A longitudinal bifactor approach to modelling somatic symptom development in psychosomatic treatment

Persistent somatic symptoms display a large public health problem and are often treated or addressed alongside other conditions in psychosomatic medicine units. With growing evidence pointing to unfavorable central nervous processing in persistent symptom perception, we aimed at modelling somatic symptom clusters together with an overarching general factor (g-factor) longitudinally during psychosomatic treatment.

This study analyses data from the Multicenter Effectiveness Study of Inpatient Psychosomatic-Psychotherapeutic Treatment in German University Hospitals (MEPP, 19 academic psychosomatic medicine departments with inpatient and day clinics, N = 2094 patients with ICD diagnoses between F30 and F69). The PHQ-15 was used to construct bifactor models of symptom perception. We used structural equation models to investigate how the g-factor and symptom-group clusters developed from treatment intake over discharge to a one year follow-up. Factor scores were used to evaluate treatment effects on the individual components.

As expected, the highest treatment effects, from a bifactor model perspective, were found for the overarching affective-motivational component (η_part²=0.277) and for fatigue (η_part²=0.250). All components predicted themselves at future timepoints: autoregressions were high, significant, and explained large proportions of the variance (all β>0.69,p<.001).

The bifactor model of symptom perception can be longitudinally applied in a large clinical sample with various diagnoses. Psychosomatic treatment showed effects on all components, both those similar to an overarching g-factor and those related to the perception of physiological signals, even when somatic symptoms were not the main target of the treatment.