Much concern has arisen over the number and types of medically unexplained somatic symptoms reported by Gulf War Veterans. In order to better understand these concerns, at least six teams of researchers have each analyzed a separate data set of symptoms reported by separate cohorts of Gulf War veterans using slightly different implementations of factor analysis. To compare the reported results and understand their substantive importance, we have begun a series of evaluations to assess the differences in: (1) the populations studied; (2) the subjects within a data set (e.g., deployed vs. not deployed); (3) the data structure (e.g., binary vs. ordinal); and, (4) the specific implementations of factor analysis (e.g., principal components vs. principal factors, orthogonal vs. oblique rotations). In this study, we summarize the data and methods used in each previous study and apply the six published factor analysis implementations to symptom data reported by two different populations, a random sample of the Veterans Administration's Gulf War Registry members and a sample of US Air Force troops. In the first data set, the results from all methods are very similar, showing factors that reflect psychological-fatigue symptoms, musculoskeletal symptoms, gastrointestinal symptoms and respiratory-throat symptoms. Results based on the CDC data are slightly more variable among methods, but remarkably similar overall for descriptive purposes. Variations may results if these factors are used to classify individuals. We find the general agreement among results reassuring methodologically.
Learning Objectives: N/A
Keywords: Methodology, Gulf War
Presenting author's disclosure statement:
Organization/institution whose products or services will be discussed: None
I do not have any significant financial interest/arrangement or affiliation with any organization/institution whose products or services are being discussed in this session.