Suicidal behavior in Mexican adolescents: A test of a latent class model using two independent probability samples

Abstract

Suicidal behavior is a serious health issue that affects the adolescent population and that may be prevented through evidence-based approaches. There are many risk factors involved in suicidal behavior, but there are few studies encompassing the complex relationships between them. This study tested the similarity between two latent class models from two different epidemiological samples of middle-school students and evaluated if the multinomial regression model replicated the associations between the classes and the psychosocial variables. Data of 4013 adolescents from the Mexican states of Campeche and Querétaro were included in LCA to characterize suicide behavior and associated factors. The most likely latent class membership was used as the outcome in multinomial regression models. The model with data from Querétaro was consistent with the previous LC model from Campeche. The four latent classes were nearly identical for both populations and can be defined as follows: 1) “No Problems,” included 73% of the adolescents; 2) “Drug Use Only,” comprised 7% of the adolescents; 3) “Suicidal Behavior, No Depression,” captured 8% of the adolescents (25% reported binge drinking in the past month, 50% self-inflicted injuries, and 43% low-lethality suicide attempt); 4) “Depression and Suicidal Behavior,” comprised 12% of the sample (66% met criteria for Major Depressive Episode, 28% reported binge drinking in the past month, and 11% had a high-lethality suicide attempt). The congruence between the models and the consistency with identified factors highlight the need for appropriate and effective prevention strategies to minimize risk factors and reinforce protective factors in the adolescent population.