To the Editor: Since the article by M. Tracie Shea, Ph.D., et al. (1) is not the only recent report in the literature of rapid change in personality found with the use of semistructured DSM personality instruments (2), it is important to discuss what may be important implications. Although the article by Dr. Shea et al. discussed some of these issues, I feel it is important that they be fleshed out.

The first issue to raise is whether semistructured DSM personality instruments are as accurate as we believe. Although much has been made of the difference in outcomes between semistructured and self-report instruments, I am not aware of any reports indicating that semistructured interviews agree with each other much beyond the diagnosis of any personality disorder. These instruments may have a wide variance in their measurements.

Next, we know that personality measurements are often influenced by state affects (3, 4). The variation we may be seeing may be the result of patients coming to clinics with more severe axis I symptoms that decline over time, taking down the state-dependent personality measure with it.

Of course, the most interesting possibility is that we may have wrongly conceptualized personality disorders as long lasting and immutable; they may in reality show a considerable fluctuation of symptoms. Related to this, we may not be dealing with one set of disorders but two. One set is long lasting and the second, "stress-induced" or "state" personality disorders, may be relatively transient (5).

Finally, there is the issue of treatment. Although the article by Dr. Shea et al. did not report such a finding, there are many reports of personality traits being treated by medications (6, 7), and this should be kept in mind in the interpretation of the results. I hope that this letter furthers productive discussion in this area.