Bridging Disciplinary Boundaries (January 11 - 14, 2007)

Introduction: The co-occurrence of Major Depressive Disorder (MDD) and General Anxiety Disorder (GAD) has been well researched. Most of the studies to date have used odds ratios and correlations between these diagnostic categories to understand their relationship. However, these diagnostic categories represent heterogeneous groups. For example, several subtypes of depression, such as endogenous or atypical, have been identified. In additional to heterogeneity of presentation, there are also different levels of severity within diagnostic categories. Evidence of high rates of co-occurrence also suggests a degree of homogeneity between the diagnoses. Many of the diagnostic indicators for depression such as sleep disturbance, lack of energy, and problems with concentration are also included in the algorithm for GAD. Improving understanding of the relationship between these disorders requires analyze the relationship between symptoms rather than just between diagnoses. Methods: Data from the National Co-morbidity Survey (1990-1992) was used to analyze the symptoms of MDD and GAD. All individuals who endorsed the screening questions for MDD and GAD were included (n=1009). 12 depression and 11 general anxiety symptoms were coded as dichotomous indicators. Individuals were classified by their patterns of symptom responses using a Latent Class Analysis conducted in Mplus 3.1. The fit of models with 1 to 10 classes were compared using Bayesian Information Criterion (BIC) statistics and interpretability. The best fitting model was then used to predict individual class membership. An individual's class membership was then used to analyze different clinical indicators including mental health hospitalization, service utilization, perceived mental health, suicide attempts and co-morbidity with other diagnoses. Regression and logistic regression models were run, predicting each clinical indicator, while controlling for demographic and co-morbid diagnoses. These models were then used to test whether the symptom-based class membership improved the prediction of the clinical indicators compared with the DSM diagnostic categories of MDD and GAD. Results: According to the BIC and interpretability criteria the seven-class model had the best fit. Each of these seven classes displayed not only differences in severity but also heterogeneity in the types of symptoms. All but one class had high levels of psychological anxiety symptoms (i.e. feeling restless). Other classes were distinguished by the amount of physical anxiety symptoms (i.e. dizziness), somatic depression symptoms (i.e. psychomotor retardation), or psychological depression symptoms (i.e. feelings of worthlessness). Membership in classes with high rates of physical anxiety symptoms predicted higher rates of mental health hospitalization, while membership in classes with high rates of psychological depression symptoms predicted higher rates of suicide attempts. When compared to the DSM diagnostic categories of MDD and GAD, class membership was a better predictor of outcome. Discussion: These results suggest that there are several distinct patterns of co-morbidity between depression and general anxiety. These subtypes also have important clinical implications. This analysis also provides evidence that examining the patterns of symptoms of disorders rather than just the diagnostic categories can improve our understanding of mental distress and help with targeting services and interventions.