Daily prediction of inpatient suicide attempts using routinely collected theory-driven data

We lack knowledge about the short-term predictors of suicide attempts (SAs) among treatment-seeking individuals. The current study evaluated whether (a) interpersonal difficulties, hopelessness, and affective states are associated with an increased risk of SAs on the same and the next day; (b) these daily states are interconnected differently over time among inpatients who attempt suicide compared to those who do not. In total, 110 psychiatric inpatients who attempted suicide during their stay at a psychiatric hospital self-reported their suicidal ideation, negative affect, positive affect, wish to live, interpersonal needs, and hopelessness each day (3,018 daily reports). Multilevel structural equation modeling was used to examine same-day and next-day predictors of SAs. Multilevel temporal network models assessed interconnectedness between daily predictors and were compared to network models from a matched sample of 110 psychiatric inpatients who did not attempt suicide. In multivariate models, increases in perceived burdensomeness were significantly associated with same-day SAs, whereas increased hopelessness was associated with next-day SAs. Network models for patients who attempted suicide indicated that hopelessness and suicidal ideation were central to change, leading to next-day deteriorations in mental health. In subsequent models, feeling calm and relaxed, and feeling fresh and rested were centrally connected to other variables. The centrality of these metrics tended to be higher than in the network models for patients who did not attempt suicide, suggesting differences in the interplay between risk and protective factors. This study suggests routinely monitoring interpersonal factors and hopelessness may help identify increased short-term risk of SAs among psychiatric inpatients.