Measurement and Stratification of Nonsuicidal Self-injury in Adolescents

Background: Nonsuicidal self-injury (NSSI) is highly prevalent in adolescents. In survey and interview studies assessing NSSI, methods of assessment have been shown to influence prevalence estimates. However, knowledge of which groups of adolescents that are identified with different measurement methods is lacking, and the characteristics of identified groups are yet to be investigated. Further, only a handful of studies have been carried out using exploratory methods to identify subgroups among adolescents with NSSI.

Methods: The performance of two prevalence measures (single-item vs. behavioral checklist) in the same cross-sectional community sample (n = 266, age M = 14.21, 58.3% female) of adolescents was compared regarding prevalence estimates and also characterization of the identified groups with lifetime NSSI prevalence. A cluster analysis was carried out in the same sample. Identified clusters were compared to the two groups defined using the prevalence measures.

Results: A total of 118 (44.4%) participants acknowledged having engaged in NSSI at least once. Of these, a group of 55 (20.7%) adolescents confirmed NSSI on a single item and 63 (23.7%) adolescents confirmed NSSI only on a behavioral checklist, while denying NSSI on the single item. Groups differed significantly, with the single-item group being more severely affected and having higher mean scores on difficulties in emotion regulation, self-criticism, number of methods, higher frequency of NSSI, higher rates of suicidal ideation and suicidal behavior and lower mean score on health-related quality of life. All cases with higher severity were not identified by the single-item question. Cluster analysis identified three clusters, two of which fit well with the groups identified by single-item and behavioral checklist measures.

Conclusions: When investigating NSSI prevalence in adolescents, findings are influenced by the researchers' choice of measures. The present study provides some directions toward what kind of influence to expect given the type of measure used, both with regards to the size of the identified group and its composition. Implications for future research as well as clinical and preventive work are discussed.