The Psychiatric Record

Practical intelligence for psychiatric practice: prescribing, documentation, clinical risk, regulation, technology, and the workflows that fail quietly.

[The Psychiatric Record @PsychRecord]

A migraine diagnosis does not require an MRI. The absence of a structural finding was never how a primary headache disorder gets diagnosed. So "no neurologic cause identified" is a fact about the tests, not a psychiatric diagnosis.

New in The Psychiatric Record, with a handoff worksheet:

https://t.co/lSRnu6Z1IZ

[The Psychiatric Record @PsychRecord]

@PsychiatricNews New mechanism, same chart. Writing "started Cobenfy" tells the next clinician nothing. What matters is why this dose, what you are watching for and when, and what you do if the patient cannot tolerate it. That is the part that protects the patient and the prescriber.

[The Psychiatric Record @PsychRecord]

@JAMAPsych Approval on this trial ratio is a regulatory fact, not a consent script. The clinical act is the note showing the patient heard that evidence base before this agent was chosen over what already failed. Approved and explained are not the same status.

[The Psychiatric Record @PsychRecord]

@NTFabiano A single session shifting neural-efficiency markers is a lab finding. An exercise prescription for a depressed patient is a different act: dose, frequency, tolerability, and a note that records it as treatment rather than encouragement. Patients report back on the second one.

[The Psychiatric Record @PsychRecord]

Worth reading the moderators. The paper reports the association is moderated by which depression measurement tool was used and by which component of meaning was measured, with presence, search, and purpose behaving differently.

So r = -0.332 is for meaning in life composite, not purpose specifically, and it moves depending on the instrument. The authors frame it as guidance for optimizing intervention design. That is a different claim from potent antidepressant.

[The Psychiatric Record @PsychRecord]

Point 11 has a documentation consequence worth naming. Once akathisia is charted as "agitation," the note licenses the wrong act: the next clinician reads it and escalates the antipsychotic.

And "no psychomotor agitation observed" can be a true MSE finding sitting alongside severe subjective akathisia. The temporal link to the last dose change is often the only thing in the chart that would have caught it.

[The Psychiatric Record @PsychRecord]

A clinician asked a patient: "could you hear any voices?"

The AI scribe wrote: "denied auditory hallucinations."

Nothing was invented. But the note now implies a symptom was explicitly assessed in those terms.

Reported in a small simulation study with eight psychiatrists.

https://t.co/U0ciEPPqwd

[The Psychiatric Record @PsychRecord]

A psychiatrist charges $400 for a 30-minute visit.

$400 × 2 × 8 × 5 × 50 = $1.6 million a year.

The arithmetic is right. The conclusion is not.

New issue: what a posted fee actually tells you about a clinician's hourly rate. Plus a one-page worksheet.

https://t.co/UZUU9zJnly

[The Psychiatric Record @PsychRecord]

The definitional layer has the same problem one level up: "response" and "remission" get invoked in charts without naming which rule. The 2002 Kroenke–Spitzer rule of thumb, the 2022 VA/DoD appendix, and the 2001 validation paper can label the same score change three different ways. Wrote that up here: https://t.co/kr3la70qnd

[The Psychiatric Record @PsychRecord]

Worth adding: the double-barrel ambiguity carries into change scores. An insomnia to hypersomnia switch scores identically pre and post, so "PHQ-9 stable" in a chart can conceal a symptom migration that should change management. The item-level split fixes trajectory tracking, not just cross-sectional assessment. Nice work.

[The Psychiatric Record @PsychRecord]

The first sentence is what the paper shows. The second is what everyone wants it to show. r = −0.332 in a correlational meta-analysis is an association, not an antidepressant, and the arrow plausibly runs backward: depression suppresses meaning-making, so part of that r is the illness lowering the meaning score. "Correlate" and "treatment" are different words for a reason.

[The Psychiatric Record @PsychRecord]

The buried lede is tweet three: sometimes the correct movement prescription is less movement. "Exercise" is a status word, not a clinical act. The act is matching dose to the patient's state and capacity. A note that says "encouraged exercise" documents that something was said, not that anything was prescribed.

[The Psychiatric Record @PsychRecord]

Useful description of an actual system: weekly multidisciplinary review, shared high-risk list, EMR flag visible to the whole team, care continued after removal from the list.

But the headline finding needs care. 19 students were evacuated, 17 never returned, and the paper says their status is unknown. "No recorded suicides at the institution" is a statement about the records, not about the cohort. The paper's own intro notes that suicides at Caribbean schools may be underreported for institutional reasons.

The system may well be protective. This design can't show it.

[The Psychiatric Record @PsychRecord]

Worth reading the diary section of this one. The participant's own conclusion is that what helped was starting the work, not pausing it. That's behavioral activation, not paradoxical withdrawal.

Also: RCI tells you whether one person's change exceeds measurement error. It isn't a test of whether an intervention works. n=1 with no control can generate a hypothesis; it can't support "the intervention alleviates anxiety."

[The Psychiatric Record @PsychRecord]

Psychiatrists and PMHNPs prescribing clozapine: the Sept 2024 and June 2025 labels have the identical ANC schedule. Weekly for 6 months, every 2 weeks to month 12, then monthly.

One word changed. "Must" became "recommended."

The REMS is gone. The monitoring recommendation is not.

https://t.co/fQBnB8CqLx

[The Psychiatric Record @PsychRecord]

“Lamotrigine Is Not a Diagnosis.”

Some patients arrive saying, “I have ADHD,” when the history is dominated by mood reactivity, rejection sensitivity, irritability, and emotional storms that do not reach bipolar disorder.

Sometimes I use lamotrigine off-label for that phenotype.

The evidence is limited. The largest adjacent randomized trial was negative. And neither response to lamotrigine nor response to a stimulant proves the diagnosis.

That tension is the point.

https://t.co/mSO4aR0kBy

[The Psychiatric Record @PsychRecord]

A PHQ-9 drops 18 → 12.

As the 2002 Kroenke–Spitzer rule of thumb is commonly applied, that's a response. Under the 2022 VA/DoD definitions, it's a partial response.

Same six points, different labels. Most charts name neither.

New in The Psychiatric Record:
https://t.co/kr3la70qnd

[The Psychiatric Record @PsychRecord]

Neither "banned" nor "back" is the status of a clause.

Published. Set aside. Appeals dropped. Removed from the CFR. Four dates, four events — not one headline.

A clause has a governing law. A headline is not it.

https://t.co/gMgr0nnEpF

[The Psychiatric Record @PsychRecord]

"Referral placed" is a status, not completion.

Under CMS's own closed-loop measure, even a written no-show sent back to the referrer can close the loop. A chart that still says "placed" cannot.

New in The Psychiatric Record:

https://t.co/quGP8SYBqs

[The Psychiatric Record @PsychRecord]

Using an LLM on a psychiatry note is a draft step. It is not a clinical act.

“I Used the Model” Is Not a Clinical Act — new in The Psychiatric Record.

A model output is not a cited source, and it is not a supervised decision.

https://t.co/sb2fqjy8EL