoia	Data CSF studies • 7–9 WBC per tube, protein 72 mg/dL • CSF culture no growth • Anti-NMDA-receptor antibody negative • VZV and HSV negative Medications • Solumedrol 1 g IV • Lorazepam 1 mg PO every 8 hours	Assessment • CSF findings represent possible encephalitis Plan • Start steroid therapy per neurology	Psychiatry Assessment and Plan • Evidence of inflammation in CSF • Presumed COVID/neuroinflammatory encephalopathy • Trial of steroids per neurology Neurology Assessment and Plan • Presumed postinfectious encephalitis • Solumedrol 1 g IV
HD 4	Cognitive slowing present but improved • Vague tactile and auditory hallucinations overnight, visual perceptual distortion	Medications • Solumedrol 1 g IV • Lorazepam 1 mg PO every 8 hours	Assessment • Presumed post-COVID-19 encephalitis Plan • Steroid therapy per neurology	Psychiatry Assessment • Neuroinflammatory COVID-19 encephalopathy Psychiatry Plan • Continue steroid therapy
HD 5	• Cognitive slowing greatly improved	Medications • Solumedrol 1 g IV • Lorazepam 1 mg PO every 8 hours	As above	As above
HD 6	• Minimal cognitive slowing	Medications • Solumedrol discontinued during the day • Lorazepam 1 mg PO every 8 hours	Assessment • Presumed post-COVID-19 encephalitis, improved Plan • Discontinue steroid therapy per neurology	Psychiatry Assessment and Plan • Neuroinflammatory COVID-19 encephalopathy • Marked improvement • Discontinue steroid therapy per neurology Neurology Assessment and Plan • Discontinue steroids
HD 7 • Symptoms resolved	Medications • Discharged with lorazepam taper	Assessment • Neuroinflammatory COVID-19 encephalopathy • At baseline Plan • Outpatient lorazepam taper per psychiatry	Psychiatry Assessment • Neuroinflammatory COVID-19 encephalopathy • At baseline Psychiatry Plan • Outpatient lorazepam taper

BFCRS = Bush-Francis Catatonia Rating Score; COVID-19 = coronavirus disease 2019; CRP = C-reactive protein; CSF = cerebrospinal fluid; ESR = erythrocyte sedimentation rate; HD = hospital day; HSV = Herpes Simplex Virus; LP = lumbar puncture; NMDA = N-Methyl-D-Aspartate; P = plan; UDS = urine drug screen; VZV = varicella zoster virus; WNL = within normal limits.

On the day of his presentation (hospital day [HD] 0), he was at a memorial service for his father who died from coronavirus disease 2019–related complications 2 years prior. His partner described finding him outside of the memorial service behaving oddly and then speaking incoherently throughout dinner. In the emergency department, he had disorganized speech, including referring to himself in third person, intermittent verbigeration (e.g., “I need to go, I need to go”), and staring. Psychiatry was consulted for a “bizarre interview with a concern for psychosis” with the presumption that he would be admitted involuntarily to a psychiatric unit for psychosis.

On initial psychiatric evaluation, he had large amplitude, rhythmic shaking of his limbs and head, a positive grasp reflex, and perseveration with the Luria test. Vitals included temperature of 98.3° F, blood pressure 163/110 mmHg, heart rate 108 bpm, respiratory rate 18, and oxygen saturation 100%. He initially responded to voice and calmed, then became agitated with increased movements, yelling, and shrieking. Due to his severe agitation with psychosis, he was given intravenous haloperidol 5 mg and lorazepam 2 mg by the emergency department team. Afterward, he could shake his head in response to questions and follow 1-step commands. After receiving these medications, his initial Bush-Francis Catatonia Rating Score was 17, and when reassessed by psychiatry several hours later, his Bush-Francis Catatonia Rating Score was 12. Given the concern for catatonia, psychiatry recommended intravenous lorazepam 1 mg every 8 hours and medical admission. Shortly after the first dose, he was significantly more alert and could follow 1- and 2-step commands. He then continued scheduled lorazepam and given his sudden onset and new psychiatric symptoms without a known etiology, psychiatry successfully advocated for his admission to medicine.

Initial psychiatry recommendations (on HD 0) included neurology consultation and a lumbar puncture (LP) due to concern for a neuroinflammatory process as the underlying etiology, though these interventions were not pursued by medicine given the absence of fever or leukocytosis. The medicine team attributed the psychosis and catatonia to a primary psychiatric condition, did not feel further workup was warranted, and advocated for immediate psychiatric disposition. On HD 1, the medicine team assessed his mental status to be back to baseline and prepared to discharge him. While his catatonia had improved (Bush-Francis Catatonia Rating Score of 2), he continued to experience auditory and visual hallucinations and had notable cognitive impairment with speech latency and inattention. He denied that the memorial service had been overly stressful. The medicine team requested that psychiatry determine his disposition. However, psychiatry was not able to make a diagnosis of primary psychiatric illness given the acute onset of symptoms without further medical workup and required that the medicine team first obtain neurology consultation and LP prior to making a recommendation for psychiatric disposition.

Neurology was consulted later on HD 1 and recommended an LP, which was not immediately pursued, given the medicine team’s continued assessment of a likely primary psychiatric disorder. On HD 2, he reported ongoing hallucinations and was noted to have increased response latency on mental status examination. After multidisciplinary discussion, the LP was performed on HD 2. It showed pleocytosis and elevated protein, consistent with SARS-CoV-2-related encephalitis. Given his known recent infection with SARS-CoV-2, this was felt to be the most likely source of his encephalitis, though confirmatory cerebrospinal fluid testing is not generally available to confirm this diagnosis. Testing for other sources of encephalitis was negative (see Table 1). After further advocacy from psychiatry and neurology, a 3-day course of intravenous methylprednisolone was started on HD 3 with significant improvement in his symptoms.

He was discharged on HD 7 with no evidence of catatonia or psychosis. He successfully completed an outpatient lorazepam taper, and when he was followed up with outpatient psychiatry, his psychosis and cognitive symptoms had resolved. He did develop significant anxiety related to illness, fear of dying, and reexperiencing of periods of his hospitalization. He was ultimately diagnosed with PTSD. His symptoms improved over a period of about 6-month treatment with sertraline, and he was discharged from outpatient clinic.

CASE CONSIDERATIONS

Work-Up of SARS-CoV-2 Encephalitis (Dr. Erica Baller)

In this case, a previously high-functioning patient was presented with altered mental status 2 weeks after testing positive for SARS-CoV-2. Although mild respiratory symptoms had been resolved by the day of presentation, the patient had ongoing and severe behavioral, cognitive, and neurological symptoms. While these symptoms could reflect an underlying primary psychiatric illness, the constellation of psychosis, disorganized behavior, catatonia, inability to follow 2-step commands, perseveration on the Luria examination, and unexplained tachycardia in a patient with no previous psychiatric history should prompt the consideration of SARS-CoV-2-related encephalitis.

Due to the avoidance of extensive neurological workups in the early parts of the coronavirus disease 2019 pandemic, data on SARS-CoV-2-related encephalitis are only recently emerging. Case reports and a growing literature suggest that mental status changes associated with acute SARS-CoV-2 are common.^1–3 In one study, altered mental status was the 6th most common presenting complaint of older adults who presented to the ED with coronavirus disease 2019, and a third of these patients had no other symptoms.⁴ As with other encephalitis, the first presenting symptom of SARS-CoV-2 encephalitis may be mood or behavioral changes, with neurologic symptoms presenting later.⁵ Multiple mechanisms have been suggested including direct invasion of the virus via the nose or hematologic spread via infection of endothelial cells.² Molecular mimicry has also been proposed, where the body generates immune cells that target neuronal proteins in the process of developing a robust immune response directed at SARS-CoV-2 specifically.⁶

Psychiatrists play a vital role in the management of patients with altered mental status in the setting of current or recent SARS-CoV-2 infection. On initial presentation, neuropsychiatric symptoms may be misattributed to a primary psychiatry disorder. Therefore, a structured assessment and differential diagnosis is crucial for constructing an accurate formulation, and directly impacts management and often, disposition. Given the overlapping features of primary psychiatric syndromes with encephalitis, eliciting subtle neurological abnormalities are of particular importance when advocating for invasive testing such an LP.

A careful clinical history is the first step in identifying SARS-CoV-2 encephalitis. Neuropsychiatric symptoms usually appear acutely, within days to weeks after the infection.^2,6 Changes in mood, sleep, thought organization, and cognition are often noted by collateral informants and represent a significant departure from baseline functioning. As opposed to the onset of primary psychotic illnesses, which tend to manifest in the teenage years and the early twenties, SARS-CoV-2–related encephalitis can occur at any age. Encephalitis symptoms can emerge concurrently with coronavirus disease 2019, after severe illness, or in otherwise asymptomatic persons.

On examination, certain findings should elicit suspicion for a neurologic etiology. Asymptomatic tachycardia is often documented, which we saw in our patient. Bedside testing of orientation, evaluating concentration, 3-object recall, and a clock drawing can quickly screen for encephalopathy. Additional assessments of frontal lobe function can also be informative. Our patient’s inability to follow 2-step commands and extensive perseveration when performing the Luria task suggested frontal lobe network dysfunction.

On neurologic assessment, providers should perform a focused motor examination, noting abnormalities in tone and assessing for cogwheel rigidity and waxy flexibility. Testing for primitive reflexes (e.g., grasp, glabellar) may provide more evidence to suggest encephalopathy. A provider should have a low threshold to formally assess for catatonia, including the commonly used Bush-Francis Catatonia Rating Score. A lorazepam challenge (typically 2 mg of IV lorazepam) should be undertaken immediately if the patient screens positive for catatonia. Though some literature defines a positive screen as the presence of 2 or more items in screening questions 1–14 of the Bush-Francis Catatonia Rating Score,⁷ there is no agreed upon threshold for screening for catatonia. Due to the underdiagnosis of the condition, any elevated suspicion should lead to the consideration of a lorazepam challenge. Once there is clinical suspicion of SARS-CoV-2 encephalitis, the patient should be medically admitted and managed primarily by or in consultation with the neurology team. Recommended tests include electroencephalogram, which often identifies generalized slowing and may also rule out seizures, magnetic resonance imaging (though a normal magnetic resonance imaging does not rule out encephalitis⁸) and LP. SARS-CoV-2 encephalitis is associated with cerebrospinal fluid results of elevated protein and mild pleocytosis, with white blood cells present but less than 10/uL.⁹

Management of SARS-CoV-2 encephalitis has so far has been extrapolated from other encephalitis treatments, including methylprednisolone 1 mg intravenous daily for 5 days, and in more extreme cases, intravenous immunoglobulins.^10,11 Case reports have suggested that patients respond rapidly to this treatment.¹¹ Lorazepam lysed catatonia symptoms quickly for our patient and he had a robust response to steroids. He was able to discharge home to family after 5 days.

Cognitive Biases in Medical Decision-Making (Dr. Samantha Zwiebel)

A psychiatrist is often responsible for the initial assessment and management of encephalitis presenting with predominantly psychiatric symptoms. We must be aware of our clinical biases when asked to evaluate young adults with new onset psychosis. Such biases in clinical reasoning can impede our differential diagnosis, as evidenced in this case.

Across medicine and all areas of repetitive decision-making, we develop advanced pattern recognition as part of “fast” or “system 1” thinking and beneficial cognitive shortcuts termed heuristics.^12-14 Fast thinking facilitates rapid diagnosis and forms the basis for our maturation as physicians.¹⁵ When heuristics go awry and become unhelpful, they are termed cognitive errors or biases.¹³ These cognitive biases are associated with higher rates of diagnostic inaccuracy as well as management and treatment errors.¹⁶ In a study by Graber et al.¹⁷ of 100 diagnostic errors in internal medicine, 74% of cases had cognitive errors.¹⁸ Because these biases are highly prevalent, providers should routinely reassess their diagnostic thought processes. Table 2 defines common cognitive errors.¹⁹ In pursuing the diagnosis for this patient, the teams demonstrated the errors of premature closure, anchoring bias, and confirmation bias.

TABLE 2.

Common cognitive errors and biases

Cognitive bias	Definition
Anchoring bias	Sticking with an initial diagnostic impression despite new information
Availability heuristic	Information that comes to mind most readily is believed to be the most relevant
Commission bias	Tendency to prefer action over inaction
Confirmation bias	Selectively pursuing and interpreting data to confirm a suspected diagnosis, including discarding non-confirmatory information
Framing effect	Decision is influenced by how the question is framed
Overconfidence	Overestimation of one’s own abilities
Omission bias	Tendency to more readily accept harm due to inaction than harm caused by taking action
Outcome bias	Judging the appropriateness of a decision based on the outcome rather than the rationale that guided the decision
Premature closure	Premature diagnostic satisfaction when additional work-up should be completed and additional diagnoses should be considered
Representation error	Tendency to be guided by prototypical features of a condition and overlook atypical scenarios or other contextual clues

Adapted from Croskerry 2003.

In this case, a likely explanation for a young adult with new onset psychosis was a primary psychiatric disorder, which was satisfactory to some of the providers on the treatment team. However, considering the possibility of an alternative diagnosis and performing a neurologic examination to expand the differential diagnosis proved critical for accurate diagnosis. At this stage of the case, some providers demonstrated the cognitive error of premature closure, which occurs when a provider has identified a diagnostic explanation and prematurely terminates the evaluation. Such termination can result in either misdiagnosis or missed diagnosis. In the study of errors by Graber et al.¹⁸ in internal medicine, premature closure was the most prevalent cognitive error. Premature closure can be difficult to avoid, as the clinician often simply fails to consider the correct diagnosis,²⁰ as occurred in this case when some providers did not consider the possibility of psychosis due to medical condition.

Another pivotal decision-making point in the case was the interpretation of the patient’s attendance at a memorial service, with some believing that this event played a critical role in triggering psychiatric symptoms. In hindsight, the patient’s arrival to the hospital from a memorial service was noncontributory. This cognitive error is an example of anchoring bias. Anchoring bias in medicine describes when the provider becomes tethered to an initial diagnostic impression, formulates the case from only that perspective, and does not consider new information. A study of internal medicine residents found that anchoring bias was most often identified by residents as the reason for their diagnostic error.²¹ Irrelevant anchors significantly affect judgment²² and are red herrings masquerading as diagnostic clues. For this case, the memorial service was an irrelevant anchor that heavily influenced the formulation of the case, even after the patient denied that the event was stressful.

The combination of premature closure and anchoring bias likely delayed performance of the LP. The decision to not pursue the LP was informed by the assumption that the patient had a primary psychiatric condition (reinforced by the irrelevant anchor of a recent stressor). When the cerebrospinal fluid results revealed a mild pleocytosis that is characteristic of SARS-CoV-2 encephalitis, there was an initial reluctance to acknowledge these results because they did not conform to the expected results of a primary psychiatric disorder. Such selective pursuit of a workup, and dismissal of unsupportive information, defines the cognitive error of confirmation bias.

For this patient, the premature closure, anchoring bias, and confirmation bias were mutually supportive and delayed management. There is limited evidence regarding how one can minimize the frequency or mitigate the effects of cognitive errors and biases. Reviews have highlighted the utility of guided reflection,²³ metacognitive training,²⁴ focused workshops,²³ and technological strategies²⁴ with modest success. We therefore recommend accepting that cognitive biases are a common pitfall to which all providers, regardless of career stage and expertise, are vulnerable, and must therefore actively seek to recognize.

The Role of Race (Dr. Adrienne Taylor)

A misdiagnosis in a minority patient may include failing to identify a diagnosis when it is present (underdiagnosis), identifying a disorder when none is present (overdiagnosis), or mistaking the diagnosis for another condition (misidentification), particularly when culturally normative behavior is mistaken for psychopathology.²⁵ As with this case, the risk of misdiagnosis is most acute when clinicians fail to elicit crucial and accurate diagnostic information because of insufficient attention to contextual factors. Historically, Black communities are diagnosed with schizophrenia-spectrum disorders at a rate that is 3 to 4 times higher than White communities.²⁶ A national epidemiology study found that Black patients had higher lifetime rates of disorders that included psychotic symptoms (15.3%) than Latino (13.6%), White (9.7%) and Asian Americans (9.6%).²⁷

To understand the connections between race, racism, racial discrimination, and psychosis, it is important to recognize how the conflation of race, ethnicity, and genetics has led to a misunderstanding that certain races are biologically inclined to have increased incidence of schizophrenia.²⁸ Race is not a biological construct but rather a sociopolitical construct that has no relevant genetic basis. The term racism refers to a system of beliefs (racial prejudices), practices (racial discrimination), and policies based on race that advantage individuals with historic power.^29-31 Race has also been shown to influence the level of empathy that individuals have toward each other with White clinicians in studies being shown to have reduced empathy toward their Black patients.^28,32,33 This reduced empathy may partially explain how disturbances in cognition in Black patients are often misinterpreted or why psychotic disorders are often over diagnosed in this population. Racial discrimination may be overt or covert and can be perpetuated by anyone, regardless of race or conscious intention.^28,30

Within the United States, most people hold culturally derived implicit biases which are attitudes or stereotypes that affect our understanding, actions, and decisions in an unconscious manner.³⁴ Our individual biases operate within larger social, cultural, and economic structures whose biased policies and practices can perpetuate systemic racism and other forms of discrimination. They can impact provider behavior, proper assessment, and treatment choices. Black individuals are often more likely to be perceived as scary, violent, unreliable, less educated, and noncommunicative.³⁵ Moreover, racial bias is reflected not only in the application of diagnostic frameworks but also in the definition of the categories themselves. Throughout the 1960s and 1970s, racial bias shaped the definition of psychosis in the United States, as psychiatric nosology shifted to address the anxieties of the Civil Rights era.³⁶ In the case highlighted here, there was likely a racial bias among treatment providers which resulted in the initial push towards psychiatric hospitalization and delay in neurologic work-up due to assumptions that the presentation was due to a primary psychotic disorder.

In addition, Black people may be misdiagnosed with schizophrenia-spectrum disorders due to presence of healthy suspicion, also known as responsive paranoia or cultural mistrust.^28,35 Healthy suspiciousness refers to “normative reactions” such as guardedness and mistrust that occur in response to discrimination.³⁷ Studies have found that Black individuals endorse higher levels of subclinical paranoia when compared to White control groups. Perceived racism predicts cultural mistrust and nonclinical paranoia on the part of the Black patient which can be viewed as a healthy and adaptive strategy for Black people.³⁸ However, given the historical presence of medical racism and biases, this suspiciousness is misunderstood by clinicians and considered to be delusional. Although the presence of encephalitis in this case confounds this phenomenon, the misinterpretation as a result of racial bias of the patient’s odd behaviors and request for discharge were seen as symptoms of primary psychosis rather than another underlying pathology.

While it is unclear when implicit biases develop during medical training, these unconscious thoughts tend be automatically activated and applied most often when people are busy, distracted, tired, and under pressure.³⁹ As physicians, one may find that the cognitive effort to assess and process a person’s individual characteristics appears to be greater than that required to quickly categorize a person into a group with specific characteristics. While these shortcuts in thinking may be useful in certain situations and may have been strengthened in medical training, providers should strive to establish genuine working relationships with their patients and deliver equitable health care. The use of cognitive strategies through stereotype replacement, individuation, and perspective taking can be useful in reducing the risk of acting out on implicit biases and challenging stereotypes.⁴⁰ There is also an opportunity to incorporate implicit racial bias training in undergraduate, graduate, and continuing medical education to undermine the effects of racial bias and discrimination in the health care environment.

CONCLUSION

Patients presenting with psychiatric complaints merit a thorough evaluation for underlying medical etiologies. Neuropsychiatric symptoms can be misattributed to primary psychotic disorders; this is particularly true, as in the case presented here, when implicit biases delay accurate diagnosis and targeted treatment. Careful assessments to reduce the risk of misdiagnosis and ineffective treatment are specifically recommended for patients with acute onset of symptoms, those without a history of psychiatric conditions, and when there are abnormal findings on neurologic examination. Once the diagnosis in this case was solidified and appropriate treatment commenced, the patient quickly improved.

Among multiple factors that may have contributed to a delay in diagnosis and appropriate treatment, racial bias was likely a key factor. Racial bias can lead to both overdiagnosis and misidentification of psychiatric presentations. Implicit and explicit cognitive biases can also shape how members of the care team interpret the patient’s presentation, in this case, misinterpreting key elements of the examination that suggested a nonpsychiatric etiology. This case shows the importance of evaluating medical etiologies of psychosis and of vigilance for the effects of cognitive and racial biases.