Background: Violence against women is multifaceted, and multidimensional global phenomenon; but highly endemic in Pakistan. Gender roles and socio-economic factors put women in subordinate status that makes them prone to all kinds of violence. Violence influences personality, lifestyle, and health; particularly reproductive practices. The current paper aimed to assess the prevalence and factors associated with violence against women in Pakistan.

Methods: Secondary data analysis of the Pakistan Demographic and Health Survey (PDHS) 2012-13 was done. A total of n= 2941 married women between the ages of 15-49 were included; of these 841 were asked about violence. The variables included socio-demographic characteristics, violence, and its typology, fertility and contraceptive use. Data were analyzed using SPSS version 19. Descriptive statistics were run to describe the population socio-demographic characteristics and violence. Binary logistic regression and multivariate model were used to assess the possible association of violence with socio-economic characteristics.

Results: Of the 2941 ever-married women of the reproductive age group (15-49 years) from Sindh included in the analysis, 58.35% were illiterate, 30.09% were employed workers, 28.24% were current contraceptive users and 34.70% ever terminated pregnancy. Of those who were asked about violence (n=841), 38.34% reported experiencing any type of violence, however, 58.90% neither sought help nor informed anyone. On Univariate analysis violence was found to be associated with: women's illiteracy (OR 3.0, 1.7-5.3), poverty (OR 3.0, 2.0-4.7) non-working status (OR 1.8, 1.3-2.5), lack of house ownership (OR 1.7, 1.1-3.0), Husbands' characteristics that were found to be associated with violence against women were illiteracy (OR 2.5, 1.6-4.0) and skilled manual occupation (OR 2.2, 1.1-4.2). Three factors that remained associated with violence on multivariate analysis were; women's non-working status (OR 1.8, 1.3-2.5) and high parity (OR 1.7, 1.2-2.4) and husbands' education (OR 1.8, 1.1-2.5).

Conclusion: Violence against women is prevalent in Pakistan; most victims are not reporting or seeking help. Victims of violence are uneducated, unemployed, high parity women who were non-contraceptive users and whose husbands were illiterate. To decrease the magnitude and intensity of violence against women, short and long term promotive to rehabilitative actions are required ranging from establishment of shelters to provide relief to violence victims to structural reforms for improving the status of women to control and prevent violence.

Keywords: Violence; women; gender roles; socio-economic factors; Pakistan

Across the globe, in all societies, women and girls are subjected to physical, sexual and psychological abuse that cuts across lines of income, class, and culture [1]. The worldwide average of physical and/or sexual intimate partner violence or sexual violence by a non-partner at some point in women's lives is estimated as 35 percent through [2], in some countries this burden is up to 70 percent [3]. Violence against women not only violates women's rights and fundamental freedom, but it also impairs their health particularly sexual and reproductive health.

Although violence against women is a complex multifactorial and multi-dimensional phenomenon; literature, however, attributes it to women's socio-economic positions. Nevertheless, the pervasiveness of violence across class, religion, ethnicity and the urban/ rural divide, strongly indicates towards this being gender-based [4]. Gender-based violence has its roots in a patriarchal social structure in which women are subservient to men, and are viewed as property; many studies have identified gender inequity as one of the key determinants of violence against women [5]. Low socio-economic status intertwined with patriarchal culture put the women in a subordinate position that have repercussions for their three important spaces namely; bodily personal and cognitive [6].

On the same premise, Pakistan provides a conducive environment for violence against women; the country ranks extremely low in Gender Development Index (GDI) (125th out of 169 countries) [7], and Gender Empowerment Measurement (GEM) (99th out of 109 countries) [8]. According to an estimate, 70 to 90 percent of Pakistani women are subjected to domestic violence [9]. Pakistani girls and women are far worse off than their global counterparts; women in Pakistan are facing discrimination and inequality in almost every aspect of life [10]. Consequently, Pakistan is reported as the third most dangerous country in the world for women, after Afghanistan and the Democratic Republic of Congo [11]. Approximately 5000 Pakistani women are killed per year from domestic violence, with thousands of others maimed or disabled [12]. Spousal abuse is rarely considered a crime socially unless it takes an extreme form of murder or attempted murder; examples range from driving a woman to suicide or engineering an accident such as the bursting of a kitchen stove [13]. It is important to estimate the magnitude and extent of violence against women in Pakistan through however, what is more, crucial is to find out the role societal practices and context play on the occurrence of violence and its consequences on women's health. The aim of this paper is therefore to determine the socio-cultural, individual and relationship level factors associated with violence against women and the effects of violence on reproductive health, especially contraceptive use.

We present results on the secondary data analysis of Pakistan Demographic and Health Survey (PDHS) 2012-13. The PDHS survey presents a randomly selected nationally representative sample of all four provinces. The sample from each province is selected through a multistage cluster sampling technique using uniform methodology. We present results from the province of Sindh only.

The PDHS survey mainly collects information related to housing conditions; socioeconomic characteristics; education; nutrition; maternal/reproductive, adult and child health; mortality and its causes; knowledge regarding Tuberculosis, Hepatitis and Human Immuno-Deficiency; and healthcare practices [14]. In the 2012-13 PDHS, for the first time, information was also gathered regarding women's empowerment and violence against them.

All married women between the ages of 15-49 (n=2941) were included in the descriptive analysis. The Socio-Demographic variables included age, gender, education, employment status, occupation of woman and husband, income, possession of house-hold goods, residence in terms of urban/rural, city and province, access to media and age at first marriage. For the determination of the prevalence of violence and the assessment of the possible association of violence with socio-demographic characteristics, only those women were analyzed (n=841) to whom questions about violence were asked. The variables related to violence included types of violence, spousal violence, the experience of physical violence, and violence during pregnancy.

The weighted analysis approach was used to adjust for sampling technique that was in the demographic survey (Multistage Cluster Sampling). Frequency distribution of demographic and socio-economic variables including the age of women, area of residence, education level of women and husband occupation and ethnicity are reported (Table 1). Weighted Simple and Multiple Binary Logistic Regression was used to determine the factors related to violence. In univariate analysis, crude Odds Ratio (OR) with 95% confidence intervals were reported. All significant variables (p-value< 0.05) at the time of univariate analysis or biologically plausible with p-value< 0.2 were considered eligible for multivariable analysis. In multivariable analysis, Adjusted Odds Ratio (AOR) with 95% confidence interval were reported. All variables whose p-value were less than 0.05 were considered as a factor at a multivariable level to get a parsimonious model. Data were analyzed using SPSS version 19.

Of 2941 ever-married women of the reproductive age group (15-49 years) from Sindh included in the analysis, 1652 were from urban and 1283 from rural areas. Around two third of these women were between 20-39 years (69.80%) and more than half had no education at all (58.35%). One third (35.29%) of women were working at the time of the survey. The most frequent ethnicity was Sindhi (35.73%) followed by Urdu speaking (23.12%) and Punjabi (19.87%). In terms of wealth index, the major proportion of women were from the poorest (31.19%) families followed by the richest (28.61%). Concerning their husband's characteristics, most (64.70%) of the husbands had some education and nearly all of them (98.8%) were working. Of those who were working, nearly half (45.81%) of them were manual workers either skilled (23.53%) or unskilled (22.28%). Of all the women interviewed, 28.24 percent were current contraceptive users, however, 34.70 percent ever terminated pregnancy (Table-1).

Of these ever-married women of the reproductive age group (15-49 years) from Sindh included in the analysis (n=2941), 841 were asked about violence. Of these (n=841), 38.40 percent reported experiencing any type of violence. Physical violence was the commonest (23.53) followed by emotional one (14.81%); 7.43 percent of women even reported violence during pregnancy. Those who reported experiencing violence, 58.9 percent neither sought help nor informed anyone. Weighted analysis of ever-married women between 15-49 years ever experiencing violence revealed that rural uneducated high parity non-working women belonging to poor families between the ages of 25-39 mostly experienced violence. Husbands of these women were uneducated and skilled manual workers (Table-2).

On Univariate analysis, violence was found to be associated with certain women's characteristics including; lack of education (OR 3.0, 1.7-5.3), poverty (OR 3.0, 2.0-4.7) non-working status (OR 1.8, 1.3-2.5) and lack of house ownership (OR 1.7, 1.1-3.0). Husbands' characteristics that were found to be associated with violence on women were lack of education (OR 2.5, 1.6-4.0) and skilled manual occupation (OR 2.2, 1.1-4.2) (Table-2). On multivariate analysis three factors that remained associated with violence were; women's non-working status (OR 1.5, 1.1-2.5), and high parity (OR 1.7, 1.2-2.4) and husbands' illiteracy (OR 1.8, 1.3-2.5) (Table-3).

In our study the reported prevalence of any type of violence (38.40%) is slightly higher than world average (35.00 %) [2], same as in Sub-Saharan Africa (38.40) [15] and almost similar to the South-East Asian average (37.7 %) a population-based study conducted in Pakistan has however reported a higher prevalence (57.6 percent) [16]. The high prevalence of violence in the country suggests the need for immediate interventions to serve the immediate needs of violence victims, such as the establishment of specific service delivery points like shelters, safe retreats, and counseling services [17, 18]. However, to control and prevent violence, structural changes are required to address underlying causes and risk factors of violence. In this regard, Pakistan is a signatory of all international treaties [1] that aim to ensure human rights and minimize violence against women. The Pakistani government has transformed its commitments into national priorities and translated these priorities into legal and policy initiatives. Consequently, several ministries/authorities/institutions have been formed to protect women's rights. These ministries/authorities/institutions have improved the advisory and research role of government institutions about women's rights though, yet however, nearly all gender equality and women empowerment indicators in the country show slow progress [13]. Furthermore, there is a rise in the magnitude, extent, and intensity of violence against Pakistani women during the last decade [19]. The two major reasons for the inability of these institutions to control violence against women are inadequate finances and administrative authority though, however failure to functionally integrate gender into other governmental sectors is also an important gap [4]. To effectively decrease the magnitude of violence against women, institutions mandated to promote gender equity and prevent gender-based violence have to work in close collaboration/coordination with other sectors to mainstream gender in health, education, security, legal and finance.

The current study also found that around 59.90 percent of Pakistani women experiencing violence neither sought care nor informed anyone. Even globally, the same proportion of women do not seek help and those who do, most look to family and friends and less than 10 percent appeal to formal institutions and mechanisms [3]. In Pakistan, laws are though protecting women's rights [20], women are still not seeking legal help possibly because of the unfair punishment received by women who claimed legal rights. These punishments were given by existing parallel justice system like Jirga and panchayat; these structures prescribe tribal/community norms that consider that violence against women is a private matter, and women's demand for individual rights is a violation against the norms [21]. Within this dogmatic society, adequate implementation of laws alone would not be enough, legal assistance needs to be brought closer to women to facilitate them in seeking legal support at the time of need; establishment of national and district level legal assistance centers could be an option [22]. Women's lack of knowledge about their rights and recourse options is another important reason for women's silence over abuse and for not seeking legal help; lack of knowledge in fact is a fallout of absence/limited education and economic opportunities for them [10]. Short-term training programs can make women aware of their constitutional and religious rights, and equip them with competencies to demand these.

Our study found a higher occurrence of violence among non-working women. Other studies also have found a similar relationship between violence against women and their earning [23, 24], financial independence [25] and socio-economic positions [26, 27]. This suggests the need for enhancement of educational and employment opportunities to help reduce violence against women. In Pakistan and other similar countries, rural women spend a great deal of time and energy in unpaid domestic and fieldwork [28] establishment of cottage industries can help rural women earn and raise their status [29]. However, it is not that simple! Women's role and hence scope of work in the society is determined by cultural norms in general and the male perception about women's role in particular. Behavior Change Communications are therefore crucial to gender sensitize communities and create gender equity in societies so that women are not considered inferior anymore and given equal education and employment opportunities. This will enable women to get an education, employment and hence protect them from violence.

We also found that women experiencing violence were low contraceptive users as also shown in studies from developed and developing countries for several decades [30-34]. Various researchers have identified a possible reason for the relationship between violence and non use of contraceptive including; women's lack of control over her reproductive functions [35-38], fear of abuse among women that make birth control negotiation difficult [33, 35, 39-43] and use of verbal and physical violence by men to force women to become pregnant [44-46]. Studies have also shown that non-use of contraception results into unintended pregnancies which are more than twice as likely to end in abortion [15, 33, 34, 44, 47-50]. Moreover, both population- and clinic-based studies have reported a direct relationship between unintended pregnancy and physical violence [30, 31, 34, 41, 45, 51-60]. Similarly, in our study, 33.70 percent terminated pregnancy a least once. Since, women frequently take long time for deciding to leave violent relationships [39, 61, 62], it is important to guide these women to plan their pregnancies. A positive move in this regard is the recognition of violence against women a major public health problem [63] and hence the recommendation of professional bodies to address this issue in clinical practice through screening patients populations[64, 65]. For this, clinical protocols including family planning counseling guidelines should include a section on violence. Experience of violence can be used as a risk marker to identify women at risk of contraceptive non-use, unintended pregnancy, and abortion. Training of healthcare providers will also be required to interact and counsel this vulnerable group since counseling on this sensitive personal issue would be challenging. Within the Pakistani milieu where clinical protocols and guidelines are not universally available and healthcare providers rarely counsel, efforts are needed to equip women with communication, negotiation and decision making skills especially with regard to reproductive health decision making.

The current paper is based on the secondary analyses of the data of the Pakistan Demographic and Health survey 2012. Being a large national survey the sample size was adequate to conduct univariate and multivariate analysis; the sample was representative of the target population and findings had greater validity and generalizability [66, 67]. Availability of substantial breath of information on socio-economic variables of the respondents and their families helped in determining the association of these factors within violence against women. However, the study could not identify any novel locally prevailing context and culture-specific factors that might be contributing in violence against women since the data were collected for some other purpose; one of the limitations of the secondary data analysis method approach [68]. Moreover, since the paper has been based only on a cross-sectional survey it could not found out why the trend of violence against women has risen during the last decade and how the factors such as women's unemployment work and husband's education are contributing in violence against women. Also, the sample was restricted to one province. Future research using a mixed-method approach may be required to answer these questions.

We conclude that despite the government's commitments and extensive efforts, violence against women is still prevalent in Pakistan and women are not reporting or seeking help. Most common victims of violence are uneducated, unemployed, high parity women who were nonusers of contraceptives and whose husbands were illiterate and skilled workers. To relief violence victims and also decrease the occurrence of violence against women, short and long term promotive to rehabilitative actions are required. Curative and rehabilitative short and medium-term actions could be: inclusion of a section on violence in gynecological/obstetric/family planning protocols/guidelines to identify and guide women at risk of non-use of contraceptives, unintended pregnancies and unsafe abortions; training of healthcare providers to communicate, counsel and guide violence victims; establishment of shelters, safe retreats and counselling services; creation of national, provincial and district level legal assistance centers; building public pressure by human rights groups and civil society to ensure effective implementation of law against perpetrators of violence against women; and review and revision of laws.

Promotive and preventive long-term actions include:attainment of gender equity through gender mainstreaming in all the sectors such as health, education, security, or other services; effective coordination and collaboration between government and non-governmental organization working for women's rights and protection; designing and implementation of socially acceptable and culturally appropriate behavior change communication interventions to modify behaviors, practices and attitudes towards women and role of women; sensitization and awareness-raising about gender and gender roles, women's rights and mechanisms to demand these rights; trainings to equip women with communication, negotiation and decision making skills; and enhancement of educational and employment opportunities for women such as establishment of cottage and other small industries.

The study was conceptualized by NR and drafted by NR. All authors have read and approved the final version of the manuscript.

Authors would like to acknowledge Mr. Iqbal Azam, who has contributed to quantitative data analysis.