Subgroups of non-suicidal self-injury in a large diverse sample of online help-seekers

TL;DR

This study identifies four distinct subgroups of individuals who engage in non-suicidal self-injury, based on online survey data, to better inform targeted interventions.

Contribution

The study empirically identifies subgroups of NSSI individuals using a large, diverse online sample and compares their characteristics.

Findings

A four-class model best represented the heterogeneity in NSSI behaviors and characteristics.

Class 1 had the highest severity and earliest age of onset, while Class 3 had the lowest severity and latest age of onset.

Classes 2 and 4 showed moderate severity but differed in suicidal ideation levels.

Abstract

Many young people access information and resources for nonsuicidal self-injury (NSSI) online; yet our understanding of who accesses such information is limited. NSSI is a behavior with varied presentations. Understanding heterogeneity can help guide person-centered intervention. The present study aimed to (1) empirically identify classes of individuals with NSSI and (2) compare the classes according to demographic and clinical characteristics.

Data were collected from a survey posted to a national advocacy group website. Latent class analysis was used to derive classes based on characteristics associated with NSSI severity. Relationships between the latent classes and variables along five dimensions (behavior change, consequences or life interference, expectancies, functions, and NSSI across lifetime) were explored via logistic regression models.

11,262 individuals reporting past month NSSI were included in analyses. The 4-class model provided the most clinically interpretable groups. Class 1 was the smallest (16.8%), scored highest on all items and reported the youngest age of onset. Class 3 was the largest (31.8%), scored lowest on all items and reported the latest age of onset. Classes 2 (29.3%) and 4 (22.2%) had moderate scores on most items and differed in levels of suicidal ideation.

Classes presented with more severe symptoms than what is typical in samples in extant literature underscoring the importance of tailoring interventions for dissemination in online contexts.