ABSTRACT
PURPOSE: To investigate whether the Patient Health Questionnaire-9 (PHQ-9) possesses the essential psychometric characteristics to measure depressive symptoms in people with visual impairment. METHODS: The PHQ-9 scale was completed by 103 participants with low vision. These data were then assessed for fit to the Rasch model. RESULTS: The participants' mean +/- standard deviation (SD) age was 74.7 +/- 12.2 years. Almost one half of them (n = 46; 44.7%) were considered to have severe vision impairment (presenting visual acuity <6/60 in the better eye). Disordered thresholds were evident initially. Collapsing the two middle categories produced ordered thresholds and fit to the Rasch model (chi = 10.1; degrees of freedom = 9; p = 0.34). The mean (SD) items and persons Fit Residual values were -0.31 (1.12) and -0.25 (0.78), respectively, where optimal fit of data to the Rasch model would have a mean = 0 and SD = 1. Unidimensionality was demonstrated confirming the construct validity of the PHQ-9 and there was no evidence of differential item functioning on a number of factors including visual disability. The person separation reliability value was 0.80 indicating that the PHQ-9 has satisfactory precision. There was a degree of mistargeting as expected in this largely non-clinically depressed sample. CONCLUSIONS: Our findings demonstrate that the PHQ-9, when scaled with Rasch analysis, forms a linear interval measurement of depressive symptoms suitable for use in a vision impaired population.
