Rosen et al observed that the clinical presentation of PSTD is not restricted to those who have experienced severe trauma, that patients who are traumatised do not necessarily develop PTSD and that PTSD is often misdiagnosed. ^{Reference Rosen, Spitzer and McHugh1} We would add that there is almost no evidence that PTSD is reliably diagnosed in ordinary clinical settings. In our naturalistic study of expert reports about psychological injury after motor vehicle accidents, we that found that the agreement about the presence of PTSD by experts engaged by the same side in the court case was little better than by chance. ^{Reference Large and Nielssen2} Most of the disagreement seemed to be due to selective use of the diagnostic criteria, although there was also difference in opinion about the severity of the patients' experiences and hence whether they met the ‘A’ criteria.

A search of PubMed, PsychLit and CINHAL did not locate any studies to show that PTSD can be reliably diagnosed without the use of a structured or semi-structured interview. The DSM–III and ICD–10 field trials did not report the interrater reliability of PTSD ^{Reference Spitzer, Forman and Nee3,Reference Sartorius, Ustun, Korten, Cooper and van Drimmelen4} and the DSM–IV trials restricted the examination of the reliability to the rating of audiotapes of 25 patients' responses to the PTSD module of the Structured Clinical Interview for DSM. ^{Reference Kilpatrick, Resnick, Freedy, Pelcovitz, Resick, Roth, van der Kolk, Widiger, Frances, Pincus, Ross, First, Davis and Kline5} Furthermore, we have not been able to ascertain whether the very high kappa scores reported in the DSM–IV trials (κ=0.85) included a correction for the loss of degrees of freedom arising from the use of the same ratings in multiple-rating pairs.

Although there are numerous studies confirming the interrater reliability of various diagnostic instruments, many of the instruments are only administered when the patient is suspected of having the disorder, and their ability to reliably distinguish PTSD from other disorders is not well established. Despite their limitations, we support the call of Miller ^{Reference Miller6} for the routine use of diagnostic interviews, as there is no evidence the disorder can be reliably diagnosed in any other way.

Rosen, Spitzer & McHugh call for DSM–V criteria that reflect research findings and limit the potential for misuse of the diagnosis. We believe that the logical step would be the complete removal of the A criteria. This would separate the clinical assessment of the patients' psychological state from issues of causation and minimise pre-emptive decisions about the cause and nature of the patient's distress. This new disorder, which could be called ‘phobic memory disorder’ or another name that does not imply a particular cause, could then be diagnosed in the usual way. As there are likely to be few objective features of the disorder, the diagnosis should be made using a semi-structured interview for the new criteria. Causative factors, including the role of trauma, premorbid conditions and litigation, would be considered in the same way as in other disorders.