Prévia do material em texto
Pergamon Child Abuse & Neglect, Vol 20, No. I, pp. 7-21, 1996 Copynght© 1995 Elsevier Science Ltd Printed in the USA. All rights reserved 0145-2134/96 $15 00 + .00 0145-2134(95)00112-3 THE LONG-TERM IMPACT OF THE PHYSICAL, EMOTIONAL, AND SEXUAL ABUSE OF CHILDREN: A COMMUNITY STUDY P. E. MULLEN, J. L. MARTIN, J. C. ANDERSON, S. E. ROMANS, AND G. P. HERBISON Departments of Psychological Medicine and Preventive and Social Medicine, University of Otago, New Zealand and Department of Psychological Medicine, Monash University, Australia Abs t rac t - -The associations between giving a history of physical, emotional, and sexual abuse in children and a range of mental health, interpersonal, and sexual problems in adult life were examined in a community sample of women. Abuse was defined to establish groups giving histories of unequivocal victimization. A history of any form of abuse was associated with increased rates of psychopathology, sexual difficulties, decreased self-esteem, and interpersonal problems. The similarities between the three forms of abuse in terms of their association with negative adult outcomes was more apparent than any differences, though there was a trend for sexual abuse to be particularly associated to sexual problems, emotional abuse to low self-esteem, and physical abuse to marital breakdown. Abuse of all types was more frequent in those from disturbed and disrupted family backgrounds. The background factors associated with reports of abuse were themselves often associated to the same range of negative adult outcomes as for abuse. Logistic regressions indicated that some, though not all, of the apparent associations between abuse and adult problems was accounted for by this matrix of childhood disadvantage from which abuse so often emerged. Key Words---Sexual abuse, Physical abuse, Emotional abuse. INTRODUCTION THE EXPERIENCE OF abuse during childhood not only generates distress and disturbance at the time but may produce long-term deleterious effects on the victims' health and functioning (Beitchman, Zucker, Hood, DaCosta, Akman, & Cassavia, 1992; Browne & Finkelhor, 1986; Kendell-Tackett, Williams, & Finkelhor, 1993; Reiss, Richters, & Radke-Yarrow, 1993). The role of child sexual abuse (CSA) in the genesis of adult disorders has been the focus of most research with the physical, and in particular the emotional, abuse of children attracting less attention (Claussen & Crittenden, 1991; Rosenberg, 1987). There are suggestions that the impact of different types of abuse can produce overlapping patterns of adverse sequelae in adult life (Briere & Runtz, 1988; Bryer, Nelson, Miller, & Kroll, 1987; Chu & Dill, 1990; Kiser, Heston, Millsap, & Pruitt, 1991; Moeller, Bachmann, & Moeller, 1993; Mullen, Romans- Clarkson, Walton, & Herbison, 1988; Muenzenmaier, Meyer, Strnening, & Ferber, 1993). Briere and Runtz (1990) produced data indicating abuse specific outcomes, with physical abuse linked to aggression towards others, emotional abuse to low self-esteem, and sexual abuse to maladaptive sexual behavior. Rieker and Carmen (1986) hypothesized that child sexual abuse was particularly damaging to the sense of self which heightened the risks of self-destructive Received for publication February 17, 1995; final revision received July 10, 1995; accepted July 1 I, 1995. Requests for reprints should be addressed to Professor P. E. Mullen, Forensic Psychiatry Service, P.O. Box 266, Rosanna, Victoria 3084, Australia. P. E. Muilen, J. L. Martin, J. C. Anderson, S. E. Romans, and G. P. Herbison and suicidal behaviors, with physical abuse in childhood producing similar though less severe damage to adult functioning. In a study focussing on bulimia nervosa a somewhat different hierarchy emerged with emotional abuse in childhood having the strongest association, to adult eating disorder, physical abuse a significant but weaker association and sexual abuse only contributing in combination with the other forms of abuse (Rorty, Yager, & Rossotto, 1994). Thus though there is general agreement that child abuse is associated to a range of problems in adult life, the extent to which sexual, physical, and emotional abuse each generate either common or specific patterns of damage remains an open question. The definition of what constitutes child abuse is fraught with difficulties. Children in our societies are all too often subjected to physical violence, emotional deprivation, and inappropri- ate sexual behaviors. In research into abuse the delineation of such abuse from the mundane, though undesirable, vicissitudes of childhood can present problems. Wide definitions of abuse, be it sexual, physical, or emotional, tend to capture large sections of the community. This may have educational or rhetorical power but in research it makes it less likely that clear correlations will emerge to long-term effects. More restrictive definitions can reasonably be criticized for by implication suggesting a range of unpleasant and even criminal acts are somehow not abusive. To further complicate the ascertainment of abuse subjectively identifying oneself as a victim of abuse produces a significantly lower rate than asking about specific experiences and categorizing on the basis of these responses (Carlin, Kemper, Ward, Sowell, Gustafson, & Stevens, 1994). In practice the nature of the research question and the data base is likely to profoundly influence both the operational definitions applied and the method of inquiry. As long as it is clear how abuse is being delineated and ascertained it is possible to generate an adequate evaluation and comparison of the results. Children who are victims of one form of abuse are more likely to also experience other forms of abuse. Focussing exclusively on one form of abuse is to risk giving a spurious prominence to the chosen form of victimization. Abuse is not randomly distributed through the community but is more likely to be found in disrupted and disturbed families and in those families subject to economic and social disadvantage (Bagley & Ramsey, 1985; Finkelhor & Baron, 1986). This creates difficulties for the attribution of an association between the abuse and any deleterious adult outcome for the risk factors for the abuse may also be risk factors for the negative outcomes. There are already suggestions in the literature with regard to sexual abuse that problems with adult psychological and sexual adjustment may be due as much to the associated factors in the family background as to the abuse (Fromuth, 1986; Kinzl, Traweger, Guenther, & Biebl, 1994; Mullen, Martin, Anderson, Romans, & Herbison, 1994; Pribor & Dinwiddie, 1992). The prominence accorded CSA as a cause of adult psychopathology is misplaced if CSA operates primarily as a marker for a range of coexisting deleterious influences rather than itself carrying the weight of causality. This paper attempts to examine both the specific associations between histories of sexual, physical, or emotional abuse to negative adult outcomes as well as the potential interaction between such experiences. The possibility that child abuse may occur more frequently in disturbed and disadvantaged home backgrounds necessitates examining family and social factors as potential confounders of any putative relationships between abuse and long-term outcomes. METHODS Subjects and Procedure The study, which was approved by the University and Health Boards' Ethics Committee, employed a two-stage design. The first involved sending a postal questionnaire to 2,250 women, Long-term impact of child abuse 9 randomly selected from the electoral rolls of Dunedin, a university city with a population of 130,000. The electoral rolls, used in national Parliamentary elections, list over 95% of the population 18 years and above; and adult literacy is over 97% in the New Zealand population. The sample derived from a random sampling programwas checked to ensure that it did not overlap with an earlier sample used to investigate sexual and physical abuse (Mullen, Romans- Clarkson, Walton, & Herbison, 1988). The postal questionnaire included questions on general health, mental health, demographics, social support, and experience of sexual and physical abuse both as a child and during adult life. Those who did not return the questionnaires were followed up by telephone and letter to clarify whether they wished to participate. The second stage was an interview phase. All 298 women under the age of 65 years who had indicated on their postal returns that they had experienced unwanted sexual advances prior to the age of 16 years were invited for interview. A group of the same size was randomly selected from the 716 subjects who did not report sexual abuse either as a child or as an adult and this group was also invited for interviews. The interviews were conducted over 14 months in 1989 and 1990, either in the women's home or at the research unit, depending on the women's preference. The interviews were largely con- ducted by J. M. and J. A. and took an average of 1 and 1/2 hours. Women were informed that there would be inquiries relating to abuse in the last section of the interview and consent was obtained once more prior to the onset of questioning. The women were assured that they could discontinue the interview at any stage and withdraw from the study. No financial or other induce- ment was offered to participants who gave generously of their time and energy. Counselling and appropriate referral was offered to those where it was acceptable. The interviewers were blind to the subjects' responses to the postal questionnaire when conducting the interviews. (For further details see Martin, Anderson, Romans, Mullen, & O'Shea, 1993.) Measures of Family and Social Background The interview included a detailed schedule which collected data on variables that have been associated with an increased risk of abuse during childhood (copies of the schedule may be obtained from the authors). Inquiries were made into family factors such as family structure, family and parental relationships, discipline and family violence. At the time of the interview subjects also completed the Parental Bonding Instrument (PBI), a questionnaire elucidating memories of their relationship to their parents (Parker, 1983; Parker, Tupling, & Brown, 1979). This instrument enabled the generation of a measure of emotional deprivation during childhood. Measures of Abuse 1. Childhood sexual abuse (CSA) was explored using a series of direct questions about un- wanted sexual advances prior to the age of 16 years which covered noncontact abuse, including being exposed to indecent suggestions and exhibitionism; contact abuse, including any sexual touching, genital contact, involving either the perpetrators or victims genitals; and abuse involving actual penetration. CSA of some form was reported by 32% of our sample with 25% describing abuse involving physical contact, 20% contact with the victims or perpetrators genitals of which 6% involved actual penetration (for further details see Anderson, Martin, Mullen, Romans, & Herbison, 1993). It must be emphasized however that for the purposes of this analysis we used a subgroup who reported that prior to the age of 16 years they had been subjected to abuse involving actual penetration and or genital contact on 10 or more occasions. 2. Physical abuse was defined as reporting physical punishment or hitting by a parent of a severity to produce at least lasting soft tissue injury on a regular basis (i.e., more than two 10 . P. E. Mullen, J. L. Martin, J. C. Anderson, S. E. Romans, and G. P. Herbison times a year) including those incidents of abuse of sufficient severity to require hospitaliza- tion. These included injuries caused by hitting with open hands or fists or with a weapon, such as a belt or stick. Emotional abuse or deprivation was defined using the responses to the Parental Bonding Instrument (PBI) (Parker, 1983; Parker, Tupling, & Brown, 1979) which generates two dimensions of the subjects memory of their parenting, the first being care and affection and the second the degree of controlling behavior. The mean scores on the PBI for our sample were virtually identical to those reported by Parker and colleagues (1979) and Parker (1983). The emotionally abused subgroup was composed of subjects whose ratings of either parent fell at least one standard deviation from our sample mean on both the "ca re" and overprotection subscales. The Control Group The control group employed in this analysis were the 390 interviewees who did not meet the criteria used in this particular analysis for any of the three forms of abuse. Measures of Adult Outcomes Mental health. The short form of the Present State Examination (PSE) (Wing, Cooper, & Sartorius, 1974; Wing, Nixon, Mann, & Left, 1977), which is a structured interview exploring a range of psychopathology, was administered by the interviewers who were trained in its use. Additional questions were also asked to generate longitudinal data on psychiatric disorder (McGuffin, Katz, & Aldrich, 1986). The PSE generates a total score indicative of the overall level of psychiatric symptomatology in the previous month as well as being able to generate diagnostic categories. The CATEGO program~was employed to identify from the PSE responses a group whose symptoms suggested they would have been identified as "a case" and received a clinical diagnosis and treatment had they been seen at a mental health clinic or by a mental health professional. Additional questions were also included on eating disorders, drug dependence, and self damaging and suicidal behaviors. The WHO alcohol questionnaire was employed to explore hazardous and abusive patterns of alcohol consumption (Saunders, Aasland, & Grant, 1987). A variable was derived from this data to identify those women who drank at a hazardous level defined as more than 14 standard drinks (140 gms alcohol) per week. Self-esteem. The self-esteem questionnaire of Robson (1989) was employed as a measure of this dimension as it covers a wider range of components of self-esteem than other instruments such as that of Rosenberg (1965). This instrument attempts to quantify the sense of contentment and self-acceptance that result from a persons appraisal of their own worth, significance, attractiveness, competence, and ability to satisfy their aspirations. Low self-esteem was defined as one standard deviation below the mean score for the entire sample. Social class. Subjects' socioeconomic class (SES) was estimated from the current occupation and, where relevant, that of the partner using the standard New Zealand tables (Elley & Irving, 1985). Educational status was assessed as a separate variable according to the highest formal qualification gained. A variable employed in this analysis was a decline in SES from that of the woman's family of origin to that of her current family. Sexual history. The history was obtained of the age at first voluntary intercourse, and any premarital pregnancies and their outcomes. Previous marriages and cohabitations and their Long-term impact of child abuse 11 outcome were also recorded. A measure of current sexual functioning was developed specifi- cally for this study and included inquiries about the frequency of sexual activity, levels of satisfaction with their current sex lives, and present and past sexual difficulties together with their perceived origins. A mixture of direct questions, open-ended questions, and Likert scales were employed. The variables employed in this analysis were the current marital status and reported sexual problems which were attributed to their own attitude to sexuality rather than to medical problems, partner shortcomings, or other practical difficulties. The quality of the women's currentrelationships was also explored using the Intimate Bond Measure (Wilhelm & Parker, 1988). Measures of Family and Social Background The interview included a detailed schedule exploring the structure and relationships within the family of origin, the mental and physical health of the parents during the subjects childhood and the quality of the relationships during childhood with other family members and peers. The factors derived from these problems were employed to construct potential risk factors which have been claimed to be associated with increased risks of abuse (Finkelhor & Baron, 1986). Statistical Analysis Data was entered on the University's mainframe computer and analyzed using the SPSSX statistical program. Control subjects, as defined for the purposes of this analysis, were compared for all analysis with the three forms of abuse separately and those three in combination. Continuous variables were transformed, where appropriate, for non-normality and analyzed using student's t-test, while categorical variables were tested for significance by X 2 and odds ratio tests. Logistic regression carried out on the SPSSX was used to test the importance of sexual, emotional, and physical abuse in combination with a range of other variables relating to the subjects childhoods which have been suggested as increasing the risk of being abused. The result of these models are expressed in terms of adjusted odds ratios. The percentage of the deviance accounted for in the models by the relevant factors are quoted as a measure of the variance in the outcome explained. In analyzing an extensive database problems may arise from multiple analysis (Jones & Rushton, 1982). In reporting odds ratios, the usual convention of providing 95% confidence limits has been adhered to, but when the lower limit approaches unity the results should be treated with reservation. This caveat does not apply to the results of logistic regression where multiple analysis is not an issue. RESULTS Sample The number of women who returned usable questionnaires was 1,376, which was 61% of the whole sample and 71% of the women under 65 with whom we were concerned in this study. Information available from the electoral role suggested that our nonresponders did not differ from respondents in socioeconomic status. The sample employed in this study was drawn from the 610 women invited for subsequent interview, 497 (80%) of whom completed the interview (for further details see Anderson, Martin, Mullen, Romans, & Herbison, 1993; and Martin, Anderson, Romans, Mullen, & O'Shea, 1993). There were 107 (21.5%) women in the interview sample who reported experiencing some 12 P. E. Mullen, J. L. Martin, J. C. Anderson, S. E. Romans, and G. P. Herbison form of abuse during childhood which fell within our criteria for this analysis. The sexually abused group consisted of 53 (10.7%) women, 32 of whom had during childhood been subjected to penetration and 21 to repeated genital contact. The physically abused group contained 39 women (7.8%) and the emotionally abused 57 (11.5%). (The pattern of overlap between the three types of abuse is presented in Table I.) The physical and emotional abuse was all at the hands of parents or primary caregivers but fathers or stepfathers accounted for 23.1% of sexual abuse, and in 45.8% of such abuse the perpetrator was outside of the immediate family. Those reporting physical abuse identified the mother as sole abuser in 15 cases, the father in 15 cases, and 9 subjects were physically abused by both. Emotional abuse was associated primarily with the father in 23 cases, the mother in 24 cases, and both father and mother in 10 instances. Long-term Impact of Abuse Sexual abuse in childhood. Those reporting child sexual abuse had significantly poorer mental health. They were more likely to be identified as cases on the PSE (35.8% vs. 9.5% X 2 = 30.32 pfrom 12.1% for any single form of abuse to 40.0% for all three types of abuse (X 2 38.202 df 3 pto below together with the drop in deviance, which is a measure of the percentage of possible variance explained by each model. A decline in SES was no longer significantly related to abuse after logistic regression. Separation and divorce remained signifi- cantly associated to physical abuse [O.R. 1.87 (95% C.L 1.2--2.89)] as well as having experi- enced frequent shifts in residence during childhood [O.R. 1.53 (95% C.I. 1.06--2.23)] but not to sexual or emotional abuse (drop in deviance 1.8%). Sexual problems remained associated with both sexual [O.R. 1.52 (C.I. 1.09--2.07)] and emotional abuse [O.R. 1.46 (95% C.I. 1.02--2.05)], as well as with violence in the parental relationship [O.R. = 1.42 (95% C.I. 1.05--1.92)] and frequent shifts in childhood [O.R. = 1.31 (95% C.I. 1.02--1.68)]. These four factors together accounted for 2.2% of the drop in deviance. Low self-esteem continued to be significantly related to sexual abuse [O.R. 1.53 (95% C.L 1.08--2.14)] and emotional abuse [O.R. 1.56 (95% C.I. 1.11 - -2 .22)] together with having described oneself as a shy child [O.R. 1.38 (95% C.L 1.07--1.77)] and to having had an invalid or chronically physically ill mother (O.R. = 1.46 95% C.L 1.09--1.95). These four factors accounted for 3.4% of the drop in deviance. Being identified as a psychiatric case on the PSE continued to be significantly associated with sexual abuse [O.R. = 2.02 (95% 1.40--2.94)] and emotional abuse [O.R.= 1.75 (95% C.L 1.19--2.56)] as well as with having grown up in a family where the parents frequently moved (O.R. = 2.09 95% C.I. 1.46--2.72) and having had a physically sick mother 16 P. E. Mullen, J L. Martin, J. C. Anderson, S. E. Romans, and G. P. Herbison Table 4. The Odds Ratios Following Logistic Regression for Negative Adult Outcomes in Those with Histories of Abuse Emotional Negative Outcome Sexual Abuse Physical Abuse Abuse Decline m SES NS NS NS Separation or Divorce NS 1.87 ( 1.20-2.89) NS Pregnant Prior to 19 years 1.64 (I.43-2.36) NS 1.83 (1.30-2.61) Sexual Problems 1.52 (1.09-2.07) NS 1.46 (1.02-2.05) Low Esteem 1.53 (1.08-2.14) NS 1.56 (1.1 ! -2.22) PSE Case 2.02 (1.40-2.94) NS 1.75 (1.19-2.59) Eating Disorder 1.75 (1.15-2.59) NS 1.76 (I.17-2.61) Ever Suicide Attempt 3.57 (2.11-5.99) NS 2.42 (1.34-4.39) Heavy Drinker 1.77 (1 22-2.56) NS NS Ever Depressed 1.80 (1.29-2.51) NS 1.45 (1.03-2.03) Ever Inpatient 2.46 ( 1.41-4.10) NS 1.77 (1.05-2.97) (O.R. = 1.56 95% C.L 1.09--2.20). The drop in deviance associated with these in combination was 7.7%. Ever having had a depressive disorder continued to be significantly associated with sexual abuse [O.R. = 1.80 (95% C.I. 1.29--2.51)] and emotional abuse [O.R. = 1.45 (95% C.L 1 . 0 3 - - 2 . 0 3 ) ] as well as with a history of domestic violence in the parental relationship (O.R. = 1.45 95% C.L 1.11--1.90), though these three variables together only accounted for 1.2% of the deviance. Eating disorders were associated with sexual [O.R. = 1.75 (95% C.L 1.15--2.59)] and emotional abuse [O.R.= 1.76 (95% C.I. 1.17--2.61)] as well as one 's parents separating during one's childhood (O.R. = 1.86 95% C.L 1.17--2.94) and these three factors produced a 2.5% drop in the deviance. Heavy drinking was associated only with sexual abuse (drop in deviance 3.3%). Suicidal behavior continued to be associated with sexual [O.R.= 3.57 (95% C.I. 2.11--5.99)] and emotional abuse [O.R. = 2.42 (95% C.I. 1 .34- - 4.39)] together with frequent parental rows (O.R. = 2.03 95% C.I. 1.19--3.49). These three variables accounted for 9.3% of the deviance. Having been admitted to a psychiatric hospital was associated with both sexual [O.R. = 2.46 (95% C.L 1.41--4.10)] and emotional abuse [O.R. = 1.77 (95% C.I. 1.05--2.97)] as well as to having had a physically ill mother (O.R. = 2.01 95% C.L 1.20--3.39), and these three factors accounted for 4.8% of the deviance. The interactions between the three forms of abuse and other negative developmental factors was further examined using logistic regression. This added little to the original logistic regres- sion for sexual and emotional abuse but demonstrated that physical abuse became a significant association to a number of negative adult outcomes when combined with other developmental variables. Physical abuse in combination with parental separation was significantly associated with low self-esteem (O.R. 1.54 95% C.L 1.11--2.14), ,in combination with frequent shifts in childhood to sexual problems (O.R. = 1.37 95% C.I. 1/08--1.75), and to being a case on the PSE in combination with parental rows (O./~. = 1.55 95% C.I. 1.08--2.23). Sexual abuse became a significant contributor to current separation and divorce when combined with frequent parental rows (O.R. 1.51 95% C.I. = 1.08--2.10). Relationship to Gender of Abuser Those subjects who report being abused physically or emotionally by both maternal and paternal caregivers tended to have higher rates/of all negative outcomes measured than those who nominated only one of the primary ~aCegivers as abusive. Emotional abuse by male caregivers, as against that by females, was l~articularly associated with difficulties with regard to sexuality in adult life (X 2 33.4 df3 pminimized in a community sample such as this in which no obvious advantage accrued to those reporting abuse and in which only four subjects had discovered or "clarified" their memories of abuse as a result of participating in therapy or abuse groups. It should be noted that the sexually abused women made relatively few attributions of long-term problems to these childhood experiences and that our. analysis suggests a greater degree of association between the reported abuse and current difficulties than their subjective assessments of its impact (see Mullen et al., 1994). This is against an exaggeration by our subjects of the role, at least of sexual abuse, in any current difficulties. In the final analysis, however, this remains a study of histories of abuse provided by adults. Alternative research strategies which are based on following up children known to have been victimized provide important insights into abuse, but they also have to contend with the ascertainment problem as less than 10% of even the most serious forms of abuse come to official or medical notice at the time (Anderson et al., 1993; Green, 1988). Methodologically ideal studies in this area may always escape us, but good enough studies will, when taken together, illuminate the central theoretical and clinical issues. Those women who reported having been exposed to some form of abuse in childhood, be it sexual, physical, or emotional were in adult life more likely to have mental health, interper- sonal, and sexual difficulties. The current psychopathology revealed by the PSE in those reporting abuse fell almost exclusively in the area of depressive and anxiety symptoms. It was similarly histories of previous depressive disorders which were so prominent among the abused. 18 P. E. Mullen, J. L. Martin, J C. Anderson, S. E. Romans, and G. P. Herbison Eating disorders were increased in all three forms of reported abuse, though most obviously in those with histories of emotional and sexual abuse rather physical abuse as suggested by Rorty, Yager, and Rossotto (1994). The elevated rates of prior psychiatric treatment in those reporting abuse of any type is consistent with studies on those with mental illness which have demonstrated very high rates of reporting childhood abuse and neglect (Meuenzenmaier, Meyer, Struening, & Ferber, 1993). There was some support for previously reported differences between the long-term impact of differing categories of abuse. Those with histories of sexual abuse, as opposed to the physically abused, were more likely to report sexual problems, but the emotionally abused group had very similar levels of sexual difficulties. The emotionally abused group had somewhat worse self-esteem than the physically or sexually abused but the differences were modest. Though these subtle differences are discernible in the pattern of adult problems associated with particular forms of abuse, the most striking finding is the similarity in the pattern of difficulties and disabilities. The postulated association between physical abuse and emotional neglect and later aggressive and criminal behavior (Widom, 1989) unfortunately could not be investigated, as no adequate measure of violent or irresponsible conduct had been included in the study design. The failure to find a clear separation between the associations to different types of abuse could simply reflect an inadequate sample size or a failure to address the specific correlates of such abuse. Conversely, it could be that any serious form of abuse in childhood produces a generalized, and largely nonspecific, disruption in the child's development. Even if subtle distinctions do exist, the hypothe- sis that there are highly specific long-term post abuse syndromes is difficult to accommodate to both such a diverse range of potential long-term sequelae and to such an extensive overlap between the types of abuse. As might be expected those reporting more than one form of abuse in childhood fared even worse than those reporting single types of abuse. Given the frequency of multiple abuse, this is of potential clinical importance. The impact of emotional abuse varied with the gender of the abuser. Those emotionally abused by their female caregivers were more prone to psychiatric difficulties in adult life, whereas such abuse from the male parent figure was particularly deleterious to adult sexuality, thus suggesting that in this culture there are still some gender specific roles with regard to the developing child. The predicted greater impact of a physically abusive male caregiver was not found. Sexual abuse is almost exclusively perpetrated by males, and with only two female sexual abusers in this series no analysis by gender was possible. Abuse was not randomly distributed through our population but occurred more often in those from disturbed and disrupted home backgrounds. Those who had grown up in intact nuclear families were significantly less likely to report any form of abuse. Similarly, having a close and confiding relationship with the mother seemed to confer a degree of protection. Those from socially and economically privileged backgrounds were also less likely to report any form of abuse. Intact families do not guarantee stability and those children exposed to an environment of severe parental discord, with or without overt violence, were also at high risk. In physical and emotional abuse, the perpetrators of the abuse are usually also responsible for the disrupted and violent home backgrounds, whereas in the sexually abused the abuse is often at the hands of men outside of the immediate family. A lack of effective protection and a child drawn to seek affection and caring outside of the family probably mediates the connection between sexual abuse and a disrupted home life. Those elements in the child's background that could reasonably be assumed to have been operative prior to the abuse constitute potential risk factors. Obtaining reliable chronologies of childhood experiences is plagued with difficulties. We counted as risk factors those elements which certainly predated abuse, such as parental social class, but also included factors where the history gave rise to a reasonable supposition of precedence. In factors such as having a confidante as a child or confiding in the mother, a degree of skepticism is appropriate for these Long-term impact of chdd abuse 19 could have followed the abuse or the abuse could have led to an understandable reappraisal and adjustment of memories. The risk factors for abuse were broadly similar for all three forms of abuse though physical abuse was particularly strongly related to violence in the parental relationship, and to alcoholism or other mental disorder in one of the primary caregivers. Emotional abuse tended to go together with a lack of a confident among one's childhood peers and, not surprisingly, with a lack of a confiding relationship with the mother. The risk factors which were associated with higher levels of abuse were themselves associ- ated in many instances with the same negative outcomes in adult life as were the different types of abuse. This raised the possibility that the associations between abuse and adult difficulties could have been accounted for not by the abuse itself, but by those factors connected to the increased risks of abuse. Logistic regression was used in an attempt to disentangle the impact of the three forms of abuse both from each other and from the potentially confounding influences of problematic family and social backgrounds. The logistic regression suggests that although part of the apparent relationship between abuse and problems in adult life is accounted for by such risk factors, there remains in most instances an independent association between sexual and emotional abuse and problems in the areas of mental health, sexuality, and social functioning. In those reporting sexual abuse the risksof suicidal behavior remains three times higher, and the chances of being a case on the PSE was still over twice that of nonabused subjects. Similarly, in emotional abuse the risks of suicidal behavior and psychopathology are in the region of twice as high. Physical abuse appeared, however, to disappear following logistic regression as a contributor to all outcomes except separation and divorce. This could be the result of a lack of malignancy in the long-term effects of physical abuse or could reflect variance stealing in which the effects of physical abuse are obscured by its close relationship to alternative pathogens in the developmental history. That this is indeed the case is supported by the logistic regression which looked at the effects of the three types of abuse in combination with a range of other childhood risk factors. Physical abuse was once more a significant contributor to the associations with lowered self-esteem, sexual problems, and being a case on the PSE when studied in combination with such factors as parental rows and those frequent shift of residency in childhood suggestive of family and social instability. The overall contribution of abuse to the variables studied in this population were, however, modest; for even if we assume such associations represent causal connections, the drops in deviance in the various models indicate the contributions of abuse to lie in the range of 1% to 5%. This should caution against overestimating the contribution of childhood abuse in general, and sexual abuse in particular, to explaining mental health and interpersonal difficulties in our community. The contribution to the overall level of distress and disorder in the commu- nity should not, however, be confused with the possible contribution of abuse to difficulties and disabilities in an individual victim. Child abuse, and sexual abuse in particular, has come to be regarded by many clinicians as making a powerful contribution to adult psychopathology. A study such as this on a random community sample inevitably examines a wider spectrum of abuse than will be encountered in clinics, and is unlikely to include some of the severely abused and distressed victims driven by their problems to seek professional assistance. Extrapolation from epidemiological studies to the consulting room require caution, but it does potentially provide a useful additional dimension, if not necessarily a corrective one, to clinical impression and clinically based research. This study once more highlights the close relationships between sexual, emotional, and physical abuse reinforcing the findings of previous workers (Briere & Runtz, 1988; Bryer, Nelson, Miller, & Kroll, 1987; Moeller et al., 1993; Rorty, Yager, & Rossotto, 1994). A focus on sexual abuse should not be allowed to obscure the possible relevance of emotional and physical abuse to adult problems. Further, this study places the adverse outcomes associated 20 P. E. Mullen, J. L. Martin, J. C. Anderson, S. E. Romans, and G. P. Herbison wi th his tor ies o f any f o r m o f abuse f i rmly in the con tex t o f a mat r ix o f ch i ldhood d isadvantage , f r o m wh ich such abuse so of ten arises and f r o m which some, at least, o f the apparent associa- t ions to abuse are l ike ly to be at tr ibutable. Ch i ld abuse is an ev i l to wh ich no ch i ld should be exposed . Th is s tudy supports an assoc ia t ion b e t w e e n g iv ing a his tory o f such abuse and an increased vu lnerab i l i ty to a range o f menta l heal th, persona l and socia l p r o b l e m s in adult l ife, but p laces abuse a longs ide a range o f o ther d e v e l o p m e n t a l t raumas and d i sadvan tages as potent ia l causal agents. T h e m e s s a g e for therapists is that w h e n eva lua t ing the r e l evance o f ch i ldhood abuse to b e w a r e an exc lus ive , and po ten t ia l ly exagge ra t ed focus on the t raumas o f sexual abuse wh ich m a y obscure both the r e l e v a n c e o f o ther fo rms o f abuse and the un fo ld ing o f o ther d a m a g i n g d e v e l o p m e n t a l inf luences . R E F E R E N C E S Anderson, J. C., Martin, J. L., Mullen, P. E., Romans, S., & Herbison, P. (1993). The prevalence of childhood sexual abuse experiences in a community sample of women. Journal of the American Academy of Child and Adolescent Psychiatry, 32, 911-919. Bagley, C., & Ramsey, R. (1985). Sexual abuse in childhood: Psychosocial outcomes and implications for social work practice. Social Work and Human Sexuality, 4, 33-47. Beitchman, J. H., Zucker, K. J., Hood, J. E., DaCosta, G. A., Akman, D., & Cassavia, E. (1992). A review of the long term effects of child sexual abuse. Child Abuse & Neglect, 16, 101-118. Briere, J., & Runtz, M. (1988). Multivariate correlates of childhood psychological and physical maltreatment among university women. Child Abuse & Neglect, 12, 331-341. Briere, J., & Runtz, M. (1990). Differential adult symptomatologies associated with three types of child abuse histories. Child Abuse & Neglect, 14, 357-364. Browne, A., & Finkelhor, D. (1986). Impact of child sexual abuse: A review of the research. Psychological Bulletin, 99, 66-77. Bryer, J. B., Nelson, B. A., Miller, J. B., & Kroll, P. A. (1987). Childhood sexual and physical abuse as factors in adult psychiatric illness. American Journal of Psychiatry, 144, 1426-1430. Carlin, A. S., Kemper, K., Ward, N. G., Sowell, H., Gustafson, B., & Stevens, N. (1994). The effect of differences in objective and subjective definitions of childhood physical abuse on estimates of its incidence and relationship to psychopathology. Child Abuse & Neglect, 18, 393-399. Chu, J. A., & Dill, D. L. (1990). Dissociative symptoms in relation to childhood physical and sexual abuse. American Journal of Psychiatry, 147, 887-892. Claussen, A. H., & Crittenden, P. M. (1991). Physical and psychological maltreatment: Relations among types of maltreatment. Child Abuse & Neglect, 15, 5-18. Elley, W. B., & Irving, J. C. (1985). The Elley-Irving socioeconomic index. New Zealand Journal of Education Studies, 20, 115-228. Femina, D. D., Yaeger, C. A., & Lewis, D. O. (1990). Child abuse: Adolescent records versus recall. Child Abuse & Neglect, 14, 227-231 Finkelhor, D, & Baron, L. (1986). High risk children. In D. Finkelhor, S. Arajc, A. Browne, S. Peters, & G. Wyatt (Eds.), A sourcebook on child sexual abuse (pp. 60-88). Beverly Hills, CA: Sage. Fromuth, M. E. (1986). The relationship of childhood sexual abuse with later psychological and sexual adjustment in a sample of college women. Child Abuse & Neglect, 10, 5-15. Green, A. H. (1988). Child maltreatment and its victims. Pediatric Clinics of North America, 11, 157-165. Jones, D. R., & Rushton, L. (1982). Simultaneous inference in epidemiological studies. International Journal of Epidemiology, 11, 276-282. Kendell-Tackett, K., Williams, L. M., & Finkelhor, D. (1993). Impact of sexual abuse in children: A review and synthesis of recent empirical studies. Psychological Bulletin, 113, 164-180. Kinzl, J. F., Traweger, C., Guenther, V., & Biebl, W. (1994). Family background and sexual abuse associated with eating disorders. American Journal of Psychiatry, 151, 1127-1131. Kiser, L. J., Heston, J., Millsap, P., & Pruitt, D. B. (1991). Physical and sexual abuse in childhood: Relationship with post traumatic stress disorder Journal of American Academy of Child and Adolescent Psychiatry, 30, 776-783. Martin, J. L., Anderson, J. C., Romans, S. E., Mullen, P. E., & O'Shea, M. (1993). Asking about child sexual abuse: Methodological implication of a two stage survey. Child Abuse & Neglect, 17, 383-392. McGuffin, P., Katz, R., & Aldrich, J. (1986). Past and present state examination: The assessment of "lifetime ever" psychopathology. Psychological Medicine, 16, 461-465. Meuenzenmaier,K., Meyer, I., Struening, E., & Ferber, J. (1993). Childhood abuse and neglect among women outpatients with chronic mental illness. Hospital and Community Psychiatry, 44, 666-670. Moeller, T. P., Bachmann, G. A., & Moeller, J. R. (1993). The combined effects of physical, sexual and emotional abuse during childhood, long term health consequences for women. Child Abuse & Neglect, 17, 623-640. Long-term impact of child abuse 21 Mullen, P. E., Romans-Clarkson, S. E., Walton, V. A., & Herbison, G. P. (1988). Impact of sexual and physical abuse on women's mental health. The Lancet, 841-845. Mullen, P. E., Martin, J. L., Anderson, J. C., Romans, S. E., & Herbison, G. P. (1993). Childhood sexual abuse and mental health in adult life. British Journal of Psychiatry, 163, 721-732. Mullen, P. E., Martin, J. L., Anderson, J. C., Romans, S. E., & Herbison, G. P. (1994). The effect of child sexual abuse on social, interpersonal and sexual function in adult life. British Journal of Psychiatry, 165, 35-47. Parker, G., Tupling, H., & Brown, L. B. (1979). A parental bonding instrument. British Journal of Medical Psychology, 52, 1-10. Parker, G. (1983). Parental overprotection: A risk factor in psychosocial development. New York: Grune and Stratton. Pribor, E. F., & Dinwiddie, S. H. (1992). Psychiatric correlates of incest in childhood. American Journal of Psychiatry, 149, 52-56. Reiker, P. P., & Carmen, E. (1986). The victim to patient process: The disconfirmation and transformation of abuse. American Journal of Orthopsychiatry, 56, 360-370. Reiss, D., Richters, J. E., & Yarrow, R. (1993). Children and violence. New York: Guilford Press. Robson, P. (1989). Development of a new self report questionnaire to measure self-esteem. Psychological Medicine, 19, 513-518. Rorty, M., Yager, J., & Rossotto, M. A. (1994). Childhood sexual, physical and psychological abuse in bulimm nervosa. American Journal of Psychiatry, 151, 1122-1126. Rosenberg, M. (1965). Society and adolescent self image. Princeton, N J: Princeton University Press. Rosenberg, M. S. (1987). New directions for research on the psychological maltreatment of children. American Psychologist, 49, 166-17 I. Saunders, J. B., Aasland, O. G., & Grant, M. (1987). The world health organization screening instrument for harmful and hazardous alcohol consumption. Geneva, Switzerland: World Health Organization. Widom, C. S. (1989). Child abuse, neglect and adult behavior: Research design and findings on criminality violence and child abuse. American Journal of Orthopsychiatry, 59, 355-367. Wilhelm, K., & Parker, G. (1988). The development of a measure of intimate bonds. Psychological Medicine, 18, 225-234. Williams, U M. (in press). Recall of childhood trauma: A prospective study of women's memories of child sexual abuse. Journal of Consulting and Clinical Psychology. Wing, J. K., Cooper, J. E., & Sartorius, N. (1974). The measurement and classification of psychiatric symptoms. London: Cambridge University Press. Wing, J. K., Nixon, J. M., Mann, S. A., & Left, J. P. (1977). Reliability of the PSE used in a population study. Psychologwal Medicine, 7, 505- 516. R~sum6----French abstract not available at time of publication. Resumen---En esta investigaci6n se examinaron, en una muestra comunitaria de mujeres, las asociaciones entre la existencia de una historia de maltrato ffsico, maltrato emocional o abuso sexual en la infancia y una serie de problemas sexuales, interpersonales y de salud mental en la vida adulta. El maltrato fue definido de forma que se establecieran grupos con historias de victimizaci6n inequfvocas. Una historia de cualquier forma de maltrato estaba asociada con aumento en las tasas de psicopatologfa, dificultades sexuales y problemas interpersonales y un descenso en la autoes- tima. Las semejanzas entre las tres formas de maitrato en cuanto a su asociaci6n con consecuencias negativas en ia 6poca adulta eran m~is aparentes que cualquier diferencia. Sin embargo, se observaba una tendencia a que el abuso sexual estuviera particularmente asociado con problemas sexuales, el maltrato emocional con una baja autoestima y el maltrato ffsico con fracaso matrimonial. El maltrato de cualquier tipo se daba de manera mils frecuente en aqueilos sujetos con historias familiares alteradas y desorganizadas. Los factores de la historia familiar asociados con los informes de maltrato estaban, elms mismos, a menudo asociados con el mismo tango de consecuencias negativas adultas que se habfan asociado con el maltrato. Los anfilisis de regresi6n logfstica indicaron que algunas, aunque no todas, de las aparentes asociaciones entre maltrato y problemas adultos eran explicadas por dicha matriz de desventaja infantil, a partir de la cual emerge el maltrato tan frecuentemente.