Differential impact of anxiety symptoms and anxiety disorders on treatment outcome for psychotic depression in the STOP-PD study

Abstract

Background: There are conflicting results on the impact of anxiety on depression outcomes. The impact of anxiety has not been studied in major depression with psychotic features ("psychotic depression").

Aims: We assessed the impact of specific anxiety symptoms and disorders on the outcomes of psychotic depression.

Methods: We analyzed data from the Study of Pharmacotherapy for Psychotic Depression that randomized 259 younger and older participants to either olanzapine plus placebo or olanzapine plus sertraline. We assessed the impact of specific anxiety symptoms from the Brief Psychiatric Rating Scale ("tension", "anxiety" and "somatic concerns" and a composite anxiety score) and diagnoses (panic disorder and GAD) on psychotic depression outcomes using linear or logistic regression. Age, gender, education and benzodiazepine use (at baseline and end) were included as covariates.

Results: Anxiety symptoms at baseline and anxiety disorder diagnoses differentially impacted outcomes. On adjusted linear regression there was an association between improvement in depressive symptoms and both baseline "tension" (coefficient=0.784; 95% CI: 0.169-1.400; p=0.013) and the composite anxiety score (regression coefficient = 0.348; 95% CI: 0.064-0.632; p=0.017). There was an interaction between "tension" and treatment group, with better responses in those randomized to combination treatment if they had high baseline anxiety scores (coefficient=1.309; 95% CI: 0.105-2.514; p=0.033). In contrast, panic disorder was associated with worse clinical outcomes (coefficient=-3.858; 95% CI: -7.281 to -0.434; p=0.027) regardless of treatment.

Conclusions: Our results suggest that analysis of the impact of anxiety on depression outcome needs to differentiate psychic and somatic symptoms.

Fig. 1

Interaction between baseline BPRS item 6 (tension behaviour) and treatment group in predicting improvement in depressive symptoms. Error bars reflect standard error. Coefficient for interaction terms (treatment group * baseline BPRS tension behaviour) after adjustments for age group, gender, education, benzodiazepine usage at baseline and at end of trial = 1.309 (95% CI: 0.105–2.514; p = 0.033).

Fig. 2

Interaction between baseline BPRS item 6 (tension behaviour) and treatment group in predicting probability of remaining in the trial for 12 weeks. Red squares: olanzapine + placebo. Blue diamonds: olanzapine + sertraline. Coefficient for interaction term (treatment group * tension behaviour) after adjustments for age group, gender, education, benzodiazepine usage at baseline and at end of trial = 0.356 (95% CI: −0.039 to 0.674; p = 0.028).