Background
The Wixárika ethnic group of Northwestern Mexico is accustomed to high rates of maternal and infant mortality, and until the 1970s they had no vehicle access, schools or medical care of any kind. Although today their most accessible villages can be reached within 2–5 h by unpaved track, approximately half of their population of approximately 35,000 continue to live a further six to 8 h away on foot [
1‐
3]. As we demonstrate and discuss below, distance from medical facilities is only one of the structural factors that continue to lead to high rates of perinatal mortality in their communities.
Maternal and infant mortality rates are unacceptably high in Mexico’s indigenous regions compared to non-indigenous, with the largest proportion of mortality now occurring in the perinatal period [
4]. This reflects a global trend, neonatal mortality accounted for 42% of under-five mortality in 2013, an increase from 37% in 1990 [
5]. Using the most recent comparative data, the Mexican Infant Mortality Rate (IMR) was 14 per thousand live births in the non-indigenous population, yet 23 in indigenous populations [
6]. Although the national under-five mortality rate (U5MR) is now only slightly above the fourth Millennium Development Goal target of 15 per thousand [
7], Mexican indigenous people fare far worse. The U5MR was 76 [
8] in the indigenous municipality of Mezquitic where this study was conducted (there were no municipal data available on perinatal, neonatal or infant mortality), reflecting the global tendency for worse health outcomes among indigenous people in contrast to non-indigenous people within the same nations [
9‐
11].
Lack of skilled birth attendance is one of the major causes of infant mortality in indigenous regions, with only 76% of indigenous compared to 94% of non-indigenous women delivering in health facilities [
12]. Although 93% of births in Mexico were assisted by a skilled birth attendant (SBA) in 2014, only 30% of births to women in the indigenous muncipality of Mezquitic were attended by a SBA and an estimated 55% were home births [
13]. Unlike some southern Mexican indigenous groups, it is known that the Wixárika do not have a culture of traditional midwifery, and that women who birth at home often do so alone [
3,
14]. This is also the case with the Rarámuri, who live further north in the same mountain range, and is likely due to the spatial organisation of living arrangements [
15]. Both ethnic groups traditionally live in extended family ranches at considerable distance from each other, and if not alone, women may be assisted during birth by a family member or mará’kame (shaman) [
15].
A recently published systematic review suggests that skilled birth attendance could reduce neonatal mortality by 25% [
16]. Although in theory Mexico now has universal health coverage through the
Seguro Popular (SP) health programme and enrolment in public insurance programmes increased from 55 million to almost 120 million between 2004 and 2011 [
17], there has not been a concomitant increase in service provision nor financial resources [
18] and there remain serious structural barriers to accessing maternal health services. In Mexico’s indigenous regions, these include anti-indigenous discrimination, quality of care, gender inequality, language and cultural barriers, poverty, mistreatment and abuse as well as physical distance and means of communication [
19‐
21].
Considerable ethnographic research has documented the role of
structural violence in maternal deaths among indigenous women, where the intersections of gender inequality, ethnic hierarchies and poverty compound the problem of access to health facilities [
22‐
25]. Structural violence is an umbrella term used by medical social scientists to denote the harm done by social, political and economic structures and institutions, or “social arrangements that put individuals and populations in harm’s way” ([
26], p1686). The concept of ‘everyday violence’, denotes the normalised or daily (‘everyday’) and therefore unnoticed occurrence of structural inequalities leading to increased morbidity and mortality – or structural violence. Throughout her work documenting the ‘violence of everyday life’ in in a Brazilian shantytown, Scheper-Hughes discusses how infant deaths, forms of everyday violence, were ‘generally thought of as dispensable, as hardly worth counting at all’ ([
27] p216). Here, we seek not to focus solely on individual actions or inactions to explain morbidity and mortality in a context of poverty, but rather to link experiences of sickness and death to the social, economic and political conditions that structure the risk of people in different social positions. While Mexico has seen a reduction in maternal and perinatal mortality over the past decade, this decline has been considerably weaker and has stagnated in indigenous regions [
28,
29]. In this paper, we take a structural violence approach to exploring this inequality by focussing on the ways in which maternal and perinatal mortality track along structural hierarchies, with correlated lack of skilled birth attendance leading to increased mortality risk for both mother and baby. We share in- depth interview and ethnographic observation data on births, deaths, and maternal experiences in clinics and hospitals. We analyse this ethnographic case data alongside contextual data in order to understand how structural violence operates in both interpersonal relationships and the structure of health services provision to put indigenous women and their babies at risk during childbirth and the perinatal period.
Methods
This study aimed to explore the structural determinants of maternal and perinatal health in a Wixárika community located in Northwestern Mexico. Our objective was to understand women’s experiences of care in clinics and hospitals and how these relate to decreased rates of institutional delivery.
Setting
Our study took place in the towns and villages that make up the indigenous governorship of Yuáwime (not real name), with a population of approximately 2500. Spread over three highland governorships, 77% of the municipality’s population of 18,084 are indigenous [
13]. About half of the population of Yuáwime live in one of two vehicle-accessible towns located at up to 5 h from the nearest hospital, the other half live in hamlets up to a further 6 h walking distance. Covering 767 km
2, the governorship is divided by a deep valley between the two towns, each of which has a rural health clinic (URM:
Unidad Medica Rural). Officially, all highland clinics under the same medical jurisdiction are staffed from the 11th to the 30th of each month by a general practitioner, a nurse, and a travelling doctor who visits villages on foot. In addition, a student doctor attends the clinic from the 21st of each month for 20 days, running it alone from days 1 to 11.
Design
Anthropological studies including gaining community trust, gathering birth stories and verbal autopsies are recognised as important strategies for research with indigenous populations in settings where epidemiological data are lacking, unreliable or very difficult to obtain [
30]. This broad ethnographic study included interviews with women while they were pregnant and after delivery as well as long-term observations of family and community dynamics. During pregnancy we asked about socio-demographics, gender equality, previous pregnancies, antenatal care, and delivery intentions. Postpartum interviews were semi-structured and used a checklist of questions about birth attendance, outcomes and experiences (Additional file
1). The interview guide included prompts for women to report on choices regarding delivery, treatment by health professionals, and institutional care. Observations were performed and field notes collected with pregnant women and the wider community. We report on the findings of these data according to the Consolidated Criteria for Reporting Qualitative studies, COREQ [
31].
Research team and reflexivity
The principal investigator, who conducted all observations, is a medical anthropologist with expertise in health inequalities. She had worked in Yuáwime since 2008. Eight female bilingual (Wixárika/Spanish) women were trained as interviewers using pilot questionnaires with non-study women, and the forward-back translation method was used to ensure shared understanding in Wixárika or Spanish. The interviewers were all residents of Yuáwime and worked part-time on the study over the course of 2 years. Postpartum interviews were conducted by the principal investigator with translation assistance.
Procedures
Interviews and observations were conducted as part of a wider ethnographic study of the structural determinants of maternal and child health in Wixárika communities. We used ethnographic methods to explore the structural origins of maternal morbidity and mortality by analysing interactions among people and between people and institutions. The study was granted ethical approval by University College London Ethics committee and the governorship’s General Assembly meeting. Interviewees consented verbally according to these approvals and all people’s names and place’s names have been changed for anonymity.
Participant selection
All women who were known to be pregnant and could be contacted were invited to participate in the study at community meetings. Interviewers then made visits to the homes of pregnant women, who had been identified by word of mouth, to extend a personal invitation. Sixty-seven women were interviewed at baseline and there were eight refusals, with most citing a reluctance to share personal information. We were interested in all women who were pregnant, but due to the practical difficulties of finding women from distant valley communities in their homes, we expect that this group are underrepresented in our sample.
Data collection
The pregnancy interview asked participants about antenatal care, birthing intentions, fertility history and household division of tasks. Interviewers were encouraged to use their local knowledge to conduct interviews when the husband or other male relatives were not present. Women were followed up in their homes after delivery with audio-recorded semi-structured interviews lasting between 15 min and 1 h. The 2-year process was supported by observational data recorded in field notes, and informal interviews with health providers, community elders and traditional healers for data triangulation. Interviews were transcribed directly into Spanish and findings were presented to women at later community meetings and a public engagement event.
Data analysis
Numerical data were extracted from baseline and in-depth interviews to generate descriptive data on parity, age and birthing intentions and pregnancy outcomes. A similar process of extracting numerical data from ethnographic research has been used by anthropologists Scheper-Hughes and Briggs to investigate perinatal deaths where accurate data was not previously available [
27,
32]. We conducted an inductive thematic analysis of all qualitative data on N-Vivo (version 8). Following the general practice of grounded theory [
33], data that was coded together included interview transcripts and observational field notes. We first identified general codes (themes) and then added sub-codes to each theme. Finally codes were collapsed into a limited number of themes for the purpose of writing. Data were coded separately in Spanish by one bilingual (Wixárika-Spanish) and one Spanish-speaking research assistant and checked for accuracy by a third scholar. Emergent themes included social and cultural issues related to maternal and child health, Wixárika traditions and customs, gender inequalities, and health systems issues. Within the wider ethnographic dataset, we focus here especially on pseudonymised information on health systems and interpersonal relationships. We focus on both community-wide data and on in-depth ethnographic case descriptions in order to link individual experiences with larger social structures.
Discussion
In this paper, we have explored indigenous women’s experiences of birth in a context characterised by profound historical structural inequalities, that are the consequence of their position as an ethnic minority group economically and socially marginalized within a relatively wealthy nation state (see for example Bonfil Batalla [
34] or Batolomé [
35] for historical accounts of the well-documented anti-indigenous discrimination and impoverishment in Mexico). Racial, economic and geographical marginalisation have intersected throughout history and, as our data show, today expose women and babies to structural violence in the form of preventable morbidity and mortality.
We acknowledge that the study is limited by our focus on depth over breadth. Ethnographic research about structural violence seeks to link broad structural inequalities to the local, micro-environment, including personal and group experiences of sickness, injury, and death. For this reason, we focus on a small regional population as opposed to a geographically broader and potentially more generalizable sample. A further limitation is that the geographical dispersion of the population is likely to have weighted our sample towards women living in more accessible locations who are more likely to have given birth in facilities, and, therefore, to underrepresent populations without skilled birth attendance. In addition, other social and economic factors including higher rates of seasonal migration among families in the most distant communities may exclude a higher proportion of marginalised families. These limitations suggest that our findings may underestimate the true proportion of women who give birth completely alone, and, therefore, also the extent of preventable deaths. This data vacuum warrants rigorous ethnographic and epidemiological studies with in-depth, implementation of verbal autopsy [
27,
30] to identify causes of death where it has not been possible to establish this clinically.
In spite of these limitations, this study echoes public health and social science literature documenting physical distance and limited connectivity, cultural and language barriers, and extreme poverty as challenges to healthcare access faced by a majority of Mexican indigenous communities that also impacts perinatal and maternal mortality [
36‐
38]. Analyses of the structural determinants of maternal mortality in the state of Chiapas highlight the role of racism, gender inequality and poverty in producing higher mortality among indigenous populations [
24,
29]. An ethnographic study with indigenous communities in the state of Oaxaca, which together with Chiapas and Guerrero shares the double vulnerability of widespread extreme poverty and a high proportion of indigeneity [
3], describes structural and gender violence, as well as mistreatment and abuse by healthcare providers operating as deterrents to service uptake [
21‐
24]. These studies demonstrate how structural violence operates at the intersection of poverty, gender, and racial inequality to influence access to healthcare as well as morbidity and mortality. Considerable research also highlights more generally the need to address social exclusion and health inequalities in Mexican indigenous regions [
12,
39,
40].
Our data extend structural analyses of gender, race, and poverty by exploring the ways in which the organization of service delivery adds an additional layer of structural violence. Our data show that the system of healthcare delivery may function to deter women from seeking skilled care during pregnancy and delivery with implications for perinatal mortality [
41].
An in-depth ethnographic study in Oaxaca state reached similar conclusions about the unavailability of medical staff during clinic hours in indigenous communities [
42].
The discrepancy between clinics suggests that higher quality of care is possible if there is continuity of clinic staff. Notwithstanding the limitation of our data, the numbers of women who delivered in clinic “A” (n = 4, 17%) where there was a high turnover of staff, compared to clinic “B” (n = 17, 44%) where there was greater confidence in the availability a doctor would appear to corroborate our ethnographic findings. In addition, women explained that the routine practices of unconsented episiotomy, supine delivery and insistence in removal of clothes were reasons for avoiding delivery in a health facility. These unnecessary interventions and practices, coupled with experiences of verbal abuse, the use of humiliating language and reprimanding of indigenous women were also cited as reasons for giving birth at home.
In addition to identifying how structural barriers to care lead to preventable mortality, our paper also elucidates gaps in care and interventions that could address some of these issues. Although not stated in these precise words by women, the lack of cultural competence and cultural humility on behalf of doctors and the absence of a system for dealing with obstetric emergencies that occur outside health facilities are evident failings in service provision. The intercultural medical practices proposed for work in Mexico indigenous communities [
43] would greatly improve clinician-patient dynamics here. Women were told to travel to the hospital to avoid potential complications although there is not system in place to support them logistically or financially so that they are able to achieve this. A strategy named ‘Madrinas Obstéricas’ (Obstetric Godmothers) was in place in other states with the purpose of accompanying women to distant hospitals, as both social and cultural mediation support. Such a service could help bridge the gap between the doctor’s limited understanding of the realities and demands of Wixarika lives, women’s expressed desperation at not knowing what to do in case of emergencies and some of the barriers to accessing care that are faced by women who are either monolingual or speak only very limited Spanish.
However, infrequent bus availability and lack of funds for the bus would make even this intervention fail for many women. Thus, not only culturally appropriate and supportive projects like Madrinas Obstéricas, but also structurally responsive interventions would be needed. In many areas of medicine, “structural competency” is being taken on in order to address the health effects of structural violence [
44,
45]. This approach within medical training consists of developing a core series of skills for working in setting of social and economic inequalities that affect both health outcomes and health care itself. Recognising the structures that shape clinical interactions, observing and imagining structural interventions, and re-articulating common frameworks for understanding health in structural terms are three of these skills that are of particular relevance to the perinatal care of indigenous women in Mexico, and could make an important contribution to improving clinical care.
Gender inequality impacts indigenous women’s care seeking decisions in Mexico, as has been demonstrated well [
22‐
24]. Our data suggest also that men may have an important role to play in ensuring women have access to care, and must ultimately play a role in future interventions aimed at addressing gender inequality. In the current social context, men are better positioned to procure financial resources for transportation and are often required in practice for important decisions about care. Gender inequality is a structural barrier to healthcare that will require multiple culturally-appropriate strategies, first to increase the social and economic power of women and, second, to encourage male relatives to support healthy birth. Short-, medium- and long-term strategies must be developed to address this structural inequality.
A recent study of access to antenatal care in six Mesoamerican (Mexico and Central America) countries found that many women in the poorest regions were not receiving adequate antenatal care, [
46] while a study of the Salud Mesoamérica initiative reported institutional delivery (with a skilled attendant) variation between 24% in Guatemala and 88% in Nicaragua [
47], suggesting that our findings may be relevant to a wide geographical region.
Conclusions
As a member of the Organisation for Economic Cooperation and Development (OECD) Mexico is not poor: it is rather very unequal. As it approaches universal health coverage, Mexico is in a position to improve care for marginalized groups in the short- to medium-term. A recent WHO press release describes investment in health employment as key to attaining the Sustainable Development Goals [
48]; our findings suggest that investing in local service provision could bring important improvements to perinatal survival. Accurate vital registration will also be needed if we are to know whether the health system is working effectively.
Structural factors determine how and where women decide to give birth, why they choose unassisted births at home over health facility births, and ultimately how likely they are to experience perinatal deaths. While common frameworks for understanding maternal health in medicine and public health indicate that women must be educated to seek prenatal and obstetric care in a timely fashion, our research indicates that many indigenous women in Latin America have good reason for avoiding medical care. Thus, medical and public health interventions to improve childbirth and maternal survival rates must not focus only on the knowledge, beliefs and behaviours of mothers, but perhaps more importantly on the organization of health care, the attitudes and practices of health professionals in relation to indigenous women, and societal gender equity. These are important, achievable targets for a middle-income country.