The suicidal process after a recent suicide attempt A combined undirected and directed network approach on cross-sectional data

Suicide is a complex phenomenon arising from the interactions of multiple risk factors, with risk particularly elevated following a recent suicide attempt. The Integrated Motivational-Volitional (IMV) model provides a comprehensive theoretical framework for understanding the onset of suicidal behaviour, and its core constructs are hypothesized to remain salient following an attempt. A central premise of the IMV model is that distress is greater in individuals who engage in repeated suicidal behaviour and, over time, intention is translated into behaviour with increasing rapidity. Network analysis offers a useful approach for examining the complexity of suicidal by modeling both direct and indirect associations between relevant variables. Traditionally, such analyses rely on undirected networks applied to cross-sectional data, in which nodes represent psychological symptoms and edges represent statistical associations between them. However, these edges do not indicate the direction of influence, limiting interpretation. Recent methodological developments make it possible, albeit under strict assumptions, to estimate directed networks from cross-sectional data. In these models, edges are estimations of potential directional relationships between nodes, thereby partially addressing this limitation and enabling a broader range of inquiries. The present study aimed to investigate the complexity of suicidal behaviour in a cross-sectional sample of 366 patients treated for a recent suicide attempt, using both undirected and directed network approaches. In the undirected network, defeat showed the strongest association with suicidal ideation, and both defeat and entrapment emerged as the most central variables. Consistent with these findings, the directed network suggested potential directional relationships between defeat, entrapment, and suicidal ideation, as well indirect effect of entrapment on suicidal ideation via interpersonal variables. Suicidal ideation functioned as a collider in the directed network, suggesting that it is primarily influenced by multiple upstream variables rather than exerting direct influence on other nodes. Additionally, a directional path from a pre-motivational factor (i.e., perfectionism) through motivational factors (i.e., entrapment, defeat), to suicidal ideation was identified, providing further empirical support for the IMV model. Strengths and limitations of applying directed network analysis to cross-sectional data are discussed.