The AI Psych Brief

Association between CYP2D6 and CYP2C19 genotypes and venlafaxine treatment failure: a retrospective study on two cohorts from Norway and the UK

Park Y, Lenk HÇ, Zhou Y, Lauschke VM, Molden E · The Lancet Psychiatry · October 2026

Venlafaxine is converted to its active metabolite O-desmethylvenlafaxine (desvenlafaxine) mainly by CYP2D6, with CYP2C19 playing a secondary role. This study asked whether genotype predicts real-world failure, defined as switching to another antidepressant within a year. The authors used two independent naturalistic cohorts: 5,443 genotyped patients from a therapeutic drug monitoring service in Oslo and 3,624 venlafaxine users from UK Biobank, excluding anyone on strong CYP2D6/2C19 inhibitors or inducers.

In the Norwegian cohort, CYP2D6 poor metabolisers had about twice the odds of switching (OR 2.05) compared with normal metabolisers. Intermediate metabolisers (OR 1.25) and ultra-rapid metabolisers (OR 1.78) were also at higher risk. The poor-metaboliser signal replicated in UK Biobank (OR 1.58). Patients who were poor metabolisers for both CYP2D6 and CYP2C19 had roughly six-fold odds of switching, although the confidence interval was wide.

Read the paper →  doi:10.1016/S2215-0366(26)00238-5

Exploring high priority potentially inappropriate prescribing cascades in older adults: population level retrospective cohort study

Rochon PA, Austin PC, Gurwitz JH, Wu W, Matai L, et al. · BMJ · September 2026

A prescribing cascade happens when a drug's adverse effect is mistaken for a new condition and treated with a second drug. Using Ontario administrative data on 2.3 million community-dwelling adults aged 66 or older, the authors tested 65 cascades named by a 2025 international Delphi panel. They used prescription sequence symmetry analysis, which checks whether the second drug is started after the first more often than before it.

Twenty-four cascades met every prioritisation threshold. Among the highest-incidence cascades was cholinesterase inhibitor → sleep agent (10.3% of incident users). The strongest temporal link in the whole study was corticosteroid → antipsychotic (adjusted sequence ratio 2.55). A cholinesterase inhibitor → antiemetic cascade was also prominent. SSRI/SNRI → overactive-bladder drug was a cascade in older men but not in older women.

Read the paper →  doi:10.1136/bmj-2026-100499

The pill-free week is a lamotrigine dose change nobody ordered

Ethinyl estradiol induces UGT1A4, the glucuronidation enzyme that clears lamotrigine. In the label's pharmacokinetic study, a combined oral contraceptive roughly doubled lamotrigine clearance and cut its AUC by 52%. The induction reverses quickly, so during the hormone-free week trough levels rise gradually, by up to about 2-fold, and then fall again when active pills restart. This creates two clinical traps. A patient who was stable on lamotrigine and then starts a combined pill can lose half her exposure and relapse into bipolar depression; the label allows up to a doubling of the maintenance dose. A patient titrated up while taking a combined pill can develop dizziness, diplopia or ataxia every pill-free week, or become toxic if the contraceptive is stopped without lowering the dose. Continuous-cycle regimens remove the monthly swing. If the patient is also on valproate, which inhibits the same glucuronidation pathway, the label does not recommend these adjustments. Check a level whenever an estrogen is started or stopped. Remember that the interaction runs both ways: lamotrigine lowered levonorgestrel exposure by about 19%.

Analyze the defense before the content it is hiding

The ego-psychology principle of working 'from surface to depth' (Fenichel; refined by Paul Gray as close process monitoring) holds that the most useful thing to interpret is usually not the forbidden wish but the moment the patient turns away from it. When a clinician names the content first ('you're angry at your father'), the patient receives a conclusion without having seen themselves arrive at it. It can be agreed with or refuted, but it does little to build the capacity to notice one's own avoidance. Pointing to the defense instead ('something shifted just now') uses data both people just witnessed in the session. That keeps the patient's observing ego in charge and lowers the risk of an intellectualized 'insight'. The technical discipline is timing: intervene at the point of discontinuity, when affect was rising and then the topic, tone or tense abruptly changed, and describe the shift before guessing at what drove it.