Saturday, October 31, 2026

Change in HbA1c and weight after initiation of antipsychotic treatment in people with severe mental illness in England: an individual-level interrupted time-series analysis

Bazo-Alvarez JC, Stone PW, Avgerinou C, Quartagno M, Boman M, Jordan KP, Cooper C, Hayes JF, Walters K, Petersen I. The Lancet Psychiatry. November 2026.

Using English primary-care records (CPRD Aurum, 2000–2023), the authors followed 96,849 adults aged 18–59 with severe mental illness who started an oral second-generation antipsychotic, excluding anyone with pre-existing type 2 diabetes. Individual-level interrupted time-series models compared the slope of HbA1c and weight before treatment, in the first 90 days, and from day 91 to two years.

The damage was front-loaded. In the first 90 days HbA1c rose about 0.12 mmol/mol per week and weight about 0.21 kg per week — roughly ten-fold the pre-treatment slopes — after which HbA1c returned to its baseline rate of rise while weight kept creeping up. Olanzapine produced the steepest rises in both, and increases were generally larger in women. Quetiapine was the instructive outlier: substantial weight gain with almost no HbA1c change, so the two markers did not move together. Among people with prediabetes at baseline, the estimated two-year incidence of type 2 diabetes was 19.0%, versus 3.0% with a normal baseline HbA1c.

Full text   doi:10.1016/S2215-0366(26)00274-9

What happens after being not-on-track? Patterns of change, profiles and prediction of at-risk patients

Moggia D, Schwartz B, Lutz W. Journal of Consulting and Clinical Psychology. September 2026.

Routine outcome monitoring flags patients as "not-on-track" (NOT) when their session-by-session scores diverge from the expected recovery curve. This study took 573 patients who had triggered a NOT alert during feedback-informed treatment at a university outpatient psychotherapy clinic and asked what happened next.

Growth mixture modelling found three distinct trajectories after the alert: severe-modest responders, severe-fast improvers, and moderate improvers. They differed in how symptomatic they were when flagged and in how they responded afterwards, and were distinguished at intake by symptom severity, personality-disorder traits, interpersonal problems and emotion-regulation difficulties. Two of the three patterns could be predicted from intake data with acceptable accuracy, and all three groups improved on average.

Full text   doi:10.1037/ccp0001032

Viloxazine is a strong CYP1A2 inhibitor, and duloxetine is on its contraindication list

Viloxazine (Qelbree) is usually presented as the non-stimulant that avoids atomoxetine's CYP2D6 problem, and it largely does: it is only a weak 2D6 and 3A4 inhibitor. The interaction that matters runs through a different enzyme. Viloxazine is a strong CYP1A2 inhibitor, and its label contraindicates co-administration with sensitive CYP1A2 substrates or 1A2 substrates with a narrow therapeutic range — the label's examples include duloxetine, ramelteon, tasimelteon, tizanidine and theophylline.

That list collides with ordinary adult-ADHD practice: duloxetine for comorbid anxiety or depression, ramelteon or tasimelteon for sleep, tizanidine from a pain clinic. A second tier, moderate sensitive 1A2 substrates such as clozapine, is "not recommended," with dose reduction if the combination is unavoidable. The inhibition raises total exposure (AUC) rather than peak level, so toxicity builds over days instead of showing up after the first dose. Check every 1A2 substrate before you start viloxazine, not just the 2D6 ones you were taught to worry about.

When the patient agrees with every interpretation, interpret the agreement

Some patients greet each interpretation with "Yes, exactly" and then change nothing. It is tempting to read this as a strong alliance, or to blame the content and keep producing better interpretations. More often the agreeing is the defense. Compliance keeps the therapist placated and at a distance, the way the patient once kept an intrusive or fragile parent placated. The interpretation is swallowed whole and never digested, and real disagreement — which would require a separate mind — never surfaces. The technical move is to stop adding content and turn to the process: notice aloud how quickly agreement arrives and wonder what disagreeing with you would risk. This is the classic point of analysing the defense before the content (character resistance). Do it with curiosity, not as an accusation, or the patient will simply agree with that too.