THE PERILS OF PREDICTION
It is a natural human instinct to seek explanations for the seemingly senseless cruelties that humans inflict on one another, particularly on the young and innocent. However, the recent outcry about paedophiles in the UK demonstrates that there is little public desire to understand or explain such behaviour at present. People who abuse children are viewed as beyond comprehension or redemption — they are ‘wicked, evil and perverted’ and should be ‘locked away’ forever. In the current climate, the paper by Glasser et al is a brave attempt to study the perpetrators of child sexual abuse and look for causal explanations for such behaviour. Many of us would shy away from such a topic or find it difficult to be dispassionate about these perpetrators. The authors propose that they have discovered “links between being a victim and becoming a perpetrator”. Indeed, a casual reading of this paper or a glance at the abstract might seem to indicate that children who are sexually abused are, in turn, likely to become sexual abusers of children in adulthood — what the authors refer to as “the cycle of child sexual abuse”. But blaming the victim is a risky business and this so-called ‘link’ may lead to further stigmatisation and distress for individuals who have been abused in childhood. Therefore, before reaching this conclusion, we should examine carefully the problems associated with establishing causality from the data presented in this paper.
The study is based on a sample of adults attending a forensic psychotherapy centre for treatment of sexual deviation and delinquency. Using mainly case note information, the authors estimate that 35% of the male child sex abusers in this sample were themselves victims of childhood sexual abuse compared with 11% of the males who had not abused children. The data are further analysed according to type of self-reported childhood abuse. The authors find that 51% of the males who reported that they had suffered incest during childhood were also perpetrators of child sexual abuse, 61% who reported childhood paedophile abuse were perpetrators and 75% of those who reported both incest and paedophile abuse in childhood were perpetrators. The authors claim that this “ stepwise” increase in the “risk” of being a perpetrator according to childhood abuse experience provides support for a “ victim to victimiser cycle”. But does it? Let us examine some other explanations for this effect.
Sources of bias
This is a highly-selected sample based on a national tertiary referral centre. Referrals came from a variety of sources including medical, psychology, psychotherapy, social and probation services. By the time of their assessment at the clinic, these individuals had managed to make their way through this complex referral system and may not be representative of all child sexual abusers.
The use of self-reports of childhood sexual abuse from individuals who are, themselves, child abusers is bound to raise questions about the reliability of these data. Reporting sexual victimisation in childhood may be an attempt on the part of the perpetrators to explain their abusive behaviour and elicit sympathy from the therapist. Glasser et al used information in social service, probation and medical reports to complement self-report data for the assessment of perpetrator status but it is not clear whether this was also done for victim status. Recall bias is usually thought of as an unconscious action, a form of ‘search after meaning’ rather than deliberate lying or fabrication, but all these processes will lead to biased associations. Additionally, a form of interviewer bias may have operated if the therapists subscribed to the idea of a cycle of child sexual abuse and were more likely to enquire about such childhood experiences during assessment of perpetrators.
Sources of confounding
The relationship between childhood sexual abuse and becoming a child abuser in adulthood may be confounded by childhood environment and family situation. Seghorn et al (Reference Seghorn, Prentky and Boucher1987) carried out a study with a similar design to that of Glasser et al but with access to more detailed objective information about childhood circumstances. They found that child sex abusers who were themselves sexually abused in childhood were more likely than those who were not abused to have had fathers with a criminal and/or substance use history, parents with psychiatric problems, sexual deviance within the family and a high incidence of childhood neglect. It is difficult to covary out the effects of sexual abuse alone from such global family pathology.
Glasser et al examined two measures of childhood adversity as potential confounders: a self-report of maternal emotional care and a measure of loss in childhood (parental divorce, separation or death). Poor maternal emotional care was related to victim status, and experience of parental loss was a significant predictor of later perpetrator status. Sexual abuse in childhood remained a significant risk factor for later perpetrator status when adjusted for both these other variables in the logistic regression analysis. However, adjusted estimates are not necessarily free from confounding and much depends on the adequacy of information about the confounder. Inadequate information will give rise to residual confounding even after statistical adjustment (Reference LeonLeon, 1993). “ Poor maternal emotional care” and “parental loss” are used as proxy measures for general childhood neglect and deprivation, but the complex mix of factors contributing to global family pathology or childhood deprivation cannot be accurately summed up in one or even two variables. In this study the higher “risk” of later perpetrator status associated with being a reported childhood victim of both incest and paedophilia, compared with incest or paedophilia separately, may actually reflect a greater degree of parental neglect and household chaos, rather than a true dose-response relationship with degree of childhood sexual abuse.
Complex causal models
There is (almost) never just one causal agent for a single outcome. Adequate prediction of later disturbance or illness requires consideration of multiple risks and protective factors and their interplay (Reference Cicchetti and SroufeCicchetti & Sroufe, 2000). We should not be asking “What is the reason that X becomes a child sex abuser?” but “What are the factors that have initiated and maintained X on a developmental pathway that is associated with becoming a child sex abuser?”. We should ask why two individuals with the same risk factor have different outcomes. An analogy of vulnerability to trauma has been described in the context of three dolls made respectively of glass, plastic and steel that were exposed to the same standard blow from a hammer (Reference Anthony, Rosenthal and KetyAnthony, 1968). Under impact the glass doll shattered, the plastic one showed a dent and the steel one remained unscarred. The retrospective design of Glasser et al's study does not allow us to estimate what proportion of children who have been sexually abused go on to become child abusers in adulthood, but it is, presumably, a small minority. Future research needs to examine cases where children appear to have overcome or been protected from negative consequences of their early childhood adversities. Such knowledge would have important implications for the development of prevention and treatment programmes for children who experience early victimisation.
The ideal way to answer these questions and look for links between childhood abuse and adult outcomes is to carry out longitudinal studies that carefully trace the developmental pathways of abused and non-abused children throughout life (Reference Kraemer, Yesavage and TaylorKraemer et al, 2000). It may be possible to speed up this long and costly process by using preexisting cohorts or official sources of information. Widom & Ames (Reference Widom and Ames1994) used a prospective cohort design to assess the criminal consequences of childhood sexual abuse. Official criminal records were traced for a large sample (n=908) of children with a validated history of sexual abuse, physical abuse or neglect, and a control group matched for age, gender, race and family socio-economic status. The authors found no evidence for a cycle of sexual abuse as proposed by Glasser et al. Rather, they found that all three of the abuse groups (sexual, physical and neglect) were significantly more likely to be arrested for a sexual offence (including prostitution) than were the controls. In fact, although children who had been sexually abused were significantly more likely than controls to have an adult arrest for prostitution, they were not significantly more likely than controls to have adult arrests for other sex crimes.
How little we know
No single study, no matter how large, can provide all the answers about the complex topic of child abuse. We need to know more about the psychological and functional outcomes of victims of childhood sexual abuse and we urgently need to know more about why some adults sexually abuse children. Epidemiological information on sexual ‘perversions’ is notoriously poor (Reference Meyer, Kaplan and SadockMeyer, 1995). The legacy of Freud encourages psychiatrists and psychologists to look to childhood for the roots of adult behaviours. We do not wish to be seen to blame the victim, but how can we protect children from adverse consequences of abuse if we do not even know what these are? How can we devise effective treatments or interventions for child abusers if we have no idea what drives their actions? Glasser et al, with this fascinating study which exposes psychoanalytic theory to epidemiological scrutiny, have opened the scientific debate on this important issue and shown us how little we really know.
eLetters
No eLetters have been published for this article.