Evidence for a Somatic and Non-Somatic Factor Structure in the Patient Health Questionnaire-8 in a Military Sexual...

Area of Science:

• Psychiatry
• Psychological Assessment
• Military Mental Health

Background:

• The Patient Health Questionnaire-8 (PHQ-8) is an abbreviated measure of depression severity, derived from the PHQ-9.
• The PHQ-8 omits questions on suicide ideation, making it suitable for specific clinical contexts.
• Previous research on the PHQ-9 suggests mixed findings regarding its factor structure, with evidence supporting both unidimensional and two-dimensional (somatic/non-somatic) models.

Purpose of the Study:

• To examine the factor structure of the PHQ-8 in a sample of military sexual assault survivors.
• To compare the fit of various structural models, including unidimensional and two-dimensional approaches.
• To determine the optimal model for understanding depression symptom presentation in this high-risk population.

Main Methods:

• A secondary analysis was conducted on data from 346 service members and veterans who experienced sexual assault.
• Confirmatory factor analysis was employed to test five structural models: one unidimensional and four two-dimensional.
• Model fit was evaluated using standard indices: Chi-squared, CFI, TLI, RMSEA, BIC, and SRMR, with established thresholds for strong fit.

Main Results:

• A two-dimensional model, distinguishing between somatic (sleep, fatigue, appetite) and non-somatic (anhedonia, mood, worthlessness, concentration, psychomotor) symptoms, demonstrated the most optimal fit.
• This superior two-factor model achieved excellent fit statistics (CFI=.98, TLI=.97, RMSEA=.06, SRMR=.03).
• The unidimensional model showed statistically inferior fit compared to the two-dimensional models.

Conclusions:

• A two-factor structure may offer a more accurate representation of depression symptoms in military sexual assault survivors than a unidimensional model.
• This finding suggests potential clinical utility in tailoring treatments to specific symptom clusters (somatic vs. non-somatic).
• Future research should validate this model in diverse populations and explore its application in outcome studies.