Somatic symptom reports in the general population: Application of a bi-factor model to the analysis of change.

Objective: To investigate the latent structure of somatic symptom reports in the general population with a bi-factor model and apply the structure to the analysis of change in reported symptoms after the emergence of an uncertain environmental health risk.

Methods: Somatic symptoms were assessed in two general population environmental health cohorts (AMIGO, n=14,829 & POWER, n=951) using the somatization scale of the four-dimensional symptom questionnaire (4DSQ-S). Exploratory bi-factor analysis was used to determine the factor structure in the AMIGO cohort. Multi-group and longitudinal models were applied to assess measurement invariance. For a subsample of residents living close to a newly introduced power line (n=224), we compared a uni- and multidimensional method for the analysis of change in reported symptoms after the power line was put into operation.

Results: We found a good fit (RMSEA=0.03, CFI=0.98) for a bi-factor model with one general and three symptom specific factors (musculoskeletal, gastrointestinal, cardiopulmonary). The latent structure was found to be invariant between cohorts and over time. A significant increase (p<.05) was found only for musculoskeletal and gastrointestinal symptoms after the power line was put into operation.

Conclusions: In our study we found that a bi-factor structure of somatic symptoms reports was equivalent between cohorts and over time. Our findings suggest that taking this structure into account can lead to a more informative interpretation of a change in symptom reports compared to a unidimensional approach.