Borderline personality disorder (BPD) is a form of psychopathology characterized by a pervasive pattern of instability with emotion regulation, impulse control, interpersonal relationships, and sense of self. While not a required diagnostic marker, the majority of individuals with BPD engage in some form of deliberate self-harm (e.g., suicide attempts, nonsuicidal self-injurious behavior) or suicide-related behavior (e.g., suicidal ideation, suicide threats). Longitudinal data from a sample of adolescent psychiatric inpatients who were hospitalized for deliberate self-harm and suicide-related behavior were followed for 5 years to investigate whether deliberate self-harm or suicide-related behaviors predicts BPD at 3-year follow-up and 5-year chart review. The extant data set consisted of 132 consecutively admitted adolescent psychiatric inpatients who completed a series of self-report questionnaires assessing deliberate self-harm and suicide-related behaviors, maladaptive familial behavior, peer victimization, and emotion regulation difficulties. Data regarding index psychiatric hospital admission diagnoses, childhood maltreatment, and BPD diagnoses were abstracted from the patient’s medical and psychiatric records and BPD was also assessed though a structured clinical interview. Suicide threats were the only variable found to be predictive of BPD at 5-year chart review. Other empirically (e.g., history of childhood maltreatment, maladaptive familial behavior, and peer victimization) and theoretically (e.g., emotion regulation difficulties) grounded constructs were also examined and were not found to be predictive of BPD in the current study. To our knowledge, this is the first study to investigate prospective associations between deliberate self-harm and suicide-related behaviors and BPD. Future directions and limitations of the research are discussed.

Fifteen hospitalized adolescents were used as subjects. An individually consequated token economy was in effect during baseline. Measures were taken of work output, attending behavior, and disruptive behavior. During the treatment phase, reinforcement was contingent upon the performance of a randomly selected subgroup. Following the treatment phase, the individual token system was reinstated for baseline-2 measures. The mean performance of the group during baseline was compared to performance under treatment conditions for work output and attending behaviors. In addition, performance of the contingent subgroup was compared to performance of the non-contingent group. No significant t values were obtained. With failure to obtain significant t values, the null hypothesis was not rejected, i.e., the two conditions were not proven significantly different.