Structural equation modeling of disability in women with fibromyalgia or multiple sclerosis.

Structural equation modeling (SEM), a popular statistical technique for analysis of multivariate data in the social sciences, is increasingly being used in the behavioral and clinical sciences. SEM is appropriate for posing complex models that evaluate the direct and indirect influence of several variables on one or more outcome variables. A biosocial model of disability, the Disablement Process Model, lends itself to evaluation by SEM. Using SEM, this study examined predictors of disability (Age, Education, Duration of Illness, and Economic Adequacy Functional Limitations, Depressive Symptoms, and Social Support) separately in women with multiple sclerosis (MS) and women with fibromyalgia syndrome (FMS) and compared the respective models across groups. Data were analyzed with Analysis of Moment Structures (Amos) 7.0. Problems identified in initial confirmatory model testing included collateral correlated errors, a negative error variance, and poor performance of the disability indicators. After specifying well-fitting confirmatory models for each group, a structural model for the larger FMS group was estimated. Model refinement resulted in the reversal of the path between Depressive Symptoms and Social Support. Further model revisions were based on comparative fit statistics and theoretical logic. The structural model developed from the FMS sample required minimal changes to fit the MS sample. The multisample model explained greater variance in disability in women with FMS than in women with MS. Social support and depressive symptoms mediated the effect of functional limitations on disability. Interventions that target modifiable characteristics, such as depression and social support, may improve outcomes such as disability.