Factor Structure of Secondary Traumatic Stress: A Comparison of Four Models

Background and Purpose: Much empirical attention has been given to understanding the underlying factor structure of PTSD. However, little has been done to examine these factor structures as they relate to secondary traumatic stress (STS). The conceptualization of PTSD has come under scrutiny in the literature as factor analytic studies have failed to provide support for the three factor structure leading to the development and testing of new models. Two four factor models have received substantial attention and support, the King et al model (1998) based on empirical evidence indicating that numbing and avoidance symptoms represent two unique set of symptoms and the Simms et al. (2002) model, containing the three factors identified in the DSM-IV-TR with the addition of a fourth factor, dysphoria, representing general negative affect. Studies examining the factor structure of STS have found validity in two models, the three factor model provided by the DSM-IV-TR and a one factor model with STS as a unidimensional factor. The present study compared the factor structure of these four models applied to STS using confirmatory factor analysis in a sample of 796 substance abuse counselors.

Methods: Data were collected through the NIDA Clinical Trials Network (CTN), a collaboration between researchers and practitioners aimed at testing the effectiveness of substance abuse treatment techniques sponsored by the National Institute on Drug Abuse. All counselors within the CTN were mailed a questionnaire with the Secondary Traumatic Stress Scale and received a US$40 incentive payment upon return.

Results: Study participants had a mean age of 46 years and were primarily female and Caucasian. Over 47 percent of participants indicated that at least half of their current caseload were trauma survivors and 48 percent indicated at least a quarter of their current caseload had posttraumatic stress disorder. Scores on the STSS ranged from 17-72 with a mean of 28.5 (SD = 10.1). Maximum Likelihood estimation was used for identifying parameter values and chi-square tests and fit indices were used to examine model fit. Results revealed the King model and the DSM-IV-TR model to be misspecified due to highly correlated factors suggesting a lack of fit for each of these models. Both the Simms dysphoria model and the undimensional model produced sufficient CFI, NNFI, RMSEA, IFI and SRMR values indicating good fit. The AIC value was examined to compare the two models. The Simms dysphoria model had the lowest AIC indicating the model best fit the underlying symptom structure of secondary traumatic stress.

Conclusions and Implications: These results are the first to apply emerging PTSD models to STS and find congruence to a four factor model. The findings also failed to provide support for separation of numbing and avoidance and for the current DSM-IV-TR symptom clusters. This has tremendous implications for the diagnosis of STS as a diagnosis relies on individuals experiencing symptoms from each symptom cluster. Future research should replicate these results and work towards developing an empirically based and adequate factor structure of secondary traumatic stress.