To the Editor: I agree with Mary C. Zanarini, Ed.D., and her colleagues R6715511CHDCEGDI that borderline personality disorder will ultimately be understood as an illness with a multifactorial etiology. In our own work, cited by Zanarini and elaborated in a series of reports R6715511CHDBIAHCR6715511CHDBEDIG, we have similarly emphasized that childhood sexual abuse is neither necessary nor sufficient for the development of borderline personality disorder but that the patient with this diagnosis has almost invariably emerged from an environment characterized by disordered attachments and histories of affective illness in one or more close relatives.

Earlier studies by Zanarini et al. R6715511CHDDDGFG and others R6715511CHDCEFGDR6715511CHDDFAIB support this etiological perspective. Still others R6715511CHDDEIFIR6715511CHDCDGIH further note that a history of sexual abuse, along with a cluster of severely self-destructive behaviors, most accurately characterizes those borderline patients who require hospitalization. Therefore, it is not surprising that for the majority (61.5%) of "severely impaired" inpatients in the study of Zanarini et al. R6715511CHDCEGDI, childhood sexual abuse would emerge as a predisposing factor in their illness. Unfortunately, the authors make no comment on the skewed nature of their study group and its possible relation to this finding.

Eagle R6715511CHDCFEFF has issued a general warning regarding the use of this class of restrospective data to establish the etiology of borderline personality disorder. Following Eagle’s logic, if researchers were to sample only hospitalized borderline patients, "they would likely develop an etiological theory seriously distorting and exaggerating the nature and strength of the relationship between childhood sexual abuse and adult personality disorder" (p. 134). Such distortion, if unquestioned, can have serious consequences for both research and clinical practice.

Zanarini and colleagues themselves call for a more complex, nuanced understanding of borderline personality disorder. In the absence of prospective developmental studies, a next step might be a careful comparison between predisposing factors in "severely impaired" inpatients like those in the Zanarini group and an equally large sample of outpatients who carry a borderline diagnosis but have not needed hospitalization. Such a comparison would help clarify those factors common to the majority of individuals suffering from borderline personality disorder, and would document the extent to which sexual abuse may be correlated with symptom severity, as opposed to diagnosis per se.