New-Onset Psychosis or Anti-NMDA Receptor Encephalitis: The Antipsychotic-First Question
A young woman's sudden, severe psychosis carries atypical features that could mean her presentation isn't a first psychotic break at all — and that antipsychotic-first treatment could be the wrong move.
S.K., a 22-year-old woman, is two months into her first job out of college, a marketing role she describes to friends as "finally the real thing" — the kind of detail her mother keeps returning to when she tries to explain how sudden this has felt, because nothing about the last month sounded like a young woman with a hidden history of mental illness quietly unraveling. Ten days ago she had a low-grade fever and headache her roommate assumed was a cold. Five days ago she became convinced her coworkers were recording her conversations, stopped going to work, and began pacing the apartment at night. Two days ago she was brought to the emergency department after her mother found her disoriented and unable to say what day it was, speaking in fragmented, sometimes nonsensical sentences interrupted by brief episodes of unresponsive staring. On exam tonight, her mother also points out something she'd dismissed as nervous habit: small, repetitive movements of S.K.'s jaw and lips that come and go independent of what she's saying.
Psychosis with this abruptness, in a patient this age, with no prior psychiatric history, is a real diagnostic fork. The straightforward reading is a first psychotic break — the most common explanation for exactly this presentation, and one that should not be dismissed reflexively just because a rarer alternative is more interesting. But the specific combination in front of the team — a viral-appearing prodrome, waxing disorientation, orofacial dyskinesias, and brief unresponsive episodes that could be non-convulsive seizures — matches the early clinical progression Dalmau and colleagues first characterized in anti-NMDA receptor encephalitis (Lancet Neurology, 2008; expanded in Titulaer et al., 2013): a prodrome, then a psychiatric and behavioral stage often severe enough to prompt psychiatric admission, before more overtly neurological findings — dyskinesias, autonomic instability, decreased consciousness — become unmistakable. Psychiatrists took part in the initial assessment of 77 of Dalmau's original hundred patients — this presentation reaching a psychiatric service first is the norm, not the exception. Her vitals tonight show a heart rate swinging between 58 and 124 over twenty minutes with no clear trigger — the kind of autonomic lability that doesn't fit a purely psychiatric illness at all. A first psychotic break does not move a resting heart rate sixty-six beats in twenty minutes; whatever else is true of her tonight, that number belongs to a body, not to a mood.
Emergency department, night of admission
She is agitated enough right now to be a safety risk to herself and to staff, and I'm not comfortable holding all pharmacologic management for a workup that could take another day or more to fully result. A low-dose second-generation antipsychotic is standard, appropriate first-line management for new-onset psychosis with agitation, and delaying it isn't a neutral choice — it has its own real cost if she injures herself or someone else tonight.
I want to be clear I'm not arguing against the workup, or against the encephalitis possibility being taken seriously. I'm arguing specifically against treating "give an antipsychotic" and "pursue the encephalitis workup" as mutually exclusive tonight.
They aren't mutually exclusive in principle, but antipsychotics specifically carry a real, described risk in this exact syndrome that a routine first-break workup doesn't. The dedicated study here is Lejuste and colleagues (Neurology: Neuroimmunology & Neuroinflammation, 2016), a French cohort of 111 adults with anti-NMDA receptor encephalitis: among those given antipsychotics, roughly a third showed frank intolerance — worsened rigidity, autonomic instability, and in some cases a neuroleptic-malignant-syndrome-like picture. The mechanism is plausible rather than proven: the same NMDA receptor hypofunction driving her psychosis may also predispose to catatonic and dystonic reactions that dopamine blockade intensifies rather than settles. Her orofacial dyskinesias and swinging heart rate are exactly the atypical features the literature flags as distinguishing this from a routine first break — this isn't a rare-disease reach, it's the described clinical picture.
I'm not proposing we leave her unmedicated and agitated overnight. I'm proposing the first-line agent for her agitation shouldn't be an antipsychotic at all — benzodiazepines don't carry that specific interaction risk and are a reasonable bridge while we push the MRI, EEG, and CSF studies to be read tonight rather than tomorrow.
I don't think this needs to be antipsychotic-versus-benzodiazepine as an absolute rule either, and I'd like us to be precise about where she actually stands, because it matters for what we do next. The criteria for probable anti-NMDA receptor encephalitis (Graus et al., 2016) have three limbs, not one: rapid onset over less than three months of at least four of six major symptom groups; plus at least one abnormal supporting study — an abnormal EEG, or CSF pleocytosis or oligoclonal bands; plus reasonable exclusion of other causes.
On the first limb she is already there — psychiatric and behavioral change, disorganized speech, orofacial dyskinesias, autonomic instability, four of six before anyone argues about her staring episodes. What she does not have is the second limb, because her EEG and spinal fluid have not resulted. So I'd resist saying she meets the criteria tonight; she meets most of them, and the missing piece is sitting in a queue rather than in doubt. That is still more than enough to make antipsychotics the wrong first reach for her specifically — but it is a reason to expedite the studies, not to treat the diagnosis as already made.
Where I'd draw the line differently is on absolute avoidance: if benzodiazepines alone don't control her agitation and safety becomes acute, a low-dose antipsychotic isn't categorically forbidden here — it's a monitored, second-line option rather than a first-line default, watched closely for the rigidity and autonomic worsening the Neurologist described. What can't happen either way is the workup being deferred while everyone waits to see which medication controls her tonight — CSF and EEG go forward regardless of what she's given for agitation.
Agreed: lorazepam as first-line management for tonight's agitation rather than a routine antipsychotic, and the encephalitis workup (MRI, EEG, CSF, pelvic ultrasound for a possible ovarian teratoma) proceeds on an expedited basis regardless of which medication controls her symptoms tonight. Her symptoms already meet the four-of-six threshold the Graus criteria set; the EEG and spinal fluid are what would complete a probable diagnosis, and if either is abnormal, empiric high-dose steroids start without waiting for confirmatory antibody testing, which can take days to result.
Not agreed: whether a low-dose antipsychotic becomes reasonable as a genuine second-line option if benzodiazepines alone don't control her agitation later tonight. The Clinical Pharmacologist would allow it, closely monitored, rather than escalating benzodiazepine dosing indefinitely; the Neurologist would rather push benzodiazepine dosing further and involve neurology-ICU support before reaching for an antipsychotic at all, given how specifically the literature flags that class in this exact syndrome. Both agreed the question is moot unless it actually arises tonight, and left it there rather than pre-deciding.