Is factor analysis useful for revising diagnostic criteria for PTSD? A systematic review of five issues ten years...

Based on factor analysis research, DSM-5 revised the diagnostic criteria for posttraumatic stress disorder (PTSD) by increasing symptom clusters from three to four.

To question whether that is an appropriate use of factor analysis.

Reviewed the literature on five issues of factor analysis relevant to diagnostic criteria: (1) discovery of factors identical to symptom clusters, (2) consensus about the number of factors in best-fitting models, (3) configural variance between subpopulations to explain inconsistent model results, (4) methods to externally validate factors after discovery, and (5) treatment response of symptom clusters to externally validate factors. Two hundred four articles using DSM-IV or DSM-5 symptoms were included.

Two of four DSM-5 clusters were discovered with exploratory factor analysis. Support for a best-fitting model was inconsistent. Models with the highest number of factors were the best mathematical fit 87% of the time. Subpopulations did not reveal a pattern of configural variance to explain inconsistent findings. External validation of factors relied entirely on questionnaires. A review of 143 randomized controlled trials did not reveal differential treatment response of any symptom cluster.

Findings did not support the usefulness of factor analysis because the findings are too disparate to be helpful.