Gut signals were harder to read in weight-restored anorexia, and the difficulty predicted relapse

Design

single-blind, within-participant randomised crossover study with six-month follow-up

Population

62 females with weight-restored restrictive anorexia nervosa and 57 matched comparators; 54 followed up

Primary outcome

interoceptive accuracy and, at six months, relapse and symptom severity

Effect

perceptual accuracy Cohen d -0.98 (95% CI -1.51 to -0.44); relapse predicted by response bias OR 5.37 (1.15-32.04)

Sixty-two females with weight-restored restrictive anorexia nervosa (mean age 18.9) and 57 matched comparators swallowed a vibrating capsule that delivered randomised blocks of normal and enhanced gut stimulation, while detection performance, EEG and peripheral physiology were recorded. Fifty-four of the anorexia group were followed to six months.

Detection was substantially worse in the anorexia group (perceptual accuracy Cohen d -0.98, 95% CI -1.51 to -0.44, P=0.001; miss rate d 1.02, 0.55-1.48, P<0.001). Computational modelling described the shape of the deficit: stronger prior expectations that no vibration was present, more unstable interoceptive precision between blocks, and asymmetric learning. Gastric-evoked potential amplitudes did not differ between groups. Capsule stimulation raised hunger ratings more in the anorexia group (d 0.94 against 0.40).

At six months, relapse was predicted by initial priors (odds ratio 3.82, 95% CI 1.02-15.91), response bias (OR 5.37, 1.15-32.04) and rated stomach unpleasantness (OR 5.73, 1.38-33.5). Those intervals are very wide - the lower bounds sit barely above 1 - and 54 participants is a small base for a prediction model. Read this as a mechanism paper that generates a plausible relapse marker, not as a test ready for clinic. What it does support today is taking a patient's report that she cannot tell whether she is hungry as a clinical observation rather than a figure of speech.

• Ask directly about the ability to detect hunger and fullness at follow-up after weight restoration, and record the answer
• Treat persistent gastric discomfort after refeeding as a symptom to explore, not one to reassure away
• Do not order or request interoceptive testing - the capsule protocol is a research tool
• Remember the relapse rate this addresses: close to half after weight restoration, which is why weight alone is a poor endpoint

Why it matters

It offers a candidate explanation for why half of weight-restored patients relapse, and a symptom to ask about while it is still a research finding.

Don't overread it

The relapse predictions rest on 54 participants with very wide confidence intervals - they are hypothesis-generating, not a clinical risk score.

The statistics, in plain English

A Cohen d of -0.98 is a large effect - roughly a one standard deviation gap between groups, enough that the distributions barely overlap in the middle. The relapse odds ratios are a different matter: an interval of 1.15 to 32.04 tells you the direction and almost nothing about the magnitude, which is what happens when a few dozen participants are split into relapsed and not.

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.