Background: Major depressive disorder (MDD) is associated with maladaptive self-reported interoception, i.e., abnormal bodily self-experience. Although diminished body trusting predicts suicidal ideation, interoceptive measures have not been considered in depressed inpatients, whose suicide risk regularly peaks post-discharge. This study aims to explore interoceptive characteristics at admission that help identify inpatients at risk for suicidal ideation at discharge, thereby preventing fatal outcomes. Methods: The observational study included 87 depressed inpatients providing self-ratings at both hospital admission (T0) and discharge (T1) on the following scales: Multidimensional Assessment of Interoceptive Awareness (MAIA-2); Beck Depression Inventory-II (BDI-II). A hierarchical logistic regression analysis estimated the longitudinal association between self-reported interoception (T0) and suicidal ideation (T1). The optimal cutpoints for predicting suicidal ideation were calculated using ROC curve analysis. Results: Suicidal ideation was found in 17.24% patients at discharge, who reported lower baseline MAIA-2 Trusting scores than non-ideators (p=0.01). Diminished body trusting (OR=0.19), somatic comorbidity (OR=16.77), and baseline suicidal ideation (OR=24.01) significantly predicted suicidal ideation (T1). For body trusting, we estimated an optimal classification of subsequent suicidal ideation for the cutpoint≤2.33 (AUC=0.70 [95% CI 0.57, 0.83], sensitivity=0.87, specificity=0.44, positive predictive value=0.25, negative predictive value=0.94). Limitations: Due to the exploratory nature of the study, the findings should be replicated in pre-registered trials with larger sample sizes. Conclusions: Diminished body trusting is, with acceptable sensitivity, a significant predictor for post-treatment suicidal ideation in depressed inpatients. This finding emphasizes the importance of incorporating body-centered approaches into multimodal treatment strategies especially in inpatients under risk to prevent suicidal incidents.
