Tuesday, October 13, 2026

The AI Psych Brief

🔬 Section 1

Psychiatric Research Article

Pharmacotherapies for Co-Occurring Stimulant Use Disorder and Attention-Deficit/Hyperactivity Disorder: A Network Meta-Analysis

Oliva HNP, Pulido-Saavedra A, Paredes Naveda AM, Forselius E, Potenza MN, Jegede OO, Angarita GA · Journal of Clinical Psychiatry · August 2026

Yale investigators pooled every randomized trial they could find of medication for adults who have both a stimulant use disorder (cocaine or methamphetamine) and ADHD. Out of more than 4,000 records, only 6 trials qualified, comparing methylphenidate, mixed amphetamine salts and bupropion against placebo, and the authors used a network meta-analysis to rank doses against each other.

Mixed amphetamine salts produced the largest gain in continuous abstinence (OR 8.45, 95% CI 2.55–27.98), with the 80 mg dose ranking first. Pharmacotherapy overall modestly raised the odds of negative urine toxicology (OR 1.65), and ADHD severity fell by a moderate amount (SMD −0.59), with methylphenidate 180 mg ranking best for symptoms. Adverse-event profiles were comparable to placebo. The evidence base is small and the confidence intervals are wide, so the rankings are hypothesis-generating rather than settled.

💡 Why it matters

For a patient with true ADHD and active cocaine or methamphetamine use, the trial evidence that exists points toward a stimulant at a higher-than-usual dose, not toward withholding stimulants or reaching for bupropion. Under-dosing seems to be where earlier negative trials went wrong, so if you do prescribe, prescribe enough, with structure (urine monitoring, limited dispensing) around it.

Read the paper → doi:10.4088/JCP.26r16478

🩺 Section 2

General Medicine Article

Antiseizure medication use in pregnancy and the risk of poor fetal growth in the EURAP international registry: a prospective cohort study

Perucca P, Malpas CB, Berry-Noronha A, Walker SP, et al.; Tomson T (EURAP Study Group) · Lancet Neurology · September 2026

EURAP is the large international registry that follows pregnancies in women with epilepsy who take antiseizure medication. This analysis used about 15,900 prospectively followed liveborn infants (1999–2023) to ask a question that has been overshadowed by malformation and neurodevelopment data: do these drugs restrict fetal growth?

Compared with monotherapy, polytherapy was linked to roughly a 50% higher risk of small-for-gestational-age birth, and birthweight centile fell further with each added drug. Among monotherapies, with lamotrigine as the reference, birthweight centile was lowest with topiramate (about 12 centile points lower), followed by phenobarbital, oxcarbazepine, carbamazepine, valproate and levetiracetam. Adding levetiracetam or valproate to lamotrigine did not lower birthweight centile further than lamotrigine alone.

💡 Why it matters

Psychiatry prescribes several of these drugs off the epilepsy indication: topiramate for weight or alcohol use, carbamazepine, oxcarbazepine and valproate for mood. In anyone who could become pregnant, add fetal growth restriction to the counselling list beside malformations, and favor lamotrigine and a single agent over stacking anticonvulsants.

Read the paper → doi:10.1016/S1474-4422(26)00216-4

💊 Section 3

Psychiatric Fact

Topiramate's pill interaction has a dose threshold, and psychiatry often prescribes above it

Topiramate is only a weak inducer of ethinyl estradiol metabolism, and the effect depends on dose. In healthy volunteers taking 50 to 200 mg/day, ethinyl estradiol AUC changed by no more than about 12%, which was not statistically significant (Doose 2003). In the earlier adjunctive study, exposure fell by 18%, 21% and 30% at 200, 400 and 800 mg/day. Progestin exposure was largely unaffected.

This matters because psychiatric use often drifts upward: alcohol use disorder trials titrate to 200–300 mg/day, and weight-mitigation regimens get pushed when results stall. Above about 200 mg/day, treat a combined pill as possibly unreliable: advise a higher-estrogen pill or, better, an IUD or implant, which no enzyme inducer touches. The stakes are higher with topiramate than with most inducers because of its own teratogenic signal (oral clefts, and in EURAP the largest reduction in birthweight among monotherapies). So a contraceptive failure on this drug is not just an unplanned pregnancy but an exposed one.

🛋️ Section 4

Psychotherapy Teaching Pearl

When a patient asks about your life, explore the question before you answer it

A personal question ("Do you have kids?" "Are you religious?") seems to force a choice between answering and seeming like an ordinary person, or deflecting and seeming withholding. Framed that way, both options lose. The question is itself material: it arrives at a particular moment, carries a fantasy about the answer, and usually says something about what the patient fears or hopes is true of you. The technique is to ask what prompted the question, and what answer they imagine, before deciding whether to answer. Then decide on purpose. A brief factual answer is often fine and keeps the relationship human, but only after the meaning has had a chance to come out, because once the fact is given the fantasy usually closes. Reflexive refusal fails in the opposite way: a fragile patient experiences it as the coldness they already expected, and the rupture swamps any insight.

🗒️ Vignette

Three months into weekly therapy, a 34-year-old woman who has been describing her mother's critical hovering suddenly asks the resident, "Do you have children?" The resident feels pulled to say "no" and move on. Instead: "I'm glad to talk about that. First, though, I'm curious what made it come up just now." She pauses, then says that if the resident had kids, they would "get it" and side with her mother. If not, they couldn't understand what mothering is like. The resident notes that either answer would mean being judged, which is exactly how she feels with her mother. She cries and says, "I guess I'm always waiting to find out whose side someone's on." The resident then answers briefly and truthfully. The fact matters less now, because what the question was really about is already in the room.