To the Editor: The article by Read and Argyle (1) in the November 1999 issue seeks to relate hallucinations, delusions, and thought disorder among inpatients to types of childhood abuse, including sexual abuse. The study method involved reading the records of 100 consecutive admissions. According to the abstract, 22 records in which a history of abuse was mentioned were selected for study. However, the veracity of the abuse histories is not clear from the information presented in the text. The report does not specify when, from whom, and by whom the history was first obtained; how the likelihood of actual abuse was judged; and whether and how it was in any way corroborated.
Physical and sexual abuse are significant problems in our society. Nevertheless, determining the role of early traumatic experience in a patient's current presentation is often extremely complex, especially with sexual abuse. Conclusions about the relationship between psychotic symptoms and a recorded history of abuse need to be based on actual abuse, the likelihood of which is not always determinable.
Given the authors' statement that "the paucity of information on symptom content and the identity of the perpetrator of abuse reduced the number of records available for analysis," it would be helpful to know what inclusion criteria, if any, beyond mention of abuse were used to establish it in the selected records. If someone once taking a history considered that abuse might have occurred, its presumed occurrence eventually might have been translated into definite abuse in the subsequent record.
The study refers to the high risk of type I errors, but this cautionary note appears to apply to interpretations based on the statistical tests used, not to the selection of cases. In people's minds, both true and false information initially tends to be represented as true (Spinoza's hypothesis), and once a proposition is so represented, it is not readily changed from true to false (2).
The authors further claim that the prevalence of abuse is significantly underestimated when it is based on records alone and that "this study involved only the minority of abused patients whose records indicated they had been abused." Such an underestimation might occur; however, several related studies of abuse, particularly sexual abuse, appear to base prevalence rates on inclusive criteria that likely also select false positive cases. For example, in an article cited by the authors, a 46 percent rate of childhood incest reported among a series of chronic female inpatients drops to 15 percent when cases scored as less than highly likely are excluded (3).
On the one hand, the occurrence of abuse may be missed; on the other hand, increasingly it may be misdiagnosed (4), perhaps as an overcorrection for past omissions. We all naturally decry overlooking cases of abuse, but for clinical and scientific reasons we also need to be judicious in drawing conclusions about its presence. Unless there is convincing reason to conclude that the selected records in this study do not include false positive abuse cases, the correlation of certain schizophrenic symptoms with types of abuse may be unfounded.
Alternatively, it is possible that patients with spectrum variants of bipolar disorder are more likely to report a history of early abuse, especially sexual abuse. If so, this tendency would be consistent with a greater likelihood of their having so-called pseudohallucinations and sexualized or suicidal mental content, as well as socially engaging behavior (3). It therefore would be useful to know how the phenomenology of reported sexual abuse among inpatients correlates with the results of a full diagnostic assessment.
Dr. Good is associate clinical professor of psychiatry at Harvard Medical School in Boston and staff psychiatrist at Westborough (Mass.) State Hospital.