Self-injurious behavior (SIB) is a severe and frequent problem in patients with a borderline personality disorder (BPD). Previous studies have shown difficulties in social cognition (SC) related to SIB, and different deficits in SC have also been observed in BPD patients. However, the role of SC in SIB has not yet been studied in people with BPD. Therefore, the aim of this work was to study the relationship between SC impairments and self-injurious behavior in patients with BPD, using machine learning methods.
MethodsThe SC of 77 patients with a BPD (71.5 % with suicide attempts and 60 % with nonsuicidal self-injury (NSSI)) was assessed using the Movie for the Assessment of Social Cognition (MASC). Other relevant clinical and sociodemographic features were also explored. Participants were classified according to the SIB presence and considering sociodemographic, SC and clinical factors, using supervised learning techniques adapted to small samples under different algorithms.
ResultsTheory of Mind (ToM) errors had a moderate predictive value on the presence of NSSI. Other variables with a relevant influence for NSSI classification were: high levels of impulsivity, anxiety, impairment in functioning, depressive symptomatology, age and theory of mind errors (AUC: 0.66). In addition, significant differences in impulsivity were found between patients with NSSI compared to those without NSSI.
ConclusionThe results of this study show that although impairments in social cognition have a moderate influence on the presence of NSSI in patients with a BPD, their impact is smaller than that of other variables such as impulsivity, functioning and emotional dysregulation.