To the Editor: We read with interest the timely article by Dr. Marshall et al. on the newly proposed diagnosis of acute stress disorder and were dismayed by the conclusion that dissociation should be eliminated as a core symptom of acute stress disorder. The usefulness of acute dissociation in predicting PTSD is a complex topic containing several possible roles for dissociation: as a sole predictor, as the most important predictor, as a valuable additional predictor that adds prognostic information, or as an associated feature of limited specific value. Prediction in itself is complicated and encompasses sensitivity, specificity, and positive and negative predictive values. Reports on the topic typically address only aspects of these questions, and the authors do not systematically flush these out, as if set on disproving the importance of many positive findings regarding dissociation.

A good number of well-designed studies, reviewed in this article and others (1), have shown that acute dissociation predicts not only a higher likelihood of PTSD but also greater severity and chronicity. A recent study (2) teased out the contribution of the various acute stress disorder symptoms and clusters to the prediction of developing PTSD. The four clusters (dissociation, reexperiencing, avoidance, and arousal) are comparable in accuracy in predicting PTSD; avoidance leads to the most accurate classification, followed by dissociation. The study concluded that there may be two independent factors increasing the risk of PTSD: one captured by high levels of reexperiencing and arousal and one captured by the acute stress disorder diagnosis itself.

Furthermore, how well it predicts PTSD should not be the sole criterion for keeping dissociation in the acute stress disorder diagnosis. We need to know how the "pathological" acute response to trauma looks, irrespective of what it predicts. If the acute response to trauma has more dissociative features than the later response, that is interesting and needs to be understood. Many PTSD studies unfortunately continue not to measure dissociation, so the temporal evolution of dissociative symptoms is poorly known. In addition, acute dissociation may predict other future psychopathology—dissociative (3) and general. This subject also needs more study.

In conclusion, the reduction of dissociation to an associated feature of acute traumatic reactions is premature and possibly erroneous. The number of required dissociative symptoms in acute stress disorder could be decreased if the current criterion is too stringent. Better still, a diagnostic broadening that more realistically captures the richness and diversity of trauma-related syndromes (4) should be considered. One possibility is two subtypes of acute and chronic stress disorders: with or without prominent dissociation, or predominantly dissociative versus predominantly PTSD-like (1). PTSD as currently defined is not all that happens after exposure to trauma.