Association between childhood trauma, parental bonding and antisocial personality disorder in adulthood: A machine learning approach. (October 2021)

• Record Type: Journal Article
• Authors: Schorr, Manuela Teixeira; Quadors dos Santos, Barbara Tietbohl Martins; Feiten, Jacson Gabriel; Sordi, Anne Orgler; Pessi, Cristina; Von Diemen, Lisia; Passos, Ives Cavalcante; Telles, Lisieux Elaine de Borba; Hauck, Simone
• Is Part Of: Psychiatry research. Volume 304(2021)
• Publication Date: 2021-10
• Subjects: Antisocial personality disorder -- Child abuse -- Parental bonding -- Machine learning -- Statistical analysis; Psychiatry -- Periodicals; Psychiatry -- periodicals; Psychiatrie -- Périodiques; 616.89
• DOI: 10.1016/j.psychres.2021.114082
• Languages: English
• ISSNs: 0165-1781

Abstract

Highlights: Trauma, bonding, substance abuse and antisocial personality were highly associated. There was a higher prevalence of mental illness in antisocial individuals. Machine learning was used to predict antisocial personality in male cocaine users. Emotional and physical abuse were the strongest predictors for ASPD. Paternal bonding was a stronger predictor for ASPD than maternal bonding. Abstract: Childhood trauma (CT) and parental bonding (PB) have been correlated with later antisocial personality disorder (ASPD). Aiming to better understand this complex interaction we analyzed the data from a cross-sectional study that evaluated 346 male inpatient cocaine users, using both traditional statistical analysis and machine learning (ML) approaches. Childhood Trauma Questionnaire (CTQ), Parental Bonding Instrument (PBI), and Mini International Neuropsychiatric Interview (MINI) were applied. We found a markedly higher prevalence of mental illness in the ASPD group. The ML method and the traditional analysis showed that emotional and physical abuse were the factors with the strongest relationship with ASPD. Also, there were discrepancies between the findings of both methods regarding physical neglect and paternal care. Although this study does not allow definitive answers in this matter, we do propose that these two methods can aid in better comprehending how multiple variables interact with each other in the development of psychological disorders.