Unpacking the 'black box' of suicide: A latent class analysis predicting profiles of suicidal ideation in a longitudinal cohort of adolescent girls from India.

INTRODUCTION: Indian women account for 37% of global suicide-related deaths. As suicide is a growing concern among adolescent girls, identifying the social determinants of suicide with this group targeted prevention. We selected social determinants that include intersectional identities and broader syndemics; we then used longitudinal data from a prospective cohort of adolescent girls from Northern India to classify them into unique profiles across multiple socioecological levels.

METHODS: Girls aged 10-19 (N = 11,864) completed self-report questionnaires measuring socio-demographic and trauma exposure variables. At three-year follow-up, they were asked to indicate current suicidal ideation (SI). We conducted latent class analysis (LCA) to classify profiles and then predicted risk of current SI at three-year follow-up.

RESULTS: LCA supported a four-class solution: a 'privileged' class (Class 1; n = 1,470), a 'modal' class (Class 2; n = 7,449), an 'intergenerational violence' class (Class 3; n = 2,113), and a 'psychological distress' class (Class 4; n = 732). Classes significantly predicted odds ratios (OR) for SI at follow up; women in Class 4 were associated with the greatest likelihood of SI (OR 1.84, 95% CI 1.38, 2.47), suggesting that psychological distress factors confer greatest risk.

CONCLUSION: Results of the distinct classes of risk and protective factors indicate targets for policy-level interventions. Disrupting cycles of psychological distress and substance use, increasing access to behavioral interventions, and intervening to mitigate intergenerational violence may be particularly impactful with this population.