Association of anxiety phenotypes with risk of depression and suicidal ideation in community youth

Abstract

Background: Anxiety symptoms are common in adolescence and are often considered developmentally benign. Yet for some, anxiety presents with serious comorbid nonanxiety psychopathology. Early identification of such "malignant" anxiety presentations is a major challenge. We aimed to characterize anxiety symptoms suggestive of risk for depression and suicidal ideation (SI) in community youths.

Methods: Cross-sectional associations were evaluated in community youths (n = 7,054, mean age: 15.8) who were assessed for anxiety, depression, and SI. We employed factor and latent class analyses to identify anxiety clusters and subtypes. Longitudinal risk of anxiety was evaluated in a subset of 330 youths with longitudinal data on depression and SI (with baseline mean age of 12.3 years and follow-up mean age of 16.98 years).

Outcomes: Almost all (92%) adolescents reported anxiety symptoms. Data-driven approaches revealed anxiety factors and subtypes that were differentially associated with depression and SI. Cross-sectional analyses revealed that panic and generalized anxiety symptoms show the most robust associations with depression and SI. Longitudinal, multivariate analyses revealed that panic symptoms during early adolescence, not generalized anxiety symptoms, predict depression and SI for later adolescent years, particularly in males.

Interpretation: Anxiety is common in youths, with certain symptom clusters/subtypes predicting risk for depression and SI. Panic symptoms in early adolescence, even below disorder threshold, predict high risk for late adolescent depression and SI.

Keywords: Philadelphia Neurodevelopmental Cohort; adolescent depression; anxiety; factor analysis; latent class analysis; suicidal ideation.

Figure 1-

Association of anxiety domains with depression, suicidal ideation, and impaired function Odds ratios were calculated based on binary logistic regression models including all anxiety symptom domains, controlling for age, sex, and socioeconomic status.

Figure 2 -

Latent classes of anxiety symptomatology in youth Anxiety classes across symptoms (A) and percentile distribution of class membership (B) based on participants’ highest probability of class assignment. The y-axis in Figure 2A represents mean z-score of the item within a class. Model entropy = 0.78; Bayesian Information Criterion = 182651.76; Akaike Information Criterion = 181528.06.

Figure 3-

Baseline panic and generalized symptoms by preadolescence and the risk for late adolescent depression and suicidal ideation (SI). Longitudinal risk of panic (A,C) and generalized anxiety symptoms (B,D) to depression (A,B) and SI (C,D). Abbreviation: SI, suicidal ideation. Error bars represent standard errors.