Psychometric Evaluation of the Swedish Version of the Prolonged Grief Disorder-13 (PG-13) in a Bereaved Mixed Trauma Sample

Background: This study aimed to examine the psychometric properties of the Swedish PG-13 in a bereaved trauma exposed sample. A second aim was to examine the latent structure of prolonged grief using the PG-13. Methods: The participants were adults (n = 123) taking part in an ongoing longitudinal study regarding the effects of potentially traumatic events. Participants had experienced a potentially traumatic event in the past 5 years and had reported a death of a significant other either as their primary traumatic event or in addition to another traumatic event. Assessment included self-report of prolonged grief, posttraumatic stress, and general psychological distress. Clinical interviews were used to assess depression, posttraumatic stress disorder, and disability level. The psychometric properties of the Swedish PG-13 were examined through reliability tests and assessment of associations with symptoms of posttraumatic stress, depression, general psychological distress, and disability level. Principal component analysis (PCA) and confirmatory factor analyses (CFA) were used to assess the latent structure. Results: The internal consistency (Cronbach's α = 0.86) and test-retest (r = 0.86) reliability were good. PCA suggested a three-factor model as descriptive of the latent structure of the instrument. Therefore, the CFA used this model, as well as two models suggested in the literature. The three-factor model had the best fit to data. Support of concurrent validity of PG-13 was shown by moderate positive associations with measures of posttraumatic stress, depression, and general psychological distress. Conclusions: The Swedish PG-13 demonstrated good psychometric properties, and its use in research and practice to assess prolonged grief was supported. The factor analyses provided stronger support for models with two or three factors, as compared with a unidimensional model of prolonged grief, with the three-factor model having the best fit.