Statistical methods

Data analysis was performed using SAS statistical software, version 6.12 (SAS, Institute, 1997). Logistic regression modelling was used to examine the associations between demographic variables and mental health measures. A stepwise procedure with backward elimination was used to select those variables that remained independently associated with the psychiatric outcome of interest. The variables selected by the stepwise procedure were used in subsequent analyses to adjust the associations of abusive experience variables with psychiatric outcome for any significant demographic characteristics.

Odds ratios relative to a baseline category were estimated for the different levels of categorical variables. The values for odds ratios are such that values greater than 1 indicate an increased risk of a psychiatric outcome and values less than 1 a reduced risk.

Demographic characteristics

The ages of the 1207 women respondents ranged from 16 years to 85 years, with a mean age of 37.2 years (s.d.=15.9). A total of 569 (49%) reported that they or their family owned the property in which they lived, and 597 (50%) owned or lived with someone who owned a car. Four hundred and seventeen (35%) were born outside the UK, 634 (54%) were White; 443 (38%) were single, 413 (35%) married, 151 (13%) divorced or separated, 35 (3%) widowed, and 123 (11%) cohabiting; 593 (51%) were currently living with a partner or spouse, 321 (28%) lived alone, and 250 (21%) were living in other circumstances; 730 (61%) had children.

Prevalence of abusive experiences

Experiences of childhood abuse (at age less than 16 years) included 88 (9%) women who reported unwanted sexual intercourse; 116 (11%) unwanted sexual activities, but not intercourse; 48 (5%) severely beaten by a parent or carer on one occasion and 112 (12%) severely beaten more than once; 53 (5%) experiencing both unwanted sexual intercourse and other unwanted sexual activities. Among the women reporting that they had been severely beaten on more than one occasion, 23 (21%) reported experiencing unwanted sexual intercourse and 29 (27%) unwanted sexual activities. A total of 17 (2%) reported all three forms of abusive experience before age 16 years.

Reported adult (16 years and above) abusive and traumatic experiences included 425 (41%) women who fulfilled the study definition for domestic violence, 61 (17%) with more than one partner; 79 (8%) reported that they had been raped; 87 (9%) reported that they had been sexually assaulted (but not raped). Excluding women who reported that they had at some time been forced to have sex by a partner, 33 (4%) reported that they had been raped. The questionnaire did not identify the perpetrator of the rape.

Within the sample surveyed, all childhood abusive experiences were closely interrelated, tending to co-occur in the lives of individual participants: 53 (67%) of the women who had experienced unwanted sexual intercourse prior to age 16 years had also experienced other unwanted sexual activities (OR=30.3; 95% CI 17.7–52.0; P<0.001); 28 (39%) of the same subgroup had been beaten on one or more occasions (OR=3.5; 95% CI 2.1–5.8; P<0.001) and 35 (35%) of women who had experienced unwanted sexual activities (not intercourse) before age 16 years had been severely beaten on one or more occasions (OR=3.0; 95% CI 1.9–4.7; P<0.001). In addition, adult experiences of abuse and trauma co-occurred: 55 (14%) of the women who experienced domestic violence also reported rape (OR=4.7; 95% CI 2.76–8.11; P<0.001); 54 (14%) sexual assault other than rape (OR=3.34; 95% CI 2.06–5.41; P<0.001). Forty-one (59%) women who reported being raped reported other forms of sexual assault (OR=30.03; 95% CI 16.99–53.09; P<0.001) (Reference Coid, Petruckevitch and FederCoid et al, 2001).

Prevalence of psychiatric morbidity

A total of 301 (27%) women scored 11 or more on the anxiety scale, and 105 (9%) scored 11 or more on the depression scale of the HADS; 144 (13%) had made a suicide attempt, and 104 (9%) had deliberately injured themselves, but without suicidal intention; 708 (62%) reported that they drank alcohol. Of this latter subgroup, 209 (30%) had at some time felt that they should cut down on their drinking, 82 (12%) reporting that people had annoyed them by criticising their drinking; 112 (16%) had at some time felt bad or guilty about their drinking, and 46 (6%) had drunk alcohol first thing in the morning to steady their nerves or get rid of a hangover. Within the subgroup of drinkers, 27 (4%) also reported that they had had a problem with their drinking in the past 12 months, 35 (5%) before the past 12 months. Of the total sample of women, 132 (12%) screened positive for alcohol misuse (affirmative response to two or more of the CAGE questions); 222 (20%) women reported that they had at some time used illicit (street) drugs. Of these, 31 (14%, which is 3% of the total sample) had at some time been unable to reduce their use of these drugs. A total of 95 women (8%) were observed to be suffering from PTSD based on the criteria of Breslau et al (Reference Breslau, Peterson and Kessler1999).

Within this population all the mental health measures were closely associated, confirming a high level of comorbidity between anxiety, depression, illicit drug use, alcohol problems, parasuicide, selfharm and PTSD. All associations demonstrated statistical significance at the P<0.001 level, except for depression and illicit use of drugs, for which the association was not significant.

Demographic associations with mental health measures

Table 1 shows the independent associations between the demographic characteristics of the women in the survey and mental health measures. All models were adjusted for age, which was significantly associated only with reported use of illicit drugs (more prevalent in the 25–34 year age group), and alcohol misuse (more prevalent between the ages of 25 years and 44 years). Marital status had several important independent associations with mental outcomes: being divorced or separated was independently associated with anxiety, depression, PTSD and illicit drug use. Illicit drug use was also associated with single marital status and cohabitation. Being widowed was associated with a lower prevalence of anxiety.

Table 1

Demographic variables independently associated with mental health outcomes for women primary care attenders

Variable	Anxiety (n=978)		Depression (n=901)		PTSD (n=1067)		Suicide attempt (n=1036)		Deliberate injury (n=925)		Illicit drugs (n=907)		CAGE ≥ 2 (n=1050)
	OR	95% CI	OR	95% CI	OR	95% CI	OR	95% CI	OR	95% CI	OR	95% CI	OR	95% CI
Age (years)
16-24	1.00		1.00		1.00		1.00		1.00		1.00		1.00
25-34									0.44	0.23-0.83	1.86	1.12-3.11	2.02	1.06-3.84
35-44													2.05	1.01-4.16
45+									0.35	0.15-0.80
Marital status1
Married	1.00		1.00		1.00						1.00
Divorced/separated	1.73	1.11-2.71	3.02	1.63-5.60	2.69	1.28-5.65					2.19	1.14-4.21
Widowed	0.24	0.07-0.82	1.00		3.46	1.03-11.6					1.00
Single	1.00		1.00		2.05	1.10-3.83					2.62	1.65-4.15
Cohabiting	1.00		1.00		2.02	0.88-4.62					3.09	1.70-5.61
Education1
≥ 13 years			0.43	0.23-0.80					0.41	0.24-0.70	3.44	2.32-5.08
Home ownership1
Council/temporary/HA	2.02	1.47-2.76	2.26	1.23-4.13			2.58	1.70-3.91
CAGE ≥ 2	1.92	1.25-2.95	2.59	1.35-4.95	3.55	2.12-5.94	3.01	1.85-4.90	2.55	1.45-4.47	4.16	2.61-6.63
Drug use	1.69	1.16-2.46					2.17	1.37-3.42	4.36	2.55-7.45			4.75	3.15-7.16
UK born									2.81	1.40-5.64	1.69	1.10-2.61	2.37	1.42-3.96
Black ethnicity	0.51	0.34-0.77	0.27	0.10-0.68

Having further education (more than 13 years) was associated with self-reported use of illicit drugs, but was negatively associated with depression and deliberate self-injury. Living in council or temporary accommodation was associated with anxiety, depression and a history of parasuicide. Women who had been born in the UK were more likely to have deliberately injured themselves and used illicit drugs than women born outside the UK. Black women were less likely to suffer from anxiety or depression than other ethnic groups.

Associations between abusive experiences and mental health measures

Table 2 demonstrates the independent associations between abusive experiences reported by the women and seven mental health outcomes after adjusting for significant demographic characteristics. Bivariate analyses showed associations with all mental health measures. Following logistic regression, however, neither form of childhood sexual abuse was associated with any of the mental health measures, except for PTSD where women reporting sexual activities before age 16 years were five times more likely to be rated as a PTSD sufferer. However, being beaten by a parent or carer in childhood was strongly and independently associated with several measures: beating on more than one occasion was independently associated with anxiety, depression, PTSD, parasuicide and deliberate self-injury, but not drug or alcohol misuse. Table 2 also highlights the influence of this repetitive form of childhood physical maltreatment. The odds ratios are higher for those reporting beating on more than one occasion.

Table 2

Independent associations between childhood maltreatment, adult abusive experiences and mental health outcomes for women primary care attenders

Variable	Anxiety1 (n=679)		Depression2 (n=629)		PTSD3 (n=716)		Suicide attempt4 (n=719)		Deliberate injury5 (n=716)		Illicit drugs6 (n=641)		CAGE ≥ 27 (n=718)
	Adjusted OR	95% CI	Adjusted OR	95% CI	Adjusted OR	95% CI	Adjusted OR	95% CI	Adjusted OR	95% CI	Adjusted OR	95% CI	Adjusted OR	95% CI
Sexual activity before age 16 years	1.12	0.56-2.23	1.61	0.56-4.66	5.40	2.50-11.7	1.71	0.63-3.83	1.32	0.58-2.99	1.97	0.99-3.90	1.72	0.83-3.55
Sexual intercourse before age 16 years	1.09	0.45-2.63	0.23	0.05-1.00	0.56	0.19-1.61	2.32	0.94-5.75	1.42	0.56-3.61	1.37	0.58-3.24	0.54	0.21-1.36
Beaten once	2.69	1.19-6.05	1.54	0.45-5.32	1.37	0.41-4.59	2.24	0.82-6.11	1.75	0.56-5.44	0.99	0.37-2.67	1.03	0.35-3.03
Beaten more than once	3.89	2.16-7.01	2.34	1.00-5.47	3.62	1.74-7.56	4.74	2.44-9.19	2.99	1.42-6.31	1.03	0.53-2.00	1.37	0.67-2.80
Sexual assault aged 16+ years	0.49	0.21-1.13	1.32	0.42-4.16	1.07	0.39-2.91	1.08	0.43-2.72	1.43	0.59-3.45	2.69	1.30-5.59	2.61	1.25-5.46
Raped aged 16+ years	2.18	0.97-4.88	2.71	0.91-8.05	3.95	1.51-10.3	4.41	1.80-10.8	2.30	0.89-5.92	1.75	0.74-4.15	1.75	0.75-4.07
Domestic violence (ever)	3.31	2.25-4.86	2.76	1.41-5.41	3.27	1.73-6.18	2.18	1.27-3.73	1.40	0.77-2.52	2.19	1.41-3.41	2.44	1.48-4.02

Independent associations between sexual abuse in adulthood and mental health problems also differed according to the severity of the experience. Table 2 shows contrasts between the odds ratios for sexual assault and rape and individual mental health measures: independent associations with sexual assault included illicit drug and alcohol misuse, but not anxiety, depression, PTSD, parasuicide or deliberate self-injury; rape was independently associated with PTSD and parasuicide, but not with anxiety, depression, deliberate self-injury, or misuse of drugs or alcohol.

Women experiencing domestic violence were more likely to report all forms of psychiatric morbidity except for deliberate self-injury. The effect of current abuse was examined by separating women who reported domestic violence in the past 12 months from those who reported it only in the period prior to the past 12 months. Adjusted odds ratios demonstrated stronger associations with most mental health measures for domestic violence occurring during the past 12 months than for violence occurring prior to the past 12 months, including anxiety (OR=3.6, 95% CI 2.2–6.0 v. OR=2.4, 95% CI 1.5–3.9), depression (OR=3.9, 95% CI 1.7–8.5 v. OR=2.1, 95% CI 0.9–4.6), PTSD (OR=2.52, 95% CI 1.15–5.52 v. OR=1.94, 95% CI 0.91–4.14), parasuicide (OR=2.23, 95% CI 1.14–4.38 v. OR=1.95, 95% CI 1.01–3.75) and alcohol misuse (OR=2.58, 95% CI 1.39–4.79 v. OR=1.86, 95% CI 1.04–3.33), but with associations in the opposite direction for illicit drug use (OR=1.83, 95% CI 1.01–3.33 v. OR=2.52, 95% CI 1.49–4.24).

Child maltreatment

In contrast to previous research, our study demonstrated that childhood abuse in the form of a severe beating by a parent or carer was independently associated with several adverse mental health measures, whereas childhood sexual abuse was not associated. These findings also confirmed a dose–response relationship between repetition of physical abuse and anxiety, depression, PTSD, parasuicide and other deliberate self-harm. There is an important divergence between our findings and other studies which found associations between childhood sexual abuse and psychiatric morbidity in adulthood (Reference Bagley and RamsayBagley & Ramsay, 1986; Reference Burnam, Stein and GoldingBurnam et al, 1988; Reference Mullen, Romans-Clarkson and WaltonMullen et al, 1988; Reference Peters, Wyatt and PowellPeters, 1988; Reference Stein, Golding, Siegel, Wyatt and PowellStein et al, 1988; Reference Winfield, George and SwartzWinfield et al, 1990; Reference Bifulco, Brown and AdlerBifulco et al, 1991; Reference Bushnell, Wells and Oakley-BrowneBushnell et al, 1992; Reference Mullen, Martin and AndersonMullen et al, 1993; Reference Fergusson, Horwood and LynskeyFergusson et al, 1996; Reference Kessler, Davis and KendlerKessler et al, 1997; Reference Hill, Davis and ByattHill et al, 2000). It is possible that our measures of childhood sexual abuse were insufficiently precise and were different from several of these studies. We obtained our sample from a primary care setting in the community and did not carry out clinical interviews with the participants to explore in depth the nature of their abusive experiences. However, previous studies might have missed a confounding effect of childhood physical abuse, either by not measuring it (Reference Mullen, Romans-Clarkson and WaltonMullen et al, 1988; Reference Bushnell, Wells and Oakley-BrowneBushnell et al, 1992; Reference Fergusson, Horwood and LynskeyFergusson et al, 1996) or by not including it in a logistic regression model (Reference Bagley and RamsayBagley & Ramsay, 1986; Reference Mullen, Martin and AndersonMullen et al, 1993; Reference Fergusson, Horwood and LynskeyFergusson et al, 1996; Reference Hill, Davis and ByattHill et al, 2000), although some controlled for childhood neglect (Reference Bifulco, Brown and AdlerBifulco et al, 1991; Reference Mullen, Martin and AndersonMullen et al, 1993; Reference Fergusson, Horwood and LynskeyFergusson et al, 1996; Reference Kessler, Davis and KendlerKessler et al, 1997). Our results are concordant with a study by Mulder et al (Reference Mulder, Beautrais and Joyce1998), which found a relationship between childhood physical abuse and adult dissociative symptoms but did not find an association with childhood sexual abuse. Similarly, Ernst et al (Reference Ernst, Angst and Foldenyi1993) found that adult depression was more strongly associated with a range of early familial risk factors, including physical abuse, than with early sexual abuse.

We did not make any prior assumption that childhood sexual abuse was the primary abusive experience associated with the psychopathological symptoms measured in adulthood. The analysis in our study assumed that all abusive experiences had an equal chance of association with these measures. As there was a high level of diagnostic comorbidity, and a marked tendency for the abusive experiences to co-occur over the lifetime for certain participants, it was essential to examine the differential associations. In doing so, the effect of childhood sexual abuse on adult mental health proved to be considerably less important than we had expected.

Associations between childhood sexual abuse and measures of morbidity in previous studies may not reflect causation. Social, family and contextual factors associated with childhood sexual abuse and increased risk of adult psychopathological disorder may have been confounded because childhood sexual abuse is more prevalent in dysfunctional families, with impaired parent–child relationships and parental psychopathology (Reference Mullen, Martin and AndersonMullen et a 1994; Reference Paradise, Rose and SleeperParadise et al, 1994; Reference Stern, Lynch and OatesStern et al, 1995; Reference Fergusson, Horwood and LynskeyFergusson et al, 1996). Associations between childhood sexual abuse and certain forms of adult psychiatric disorder could have been due to these factors (of which severe beating by carers constituted an additional and more important component) rather than a result of direct effects of sexual abuse. However, studies that have examined the severity of sexual abuse, indicated by attempted or completed penetration, long duration, and where physical violence and restraint are an integral part, tend to show an association with later psychopathological disorder (Reference Mullen, Martin and AndersonMullen et al, 1993; Reference Fergusson, Horwood and LynskeyFergusson et al, 1996; Reference Dinwiddie, Heath and DunneDinwiddie et al, 2000; Reference Bulik, Prescott and KendlerBulik et al, 2001).

An alternative hypothesis could be tested in future research which might explain the absence of the expected association. Both childhood physical and sexual abuse substantially and independently increase the risks of further physical and sexually abusive experiences in adulthood (Reference Coid, Petruckevitch and FederCoid et al, 2001). The apparent associations found in studies that have not measured adult abusive experiences could therefore be confounded within a three-stage process, in which the childhood abusive experience independently leads to the adult abusive experience, which in turn independently leads to adult psychopathological disorder.

Sexual assault and rape in adulthood

Sexual violence towards women in adulthood contrasted with sexually abusive experiences in childhood in demonstrating associations with the mental health outcomes. The more severe experience of rape in adulthood demonstrated stronger associations than sexual assault. Associations between sexual assault (but not rape) and substance misuse are consistent with other studies, suggesting that although substance misuse may be associated with unwanted physical and sexual experience in adulthood, this is independent of childhood abusive experiences (Reference Merrill, Newell and ThomsonMerrill et al, 1999). It is possible that women who had experienced sexual assault (but not rape) continued to take more risks than women who had been raped. It has been observed that sexual encounters characterised by heavy drinking are more likely to involve sexual aggression (Reference Muehlenhard and LintonMuehlenhard & Linton, 1987). Furthermore, women with more sexual partners may be at increased risk of victimisation through greater contact with a larger number of men who have the potential to become sexually aggressive (Reference Koss and DineroKoss & Dinero, 1989). We did not collect data on the sexual behaviour of respondents to test this hypothesis.

The question remains as to why rape was not associated with substance misuse. There was no difference in the mean age at which rape and sexual assault had occurred in this sample of women, both experiences occurring at a mean age of 21 years. There are two possible explanations. First, women who were raped and women who were sexually assaulted could have constituted two distinct subgroups. This is unlikely, because both experiences cooccurred in the lives of a substantial proportion of the participants, with 59% of those who had been raped reporting that they had been sexually assaulted. Furthermore, the demographic characteristics of women who had experienced these two forms of abuse were similar. The second – more likely – explanation was that rape, which constituted a more serious trauma in the lives of these women, had led to substantial changes in their subsequent lifestyle, as previously observed by Mezey & Taylor (Reference Mezey and Taylor1988). Presenting symptoms soon after rape do not predict outcome. Some women experience severe impairment, including depression (Reference Ellis, Atkeson and CalhounEllis et al, 1981; Reference Nadelson, Notman and ZacksonNadelson et al, 1982), sexual dysfunction (Reference Feldman-Summers, Gordon and MeagherFeldman-Summers et al, 1977), impaired task performance (Burgess & Holmstrom, Reference Burgess and Holmstrom1974a , Reference Burgess and Holmstrom b ; Reference Walker and BrodskyWalker & Brodsky, 1976) and social adjustment (Reference Resick, Culhoun and AtkesonResick et al, 1981). Associations with PTSD and suicide attempts in our respondents who had been raped suggest that many had been unable to resolve and recover from these traumatic experiences, in contrast to women who had been sexually assaulted but not raped. Their psychiatric sequelae are likely to have further impaired their social adjustment, so that they were less likely to engage in social situations that might involve substance misuse. The strong association between parasuicide and rape in this survey underlines the importance of identifying this traumatic experience for future interventions to prevent suicide.

Domestic violence

The findings of our study are generally consistent with a recent meta-analysis of the literature on intimate partner violence and mental disorder, which suggests that depression, parasuicide, PTSD, and alcohol and drug misuse are common among battered women (Reference GoldingGolding, 1999). Domestic violence was associated with a wider range of mental health measures than any other abusive experience in this study. Domestic violence had a stronger association than rape with anxiety and depression. It is possible that these findings were an artefact, in that the trauma of domestic violence had been more recent, or was ongoing, at the time of the survey (Reference McCauley, Kern and KolodnerMcCauley et al, 1995). This contrasted with rape, which had predated the survey by several years for most women and where some may have shown partial resolution of their symptoms. Golding (Reference Golding1999) suggests that depression tends to remit following cessation of violence, and both depression and PTSD are influenced by the severity of women's violent experiences.

The associations between domestic violence and illicit drug use and alcohol misuse are not explained by our study. Substance misuse in this sample could have been the outcome of the domestic violence, but can also be understood in the context of the abusive partner's substance misuse. Substance misuse may become a feature of the couple's relationship, further increasing the risk of violence between them (Reference Giles-Sims, Jasinski and WilliamsGiles-Sims, 1998).

CLINICAL IMPLICATIONS

1. ▪ Women abused in childhood and adulthood are at risk of further abuse. These experiences have differential effects on the women's subsequent adult psychopathological state.

2. ▪ Experiencing physical violence from carers during childhood may be a more important risk factor than sexual abuse, depending on its severity.

3. ▪ Women should be questioned about ongoing and recent abusive experiences as well as those that occurred in childhood.

LIMITATIONS

1. ▪ The cross-sectional method precluded examination of the direction of the associations found in the study.

2. ▪ Information on most of these abusive experiences was entirely retrospective.

3. ▪ There was limited exploration of the temporal associations between abusive experiences and the onset of psychopathological symptoms.