Evidence suggests national- and community-level interventions are not reaching women living at the economic and social margins of society in Pakistan. We conducted a 10-month qualitative study (May 2010–February 2011) in a village in Punjab, Pakistan. Data were collected using 94 in-depth interviews, 11 focus group discussions, 134 observational sessions, and 5 maternal death case studies. Despite awareness of birth complications and treatment options, poverty and dependence on richer, higher-caste people for cash transfers or loans prevented women from accessing required care. There is a need to end the invisibility of low-caste groups in Pakistani health care policy. Technical improvements in maternal health care services should be supported to counter social and economic marginalization so progress can be made toward Millennium Development Goal 5 in Pakistan.

Complications of pregnancy and childbirth remain the leading cause of death and disability for childbearing women in Pakistan. With a maternal mortality ratio of 297 per 100 000 live births,1 Pakistan is 1 of 6 countries contributing to more than 50% of all maternal deaths worldwide.2

The widely endorsed strategies for promoting safe childbirth3,4—skilled birth attendance, with timely referral for emergency care in a well-functioning health care system—are reflected in Pakistan’s formal maternal health policy.5 Significant attempts to implement this policy have occurred, first under the Safe Motherhood and later the Millennium Development Goal (MDG) initiatives, with a degree of success in improving services.6

Yet despite these efforts, Pakistan is unlikely to meet the targets of the fifth MDG, which includes the reduction of maternal mortality and universal access to reproductive health care by 2015.2,7 We suggest one reason for this failure is that the maternal health strategies used to date have failed to understand and address the factors that restrict access to care for women living at the economic and social margins of society. The most recent national level data show the persistence of large socioeconomic inequalities in access to maternal health care services across every indicator. Although 92% of women in the highest wealth quintile report antenatal care use, this percentage is 37% among women in the lowest quintile. Similarly, 74% of women in the highest wealth quintile deliver in a health facility compared with 12% of women in the lowest quintile.1

The current Pakistani National Maternal Newborn and Child Health Program (2013–2016) acknowledges these inequities and aims to provide care to the disadvantaged and vulnerable groups.5 To date, policy and practice interventions remain predominantly “more of the same,” with most effort focused on strengthening district health systems through improvements in technical and managerial capacities, and the introduction of a new cadre of community-based skilled birth attendants. Nevertheless, there has been some effort to increase demand for maternal health services through state-run targeted, socially acceptable communication strategies.5 This reflects the growing recognition that a simple, supply side, technical approach is insufficient to address inequalities in access to maternal health care services or to impact on levels of maternal mortality.7,8 Furthermore, recognizing that maternal health-seeking behavior is constrained by a range of cultural and structural factors, particularly gendered norms that devalue women’s well-being and also poverty, a number of small-scale interventional approaches have tried to reduce the demand-side obstacles. Bhutta et al.9 have described a community-based cluster randomized controlled trial in Sind that included promotion of health care seeking (including facility birth) and maternal health education, through group sessions delivered by trained women health workers. The 10 district Pakistan Initiative for Mothers and Newborns (PAIMAN) intervention between 2005 and 2010 combined upgrading of health facilities with behavior change and community mobilization interventions, including a particular focus on “birth preparedness” and “complication readiness.”6 An intervention in 2009 provided women in the Jhang district of Punjab who met poverty selection criteria with highly subsidized antenatal, delivery, and postnatal care through a low-cost voucher scheme and reimbursement of travel costs.10

The effects of these interventions on levels and disparities in maternal health care use has been variable. Bhutta et al.9 reported a 10% increase in facility births (54% vs 44%; P = .07), but these authors did not explore variations by socioeconomic status; this was a surprising omission given that their intervention was based on geographic clusters rather than targeted to particularly poor women, and therefore, introduced the possibility of differential benefits between social groups. The PAIMAN intervention was found to increase levels of skilled birth attendance and postpartum care across all wealth quintiles, but did not decrease the differential between the rich and the poor. In terms of institutional delivery, no increase was seen among the poorest wealth quintile, and there was an increased disparity between the poorest and all other wealth groups (17% vs 74% in the richest quintile postintervention).6 The voucher scheme,10 which specifically targeted poorer women, appeared to have more success in tackling inequalities, with a significant increase in institutional delivery among the poorest quintile (31%–47%) and a reduction in the disparity between this group and the richest group (33%–16% points). Nevertheless, even after the introduction of the scheme, more than 60% of the poorest quintile of women reported not receiving adequate antenatal care, and more than 50% did not deliver in a health care facility. Importantly, Agha10 reported that selling the vouchers to women was a difficult and time-consuming task, but that utilization of the vouchers for antenatal care and delivery care was very high (approximately 97% in the latter case). The limited impact therefore related primarily to enrollment in the intervention and raised questions about who remained outside of the scheme and why.

The partial success of these demand-side interventions in tackling inequalities raises doubts as to their ability to effectively identify the poorest women and address the obstacles to care that they face. It also questions whether they are adequately informed by an understanding of the factors that restrict access. Although research in Pakistan has consistently documented large socioeconomic inequalities in maternal health service access,1,11 there has been little investigation to date beyond descriptive analyses of correlates (e.g., wealth and education) that could provide insight into the underlying causes and potential remedies of these persistent disparities.

There have been sustained calls for greater attention to the “causes of the causes” of poor maternal health outcomes and a return to framing health improvement within a political context.12 Although the relationship between economic poverty and uptake of maternal health services is well documented, a growing body of research elsewhere in South Asia suggests the importance of understanding the role of nonmaterial aspects of poverty. Such research indicates that economic poverty alone does not explain the large disparities in access to maternal health care between the rich and poor and highlights the ways in which sociocultural hierarchies operating along the lines of gender, caste, religion, and ethnicity act to systematically exclude and marginalize particular groups of women.13,14 A number of Indian studies draw attention to the large variations in maternal health care use between castes, even after controlling for income and education.15,16 While currently overlooked in work on maternal health in Pakistan, the importance of understanding the multidimensional nature of poverty is also suggested by recent social development research. At the macrolevel, Pakistan has been found to show poor progress on human development and poverty reduction compared with countries with similar levels of per capita gross national product—a situation that has been attributed in part to its “high degree of ethnic and class polarization.”17(p30) Microlevel studies demonstrate how hierarchical social relationships based on caste and extended family (biradari) position some groups as subordinate to others, leading to economic exploitation, social exclusion, and political marginalization.18,19

Currently, despite the stated commitment to provide care to disadvantaged groups,5 neither national-level projects nor community-based interventions in Pakistan have been informed by a detailed understanding of how long-term economic and social marginalization operate to constrain poor women’s access to maternal health services. Understanding these structures and processes and being able to effectively identify the poor and socially excluded is crucial to developing interventions that can tackle the persistent inequalities. We report a detailed, qualitative study that begins to address these questions. Our findings paint a clearer picture of the realities of marginalized rural Pakistani women’s lives and suggest a number of key factors that must be addressed in the design and implementation of policies and interventions if they are to be more successful in tackling inequalities in maternal health.

We conducted a 10-month qualitative study (May 2010–February 2011) in 1 village in district Chakwal, northern Punjab, with a team of 5 researchers (including the lead author) with the aim of describing, in detail, the social structures and patterns of resource distribution and their links to maternal health.

Ganji (a pseudonym) is a loosely defined village, home to approximately 1296 people. It is connected by an all-weather road to the main highway of the country. Appreciating that no single village can be representative of a whole country, Ganji was selected for both methodological and practical reasons. The village exhibited heterogeneity in socioeconomic status and caste, thereby offering a suitable location in which to study these social divisions. In addition, the research team members were well-acquainted with the culture and language of the area, and importantly, we were able to negotiate entry into the fieldsite via a local nongovernmental organization. This introduction was important to ensure both the safety of the research team and the feasibility of conducting in-depth fieldwork, because Pakistan’s fragile security situation has made villagers wary of strangers. District Chakwal is a relatively well-developed district. Landholdings are small compared with southern Punjab or Sind because farming practices are rain-dependent. Men commonly seek work beyond the village to supplement incomes. The district is a part of the “martial belt” because men from this region traditionally join the army. These factors mean that poverty levels are somewhat lower than southern Punjab, Sind, and Balochistan. Two fully functional public sector rural health centers (RHCs), staffed by resident female physicians, are an approximate 15-minute drive on either side of the village and provide 24-hour services recently upgraded by the Punjab Department of Health and the United Nations. A first-level care facility is situated approximately a 10-minute drive away and has a female physician available twice a week and a labor room that is accessible during morning working hours. A district hospital about an hour-and-a-half drive away provides emergency obstetric services. The private sector consists of untrained birth attendants, trained midwives, and several physicians, all within a 1-hour drive. There are also 3 large teaching hospitals a 4-hour drive away.

We investigated four interrelated phases of data generation: (1) familiarization and rapport building (4 social-mapping exercises, home visits, and a basic sociodemographic survey covering all the village households and including information on service use for all pregnancies in the preceding 5 years); (2) exploration of social norms and everyday practices, and their relation to maternal health care seeking (observations and informal interviews, mapping of behaviors and decision-making processes for 18 pregnant women, and in-depth interviews with 34 young women, 27 older women, 20 young men, and 13 older men); (3) detailed exploration of 5 maternal deaths that occurred in the preceding 4 years (interviews with husbands, mothers-in-law, other women, neighbors, and health care providers associated with the dead women around the time of her death); and (4) respondent validation (11 focus group discussions of 6–10 individuals each organized along age, gender, and caste lines; repeated conversations, 61 in total, with 2 key informants). We selected respondents on the basis of experiences that we believed would illuminate the ways in which processes of social exclusion operated and their impact on maternal health, while also ensuring that we included a wide range of people varying in terms of age, gender, socioeconomic position, and caste. Our data generation approaches ensured a large and very rich body of data. Most data were digitally recorded. Data analysis occurred concurrently with data collection and creation of a database of transcribed notes.20 The researchers immersed themselves in the data to gain intimate knowledge of the cases and reflect carefully on their meaning.21,22 Data were coded and manually recategorized into domains, which were then analyzed to extract themes. Narratives from different sources were merged to describe typical experiences and behaviors. Atypical experiences were also accounted for and alternate explanations carefully considered through collective deliberation by the research team members.

Our fieldwork revealed that the village structure was based on the traditional South Asian occupational hierarchy or caste system (see the box at the top of page e4). The lowest caste, the Kammis, made up just less than one fifth of the village inhabitants, were largely landless, lacked access to education and income-generating opportunities, and were characterized by long-term intergenerational poverty. Around 70% of Kammi families were engaged in highly asymmetrical employment arrangements with higher caste individuals. Two such arrangements were prominent. First, all the Mussali families (the lowest subcaste of the Kammi caste) were tied to higher caste wealthy and socially powerful people in a social contract called the “seph.” The Mussalis worked year round, performing agricultural and domestic tasks. Their remuneration consisted of immediate, but small, cash payments and longer term benefits, including money for major expenses such as health emergencies. Mussali people within a seph were referred to as “belonging” to their high caste “masters,” and their entitlements were found to be informal and uncertain.

Masterji brought us here … the house and land belongs to his sons now and we live here in exchange for work. (Kammi woman, 70 years old).

The Village Structure
The village caste system consisted of 5 main castes (zaat or quaum): Chaudhrys and Rajas at the top, followed by the Mirzas, the Miannie, and finally, the Kammis. Each caste was further subdivided into extended family groups called biradaris.
We are Janjua Rajput… we are the highest caste. Of all the quaums Allah has made, we are the highest zaat. (Rajput woman, 64 years old).
The social and economic hierarchies were largely found to mirror and reinforce one another. Caste boundaries were largely rigid and caste identities resilient. Members of the Kammi caste traditionally performed low status occupations such as butchering or shoemaking. The lowest subcaste of the Kammis, the Mussalis, were found to perform the most menial tasks and to hold a similar social status to the Dalits (untouchables) in India.

Second, several Kammi families who could not secure a seph (or reported that they had left such a contract because of the exploitative arrangements) had taken up work beyond the village in a brick-making kiln (bhatta). Employment options were found to be very limited for such people, who lacked social connections and education. Remuneration offered by bhatta owners was very poor at Rupees (Rs) 300 ($3.75) for 1000 bricks; the quantity that a family of 2 adults and 4 children could produce in a day, equating to just Rs 50 per worker per day. This compares to a daily wage rate of Rs 300 for an adult man performing manual labor, such as digging or carrying bricks and Rs 350 for cooking at a wedding in the village. Because this was not enough to sustain basic needs, given that families also tend to be large, the workers borrowed money from the bhatta owners on an ongoing basis. Interviews revealed that these people struggled to keep track of their debt and depended on the owner’s accountant to maintain the record of what they owed. Most of these respondents expected to remain indebted for the duration of their lives and to pass on this debt to their children.

Dimensions of Low Caste Experience and Implications for Maternal Health

Although economic poverty was a pervasive dimension of Kammi life, it was not an isolated factor, but rather was found to be one aspect of a set of interlinked processes that marginalized these families both socially and economically. We identified 5 core dimensions of the Kammi experience in Ganji that influenced all aspects of their lives, including constraining Kammi women’s access to timely, good-quality maternal health care. We describe these characteristics in the following, highlighting their potential implications for interventions intended to enhance maternal health care use among similarly marginalized groups of women.


Our findings revealed how gender and low caste converge to make Kammi women invisible and silent—characteristics that are also expected of this group. It took our team nearly 6 months of living in the village before we identified all the Kammi women and garnered their trust. The higher caste villagers did not identify these respondents despite our requests to identify “poor people” because they did not perceive them as being poor, a point further discussed in the following. We had to constantly struggle to seek Kammi women’s perspectives and experiences during the research processes, which tended to become dominated by higher castes. The latter chided us for our interest in Kammi women and devalued our focus on them. Kammi women were so invisible that even a private nongovernmental organization working for women’s economic empowerment and human rights within the village overlooked them in its projects. The Kammi women also preferred their invisibility because it allowed them to avoid attracting attention to what is a stigmatized identity. This was evidenced in their initial reactions of fear and suspicion to our interest in them.

Maternal health care interventions that aim to reach the poorest women need to recognize that such women may not be readily identifiable and may not desire to be identified. Careful, discreet, and sensitive strategies are needed to reach them, as discussed in greater detail later.

Social Inferiority and Stigma

A key element of the Kammi identity was their inferior social status relative to the higher castes. This was symbolized in their caste label, “kam,” which translates as “less” and signals the widely held belief among the villagers that these people have a lower level of virtue or moral character (zameer) than members of higher castes. Higher caste people frequently used derogatory terms to refer to Kammis.

One consequence of the pervasive belief in Kammi caste inferiority is that Kammi women were poorly treated by government health care providers. We found that only 6 of the 26 Kammi women pregnant in the last 5 years had sought skilled birth attendants compared with nearly all higher caste women, both rich and poor. Kammi women expressed a deep mistrust of government services and cited many examples of abusive treatment (although most recognized the potential benefits of biomedical care, particularly in the case of pregnancy or delivery complications):

I’d rather die at home than go to Sukhru (the rural health center). They killed my aunt. (Kammi woman, 28 years old, in labor but refusing to go to a health facility. Her aunt had died in childbirth in the same facility 3 years earlier).

This is how it happens, Shimraz’s son was born in Sukhru (the rural health center). Ask her what happened with her; how they dealt with her. They made her suffer. Like slaughtering a goat. This is how it happened. It is good for us poor people to use government hospitals, but government hospitals are butchers. (Kammi woman, 32 years old).

Our field observations confirmed the differentially poor treatment that Kammi women commonly received from government health services. We observed how the higher caste wealthy Chaudhry and Raja women would jump the queue in the local RHC, demand care, and be welcomed by the staff, whereas Kammi women not only had to wait for long periods of time, they were ignored and often sent home without being seen. The Kammi women referred to the better treatment the higher caste women received as their “ponch,” which translates into special social access. In addition, we found that Kammi women who worked in brick-making kilns on the village outskirts were systematically excluded from the services provided by the Lady Health Worker program (see the box below).

Childbirth for the majority of Kammi women was home-based and attended by a relative who had previous delivery experience, in some cases identified as a “dai” (traditional birth attendant). In the case of complications, Kammi families tended to first seek care from unskilled birth attendants. Although motivated by their limited financial resources, the Kammis’ preference for private, unskilled birth attendants was also clearly linked to their experience of this care, which contrasted very positively with that offered in the public system. Respondents valued the way in which these providers communicated and treated them with care and respect.

Kammi Exclusion From the Lady Health Worker (LHW) Program
The Lady Health Worker program is a national program consisting of more than 105 000 LHWs who are commissioned to provide doorstep maternal and child health services that include the provision of iron supplements and folic acid during pregnancy and accompanying women to health facilities for antenatal and childbirth care. Program coverage is about 60%–70% of rural areas and urban slums of Pakistan.
The program is currently operationalized under the catchment area model, in which LHWs are assigned to households within an hour’s walk of the worker’s residence. The catchment areas are limited to the built areas of the village. Our field work revealed 5 brick-making kilns (bhattas) within 10–15 minutes walking distance from the village, but they were not part of the LHWs' catchment areas. LHWs also considered it demeaning to enter these brick kilns. Consequently, women and children in bhattas were excluded from these formal health care services.

Half the problem is solved by her interpersonal treatment. She treats everyone as though they are her own. (Kammi woman, 34 years old, talking about the care provided by a low-skilled, private midwife).

However, women’s narratives of their childbirth experiences and field observations by the primary investigator, a physician, suggested that these affordable unskilled birth attendants had a high threshold for tolerating risk. They delivered breech presentations without knowledge of possible feto-pelvic disproportion, and risked a trial of labor in women with previous C-sections and diagnoses of preeclampsia. They commonly delayed referring such women to a facility that could provide emergency obstetric services, putting these women at higher risk of maternal mortality. They were often also unwilling to transfer women to formal care because this might undermine their perceived competence and remuneration.

The technical upgrading of government facilities will not benefit the poorest women if they continue to avoid services because of poor provider treatment.

Dependence on Informal Entitlements

The employment options available to the previously described Kammi families meant that these people had very low incomes, were unable to accumulate cash reserves, and were often in debt to higher caste individuals. This long-term poverty deterred Kammi families from accessing skilled birth attendants and facility births because even the public sector services in practice involved under-the-table payments to service providers, transport costs, and the need to purchase medicines. Furthermore, we found that most Kammi families did not have any financial resources set aside to cover the costs of care in the event of birth complications. Instead, Kammi families faced with a maternal emergency tended to turn to higher caste, wealthy members of the village. Those Mussali families who had seph contracts sought money from their higher caste masters, considering this claim to be part of the payment due to them for services rendered. Other Kammi families took loans from higher caste individuals at high interest rates. However, in both scenarios, whether resources were forthcoming was highly uncertain. As a local physician commented: “When you refer these people to Holy Family hospital in the morning, they are still begging for money till evening.”

We found that the notion of saving for the potential complications of childbirth was universally recognized in the village. Respondents were also well aware of the treatment and transport options available should complications arise during delivery. However, whereas higher caste, better off families could manage to save significant sums, Kammi families were unable to do the same. Instead, these individuals adopted the strategy of investing in their social ties with higher caste individuals in the hope that these would then yield the necessary money in times of crisis. However, as the case studies illustrated, reliance on such informal entitlements increased vulnerability (see the box below).

Maternal Death Case Studies
Six maternal deaths had occurred in the previous 4 years, 5 of which were Kammi women, who made up just less than one fifth of childbearing women in the village (and among whom the birth rate for 15- to 19-year-old married or widowed women was 3.05 compared with 3.56 among higher caste women). We also identified 2 more maternal deaths among Kammi women in nearby brick-making kilns (bhattas) during the fieldwork period.
The detailed case studies that we recorded illustrated the way in which a combination of factors aligned to cause these women’s deaths. Prominent among these factors was delayed care-seeking for complications caused by difficulties in securing money and receipt of poor quality care.
Case study 1
Zahida developed antepartum hemorrhage because of an anterior placenta previa when she was 32 weeks pregnant. She had received free antenatal care from the local physician (because her aunt worked as her maid), who then referred her to a teaching hospital 4 hours’ drive away. Not having any cash reserves themselves, her Kammi family set about collecting the necessary funds from the higher caste family with whom they had a seph. However, because this family’s funeral expenses had recently been borne by the same people, there was reluctance to pay up. Zahida’s family therefore had to beg from other high caste villagers as well, and this led to a delay of more than 24 hours before Zahida arrived at hospital. She died of hemorrhage.
Case study 2
Shida, a mother of 3 lived and worked in a bhatta as a member of a indebted family, and therefore, bonded to the bhatta owner. After 2 days in labor in the care of traditional birth attendants, Shida’s family finally decided to take her to a hospital. Having no cash reserves, they borrowed money from the bhatta owner and took her to a private hospital where a C-section was performed because of a transverse lie. The baby was born dead. After 41 days postpartum, Shida complained of pain in her legs and deteriorated rapidly. The doctors from the same hospital referred her to a teaching hospital, but the family decided they could not afford to take any more loans. Both Shida’s mother and mother-in-law, in separate interviews, said “We decided to let her go.”

Interventions that aim to encourage birth preparedness and complication readiness must recognize that rather than lack of knowledge, low caste, poor families are severely constrained in their ability to accumulate cash reserves and their entitlements are uncertain and informal.

Exclusion From Formal Transfers

In addition to lack of independent income earning opportunities, Kammi families were found to be systematically excluded from formal transfers that are intended to address economic insecurity. The clearest example is the national Benazir Income Support Program (BISP), which aims to make cash transfers to the poorest and is a general benefit not tied to health service use. Our fieldwork revealed that despite formal criteria for the dispersal of funds, in practice, BISP resource allocation was decided by the local political leader, who was also the largest landlord. We found that BISP beneficiaries were predominantly individuals with family connections to the village elite. Only 4 of the 27 BISP recipients in the village were Kammi women, and only 14 could be categorized as poor. The Kammis who were resident in the bhattas were completely unaware of the BISP.

Only people with connections get BISP money. (Kammi woman, 39 years old).

The government thinks about the people, just not about poor people. (Kammi woman, 34 years old).

Interventions that aim to transfer cash to the poorest women in support of better access to maternal health care will need to adopt sophisticated approaches to ensure that resources actually reach those in greatest need.

Denial of Deprivation

A further important element of the Kammis’ marginalized position was the pervasive denial of their poverty and vulnerability by higher caste villagers. Higher caste respondents consistently challenged any suggestion that Kammi families were poor, arguing that, although it was true that Kammis did not own land or have jobs in the formal sector, they worked for higher caste families who are morally accountable for ensuring justice in payments.

How can that Kammi be poor?! He belongs to Chaudhry Big Landlord. It’s his job to look after him and he is accountable to Allah. (Raja man, 45 years old).

Kammi respondents were also found to emphasize their dependency on their masters, failing to demarcate the boundaries between their own interests and those of these higher caste families. However, while the higher caste respondents denied that the current structures acted to maintain long-term intergenerational poverty, some Kammi respondents did express awareness of their marginalization and exploitation, stating that they are kept suppressed (thalle hi) by the higher castes. Nevertheless, we witnessed no overt challenge to the social order.

Interventions that are intended to reach the poorest, most marginalized women must recognize that the social structures that perpetuate their situation are deeply embedded, supported by an ideology that legitimizes their marginalization, and advantageous to other sections of society. Improving the lives of such women necessarily involves disturbing the status quo and will likely be contested by more powerful groups.

Our study aimed to move beyond descriptive analyses of the correlates of maternal health service use to provide insight into the underlying causes of disparities that persist despite formal policy commitments and significant investment. The findings suggest that demand-side interventions aimed at shifting attitudes toward the value of skilled maternal health care and reducing financial obstacles to accessing such services, although not misplaced, will have limited impact until they are informed by a better understanding of the social and economic realities of poor women’s lives.

More specifically, our findings revealed how the caste system systematically marginalized Kammi women both economically and socially. In the absence of formal claims and state social protection, Kammis have no option but to engage in highly asymmetrical relationships with higher caste individuals that preclude income-earning and their accumulation of cash assets and are characterized by uncertainty. Importantly, an ideology that constructs Kammis as inherently inferior also pervades the provision of statutory services, thereby legitimizing their poor treatment by public health care practitioners (a finding mirrored in other South Asian countries)23 and systematic exclusion from poverty-reduction schemes. This combination of embedded structural inequality and exclusionary processes means that Kammi families typically neither can afford to pay for private maternal health care nor are willing to risk using the public sector (which, as we previously highlighted, is rarely free of charge). Most Kammi families continued to opt for untrained birth attendants whom they felt they could trust, and when emergencies arose, struggled to mobilize resources promptly to secure the required care.

These findings are based on a large and very rich body of qualitative data. Nevertheless, because the study was conducted in just 1 village in Punjab, questions of transferability necessarily arise. However, while the specifics are likely to differ by province and region, there is evidence that the systematic social and economic marginalization of lower caste women and their families is a persistent feature across Pakistan.18 In Sind, for example, low caste Hindu women have been found to be particularly disadvantaged.24 Kabeer23 described similar obstacles to progress in other South Asian countries, referring to the idea of durable inequalities that are deeply embedded in the fabric of caste-based social systems. We therefore suggest that the general, theoretical claim we make—that much maternal health research and practice currently employs inadequate conceptualizations of poverty and inequality—will be transferable across many settings.

The study highlights issues that are particularly important for current maternal health care policy and practice in Pakistan. While a robust health system is essential, our findings illustrate that improvement in technical aspects of public sector health services alone does not benefit the very poor and socially marginalized. Policymakers must recognize that health care organizations are not mechanical structures that provide services, but are culturally embedded.25 There is an urgent need for Pakistan’s current health policy to engage with the reality of entrenched caste relations. Our findings concur with those of social development commentators who have described caste in Pakistan as the “elephant in the room.”26 More specifically, our findings suggest a number of ways in which demand-side interventions need to be strengthened to effectively address barriers to timely, high-quality maternal health care for lower caste women (see the box on the previous page). These suggestions and the research findings have been shared with Pakistani parliamentarians, health policymakers, the Maternal, Neonatal and Child Health program managers, donors (Department for International Development, Canadian International Development Agency, Aga Khan Foundation) and the United Nations bodies Population Fund and the World Health Organization through both face-to-face and written briefings. The research team members are engaged in an ongoing dialogue with these key actors.

Principles for Interventions Aimed at Addressing Inequalities in Maternal Health Care in Pakistan
1. Explicitly acknowledge the special needs of lower caste women, for whom the interplay of gender and socioeconomic hierarchies produces extreme disadvantage. This acknowledgment should be incorporated in policy documents. In-service training to educate and sensitize policymakers, program managers, and frontline health care providers can raise awareness of the nuances of the caste system and the ways in which it structurally excludes the lowest castes. The training should be incorporated as part of continuing professional development for health care professionals and become a requirement for renewing credentials.
2. Develop ways to identify the most marginalized, largely invisible, women without further stigmatizing them. This can be done by developing a set of effective and context-appropriate indicators of poverty and exclusion that go beyond the standard asset-based indicators currently used.
3. Ensure obstetric services are provided to this group of women free of cost. Given the financial circumstances that constrain savings and undermine “birth preparedness” for many lower caste families, specific steps need to be taken to remove financial barriers. This may necessitate cash transfers to guarantee access to skilled care.
4. Ensure that the design and delivery of cash transfer or voucher schemes circumvent local power structures that currently undermine their receipt by the poorest families. This can be done by ensuring (1) the truly poor and excluded are identified using sensitive indicators of social exclusion and poverty and (2) by using mobile cell phone technology to transfer funds directly to the recipients.
5. Challenge and support providers of public maternity services to provide good quality, supportive care to lower caste vulnerable women, so that upgraded government facilities benefit those least able to afford private care. This should include in-service training that focuses on reorienting provider attitudes and values to provide respectful maternity care. Social and financial barriers the poor face should be addressed by penalizing abusive and corrupt providers.
6. Recognize that the poorest families are commonly not living within coherent and supportive structures, so that interventions aimed at “mobilizing” or “empowering” communities may be contested or undermined by privileged individuals. There is a need for further research to explore ways in which “community empowerment” initiatives can act to include the most socially marginalized groups.
7. Monitor access to services by indicators of caste status as well as income and wealth. Targets for equity should be developed and monitored for performance. Meeting equity targets should be incorporated into a provider's evaluation of performance.
8. Acknowledge that the responsibilities of those charged with improving maternal health extend beyond the provision of services to ensuring their equitable uptake as well as to working with other sectors to address the wider determinants of poor maternal health.

Unfortunately, even significant current investments do not look promising for the previously identified issues. For instance, the ongoing training of 12 000 community midwives seems unlikely to increase access to skilled care among the very poor, first because they are to be deployed in the private sector, and second because their training includes no explicit attention to why and how they should direct good quality care to marginalized women.27

There is little doubt that even today women often remain largely invisible,12 yet the findings presented here illustrate how some women are even more invisible than others. There is now an urgent need to match technical improvements in maternal health care services with informed action to address the social and economic marginalization of Pakistan’s most vulnerable women.


The study was funded by the Canadian Institutes of Health Research. Z. Mumtaz is funded by Alberta Innovates-Health Solutions through its Alberta Heritage Foundation for Medical Research Population Health Investigator Awards.

These findings have been presented to policymakers, program managers, and donors in Islamabad. They have also been presented in four international conferences.

We would like to thank the people of village Ganji for their hospitality and time. Ayesha Farooq, Nighat Parveen, and Manzhar Bashir assisted with data collection.

Note. The authors have no relationships or activities that could appear to have influenced the submitted work.

Human Participant Protection

Ethics approval was obtained from the National Bioethics Committee, Pakistan and the University of Alberta, Human Ethics Research, Health Panel B.


1. National Institute of Population Studies (NIPS) (Pakistan). Pakistan Demographic Health Survey, 2006-2007. Calverton, MD: Macro International Inc; 2007. Google Scholar
2. Hogan MC, Foreman K, Naghavi M, et al. Maternal mortality for 181 countries, 1980-2008: systematic analysis of progress towards Millennium Development Goal 5. Lancet. 2010;375(9726):16091623. Crossref, MedlineGoogle Scholar
3. Koblinsky M, Mathews Z, Hussein J, et al. Going to scale with professional skilled care. Lancet. 2006;368(9544):13771386. Crossref, MedlineGoogle Scholar
4. United Nations, Task Force on Child Health and Maternal Health. Who’s got the power? Transforming health systems for women and children. UN Millennium Project. 2006. Available at: http://www.unmillenniumproject.org/reports/tf_health.htm. Accessed July 22, 2010. Google Scholar
5. Government of Pakistan. PCI National Maternal, Newborn and Child Health Program (MNCH). Islamabad, Pakistan: Federal Ministry of Health; 2006. Google Scholar
6. Mahmood A. Improving maternal and neonatal health: measuring the impact of the PAIMAN project in ten districts in Pakistan. Comparing baseline and endline Survey Findings (2005-2010). Population Council, Islamabad, Pakistan. Google Scholar
7. Mahmud G, Zaman F, Jafarey S, Khan RL, Sohail R, Fatima S. Achieving Millennium Development Goals 4 and 5 in Pakistan. BJOG. 2011;118(suppl 2):6977. Crossref, MedlineGoogle Scholar
8. Chowdhury ME, Ronsmans C, Killewo J, et al. Equity in use of home-based or facility-based skilled obstetric care in rural Bangladesh: an observational study. Lancet. 2006;367(9507):327332. Crossref, MedlineGoogle Scholar
9. Bhutta ZA, Soofi S, Cousens S, et al. Improvement of perinatal and newborn care in rural Pakistan through community-based strategies: a cluster-randomised effectiveness trial. Lancet. 2011;377(9763):403412. Crossref, MedlineGoogle Scholar
10. Agha S. Changes in the proportion of facility-based deliveries and related maternal health services among the poor in rural Jhang, Pakistan: results from a demand-side financing intervention. Int J Equity Health. 2011;10:57. Crossref, MedlineGoogle Scholar
11. Agha S, Carton T. Determinants of Facility Delivery in Rural Jhang Pakistan. Working Paper No. 5. Karachi, Pakistan: Greenstar Research Department; 2010. Google Scholar
12. Horton R. The continuing invisibility of women and children. Lancet. 2010;375(9730):19411943. Crossref, MedlineGoogle Scholar
13. Narayan D. Voices of the Poor: Can Anyone Hear Us? New York: Oxford University Press; 2004. Google Scholar
14. Dodd R, Munck L. Dying for Change: Poor People’s Experience of Health and Ill-Health. Geneva, Switzerland: World Health Organization; 2002. Google Scholar
15. Raj P, Raj A. Caste variations in reproductive health status of women: a study of three eastern states. Sociol Bull J Ind Soc Sci. 2004;53(3):326346. CrossrefGoogle Scholar
16. International Institute for Population Sciences (IIPS) and Macro International. 2007. National Family Health Survey (NFHS-3), 2005-06, India: Key Findings. Mumbai, India: IIPS. Google Scholar
17. Easterly W. The Political Economy of Growth Without Development: A Case Study of Pakistan. Cambridge, MA: Kennedy School of Government, Harvard University; 2001. Google Scholar
18. Mohmand SK, Gazdar H. Social Structures in Pakistan. Islamabad, Pakistan: Asian Development Bank; 2007. Google Scholar
19. Kabeer N, Mumtaz K, Sayeed A. Beyond risk management: vulnerability, social protection and citizenship in Pakistan. J Int Dev. 2010;22(1):119. CrossrefGoogle Scholar
20. Morse JM. Myth 19: qualitative inquiry is not systematic. Qual Health Res. 1999;9(5):573574. CrossrefGoogle Scholar
21. Thorne S, Kirkhan S, MacDonald J. Interpretive description: a non-categorical qualitative alternative for developing nursing knowledge. Res Nurs Health. 1997;20(2):169177. Crossref, MedlineGoogle Scholar
22. Thorne S, Kirkham SR, O’Flynn-Magee K. The analytic challenge in interpretive description. Int J Qual Methods. 2008;3(1):111. CrossrefGoogle Scholar
23. Kabeer N. Poverty, social exclusion and the MDGs: the challenge of ‘durable inequalities’ in the Asian context. IDS Bull. 2006;37(3):6478. CrossrefGoogle Scholar
24. Shah Z. Long behind schedule: a study on the plight of scheduled Hindus in Pakistan. Information on caste based discrimination in South Asia. Karachi, Pakistan Institute of Labour Education & Research. 2007. Available at: http://www.dalits.nl/pdf/LongBehindSchedule.pdf. Accessed January 17, 2012. Google Scholar
25. Freedman LP, Waldman R, de Pinho H, Chowdhury M, Rosenfield A. Transforming health systems to improve the lives of women and children. Lancet. 2005;365(9463):9971000. Crossref, MedlineGoogle Scholar
26. Aliani S. Caste in Pakistan: the elephant in the room. Available at: http://reddiarypk.wordpress.com/2009/08/25/caste-in-pakistan. Accessed January 12, 2012. Google Scholar
27. Technical Resource Facility and Government of Pakistan. Assessment of the quality of training of community midwives in Pakistan. 2010. Available at: http://www.trfpakistan.org/LinkClick.aspx?fileticket=jNucN7xoy_E%3D&tabid=2404. Accessed January 12, 2012. Google Scholar


No related items




Zubia Mumtaz, PhD, MPH, MBBS, Sarah Salway, PhD, MSc, Afshan Bhatti, MBA, MSc, Laura Shanner, PhD, Shakila Zaman, MD, PhD, Lory Laing, PhD, and George T. H. Ellison, PhD, DSc, MScZubia Mumtaz, Laura Shanner, and Lory Laing are with the School of Public Health, University of Alberta, Edmonton. Sarah Salway is with the University of Sheffield, Sheffield, United Kingdom. Afshan Bhatti is with the Real Medicine Foundation, Islamabad, Pakistan. Shakila Zaman is with the Lahore Medical and Dental College, Lahore, Pakistan. George T. H. Ellison is with the University of Johannesburg’s Centre for Anthropological Research, Johannesburg, South Africa. “Improving Maternal Health in Pakistan: Toward a Deeper Understanding of the Social Determinants of Poor Women’s Access to Maternal Health Services”, American Journal of Public Health 104, no. S1 (February 1, 2014): pp. S17-S24.


PMID: 24354817