To Test Or Not To Test: Medical Work-Up of Psychosis
Current approaches relying on clinical suspicion leave many patients without correct diagnosis
A neurologist colleague once said to me: “If I ever become psychotic, I want you to do a lumbar puncture and test the CSF for every antibody that you can think of!” The reality is that if she were to receive the sort of care that the average patient with psychosis receives in the US, she’d be lucky if she even got an MRI scan.
Earlier in June 2023, Washington Post published a story about April Burrell, an institutionalized woman with psychosis and catatonia in the New York state mental health system who was diagnosed with neuropsychiatric lupus (thanks to a chance encounter with Sander Markx, the director of precision psychiatry at Columbia University) and received intensive immunotherapy treatment leading to a dramatic recovery. The story notes that her presentation was “clinically indistinguishable from schizophrenia” and there were no signs of lupus in organs other than the brain. She had demonstrated severe treatment resistance with no response to antipsychotics, mood stabilizers or electroconvulsive therapy. How many more patients like her remain undiagnosed in the mental healthcare system? The story says: “Researchers working with the New York state mental health-care system have identified about 200 patients with autoimmune diseases, some institutionalized for years, who may be helped by the discovery.”
Could this have been prevented by current standards of care?
If we look at the current recommendations around medical work-up of psychosis, we see a general lack of guidance on what sort of medical work up is necessary. A lot is left to clinical judgment and clinical suspicion. There is a basic set of investigations that almost everyone agrees on, but that list is pretty limited and by no means adequate at identifying the sort of case described in the Washington Post story.
The AACAP practice parameters, for example, state:
“The medical evaluation focuses on ruling out nonpsychiatric causes of psychosis and establishing baseline laboratory parameters for monitoring medication therapy. More extensive evaluation is indicated for atypical presentations, such as a gross deterioration in cognitive and motor abilities, focal neurologic symptoms, or delirium.
Assessments are obtained based on specific medical indications, e.g., neuroimaging studies when neurologic symptoms are present or an electroencephalogram for a clinical history suggestive of seizures. Toxicology screens are indicated for acute onset or exacerbations of psychosis when exposure to drugs of abuse cannot otherwise be ruled out. Genetic testing is indicated if there are associated dysmorphic or syndromic features. Similarly, tests to rule out specific syndromes or diseases (e.g., amino acid screens for inborn errors of metabolism, ceruloplasmin for Wilson disease, porphobilinogen for acute intermittent porphyria) are indicated for clinical presentations suggestive of the specific syndrome in question. Broad screening for rare medical conditions is not likely to be informative in individuals with psychosis who do not present with other neurologic or medical concerns.
At the time of first diagnosis, routine laboratory testing typically assesses blood counts, liver and renal functions, and metabolic parameters and thyroid functions, which provide a general medical screen and serve as baseline assessments for medication monitoring.” (my emphasis)
“For cases where a medically related cause of psychosis is of further consideration, we recommend the following panel of tests…” (my emphasis)
Hepatic function panel
Thyroid-stimulating hormone level
Serum treponemal test, such as fluorescent treponemal antibody absorption, to screen for syphilis
Vitamin B12 level
Erythrocyte sedimentation rate and antinuclear antibodies
Urinalysis to evaluate for urinary tract infection or other abnormalities, with reflex culture
Urine drug screen to evaluate for recent substance use
A 2020 review of multiple guidelines relating to the medical work-up of First Episode Psychosis (FEP) synthesized the literature using the categories of “universal” and “low yield.” Universal investigations were recommended by the majority of the guidelines, there was consensus regarding the usefulness of the investigation and/or there was a robust clinical reasoning for its use. Low yield investigations were those that few of the guidelines endorsed the investigation and/or there was a poor clinical justification for its use routinely and/or a low pre-test probability in the general population. Authors noted: “These low-yield investigations would be considered on a case-by-case basis and according to the clinical scenario.” The list is as follows.
First-episode psychosis medical work-up checklist
Physical examination, including neurological examination
Body mass index and waist circumference (BMI and WC)
Full blood count (FBC)
Electrolytes (including calcium), and renal function (EUC and CMP)
Erythrocyte sedimentation rate (ESR)
Liver function tests (LFT)
Urine drug screening (UDS)
Thyroid function tests
Anti NMDA and VGKC receptor antibody blood test (with CSF sample if positive)
Heavy metal screen
Brain imaging (MRI if available)
Chest X-ray (CXR)
The issue is that most psychiatric clinicians are just not very competent at recognizing when a test for a particular medical condition is indicated. This is especially the case if they work in a setting where their exposure to “organic” psychopathology is low. The clinical intuition of a consultation-liaison psychiatrist working at a tertiary academic medical center is going to be very different from a general psychiatrist who has worked at a state hospital for 20 years. Presumably, someone like Burrell could’ve been diagnosed if she had received a screening test for anti-nuclear antibodies (ANA) at some point, however, but no one thought that one was clinically indicated. This leads to the second point, which is that medical conditions can have atypical presentations which may clinically resemble psychiatric disorders and there may not be any obvious signs other than the psychiatric symptoms. The strategy of ordering an investigation only when clinical suspicion exists fails to diagnose many patients who should be diagnosed. Furthermore, clinicians working in resource depleted contexts are incentivized to use as few investigations as possible, and over time clinicians internalize these rationing judgments as clinical judgments.
The issue is that most psychiatric clinicians are just not very competent at recognizing when a test for a particular medical condition is indicated. This is especially the case if they work in a setting where their exposure to “organic” psychopathology is low. Furthermore, medical conditions can have atypical presentations which may clinically resemble psychiatric disorders and there may not be any obvious signs other than the psychiatric symptoms. The strategy of ordering an investigation only when clinical suspicion exists fails to diagnose many patients who should be diagnosed.
If we want to do a better job of identifying cases of psychosis due to autoimmune and infectious etiologies, we need to take a more systematic approach.
Erik Messamore, MD, PhD, Medical Director of Best Practices in Schizophrenia Treatment (BeST) Center at Northeast Ohio Medical University, recommends the following tests to search for medical causes of psychosis in the online resource Psychosis Care in Primary Care:
Complete blood count with differential
Renal function tests
Liver function tests
Fasting blood glucose
Thyroid stimulating hormone (TSH)
Thyroid peroxidase antibody
Erythrocyte sedimentation rate
Autoimmune encephalitis panel, serum (this includes: anti-NMDA receptor antibodies; anti-voltage-gated potassium channel antibodies; and anti-glutamic acid decarboxylase antibodies)
Tissue transglutaminase IgA and total IgA level
Gliadin antibodies (IgA and IgG)
Syphilis screen using fluorescent treponemal antibody (FTA) method
Drug test (urine or serum)
Brain imaging (MRI with contrast is preferred)
(His list contains two tests – thyroid peroxidase antibody and tissue transglutaminase antibody – that don’t appear in other published guidelines, to screen for Hashimoto’s thyroiditis and early celiac disease or non-celiac gluten sensitivity.)
The advantage of an approach like this is that it takes the idiosyncrasies of the “order if there is clinician suspicion” approach out of the equation, ensuring a more uniform assessment. The disadvantage is that i) such an approach may not necessarily be cost-effective, and ii) false positives/false negatives present problems of their own, usually requiring further intensive or invasive investigations.
For example, Wilson’s disease is so rare – one in 30,000 to 40,000 people worldwide – that the vast majority of Ceruloplasmin tests would be negative. Conversely, according to UpToDate: “Almost 15 percent of the population in the United States has been found to have a positive ANA of at least 1:80 by indirect immunofluorescence, but only 10 percent have a true autoimmune disorder.” In the case of pathogenic neuronal autoantibodies to diagnose autoimmune encephalitis, patients with antibodies in the serum may be negative for antibodies in the CSF. In one 2021 study, 3/71 patients with FEP with seropositive but none were CSF positive. However, other studies suggest benefits of screening. For example, in a 2018 study, 6 out of 113 patients with FEP had antineuronal antibodies. 5 received immunotherapy, which prompted resolution of psychosis in 4. For those who do have these disorders, uncommon though they might be, an accurate diagnosis will be life-altering and lifesaving.
Most clinical recommendations seem to focus on first-episode psychosis, however, there are two other populations where an emphasis on medical work-up is also necessary: i) individuals with psychotic disorders who don’t respond to standard treatments such as antipsychotics, and ii) individuals with serious mental illness who are institutionalized in psychiatric hospitals.
Most clinical recommendations seem to focus on first-episode psychosis, however, there are two other populations where an emphasis on medical work-up is also necessary: i) individuals with psychotic disorders who don’t respond to standard treatments such as antipsychotics, and ii) individuals with serious mental illness who are institutionalized in psychiatric hospitals. What currently happens is that an individual with first-episode psychosis may get a basic work-up in the emergency room and neuroimaging with a CT scan and get “medically cleared” for psychiatric admission, and then no one thinks of conducting further work-up (such as getting an MRI scan or screening for autoimmune conditions) unless there is a strong clinical reason to do so. If we see treatment-resistance and institutionalization as benchmarks necessitating further medical work-up, we can avoid this problem. We could consider a multi-step approach where medical investigations which are considered optional in the workup of first-episode psychosis are employed more aggressively in treatment resistance psychosis to uncover undiagnosed pathology.
Such recommendations need to come from authoritative sources for implementation to happen. Otherwise, clinicians would remain reluctant and insurance companies will continue to withhold reimbursement. For any change to happen, we first need to recognize that our current approach that relies heavily on clinician suspicion of an infectious, endocrine, or autoimmune pathology is failing a lot of patients with psychosis.
Psychiatry at the Margins is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.