Biofeedback and Training of Interoceptive Insight and Metacognitive Efficacy Beliefs to Improve Adaptive Interoception: A Subclinical Randomised Controlled Trial

Abstract

Introduction: Interoception, the sensing, awareness, and regulation of physiological states, is crucial for wellbeing and mental health. Behavioural interventions targeting interoception exist, but randomised controlled trials (RCTs) testing efficacy remain limited. The present, preregistered (ISRCTN16762367) RCT tested the novel Interoceptive iNsight and Metacognitive Efficacy beliefs (InMe) intervention. InMe uses slow breathing and cardiac biofeedback during stress to train interoceptive self-efficacy beliefs and improve self-reported interoception.

Methods: Healthy participants aged 18-30 years with low self-reported interoception were randomly assigned (1:1) to the InMe intervention (n = 50) or an active control (guided imagery; n = 52). Participants blinded to allocation were stratified by gender and disordered eating. Assessments included baseline (T0), post-intervention (T1), and 7-8 weeks post-intervention (T2). The primary outcome was the "adaptive interoception" factor of the Multidimensional Assessment of Interoceptive Awareness questionnaire.

Results: Both arms improved in the primary outcome at T1 (InMe: adjusted M difference = 5.76; 95% CI [-0.03; 11.56], p = 0.05; control: adjusted M difference = 7.90; 95% CI [1.92; 13.87], p = 0.002; marginal R2 = 0.09). However, only InMe sustained this improvement at T2 (InMe: adjusted M difference = 9.25, 95% CI [3.37; 15.13], p < 0.001; control: adjusted M difference = 2.94, 95% CI [-3.07; 8.96], p = 0.72), as indicated by a significant time*arm interaction (b = 6.31; SE = 2.92, 95% CI [0.56; 12.05], p < 0.03; marginal R2 = 0.12). Secondary outcomes showed a reduction in disordered eating scores across both arms at both time points (T1: b = -1.44, SE = 0.37, 95% CI [-2.17; -0.71], p < 0.001; T2: b = -1.05, SE = 0.37, 95% CI [-1.79; -0.32], p = 0.005).

Conclusion: The InMe intervention selectively improved self-reported interoception at follow-up but did not outperform the control for secondary outcomes. Future research should explore its efficacy in clinical populations alongside complementary therapies.

Keywords: Biofeedback; Eating disorders; Interoception; Randomised controlled trial; Self-efficacy; Slow breathing.

Fig. 1.

Overall study design, including assessment time points (T0, T1, T2), averaged time in days/weeks between the time points and intervention sessions.

Fig. 2.

CONSORT flow diagram of the progress through the phases of the randomized trial, including enrolment, allocation, randomisation, baseline (T0), post-intervention (T1) and follow-up (T2) assessments, intervention sessions, and analysis. EDE-Q, Eating Disorder Examination Questionnaire; HRD, heart rate discrimination; TSST, Trier Social Stress Test.

Fig. 3.

Timeline and items of interoceptive self-efficacy beliefs assessed throughout the TSST and intervention procedure. Baseline estimations were recorded before the TSST, prospective estimations before each stressor, retrospective estimations after each stressor, and global estimations after both stressors.

Fig. 4.

Adaptive interoception at T0 (baseline), T1 (post-intervention), and T2 (7–8 weeks of follow-up) in InMe and control arms. Jittered dots represent the score for each individual participant. The error bars denote the standard error of the arm mean. The half violin provides data distribution.

Fig. 5.

Disordered eating score at T0 (baseline), T1 (post-intervention), and T2 (7–8 weeks of follow-up) in InMe and control arms. Jittered dots represent the score for each individual participant. The error bars denote the standard error of the arm mean. The half violin provides data distribution.

Fig. 6.

General Self-Efficacy (GSE) at T0 (baseline) moderate the change in adaptive interoception scores at T1 and T2 from T0 in InMe and control arms.