How well a weight-restored patient feels her own stomach predicts relapse

Design

single-blind, within-participant randomised crossover trial with EEG and computational modelling

Population

62 females with weight-restored restrictive anorexia nervosa and 57 matched healthy comparators; 54 followed to 6 months

Primary outcome

interoceptive accuracy and modelled parameters; relapse and symptom severity at 6 months

Effect

accuracy Cohen d −0.98 (95% CI −1.51 to −0.44); relapse OR 5.73 (1.38 to 33.5) for stomach unpleasantness

Sixty-two females with weight-restored restrictive anorexia nervosa and 57 matched comparators swallowed a vibrating capsule that delivered counterbalanced blocks of normal and enhanced gut stimulation, in a single-blind within-participant randomised crossover design with EEG and computational modelling. Fifty-four of the patients were followed for six months.

The anorexia group detected gut signals less accurately (Cohen d −0.98; 95% CI −1.51 to −0.44) and missed more of them (d 1.02; 0.55 to 1.48). Modelling showed stronger prior expectations that no vibration would occur, larger shifts in interoceptive precision between blocks and asymmetric learning. Gastric-evoked potential amplitudes did not differ between groups. Capsule stimulation raised hunger more in the anorexia group (d 0.94 vs 0.40). At six months, relapse was predicted by those initial priors (OR 3.82; 1.02 to 15.91), response bias (OR 5.37; 1.15 to 32.04) and reported stomach unpleasantness (OR 5.73; 1.38 to 33.5).

This is research-stage and the apparatus is not coming to clinic. What transfers now is the clinical observation behind it: in weight-restored patients, how the stomach is interpreted — not how full it is — carries prognostic weight, and asking about gut unpleasantness at follow-up costs nothing.

• Ask weight-restored patients specifically about stomach sensation and how unpleasant they find it
• Treat persistent 'I can't tell when I'm hungry' as a relapse-relevant symptom, not a residual complaint
• Do not read a normal weight as a restored interoceptive picture
• Keep follow-up intervals tighter where gut-focused distress persists after restoration

Why it matters

Relapse after weight restoration approaches 50 per cent, and this points at a symptom nobody is currently asking about.

Don't overread it

These are prediction associations in a single research cohort — the capsule is not a clinical test, and nothing here shows that treating interoception prevents relapse.

The statistics, in plain English

The relapse odds ratios look large but their confidence intervals run from just above 1 to over 30, which is what happens with 54 patients and a modest number of relapses: the direction is informative, the magnitude is not. Several of the group differences sit at P = .05 exactly, and many measures were tested, so individual modelling findings need replication.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.