Prevalence of sexual, physical and emotional abuse in the Norwegian mother and child cohort study

Study population

Our study is based on the Norwegian Mother and Child Cohort Study (MoBa), which is a prospective population-based pregnancy cohort study conducted by the Norwegian Institute of Public Health [14]. The inclusion period was from 1999 to 2008, and 90,700 mothers and 108,000 children participated in the MoBa study. Hospitals with more than 100 births annually were invited to collaborate and 70% of all pregnant women in Norway during this period were invited to participate. The overall response rate was 38.5%. All pregnant women in Norway are offered a routine ultrasound screening at week 18 of gestation at their local hospital [14]. Together with the ultrasound appointment, the women received a postal invitation that included an informed consent form, the first questionnaire and an information brochure. A detailed protocol of the study including the consent can be found elsewhere (http://www.fhi.no/morogbarn). Women who agreed to participate received three extensive self-administrated questionnaires by post during pregnancy. The MoBa sample has been described in more detail elsewhere [14, 15]. Data from the questionnaires are linked to the Medical Birth Registry of Norway, which has kept records of all deliveries in Norway since 1967. This register is based on a standardized form completed by midwives shortly after delivery. Pregnancy was the unit of observation in the MoBa survey; while in the current study the unit of observation was the woman. Figure 1 shows a flow-chart of those excluded from the current study. We merged Questionnaires One and Three, and only women who had filled in both questionnaires were included. For women who participated with more than one pregnancy, only information from their first pregnancy was included. Only singleton pregnancies were included, and only women who had answered a minimum of one of the abuse questions (Figure 2) in Questionnaire Three were included, leaving a total of 65,393 women for the analyses. The current study is based on version 4 of the data files released for research in 2008. Written informed consent was obtained from each participant at recruitment. The study was approved by The Regional Committee for Medical Research Ethics in South-Eastern Norway.

Abuse variables

The third MoBa questionnaire was posted around Week 30 of gestation. It included four questions on abuse. Figure 2 shows the questions on abuse, and the response options provided. The two questions on emotional abuse are similar to two of the three questions on emotional abuse in the validated NorVold questionnaire [16], which is used in a review article and in other studies in the Nordic countries [17–20]. We merged the questions of emotional abuse into one variable in the analyses. The question on sexual abuse with response options in our study, is a modified version of the sexual abuse question in the Abuse Assessment Screen (ASS) [21], which is an abuse screening tool, and has been used in other Scandinavian prevalence studies and in an English study of pregnant populations [6, 10, 11]. The question of physical abuse is not validated. Women could respond “no never” to the various types of abuse, or “yes” as an adult (≥ 18 years) and/or as a child (< 18 years) to the various types of abuse. Women who answered yes to at least one of the adult abuse questions were defined as having suffered from any adult abuse. Likewise, women who responded yes to one or more of the child abuse questions were defined as having suffered from any child abuse. Those who responded yes to any abuse either as an adult or as a child were defined as suffering from any lifetime abuse. Women could also indicate whether they experienced abuse during the last 12 months.

Perpetrators

Other variables

Data analysis

Descriptive statistics were presented for all women. Logistic regression analyses were performed on any adult abuse as crude (unadjusted) and adjusted odds ratios (ORs). In the adjusted model, the various categories of socio-demographic characteristics (age, education, civil status, and parity) and other characteristics (BMI, smoking and alcohol consumption, child abuse) were included. The results from the logistic models were presented with 95% confidence intervals (95% CI) and analysed for complete cases only. The data programme PASW statistical 18 was used in the analyses.

Prevalence of different types of abuse

Overall, any lifetime abuse (which includes adult and child abuse except those exposed to both adult and child abuse) was reported by 32% of the women, 20% reported any adult abuse, 19% any child abuse, and 6% reported both any adult abuse and any child abuse. Figure 3 shows reported types of abuse according to the different age groups. Among those reporting any adult abuse, sexual and physical abuse were reported by 27% (3512) and 28% (3735), respectively, and emotional abuse by 83% (10,739). Among women reporting any child abuse, 37% (4545) reported sexual abuse, while 29% (3610) reported physical abuse and 74% (9865) reported emotional abuse. Of those reporting any adult abuse, 30% had been exposed to two or more types of abuse (Figure 4). The same occurred among women reporting any child abuse, where 31% reported two or more types of abuse. Among women exposed to emotional abuse as an adult, 23% had also experienced emotional abuse as a child; the absolute numbers are shown in Figure 3. Of the women reporting sexual abuse as an adult, 14% also reported child sexual abuse, whereas 12% who suffered physical abuse as an adult also reported child physical abuse. Of those who experienced any adult abuse, 32% reported any child abuse.

Five percent of the study population indicated having experienced any abuse in the last 12 months. Among those reporting physical abuse only as an adult, 8% reported physical abuse in the last 12 months. Furthermore, 3% of those reporting sexual abuse only as an adult also reported sexual abuse in the last 12 months, while among women reporting emotional abuse only as an adult, 22% reported emotional abuse in the last 12 months. The questions on abuse during the last 12 months had responses missing for between 73% to 88% of the different types of abuse.

Perpetrators

Thirty two percent of the women in the study reported any abuse: of whom nearly all (98%) also reported who committed the abuse. Overall, 29% reported a known perpetrator, 5% reported a stranger, and 3% reported being exposed to abuse from both a known perpetrator and a stranger.

Associations between background information and exposure to adult abuse

Figure 5 shows the reported perpetrators of the different types of abuse according to only adult abuse and only child abuse. Table 3 shows the crude and adjusted logistic regression analysis on any adult abuse according to background information. Some factors were strongly associated with being exposed to adult abuse while other factors showed less or no association. In the adjusted analyses, BMI and parity were not substantially associated with any adult abuse. Smoking daily and drinking alcohol weekly in the first trimester of pregnancy or being exposed to child abuse were associated with being exposed to adult abuse (OR =1.6, [95% CI: 1.5, 1.8]; OR 2.5, [95% CI: 1.4, 4.6]; and OR =2.4, [95% CI: 2.3, 2.5], respectively). Married or cohabiting women were less likely to report adult abuse compared with women living alone. Women at 35 years or older were more likely to have been exposed to adult abuse than the younger women. The unadjusted logistic regression was in the main confirmed by the adjusted results.

Strengths and limitations

The large number of participants and the population-based design are major strengths of our study. Furthermore, women were subjected to a broad spectrum of questions and had no information that abuse reports would be linked to other questions. It is a strength of the study that the questions give information about exposure to three types of abuse (emotional, physical and sexual) in addition to information from a long time spectrum (childhood, adulthood and last 12 months exposure), as this gives a broad picture of the exposure to abuse in this population. Three abuse measures give more possible comparisons with other studies, as does the broad time aspect of the questions; as many studies only include one or two types of abuse, seldom three, and usually a shorter time aspect than in our study. A limitation to our study is that none of the four abuse questions in our study were validated when implemented, nor at the time of the start of the survey in 1999. There has been a huge development in the past decade in improving and acknowledging the importance of using validated instruments for research and screening in this field. Nevertheless, not many abuse instruments were validated prior to the start of the MoBa study. The questions on emotional abuse in the current study are similar to those in the NorVold questionnaire which was validated in 2002, but the populations in our study and the NorAQ study are not directly comparable, as the latter study includes patients from three gynecology clinics and one population based sample. The validation study showed that the abuse variables in the NorAQ have good reliability and validity [16]. This was the first validation study of an instrument in the Nordic countries, and the aim was to create an instrument making it possible to compare prevalence rates between the five Nordic countries [16]. Furthermore, there are great similarities between the question on sexual abuse and given response option in the current study, and the question of sexual abuse in the Abuse Assessment Screen (ASS) [21]. It does not have a well-established psychometric property, but it has a broad conceptualization of abuse. According to a review on abuse screening tools, no single tool had well established psychometric properties, including the ASS [21]. The question on physical abuse in our study gives room for subjective interpretation. Nevertheless, we decided to include the question because we wanted to show the broad aspect of reported abuse among our population. Anyhow, for each of the questions, whether the abuse is described well or not, the reported abuse is subjected to the woman’s interpretation of both the questions and her own experiences. As the information was available we thought it was better to use it rather than excluding it.

Our population is based on pregnant women from all over Norway. More than 90% of the women who agreed to participate in the Mother and Child Cohort Study (MoBa) responded to Questionnaires One and Three during pregnancy [14], indicating “dedicated” responders. In addition, only 493 women, (less than one percent) of the participating women, had not responded to any of the abuse questions in the questionnaire. This shows great willingness to respond about abuse exposure. Furthermore, of those who reported one or more types of abuse, almost all (98%) also reported on the identity of the perpetrator. A limitation of the study is the high rate of missing data for the questions on abuse in the preceding 12 months. A reason for this could be the way in which the questions were expressed (Figure 2). Most of the questions in the questionnaire required that the women indicated only if she had a positive answer to the specific question. On these particular questions on abuse during the last 12 months the women were required to change the way of responding by indicating yes or no. In addition these questions were at the very end of the questionnaire that had 94 main questions, with several sub questions.

Substantially more women reported emotionally abuse than other kinds of abuse in our study. It is probably easier to report emotional abuse than sexual and physical abuse. Another reason could be that our study contained two questions on emotional abuse compared with one question of sexual and physical abuse, respectively. The questions in our study allow women to define both “forced” and “sexual acts”, and “exposed to physical acts”. Some cases of sexual and physical abuse will not be identified by this question. The low overall response rate of 38.5% in the MoBa is a limitation. Nilsen et al. investigated this possible bias in the MoBa study by comparing women participating in the study with all women giving birth in Norway, and concluded that prevalence estimates of exposures and outcomes, but not estimates of exposure-outcome, were biased [15]. The same study showed that more women in the MoBa were living alone and fewer were under the age of 25 compared with all women giving birth in Norway. We would expect that these factors and the great number of highly educated women in the MoBa study contribute to a lower prevalence of abuse than in the general population. Retrospective reporting is a challenge, but difficult to avoid in these kinds of surveys. The women were on average 30 years old when responding to exposure to abuse. Their reporting on abuse could be subject to recall bias. Being pregnant could influence their response, as negative exposures denied earlier in life, could come to awareness. The way we see it this can both have a potential impact on depression, and oppositely, being in a depressed state may have an impact of memories and hence on the retrospective reporting.

Comparing prevalence results to other studies

Pregnancy related abuse

Our study gives information about exposure to abuse in the preceding 12 months, asked at about 30 weeks of gestation (Table 1). Hence, our study provides information about exposure to abuse prior to, or during pregnancy, and the results are regarded as pregnancy-related. Our findings on last-year prevalence of any abuse were 5%, corresponding with the first national Norwegian study in a non-obstetric population, where 6% reported any partner abuse in the preceding year [12]. Our results are, however, in the lower range of the findings in WHO’s multi-country study, where between 4% (Japan and Serbia and Montenegro) and 54% (Ethiopia) of the women reported exposure to partner abuse in the last 12 months [9]. Findings in this article showed that the prevalence of abuse is usually lower in industrialized settings than in rural settings [9]. Our results correspond with the lower prevalence rates reported in the latter study, and are also in the lower range of the findings from the Gazmararian et al’s review article on the prevalence of abuse of pregnant women in developed countries, which found that exposure to abuse in the preceding 12 months in four studies varied between 6% and 24% [7]. These differences in methodology may explain why our results correspond with the lower reported prevalences, in addition to the possibility that there is a real lower exposure to abuse in Norway as an industrialized country.

Perpetrators

Our results showed that a known perpetrator is more frequent for all types of abuse (Figure 5). This finding corresponds well with other studies reported in pregnant populations [10]. WHO’s multi-country study suggests that women are at more risk of abuse from intimate partners than from any other [9]. The questionnaire in MoBa did not elicit information about a partner or former partner being the perpetrator, out of consideration for the women’s safety receiving and possibly filling out the questionnaire at home. Other research, however, suggests that this known person most frequently will have been the present or former partner [9].

Background information and relation to abuse

Living alone, exposure to child abuse, drinking alcohol in the first trimester, and being 35 or older were associated with exposure to any adult abuse in our study. Women living alone were a small group in our sample, but interestingly, the study also showed a higher exposure to abuse in the cohabiting group compared with the married group (OR 1.3, 95% CI 1.2-1.4). Our results showing that living alone or being single was associated with a higher exposure to abuse and that being married or cohabiting was a protective factor correspond with another study [11]. Our results also agree with studies showing an association of exposure to child abuse [34] and of use of alcohol [34, 35] with increased prevalence of reporting abuse, even though none of these studies can predict a causal connection between exposure to background factors and exposure to abuse. The cross–sectional design of our study provides associations and not causal relations. In the current study, women above 35 reported more exposure to any adult abuse than women in the other age groups. This may be due to accumulative effects, as the older subjects have had more time to be exposed to abuse. A Swedish clinical study showed the contrary, however, as high age was negatively associated with lifetime abuse in that study [23]. The WHO’s study on recent abuse reported higher exposure to abuse with lower age [34], and in Devries et al. study, prevalence of abuse during pregnancy was relatively constant to the age 35 and then slightly declined [22]. Younger age may reflect less opportunity to protect oneself and lower reporting from the eldest can be due to fading of memory with age. The literature is inconclusive regarding education and exposure to abuse. Norway has a generally high level of education and more women than men graduate at university level. In our study, we chose to divide higher education into two groups, those who completed four years of education at university level and those with more than four years. Our results showed that the association to any adult abuse was weaker in the group reporting four years of education at university level compared to all the other educational groups (Table 3). A low level of education is reported to be a risk factor for exposure to abuse in the populations-based WHO study on recent abuse [34], while a Swedish study from three clinical populations and one randomly selected population reported that educational level had a positive association with physical abuse but not with sexual abuse in both clinical and population samples [23]. One possible explanation is that women with higher education have higher self-esteem, are more aware of their rights, and tolerate less violation of their integrity [23]. Studies show that background factors have different impacts on different types of abuse. This indicates that the type of abuse (emotional, physical or sexual) and whether it is a single type or overlapping types are results of various patterns. Risk factors therefore vary depending on the type of abuse studied, as suggested by a study from Vietnam [30].

Public health implications

Previous research has shown that abuse of women and children is associated with morbidity for the women and the children, possibly both with short and long term consequences. Studies, including the current, have reported that abuse of women is more frequent than many other pregnancy complications [10]. Five percent of the women in our study reported exposure to abuse in the last 12 months at Week 30 of gestation. This is comparable with the prevalence of preeclampsia (2-5%) and gestational diabetes (5%) in Norway, conditions for which pregnant women are routinely screened. Several studies have shown more negative reproductive health consequences in abused than in non-abused women, e.g. reporting more pregnancy terminations [11], and more pregnancy complaints and fear of birth [20, 36]. Self-reported poor health and psycho-somatic symptoms are also more common in abused than non-abused women [19]; so also with symptoms of chronic pelvic pain, stomach pain, headache, emotional distress and depression [12, 25, 37, 38].