Psychometric evaluation and Rasch analyses of the German Overall Anxiety Severity and Impairment Scale (OASIS-D)

Abstract

The Overall Anxiety Severity and Impairment Scale (OASIS) is a 5-item self-report measure that captures symptoms of anxiety and associated functional impairments. This study evaluates a German version (OASIS-D) that was administered to a convenience sample of 1398 primary care patients of whom 419 were diagnosed with panic disorder with/without agoraphobia. Psychometric properties were analyzed using classical test theory as well as probabilistic test theory. Factor analyses suggested a unitary (latent) factor structure. The internal consistency was good to excellent. Convergent as well as discriminant validity with other self-report measures was found. A sum score (range 0-20) of ≥ 8 emerged as optimal cut-score for screening purposes. A difference score of ≥ 5 was indicative of reliable individual change. A Rasch analysis of local item independence suggested response dependency between the first two items. Rasch analyses of measurement invariance detected noninvariant subgroups associated with age and gender. Analyses of validity and optimal cut-off score were solely based on self-report measures, which may have introduced method effects. In sum, the findings support the transcultural validity of the OASIS and indicate its applicability to naturalistic primary care settings. Caution is warranted when using the scale to compare groups that differ in age or gender.

Figure 1

Confirmatory factor analysis for one-factor solution with residual correlation between items 1 and 2. Rectangles symbolize measured variables and the circle the latent construct. Standardized coefficients are given for item factor loadings and error variances.

Figure 2

Rating scale tree showing differential item functioning (DIF) within the total sample. Circles symbolize variables associated with DIF. Terminal nodes correspond to subgroups for which the null hypothesis of measurement invariance was rejected. Region plots are depicted below each terminal node to show the respective item threshold parameters.

Figure 3

Rating scale tree to assess differential item functioning (DIF) within the subsample of patients with panic disorder with/without agoraphobia. Circles symbolize variables associated with DIF. Terminal nodes correspond to subgroups for which the null hypothesis of measurement invariance was rejected. Region plots are depicted below each terminal node to show the respective item threshold parameters.