Monday, August 3, 2026

Opparounds

🔬Psychiatric Research Article

Mapping 2-Year Psychiatric and Neurologic Risks After Infections Across Body Systems and Age Groups

Maxime Taquet, Patrick Oliver, Thomas A. Pollak, Edoardo G. Ostinelli, Orestis Efthimiou, Andrea Cipriani, Paul J. Harrison, et al. · JAMA Psychiatry · July 2026

Drawing on 62 US health care organizations, the investigators propensity-matched 1,062,722 pairs of patients hospitalized with an infection versus hospitalized without one, and followed 14 psychiatric and neurologic outcomes for two years. Risk was elevated in 116 of 140 infection-by-disorder comparisons.

The pattern is more informative than the headline. Relative risk was highest for encephalitis, but the largest *absolute* excess was for cognitive deficits — 12.4 percentage points higher after a cardiac infection. Infectious encephalitides were the leading risk factor for 7 of 14 disorders and in the top three for 12 of 14. Children showed elevated relative risks but much smaller absolute risk differences than adults.

💡 Why it matters

The clinically actionable piece is the relative-versus-absolute split: the infections that most raise your *odds* are not the ones that produce the most *cases*. For a consultation-liaison service, this argues for treating a serious infection — of any organ system, not just the CNS — as a marker for two years of elevated neuropsychiatric risk, with cognitive complaints deserving formal assessment rather than attribution to deconditioning.

Read the paper → doi:10.1001/jamapsychiatry.2026.1904

🩺General Medicine Article

Ceperognastat in Early Symptomatic Alzheimer Disease: A Randomized Clinical Trial

Adam S. Fleisher, Leanne Munsie, Michele Mancini, William Kielbasa, Dustin J. Mergott, Mark Mintun, Oskar Hansson, et al. · JAMA · July 2026

Ceperognastat inhibits O-GlcNAcase, an enzyme that strips a sugar modification from tau; the hypothesis was that raising tau O-GlcNAcylation would block its aggregation. In this phase 2 trial across 72 sites in five countries, 259 participants with early symptomatic Alzheimer disease and low-to-medium tau burden received 0.75 mg, 3 mg, or placebo daily.

The result was flatly negative, and the higher dose looked worse. Change on the integrated AD rating scale was -8.39 with 0.75 mg, -13.27 with 3 mg, and -10.07 with placebo — 32 percent *greater* decline on the higher dose. Serious adverse events were more common with 3 mg (26.4 versus 15.7 percent), as were severe events. No secondary clinical endpoint showed benefit.

💡 Why it matters

For a psychiatrist fielding questions in clinic, this is a useful counterweight to the anti-amyloid enthusiasm: tau-directed small molecules remain unproven, and target engagement is not efficacy. The inverse dose-response is the part worth remembering — it is a reminder that in neurodegeneration, more mechanism is not reliably more benefit, and families deserve that framing before they pursue a trial.

Read the paper → doi:10.1001/jama.2026.12768

💊Psychiatric Fact

The lithium level you are looking at is answering a question you did not ask

A lithium level is only interpretable as a 12-hour post-dose trough, and small deviations matter more than people expect: a level drawn at 8 hours can run 20 to 30 percent higher than the same patient's true trough, which is a common source of spurious 'toxicity' in a patient who feels fine. The maintenance target is also more contested than the textbook 0.6 to 1.2 suggests. The BALANCE trial and subsequent work support 0.6 to 0.8 mmol/L for most maintenance, reserving 0.8 to 1.0 for patients with breakthrough mania or a history of severe episodes, and accepting 0.4 to 0.6 in older adults where neurotoxicity risk rises. Above 0.8 the marginal relapse protection is small and the renal and thyroid cost is not.

The interactions are worth memorizing as a short list because they are all common and all act on the proximal tubule, where lithium is reabsorbed alongside sodium. Thiazides raise levels by roughly 25 to 40 percent; ACE inhibitors and ARBs by a similar amount, sometimes more in older patients; NSAIDs by 15 to 25 percent, with indomethacin among the worst and aspirin and sulindac relatively spared. Loop diuretics are less predictable than thiazides but not safe. Anything that lowers effective circulating volume — a vomiting illness, a heat wave, a low-sodium diet started for hypertension — does the same thing without appearing on any interaction checker. And the level does not tell the whole story: a patient with cerebellar signs, coarse tremor, or confusion at a level of 1.0 has neurotoxicity that needs acting on, because chronic toxicity injures the brain at levels that look acceptable on paper.

🛋️Psychotherapy Teaching Pearl

The negative therapeutic reaction: when getting better is the threat

Freud's observation was that some patients reliably worsen precisely when the work goes well — after an accurate interpretation, after a good week, after any evidence that the treatment is helping. He attributed it to unconscious guilt and a need for punishment. Contemporary formulations are broader and more clinically usable, and the therapeutic value lies in which formulation fits this patient.

There are at least four, and they call for different responses. Unconscious guilt: improvement is forbidden because someone else did not get to improve, so the patient forfeits the gain. Identity: the symptom has become load-bearing — it explains a life, organizes a family, licenses a set of accommodations — and recovery would require answering questions the illness has been answering. Attachment: getting better means the treatment ends, so improvement is experienced as the beginning of an abandonment. Envy, in the Kleinian sense: the therapist's usefulness is intolerable, and spoiling the help restores the balance.

The common technical error is to treat the deterioration as a treatment failure and respond by working harder — intensifying, adding, reassuring — which confirms the patient's implicit demonstration that improvement is dangerous. The move is to treat the worsening itself as the communication, and to be openly curious about the timing rather than about the symptoms. And it is worth saying plainly: before invoking any of this, rule out the mundane explanations. Most patients who get worse after a good session are having a recurrence, a medication problem, or a life event, and misreading that as a negative therapeutic reaction is a genuine harm.

🗒️ Vignette

A 52-year-old woman with treatment-resistant depression has, for the first time in three years, had a good fortnight. She returned to her studio, saw two friends, slept. She reports all of this in the session and the therapist notices she reports it the way one might report someone else's news.

The following week she arrives having stopped her antidepressant, missed a deadline, and cancelled the friends. Her stated reason is that the medication was 'probably not doing anything anyway.'

The reflex — reinstate the medication, address the cognitive distortion, perhaps add a session — would treat the collapse as an obstacle to the work. The therapist instead treats the timing as the datum: 'I want to notice something before we do anything about it. You had the best two weeks in three years, and then this week you stopped the medication. I do not think that is a coincidence, and I do not think you do either.'

She becomes irritated, then quiet, and then says something she has not said in three years of treatment: that her older sister, who had the same illness, died by suicide at 49. 'She never got a good fortnight.'

There it is. The improvement was not neutral; it was a betrayal, and stopping the medication was the restitution. The therapist does not interpret this as guilt — she has just said it more clearly than any interpretation could — but names the shape of it: 'So the two weeks were not only good. Getting to have something she did not get to have costs you something.'

What the therapist explicitly does *not* do is argue that her sister would have wanted her to be well. That is true, kind, and beside the point; the loyalty is not a belief to be corrected. The work over the following months is with the loyalty itself — what it protects, what it costs, and whether a version of it exists that does not require her to stay ill. The medication is restarted, but that was never the intervention.