Saturday, August 1, 2026

Opparounds

🔬Psychiatric Research Article

Oral Semaglutide for Alcohol Use Disorder: A Randomized Clinical Trial

Joseph P. Schacht, Joseph T. Sakai, Kristen Raymond, Robert Shelton · American Journal of Psychiatry · July 2026

Fifty treatment-seeking adults with moderate to severe alcohol use disorder were randomized to oral semaglutide (3 mg daily for four weeks, then 7 mg daily for four weeks) or placebo in this eight-week phase 2 trial. The primary outcome — laboratory alcohol cue-induced craving at week 6 — was negative, as was drinks per day.

The secondary outcomes were not. Semaglutide significantly reduced heavy drinking days (b = -0.580, 95% CI -1.012 to -0.148), drinks per drinking occasion (b = -1.177, 95% CI -2.307 to -0.047), naturalistic craving reported in daily life (b = -2.195), and alcohol-related consequences (b = -4.618). More semaglutide participants dropped a WHO risk-drinking level. It also reduced cannabis use days (b = -1.434), a cross-substance signal seen in prior GLP-1 work.

💡 Why it matters

This is a small phase 2 trial that missed its primary endpoint, and it should be quoted that way. What it adds is that the oral formulation reproduces the pattern seen with injectable GLP-1 agonists — less drinking per occasion rather than fewer occasions — which fits a satiety mechanism more than an anticraving one. It is not yet a reason to prescribe off-label, but it is a reason to ask about drinking in every patient already on one of these agents for weight or diabetes.

Read the paper → doi:10.1176/appi.ajp.20260003

🩺General Medicine Article

Patient-Centered Prescription Opioid Tapering Methods: A Randomized Clinical Trial

Beth D. Darnall, Lluvia Perez, Ming-Chih Kao, et al. · Annals of Internal Medicine · July 2026

Across 11 US sites, 562 adults with chronic pain on a morphine-equivalent daily dose of at least 10 mg — with moderate or severe opioid use disorder excluded — were randomized to voluntary patient-centered tapering alone, tapering plus pain-focused CBT, or tapering plus a chronic pain self-management program. Taper success meant either a 50 percent MEDD reduction without worsening pain, or no MEDD increase with improved pain.

Success rates were statistically indistinguishable: 50.9 percent for taper alone, 48.6 percent with pain-CBT, 44.5 percent with self-management. Adding a behavioral program did not improve taper outcomes. Adverse events, largely withdrawal symptoms, were less frequent with pain-CBT (54 percent) than with tapering alone (66 percent).

💡 Why it matters

Two things follow for psychiatry. First, roughly half of patients tapered successfully with a slow, genuinely voluntary protocol and no adjunctive therapy — the taper structure itself is doing the work, so when we are consulted on a taper, the leverage is in pacing and consent, not in adding a program. Second, the value of pain-CBT here was tolerability rather than success rate, which is a more honest thing to offer a frightened patient than a promise of better outcomes.

Read the paper → doi:10.7326/ANNALS-25-04784

💊Psychiatric Fact

A carbapenem will erase a valproate level in a day, and you cannot dose around it

Meropenem, ertapenem, and imipenem drop serum valproate concentrations by 60 to 90 percent within 24 to 48 hours — one of the largest and fastest drug interactions in the formulary. The mechanism is not induction of metabolism, which is why the usual reflex fails. Carbapenems inhibit acylpeptide hydrolase, the enzyme that hydrolyzes valproate-glucuronide back to valproate, so the normal enterohepatic recycling that sustains valproate levels is abolished; they also inhibit intestinal absorption and shift valproate into erythrocytes. The result is a level that collapses even while the patient keeps taking the same dose.

The practical implication is that escalating the valproate dose does not restore the level, and multiple case series have documented breakthrough seizures and manic relapse in patients whose valproate was pushed to the maximum during a course of meropenem. If the infection genuinely requires a carbapenem, the right move is to change the anticonvulsant or mood stabilizer for the duration — levetiracetam is the usual bridge — rather than to chase the valproate. And when the carbapenem stops, levels recover over about one to two weeks, so a patient discharged on a bridged regimen needs a plan and a level, not just a return to the old dose.

🛋️Psychotherapy Teaching Pearl

Enactments are not mistakes you avoided badly — they are the material arriving in the only form it can take

The classical instruction is to notice the pull and not act on it. Contemporary relational and modern Kleinian writers make a harder claim: with certain patients, particularly those whose formative experiences were preverbal or dissociated, the pattern cannot be reported, only reproduced — and so some degree of enactment is not a technical failure but the necessary first appearance of the material. The therapist discovers they have already been doing something before they notice they are doing it.

The technical question is therefore not 'how do I avoid this' but 'what do I do once I am in it.' The sequence that works is: notice the departure from your own usual practice, stop the behavior without announcing a verdict about it, and then take joint responsibility for the pattern rather than either confessing unilaterally or interpreting it as entirely the patient's doing. A confession — 'I realize I have been colluding with you' — sounds honest but hands the patient your guilt to manage. An interpretation that locates the whole thing in the patient's projection is technically defensible and clinically false. The useful move is descriptive and shared: name what the two of you have been doing, and wonder about it together.

The tell is almost always a deviation in your own behavior — the extended sessions, the unbilled phone calls, the uncharacteristic reassurance, the flatness where you are usually curious. Supervision exists partly to catch this, because by definition you cannot see it from inside.

🗒️ Vignette

A PGY-3 has been treating a 45-year-old man with chronic depression and a history of medical neglect in childhood. Over about two months the resident has, without deciding to, begun answering his between-session emails in detail — sometimes at length, sometimes late at night — and has twice let sessions run fifteen minutes over. In supervision this comes out almost as an aside.

The supervisor asks what happens if he does not answer. The resident says, immediately: 'He'll think I don't care.' Then, hearing himself: 'That is not something he has said.'

The patient has never asked for the emails. The resident has been supplying, in advance, the responsiveness that the patient learned not to ask for — and in doing so has been enacting the caregiver who is attentive only if you never make a demand. The enactment is the treatment history, being performed.

What the resident does next matters. He does not announce a boundary change ('I've realized I need to limit our email'), which would recreate the withdrawal exactly. He waits for the next email, does not answer it at length, and opens the following session: 'I want to look at something the two of us have set up. You send me a note; I write back, usually quite a lot, often at night. Neither of us ever agreed to that — it just grew. I noticed this week that when I did not write back the way I usually do, I felt like I had done something to you. I do not know yet what that is about, but I think it is worth our attention.'

The patient's first response is to reassure the resident that it is fine. That reassurance is the same pattern in miniature — the patient managing the caregiver's discomfort — and it is now visible in the room, in the present tense, where it can be worked with rather than reported.