Harmfulness and prevalence of sex between therapist and patient are difficult to research because accurate rate of harm in the general population cannot be established. Outrage at professional-ethics violations must be separated from symptom assessment, and should not alter validity standards adopted by authors of clinical reports.
Many inferential problems are inherent in systematic research on the harmfulness and prevalence of sex between therapist and patient.1,2 Sampling bias and resultant limitations on the ability to generalize findings are substantial methodologic issues. Because of these inferential problems, conclusions regarding the degree of harm caused by therapist-patient sex can be difficult to draw. In fact, despite personal convictions to the contrary, the general ability of sex with a therapist to cause damage beyond the effects of the patient's pre-existing condition has not been substantiated empirically; neither has the ability of such sexual contact to cause more harm than that caused by sex between patient and nontherapist.1
Three familiar phenomena contribute to the inferential difficulties concerning the damage caused by therapist-patient sex either before or after termination of treatment: (1) degree of symptom expression sometimes changes unpredictably over time, which makes it difficult to attribute poorer psychologic functioning to a particular cause; (2) psychologic condition may be attributed to noncausal precursors, and such misattribution may occur in a reliable and therefore especially misleading way; and (3) sex between patients and nontherapists may be a clinically significant cause of subjectively perceived and objectively measured psychologic harm, which leads to questions about how much harm can be expected to result from any sexual relationship. These three phenomena may cause insurmountable inferential difficulties that preclude valid attribution of harm to prior sex with a therapist based on later interviews with the patient.1
Despite these inferential problems, I1 and others have asserted that a clinical picture of harm from sex with a therapist does exist. However, Brown et al.3 have led me to reconsider the extent to which clinical observations can legitimately substantiate cause-and-effect patterns of behavior. The Brown et al3 article was coauthored by nearly every important contributor to the empirical literature in English on therapist-patient sex. The article3 claimed that clinical findings substantiate what more systematic empirical studies do not: that sex with therapists after termination of therapy harms patients. Regarding Applebaum and Jorgenson's4 assertion that harm from such relations has not been substantiated by valid empirical studies, Brown et al. stated:
Dr. Applebaum and Ms. Jorgenson also argued that there is no strong evidence for harm as a result of such relationships. While we would agree that the published literature is short on such information, and consider ourselves partly responsible for not having collected and published the information we have gathered in the course of our clinical work, our impressions are otherwise. That is, we have seen the same type and severity of harm devolving from posttermination relationships as from those initiated by therapists with less effective impulse control (p. 980).
Would gathering and publishing all instances of patient harm that these authors observed clinically have made a difference? I propose that such publication would not, and should not, affect general conclusions concerning the consequences of therapist-patient sex after termination of therapy. My reasons may appear self-evident, but they require review here because, perhaps especially in the area of therapist-patient sex, concern about patient victimization may cloud otherwise clearly recognized limitations on inferences that can be drawn from clinical observation.
FACTORS THAT LIMIT USEFULNESS OF CLINICAL REPORTS
Three factors limit usefulness of clinical reports of harm for drawing general conclusions about the harmfulness of therapist-patient sex: