Posttraumatic stress disorder in abused women: research article critique

Trauma, Posttraumatic Stress Disorder Symptom Clusters, And Physical Health Symptoms in Postabused Women

Stephanie J. Woods and N. Margaret Wineman

The purpose of this research is to evaluate PTSD symptom clusters (avoidance of the situation, hyper-arousal, and intrusions) to physical health symptoms in women who have suffered abuse. The researchers are also interested in how lifetime trauma is related to the PTSD symptom clusters and physical health symptoms. The researchers make a fair case for the aims of the current study by pointing out that the majority of previous research investigating these relationships has been performed on male combat veterans (although they do discuss research with female veterans) and that a significant proportion of women in abusive relationships suffer from PTSD. Here you might be tempted to draw the conclusion that if PTSD war veterans experience health symptoms so should abused women, but this might not follow. Moreover, they do discuss inadequacies in past research and how it has not definitively demonstrated the relationship between the PTSD clusters and health symptoms.

The independent variable (IV) in the study is actually a bit hard to define. We would think that the major IV is having the symptoms of PTSD based on a history of abuse from a partner. However, this study takes an interesting turn by hypothesizing that a history of lifetime trauma affects the development of PTSD, which in turn is hypothesized to effect health. So in essence PTSD (positive or negative) is treated as both an IV and dependent variable (DV), lifetime history as IV, and health symptoms as a DV. Finally, abuse or trauma is an IV whether it is physical or non-physical. Variables are not clearly stated as such.

The majority of the articles cited within the literature review are recent to the article; they are within 10 years of the publication and are primary sources. However, the literature review is a bit scanty in that there are many citations, but few specifics. Moreover the researchers only review the literature for one PTSD cluster, hyper-arousal and its effect health. There must be more. The researchers are trying to investigate quite a view different of effects and we are looking at several measures and how they affect health and PTSD the review is inadequate.

The sampling plan is one of convenience, so we know that generalziability is poor. There was no evidence of coercion even though all participants were paid a nominal fee. The study design is purely correlational via the use of surveys. Therefore all one can determine is relationships, one cannot hypothesize cause. The measures all have Cronbach alphas over .7, except for the lifetime violent and nonviolent trauma scale, so most of the instruments are reliable. Acceptable validity coefficients are demonstrated for given for all of the measure except the lifetime trauma scale. There are few controls (e.g., the diagnoses of PTSD positive and negative is made via survey results and not on a clinical diagnosis). Therefore the outcome findings are somewhat suspicious and the conceptual framework of this study suffers a bit as a diagnosis of PTSD cannot be made on the basis of a survey finding. Other variables such as health concerns suffer the same problem. It is the participant's perceptions the researchers are getting and not actually clinical data. Moreover, there are no controls for experimenter demand.

Alpha is not explicitly stated, but we realize it is p < .05 when the authors discuss their findings. Given the number of variables and tests used the alpha is a bit too high as the number of statistical tests performed is high and this leads to a potential for Type I error in the findings. The statistical procedures include a phi coefficient (relationship between binary variables). We find that 2 of 13 symptoms differ significantly between PTSD positive and negative groups, not in line with the hypothesis that PTSD groups will have greater symptoms (by the way the authors state that one relationship with a p =.10 approaches significance, this is NOT true). Pearson correlations between trauma, PTSD clusters, and health found that lifetime violent trauma was moderately correlated with PTSD avoidance and increased health symptoms. Regression analysis revealed that violent trauma was a significant predictor of health symptoms. Although they do not mention it, but it can be determined from Table 4, nonviolent trauma was not a predictor. Then the authors performed several more regression equations with violent trauma and found that child physical and adult sexual abuse were predictors of health symptoms and six other regressions found that child physical abuse accounted for significant variance in health symptoms beyond the other 5 violent trauma types. P values are given for all analyses, but again, these are a lot of analyses with 50 subjects, typically you want 15 per variable at least and the researchers had 6 independent variables (trauma types) in the equations. Study findings are not clearly stated as they relate to earlier hypotheses.