Dr H Salihu, University of South Florida, College of Public Health, Department of Epidemiology & Biostatistics, 13201 Bruce B. Downs Blvd., MDC56, Tampa, FL 33612, USA. Email firstname.lastname@example.org
Please cite this paper as: Salihu H, August E, Salemi J, Weldeselasse H, Sarro Y, Alio A. The association between female genital mutilation and intimate partner violence. BJOG 2012;119:1597–1605.
Objective To determine whether female genital mutilation (FGM) is a risk factor for intimate partner violence (IPV) and its subtypes (physical, sexual and emotional).
Design Population-based cross-sectional study.
Setting The study used the 2006 Demographic and Health Survey (DHS) conducted in Mali.
Population A total of 7875 women aged 15–49 years who responded to the domestic violence and female circumcision modules in the 2006 administration of the DHS in Mali.
Methods Multivariable logistic regression was used to compute adjusted odds ratios (aOR) and 95% confidence intervals (CI) to measure risk for IPV.
Main outcome measures The outcomes of interest were IPV and its subtypes.
Results Women with FGM were at heightened odds of IPV (aOR 2.71, 95% CI 2.17–3.38) and IPV subtypes: physical (aOR 2.85, 95% CI 2.22–3.66), sexual (aOR 3.24, 95% CI 1.80–5.82), and emotional (aOR 2.28, 95% CI 1.68–3.11). The odds of IPV increased with ascending FGM severity (P for trend <0.0001). The most elevated odds were observed among women with severe FGM, who were nearly nine times as likely to experience more than one IPV subtype (aOR 8.81, 95% CI 5.87–13.24).
Conclusions Study findings underscore the need for multi-tiered strategies, incorporating policy and education, to reduce FGM and IPV, potentially improving the holistic health and wellbeing of Malian women.
Female genital mutilation (FGM) (also known as female genital cutting and female circumcision) is defined by the World Health Organization (WHO)1,2 as any procedure involving the alteration or excision of external female genitalia without medical indication. It has been practiced widely in parts of Africa for centuries, occurring in 28 African countries.1–10 According to the WHO, FGM is almost universal in seven of these countries (prevalence >85%); four countries have a high prevalence (60–85%); seven countries have a medium prevalence (30–50%); and the remaining ten countries have a low prevalence (ranging from 0.6% to 28.2%).1 Although FGM is recognised as an act of violence against women and a violation of human rights by many African governments, the issue is clouded in debate, because it is deeply entrenched in culture and tradition, making legislation difficult to approve and enforce.5,9,11 Previous studies highlight the considerable adverse effects of FGM on women’s reproductive health (e.g. haemorrhage, fistulae, urinary tract infections, incontinence, dysmenorrhoea, dyspareunia, more severe perineal tears at the time of delivery) and psychological well-being (e.g. depression, post-traumatic stress disorder, anxiety) that mitigate quality of life.4,10,12–15
Another equally important human rights concern affecting African women is intimate partner violence (IPV). We, and several other investigators, have found that about one of every two women in sub-Saharan Africa is a victim of IPV.16–19 In Mali, 21% of women reported ever being exposed to sexual or physical violence by a partner.19 Violence against women has been found to be a widely accepted sociocultural norm in Mali, as well as in other sub-Saharan African countries.20–22
A theoretical explanation of the association between FGM and IPV may be found in research on child violence and abuse. Previous studies illustrate that women with previous exposure to violence, particularly physical and sexual trauma during childhood, are two to five times as likely to experience violence later in life.23–27 This re-victimisation often takes the form of IPV during adulthood.25–27 As FGM is a form of violence that frequently occurs during early childhood,2,3,14,28 women who have undergone FGM may be vulnerable to IPV. This may be a reflection of the clustering of behaviours. To date, the association between FGM and IPV remains poorly understood.
Given the plausibility of a positive association between FGM and IPV, delineating the relationship between FGM and IPV is important for the following reasons. (i) If FGM is a marker for IPV, efforts targeting FGM victims could integrate IPV interventions as a comprehensive package to improve the health of women in countries where FGM has been described by the WHO as a major public health issue.1,12,28,29 (ii) Attempts to encourage legislative regulation of FGM in countries where it is a major concern have, so far, been met with either indifference or lack of public support. Identifying and disseminating information on the public health ramifications of FGM (e.g. promoting social norms in which violent acts against women are unacceptable) may aid in galvanising local support for legislative actions against the practice.
In this paper, we sought to determine whether FGM is associated with an increased odds of IPV, as well as its subtypes (i.e. physical, sexual and emotional), using data obtained from a population-based study in Mali, a sub-Saharan nation located in West Africa and described by the WHO as having one of the highest FGM prevalence rates in the world.1
We used the Demographic and Health Survey (DHS) individual recode data. The DHS is an international project conducted in 90 countries throughout the world, including sub-Saharan Africa. This project is funded by the United States Agency for International Development and implemented by Macro International for the collection and dissemination of accurate, nationally representative data. A primary purpose of the DHS is to identify sociocultural factors that negatively impact reproductive function among women of childbearing age (15–49 years).
Given the high prevalence of both FGM and IPV in the sub-Saharan region of Africa, this study focuses on combined data available from this region. The study scope was further limited to the 11 sub-Saharan African countries identified by the WHO as having almost universal prevalence (seven countries with rates of more than 85%) or high prevalence (four countries with rates of 60–85%) of FGM.1 Our inclusion criteria were that the country-specific DHS data set included the modules for both female circumcision and domestic violence. Only one sub-Saharan African nation met both inclusion criteria: Mali. Therefore, our analysis was limited to data provided in the DHS Mali survey.
The DHS survey administered in Mali was modified from basic questionnaires from the DHS programme to consider the context of the country of Mali.30 Numerous peer-reviewed publications have resulted from the DHS data with proven details regarding the validity and reliability of the methodology used.31–34 Other authors have referenced the results of many of these peer-reviewed publications.
The DHS Mali survey (2006) provides data on a nationally representative sample drawn from every region of Mali and used a two-stage clustered sampling design based on national census data.30 After stratifying regions by urban or rural status, a total of 13 695 households were selected at random, with a sampling probability that was proportional to the population of the region. Of these households, 13 160 were identified as occupied. The survey was then administered from May through December 2006 to a total of 12 998 households, resulting in a household response rate of 98.8%. Of the 15 102 women aged 15–49 years who were identified from the selected households, 14 583 women participated in the individual survey, yielding a 96.6% response rate. The high response rate may be partially attributable to the rigorous implementation of the DHS survey methodology and the provision of capacity building at the local level.35 Individuals who are knowledgeable about the geographical area and study population were enlisted as staff for the DHS and were supplemented with technical assistance during all critical stages of survey implementation to ensure all avenues for identification and recruitment of participants are exhausted.35 From survey respondents, a subset of women (n = 10 272) was selected for the domestic violence module, and 9849 (95.9%) were interviewed. This final study sample was further limited to women who responded to the main study questions on both the female circumcision and domestic violence modules, generating a total sample size of 7875. A flow diagram depicting the exclusionary process for the study is presented in Figure 1.
The survey assessed FGM by asking respondents if they had been circumcised. In addition, the DHS survey also collected information on the type of circumcision, the timing of circumcision, and who performed the circumcision. Respondents were also asked questions about their experiences with IPV, including physical, emotional and sexual forms of violence. Due to the sensitive nature of the questions on violence, the interview for the domestic violence module of the DHS was undertaken only when privacy could be achieved and maintained throughout the process. A complete description of the DHS sampling, questionnaire validation, data collection methodology and validation of data has been published elsewhere.30
We measured the main exposure category, FGM, from the survey item ‘Respondent Circumcised,’ which was a dichotomous (Yes/No) variable. We delineated the type of circumcision experienced by women based on the WHO’s four-category classification of FGM.1–3,5,6,10 Type I, traditionally referred to as sunna, involves the excision of the prepuce and the partial or total removal of the clitoris. Type II involves the complete removal of the prepuce and the clitoris and the partial or total excision of the labia minora. Type III, also known as infibulation, is the most extensive form, involving the excision of part or all of the external genitalia with stitching or narrowing of the vaginal opening. Type IV includes modifications that do not fall into the other three types, such as pricking, piercing, scraping, cauterisation, or other procedures performed on the clitoris and labia.
As in previous studies, the proportion of women who had undergone the various types of FGM was based on self-reports, which might result in some inaccuracies.36 Therefore, to reduce the possibility of misclassification, we categorised women into three broad dosage categories based on the level of severity reported: none (no FGM); mild/moderate FGM (WHO Types I, II or IV); and severe FGM (WHO Type III). The determination of these groupings was based on responses to three items regarding circumcision, including ‘flesh removed from genital area’, ‘genital area just nicked without removing any flesh’ and ‘genital area sewn closed’.
In our study, IPV among African couples was the outcome of interest. The domestic violence module of the Mali DHS Survey includes 13 items that capture violence committed by a male partner or spouse. From these questions, IPV was categorised into three main subtypes: physical, sexual and emotional. Physical IPV includes actions such as pushing, shaking, throwing, slapping, punching, kicking, dragging, burning, strangulation, threatening or attacking with a weapon, and other such acts of physical aggression. Physical IPV was determined as a woman’s report of ever experiencing a physically violent act committed by a partner. Sexual IPV was assessed with the item, ‘ever experienced any sexual violence’, which included forced sexual intercourse or other sexual acts by a partner when not wanted. Emotional IPV was assessed with multiple items, such as ‘ever humiliation’, ‘ever threatened harm’, and ‘ever emotional abuse’, or ‘ever emotional violence’, and was defined as a woman’s report of ever experiencing an act of emotional violence by a partner. Women were further categorised as those who experienced one type of IPV and those who experienced two or more types of IPV.
We compared women who had undergone FGM with those who reportedly did not based on a range of variables found to be of importance in the literature. They include: woman’s age (≤19, 20–24 and ≥25 years); woman’s age at first marriage (≤14, 15–19, 20–24 and ≥25 years); woman’s age at first intercourse (≤14, 15–19, 20–24 and ≥25 years); partner’s age (<25 and ≥25 years); woman’s educational attainment (none, primary or secondary or higher); partner’s educational attainment (none, primary or secondary or higher); liveborn parity, defined as the number of children born alive (classified as none, one to four, or five or more); religious denomination (grouped as Muslim, Christian or other/no religion); socio-economic status, using the wealth index (a composite score that assigns individuals to one of five categories: poorest, poorer, average, richer and richest); and marriage type (polygamous or monogamous). As all information was obtained from the female respondent, the age and educational level of the partner were missing in some cases. These sociodemographic characteristics describe the study population at the time of survey administration.
We used Wald chi-square tests of independence to compare differences in sociodemographic characteristics between the FGM and non-FGM groups. The prevalence of self-reported IPV was calculated overall and for each FGM subtype. To account for the complex sampling design employed by the DHS, and to obtain nationally representative estimates, all descriptive statistics were weighted using sampling weights from the domestic violence module. Following the DHS analytic approach, we excluded 87 women with nonpositive weights from all analyses.
The association between FGM and IPV was assessed using multivariable logistic regression. All covariates identified in the literature, as well as those found to be statistically significant in bivariate analyses, were retained in the final models (i.e. age, age at first intercourse, age at first marriage, partner’s age, educational level, partner’s educational level, parity, religion, wealth index, marriage type). To assess the impact of the model selection procedure chosen, we also constructed models using stepwise, forward and backward selection; however, the selection strategy adopted was not found to influence the final results. As all of these strategies yielded the same results, we are reporting the complete multivariable model results that retained all variables that were significant in the bivariate analysis. To ascertain whether a dose–response relationship exists between severity of FGM and IPV risk, the subtypes of FGM were categorised using the WHO classification mentioned earlier1–3,5,6,10 and loaded onto the model as a single variable with incremental dosage levels (i.e. no FGM, mild/moderate FGM, severe FGM).
Tests of hypotheses were two-tailed with a type 1 error rate set at 5%. All analyses were performed with sas version 9.2 (SAS Institute, Cary, NC, USA). The Institutional Review Board of the University of South Florida approved this investigation before initiation.
The sociodemographic characteristics of the sample of Malian women and their partners are reported in Table 1. The mean age of the women was 29.8 years (SD 9.0), whereas the mean reported age of their partners was 41.7 years (SD 12.0). Among women with FGM (90.4%), the average age of circumcision was about 3 years of age on average (SD 3.7). Women with FGM were more likely to be aged 19 or younger at first sexual intercourse (P = 0.003) and were comparatively poorer (P < 0.0001) than their uncircumcised counterparts. Additionally, polygamous marriage was more common among circumcised women (P = 0.004), and a higher proportion of circumcised women were of Muslim faith (P = 0.013). Partners of circumcised women were reportedly older (P = 0.009) with lower levels of education (P = 0.005).
Table 1. Sociodemographic characteristics of Malian women completing the Domestic Violence and Female Genital Cutting modules of the Demographic and Health Survey
Total (n = 7875)%a
FGM (n = 6919)%a
No FGM (n = 956)%a
*Sampling weights were used to account for the complex sampling design of the Demographic and Health Survey. Percentages may not total to 100 because of rounding.
**Partner’s age and education level as reported by the female respondents.
Age at first marriage (years)
Age at first sexual intercourse (years)
Partner’s age (years)**
Secondary or higher
Secondary or higher
5 or more children
Other or No Religion
Table 2 presents differences in reported levels of IPV among women by FGM status. Among women with FGM, more than one-quarter (26.5%) reported experiencing some form of IPV, compared with 17.3% of uncircumcised women. Of the subtypes of IPV, physical IPV was the most prevalent among circumcised women (21.6%), followed by emotional IPV (12.1%) and sexual IPV (4.9%). Nearly 10% of Malian women who reported FGM also reported experiencing more than one type of IPV. Women who experienced a more severe form of FGM reported higher prevalence rates for all types of IPV, compared with those with no FGM, whereas, those with a mild/moderate form of IPV had more elevated rates for only physical and sexual IPV. Furthermore, prevalence rates for physical and emotional IPV were generally higher among women who reported severe FGM, compared with those who reported mild/moderate FGM; however, similar rates were observed for sexual IPV.
Table 2. Rates* of IPV and its subtypes in Malian women by FGM status
Type of IPV
Total, % (n = 7875)
No FGM, % (n = 956)
Mild/Moderate FGM, % (n = 5360)
Severe FGM, % (n = 787)
Unknown FGM**, % (n = 772)
Any type of FGM, % (n = 6919)
*Sampling weights were used to account for the complex sampling design of the DHS.
**Unknown FGM includes women that reported FGM, yet determination of severity of FGM was inconclusive.
No reported IPV
Any type of IPV
One type of IPV
More than one type of IPV
We computed adjusted odds ratios (aOR) and 95% confidence intervals (95% CI) for the association between FGM and IPV (Table 3). Women reporting FGM were two to three times as likely to experience some form of IPV, regardless of subtype. Women with FGM had 2.5-fold increased odds of experiencing one type of IPV (aOR 2.50, 95% CI 1.92–3.25) and were more than three times as likely to report multiple forms of IPV (aOR 3.17, 95% CI 2.19–4.61). Of the covariates, low educational attainment (primary level) and having at least one child increased the odds of experiencing some type of IPV, in general, and, specifically, physical IPV. Polygamous marriage also increased the odds of the occurrence of physical IPV. Younger age (24 years or less) and being of a faith other than Christianity or Islam were found to elevate the odds of sexual IPV, whereas being of the poorest wealth category reduced the odds of sexual IPV by 37% (aOR 0.63, 95% CI 0.40–0.98). Having a primary level of education and a polygamous marriage increased the odds of a woman experiencing more than one type of IPV.
Table 3. Adjusted odds ratios and 95% confidence intervals for the association between female genital mutilation and intimate partner violence and its subtypes
Any type of IPV AOR (95% CI)
Physical IPV AOR (95% CI)
Sexual IPV AOR (95% CI)
Emotional IPV AOR (95% CI)
One type of IPV AOR (95% CI)
More than one type of IPV AOR (95% CI)
The odds ratio estimates are adjusted for all other variables present in the table.
Age at first marriage (years)
Age at first intercourse (years)
Partner’s age (years)
Secondary or more
Secondary or more
5 or more children
In this population, we observed a dose–response relationship between the severity of FGM performed and subsequent IPV, regardless of IPV subtype (Table 4). The dose-dependent relationship was most pronounced among women who experienced more than one type of IPV, where the odds of multiple forms of IPV increased from 2.7 times with mild/moderate FGM (aOR 2.68, 95% CI 1.84–3.91) to close to nine times with severe FGM (aOR 8.81, 95% CI 5.87–13.24).
Table 4. Adjusted odds ratios and 95% confidence intervals for the association between severity of FGM and IPV and its subtypes
Type of FGM
Type of IPV
Number of types of IPV
Any type of IPV
One type of IPV
>1 type of IPV
Discussion and conclusion
Our findings show that FGM was associated with IPV. Malian women exposed to FGM were nearly three times as likely to experience some form of IPV. Physical and sexual IPV were found to have the strongest association, followed by emotional IPV. Our results are consistent with previous research that links childhood exposure to physical or sexual trauma (e.g. FGM) to subsequent domestic violence.23–27
To our knowledge, this is the only study that examines the association between FGM and multiple forms of IPV. In the one previous study of FGM that explored its relationship with physical IPV, conducted a decade ago in Egypt, women who reported being ever-beaten by their husbands were four times as likely to be circumcised, compared with never-beaten women.37 However, in that study, the analysis and conclusions were based on crude frequency comparisons of rates of physical violence only, making it difficult to make meaningful comparison with the results in this paper. Because of methodological and design issues in that paper, the authors did not provide details on subtypes of FGM and IPV.37 Our study is, therefore, unique in that it represents a comprehensive and methodologically sound analysis of the association between FGM and IPV in an African setting.
Our study findings suggest that there may be shared sociocultural factors and established social norms that treat violence against women as acceptable and, therefore, influence the occurrence of both FGM and IPV. Women with a reported history of FGM were more likely to be younger at first intercourse. Furthermore, women with lower educational levels had a higher probability of experiencing IPV. Previous research shows that both early coitus and educational attainment have some relationship with both FGM1,2,29 and IPV,18,20,38 indicating that these public health issues may be linked through common sociocultural factors.
We determined that younger female age was associated with a heightened risk of sexual IPV. Additionally, we observed that women who reported FGM generally had older partners. Together, these findings reinforce previous research that underscores the role of male–female age disparities in sexual relationships and adverse health outcomes in sub-Saharan Africa. Numerous studies have found that younger female age is associated with increased risk of sexual violence among African women.39–43 Furthermore, research conducted within Africa has determined that younger women may be more vulnerable to sexual coercion and violence when in relationships with older men, as a large age gap may affect women’s autonomy because of power and economic imbalances.40,41,43,44
An interesting finding in our study was the reduced odds of sexual IPV among women who were categorised within the poorest wealth category. Poverty has been well-documented as a risk factor for violence perpetrated by intimate partners,21,45,46 so this finding is somewhat counterintuitive. A possible explanation of this finding is that this estimate may reflect an under-reporting of sexual assault and violence within this vulnerable population. As sexual coercion has been noted as a sociocultural norm within Africa,21,22 such practices may not have been recognised and reported among study participants as acts of sexual violence.
This is one of the first studies to yield information that shows an increased risk for IPV with increasing severity of FGM (P for trend <0.0001). However, the interpretation and application of these results should be approached with caution. The operationalisation of the type of FGM within the DHS differs markedly from the WHO typology. Additionally, the description of the type of FGM performed was based on self-reports by women, which may be erroneous, resulting in inaccurate grouping and misclassification bias.
One limitation of this study stems from the cross-sectional nature of the data. Although the data demonstrate a temporal relationship, in which FGM typically precedes marriage and, hence, the occurrence of IPV, the nature of the DHS data does not warrant a discussion of a causal link between FGM and IPV. However, our data do suggest that a childhood experience of FGM is strongly associated with subsequent IPV, similar to childhood sexual abuse.23,27 Another possible limitation of our study is the highly sensitive nature of the issues of FGM and IPV, particularly within the African setting. Although the domestic violence module of the DHS is administered within private settings to support confidential data collection, women may still feel uncomfortable discussing these personal experiences. Additionally, the data on FGM is based on women’s self-reports to trained interviewers, but there is the potential of social desirability bias in reporting. Although the DHS is an impeccably meticulous survey with regard to interviewer training and respect and assurance of privacy,35 the validity of the self-reported data on IPV and FGM cannot be guaranteed. This concern has been noted among other researchers using DHS data for analyses of IPV and FGM.47,48 Consequently, the level of self-reported FGM and IPV in this study may be underestimated, leading to a conservative estimation of the magnitude of the association between FGM and IPV, assuming non-differential misclassification.
Despite the nearly universal acceptance of the use of sampling weights in achieving representative levels of descriptive statistics including percentages and means, there are divergent opinions on the appropriateness of sampling weights in regression analyses when estimating associations at the individual level. Although we used unweighted regression models, as recommended in the Guide to DHS Statistics,35 we assessed the robustness of our findings through a series of sensitivity analyses in which all multivariable models were run using survey procedures that incorporated the domestic violence module sampling weights. The odds ratio estimates produced by the weighted models were slightly attenuated; however, the main study findings did not change appreciably.
In response to the high prevalence of FGM, the Malian government has instituted an executive order to raise awareness regarding its harmful effects; however, no advancements have been made to outlaw or restrict the practice of FGM.49 A barrier, in this respect, is the lack of public support for political or legal measures to abolish a highly respected tradition. A reasonable alternative is to frame the debate in terms of health benefits accruable to the community in the absence of FGM. It is increasingly appreciated in the African setting that IPV is associated with a range of public health issues, including unintended pregnancy,16,18,50 HIV/sexually transmitted infections,18,50,51 and fetal mortality.17,52 Hence, our study findings potentially provide a rich resource to promote a societal debate on the health consequences of FGM, which perpetuates the acceptance of violent acts against women at an early age. The results of this study add to the research regarding the deleterious impact of FGM by underscoring the association between FGM and IPV. Women who report experiences of FGM should be counselled regarding IPV, and interventions should couple preventive efforts for both IPV and FGM to challenge existing social norms and ultimately improve the overall health and wellness of vulnerable women.
Disclosure of interests
None of the authors of this manuscript have conflicts of interest with this research.
Contribution to authorship
HMS conceived the idea, developed the study objectives, provided guidance for the data analysis process, and led the preparation of the manuscript, including writing and editing. EMA aided in the conceptualisation of the study design and statistical analysis, conducted the literature review, and led the writing and development of the manuscript. JLS directed the statistical analysis, including quality assurance and control. HW conducted the statistical analysis. YSS provided expert and intellectual guidance on the manuscript. APA provided guidance regarding study questions and helped with the writing and editing of the manuscript.
Details of ethics approval
The Institutional Review Board of the University of South Florida approved this investigation before its initiation.
This study was funded by a grant from the Flight Attendant Medical Research Institute (FAMRI: 024008, Principal Investigator: Hamisu M. Salihu). The funding agency did not play any role in any aspect of the study.