Abstract
In the first decade of the 20th century, unable to distinguish between true and false memories, Sigmund Freud withdrew from investigating the link between sexual trauma and psychopathology. Freud's doubts so affected mainstream psychiatry that, until 25 years ago, incest was regarded as a rarity, and dissociative disorder an artefact created by medical interest.
The situation is very different today. Adolescent patients are routinely questioned about sexual abuse, something that would have been regarded as an outrage a generation ago. There has been a flood of scientific papers concerning the prevalence of sexual abuse and its link to psychiatric disorder. Recently, however, there has been a decided counter-reaction to this trend, with questions concerning the alleged connection between sexual abuse and psychopathology [1], doubts about the validity of the concepts of repression and dissociation [2], concerns about a sexual abuse witch-hunt said to be fomented by radical feminist therapists [3], and criticism of the validity of the diagnosis of multiple personality disorder [4].
This paper will review the scientific literature concerning the prevalence and psychopathology of child sexual abuse, discuss the methodological problems that abound in this area, and summarise theoretical models concerning the interaction between the factors that lead to adaptive or maladaptive outcome following abuse.
The prevalence of sexual abuse
Estimates of the prevalence of sexual abuse are derived from two flawed sources: contemporary official reports of child abuse; and retrospective surveys of adults. In the USA, of all substantiated cases of child maltreatment, about 15% or 150 000 per year (2.4/1000 children) relate to sexual abuse, and of these 50% involve genital penetration [5]. In 1992–1993, 59 122 cases of child maltreatment or neglect were investigated in Australia [6]; 27 196 were substantiated, and 5979 (22%) were cases of sexual abuse. If the figures available from retrospective surveys of adults are valid, these figures are a gross underestimate of true prevalence rates.
Finkelhor's [7] review of 19 retrospective surveys of adults with regard to their experience of unwanted sexual contact as minors, found prevalence rates of 1–16% in men and 2–45% in women, variations probably due to methodologic differences. A conservative estimate places the prevalence of genital penetration as 1.5% in men and 5.0% in women. In a recent New Zealand study, Anderson et al. [8], using both questionnaires and interviews with 3000 women randomly selected from electoral rolls, found an overall prevalence of 31.9%: of these, 19.8% reported genital contact, and 7.3% reported attempted or actual sexual intercourse. The age of greatest risk was 8–12 years. Only 15.0% of the abuse involved strangers; 38.3% of the abusers were relatives; 46.3% were family acquaintances. Twenty-eight per cent of the abused women had never disclosed the abuse.
Is there evidence of a recent increase in the prevalence of sexual abuse? In the study of Anderson et al., the overall rate of sexual abuse did not differ according to rural/urban residence or decade of birth, suggesting a stable incidence over 50 years [8]. Feldman et al. [9] compared data from the Kinsey report (1953) [10] with 19 recent prevalence studies, and could find no evidence of increased prevalence over 40 years.
Is there any support for the notion that public hysteria has provoked a widespread witch-hunt? Finkelhor and Williams [11] point out that, if there is a witch-hunt, in the USA at least, it is not reflected in rising substantiation rates. Over 80% of all allegations against daycare operators are dismissed. Only about 40% of substantiated abuse allegations are forwarded for prosecution, from which about 75% of defendants are convicted. Of offenders convicted, 32–46% serve no jail time.
Methodologic issues
Briere and Elliott [12] have provided a useful critique of research in this field. They point out that cross-sectional studies cannot distinguish between the impact of sexual abuse and the effect of other stressors (e.g. neglect, physical abuse, exposure to domestic violence, parental psychopathology and marital dissolution) that commonly precede or accompany sexual abuse. Nor are longitudinal designs comparing the development of abused with non-abused children free of problems (e.g. the method of selection and matching, the effect of repeated measurement, excessive attrition, and the uncontrolled effect of any subsequent treatment the subjects may have had). Retrospective cross-sectional studies are affected by report biases, different definitions of sexual abuse, the difficulty of choosing appropriate controls, the difficulty of accounting for concurrent stressors, constraints on generalisations, statistical problems, and the impossibility of determining causal direction.
Report biases may be caused by retrospective amnesia for abuse, unwillingness to disclose abuse, the loss of memory over time, and, conversely, the possibility of confabulation. Retrospective studies can seldom guarantee the validity of the criterion variable, sexual abuse.
Reliable and valid abuse-specific measures are required. Multivariate analyses and structural equation modelling of the data from longitudinal studies also assist the sifting out of the relative contributions to outcome of sexual abuse in contrast to the other stressors that often accompany it.
Adult outcome
There have been many reports of an association between child sexual abuse and adult psychopathology. All suffer from the methodologic problems already described. Sexual abuse has been linked to chronically low self-esteem [13,14], a sense of helplessness and self-hatred [15,16], and disruptive interpersonal relationships [17,18]. Chronic emotional distress, particularly chronic depression and anxiety [19,20], eating disorders [21,22], suicidal tendencies [23,24], and self-mutilation [25] have been reported as linked to sexual abuse, as have prostitution [26,27], running away from home [28], delinquency [29] and substance abuse and alcoholism [30–32].
Approximately 60–83% of patients with dissociative identity disorder are thought to have been sexually abused [33–35]. A causal link has been hypothesised between sexual abuse and borderline personality disorder [36,37], pseudoseizures [38,39], somatisation disorder [40], chronic pelvic pain [41], having had many medical or surgical procedures [42,43] and adult sexual dysfunction or dissatisfaction [44,45].
A number of researchers have found that adults with a history of sexual abuse are at risk of being raped [46,47], of being the victims of coercive sexual experiences [48] and of experiencing domestic violence [49]. Groth and Burgess [50] reported an increased prevalence of histories of sexual abuse among rapists and child molesters.
In a follow up to the New Zealand study already cited [8], 254 women in a random community study reported having been sexually abused [51]. Compared with controls, these women were more likely to have a psychiatric disorder, low self-esteem, sexual problems or difficult intimate relationships; to have become pregnant as adolescents; and to have been separated or divorced. However, child sexual abuse appeared to operate as a non-specific psychosocial risk factor acting in concert with other risk factors (e.g. parental psychopathology, poor relationship with parents, physical abuse).
In the Christchurch 1000-birth-cohort study [52], child sexual abuse involving genital penetration was found to contribute to the risk of psychopathology over and above concurrent stressors such as physical abuse and neglect.
Childhood outcome
Posttraumatic stress disorder has been found in 20–70% of children with substantiated sexual abuse [53,54]. Terr [55] has proposed that the posttraumatic stress disorder following repeated sexual traumatisation is more likely to be associated with dissociative symptoms than is posttraumatic stress disorder caused by single traumatic events. Impulsive, poorly controlled behaviour (e.g. aggression, hyperactivity and developmentally inappropriate sexual behaviour) and emotional distress (e.g. fear, anxiety, depression, somatic complaints) are often found in children repeatedly sexually traumatised [56,57]. Sexually abused children often exhibit low self-esteem, and a pervasive sense of having been damaged or of having been responsible for the abuse [58]. Dissociative symptoms [59] (amnesia, daydreaming, trances, blackouts, multiple personality) and dramatic conversion syndromes [38,39] (e.g. pseudoseizures) have been described. On the other hand, the prevalance of asymptomatic patients has varied from 21–49% [60]. One study [61] found that 30% of children who were initially asymptomatic exhibited symptoms 18 months after disclosure.
Risk factors
Not all children who have been sexually abused manifest psychiatric symptoms or functional impairment. Some may be resilient in the face of severe stress. For others, the abuse, though upsetting, may have been of minor significance; whereas, for a significant minority, an initial lack of symptoms may be succeeded by late-onset psychopathology. What are the factors that convey an increased risk of poor outcome? These will be discussed under the following headings: (i) Antecedent factors, (ii) Abuse-related stressors, (iii) Post-disclosure stressors and (iv) Mediating factors.
Antecedent factors
The quality of family functioning prior to the sexual abuse (e.g. if the child experienced emotional neglect, emotional abuse, physical abuse, exposure to domestic abuse, marital dissolution of the parents, exposure to parental psychopathology or parental substance abuse) are likely to affect the outcome of the sexual abuse. Due to the methodologic problems associated with retrospective research, little is known of this matter. However, in the Christchurch longitudinal 1000-birth-cohort study, Fergusson et al. [52] found that sexual abuse involving genital penetration contributed over and above family factors operating before and concurrently with the childhood abuse (i.e. parental conflict, impaired parent-child relationships and parental adjustment problems) to psychiatric disorder at 18 years (particularly major depression, anxiety disorder, conduct disorder, substance abuse disorder, and suicidal behaviour).
Abuse-related factors
The following characteristics of the abuse experience are associated with less favourable outcome: the propinquity of the abuser, the duration and frequency of the abuse, genital penetration and coercion or threat [56]. Either there is no evidence about the following factors, or the evidence is inconsistent: age, sex, number of perpetrators, denigration of the victim, or the offender's insistence that the victim preserve secrecy about the abuse [56].
Events subsequent to exposure
The following post-disclosure events are related to psychopathology: lack of support from the non-offending parent [62,63] and giving testimony [64]. There is no evidence to implicate as significant risk factors parental separation, out-of-home placement, or medico-legal investigations.
Mediating factors
Mediating factors refer to the child's attitude to self and others and to his or her coping style. The following attitudinal factors are associated with an increased risk of psychopathology: negative self concept [65], self-blame [66] and external locus of control [67]. The evidence is unclear with regard to other potential risk factors (e.g. self-stigmatisation, perception of physical damage, threat to family integrity, ambivalence or hostility to parental figures, and the sense of powerlessness or betrayal) [65,66].
With regard to coping style, the following defensive patterns have been associated with psychopathology: suppression, denial and avoidance of the perpetrator [68], dissociation, detachment and fantasising [69], conversion and somatisation [38,39], acting-out and externalisation [70], alcohol ingestion and drug abuse [31,32], and repetition–compulsion, re-enactment [46,47] and reversal of victimisation [50]. On the other hand, some children, but not others, find disclosure, support-seeking and the pursuit of retribution to be successful coping strategies [71]. Cognitive restructuring, the relinquishing of maladaptive coping strategies, and reappraisal of the self are associated with favourable outcome [71].
Hypothetical models concerning the effects of child sexual abuse
Early psychoanalytic and behavioural models of the psychopathology of sexual abuse have been replaced by competing models derived from family systems theory, traumagenic dynamics, developmental coping systems, attachment theory, information processing theory, and the transactional theory of development.
The child sexual abuse accommodation model
This theory was proposed by Summit [72] in order to explain why some abused children delay for long periods their disclosure of abuse, and why, having disclosed it, they often recant. Summit contended that the secrecy accompanying the abuse affects the child's self-concept. Coerced by the abuser, the child is placed in the situation of feeling responsible for protecting the family from dissolution, and of acquiescing to the abuse, shifting blame from the offender to herself. Thus, the child accommodates to the abuse at least for a time. This model does not usually apply to extra-familial abuse, and it does not account for the traumatic state often encountered in abuse victims.
The posttraumatic stress disorder model
As already described [53,54], posttraumatic stress disorder is highly prevalent in victims of sexual abuse. Posttraumatic stress disorder is a complex biopsychosocial condition caused by the stress of an inescapable, potentially lethal situation and involving a persistent derangement of central nervous system corticosteroid, catecholamine, and opioid peptide metabolism. Posttraumatic stress disorder caused by repeated abuse is said to be more likely than single-event trauma to be accompanied by dissociation [55]. Unassimilated traumatic memories are hypothesised to emerge in the form of nightmares, intrusive imagery or somatic symptoms. Posttraumatic stress disorder can be delayed, or reactivated by subsequent stressors. However, posttraumatic stress disorder is not found in all cases of sexual abuse [56,57], and it does not account for the maladaptive attributions to self and others that are so often found in abused children.
The traumagenic dynamics model
Finkelhor [73] proposed that sexual abuse victims are characterised by traumatic sexualisation, and a pervasive sense of powerlessness, stigmatisation and betrayal. However, Finkelhor's model failed to stipulate which aspects of sexual abuse are related to which aspects of the traumagenic dynamics.
The developmental coping model
Cole and Putnam [74] suggested that incest so disrupts primary attachment relationships as to jeopardise self-definition, self-integration, self-regulation, and the capacity to trust others. Incest during the preschool years may be dealt with by denial or dissociation, compromising the organisation and regulation of self. Incest during middle childhood causes intense guilt or shame and interferes with social relationships. Incest during adolescence impedes identity formation and may be dealt with by externalisation (e.g. in the form of precocious sexual behaviour, substance abuse, or running away from home). Theoretically plausible though this model may be, there is as yet no empirical evidence concerning the attachment patterns of sexually abused children.
The attachment disruption model
Alexander [75] related the impact of sexual abuse to distortion of the child's working models of attachment. The lack of attunement of the abused child's parents to the child's needs may cause the child to deny feelings about the abuse. Some children experience parental role reversal, or are affected by the parent's own insecure model of attachment. As with the developmental coping model (with which it overlaps), there is a paucity of scientific evidence for this model.
The psychobiologic information processing model
Hartmann and Burgess [76] proposed a neurobiologic information-processing model. When the limbic system is overloaded by the stimulation associated with sexual abuse, an alerting system is triggered. If the alerting system is unsuccessful in coping with overload, the survival response of dissociation is activated. Dissociation occurs when there is no escape, by flight or fight, from threat. Over time, the limbic system becomes hypersensitised, the alarm system is triggered by reminders of threat, and the normal arousal response is dysregulated. Dissociation leads to amnestic gaps or a split between traumatic memories and their accompanying feelings. Four phases are described: (i) pretrauma; (ii) trauma encapsulation; (iii) disclosure; and (iv) post-trauma outcome. At any of these stages, favourable or unfavourable influences can affect outcome. This model is an extension of the posttraumatic stress disorder model and is subject to the same criticism.
The transactional model
Spaccarelli [77] hypothesised that the outcome of sexual abuse is determined by a transactional matrix of variables. Spaccarelli contended that sexual abuse should be construed not only in terms of the abuse experience itself, but also in terms of the effect of the abuse on the child's family and community. The environment shapes the child's resources and vulnerabilities, and, in turn, interacts with them. The risk of poor outcome following abuse increases as a function of the stress involved in the abuse experience and the reaction of the family and community to the disclosure (or non-disclosure) of the abuse. Ultimately, the effect of the abuse stress and the environment's reaction to it is mediated through the child's appraisals of self (e.g. as ‘bad’) and others (e.g. as ‘betraying’) and the child's coping strategies (e.g. avoidance, support-seeking or cognitive restructuring). Abuse transactions are moderated, also, by factors that precede the abuse (e.g. developmental level, attributional style) and the prior support of the family. Thus, the outcome of abuse is determined by multiple, reciprocal transactions between moderating variables, the abuse experience, and mediating variables. This model is the most encompassing of the seven models postulated. In another paper appearing in this issue [78], a modification of the transactional model is presented on the basis of which an experimental treatment project is designed.
Summary
Child sexual abuse affects a significant number of children of both sexes. In only a minority of cases is the abuse disclosed and investigated. Although some children appear to be resilient, a significant number manifest psychiatric disorder following disclosure. The symptomatology exhibited by these children is so varied that it is not possible to sustain the concept of a specific syndrome related to sexual abuse.
Childhood sexual abuse has been linked to a number of psychiatric disorders and maladaptive lifestyles in adulthood. However, many studies of adult outcome are methodologically weak. For example, it has been difficult to disentangle the effect of sexual abuse from that of the other family problems that frequently accompany it (e.g. domestic violence, neglect, physical abuse, parental psychopathology and parental separation). Recently, however, longitudinal studies of whole-population cohorts have begun to unravel this question. It is likely that serious sexual abuse does contribute to maladaptive outcome to a degree beyond that associated with the other forms of family pathology.
A number of hypothetical models have been proposed to explain the interaction between risk variables, protective variables and outcome. None of these models is completely satisfactory. Since sexual abuse is an experience, not a psychiatric disorder, it is difficult for a single model to represent all the variations on this theme. Nevertheless, it is important to construct theoretical models, for it is on them that rational treatment programs should be based. Treatment thus becomes a test of the hypothetical model. In another paper appearing in this issue [78], a hypothetical model of the psychopathology of sexual abuse will be presented as the basis of treatment.