Introduction

Maternal health is assuming an increasingly prominent position on the stage of international health. As the World Bank’s recently published Reproductive Health Action Plan emphasises, there is renewed global consensus of the need to make significant progress on Millennium Development Goal 5: the reduction of maternal mortality and increased access to reproductive healthcare [1]. International initiatives such as the Partnership for Maternal, Newborn and Child Health, the White Ribbon Alliance for Safe Motherhood and the Partnership for Safe Motherhood and Newborn Health, aim to strengthen maternal newborn and child health (MNCH) efforts at global, regional and national levels in the context of equity, poverty reduction and human rights. Yet, maternal mortality remains a major challenge to health systems worldwide [2–4]. As Hogan et al. conclude, ‘the delivery of interventions to women when and where they need them ought to be a purposeful policy of all countries’ [5].

Ethiopia is one of the poorest and least developed countries in Africa. With a population of over 82 million [6] it ranks 174 of 187 countries listed in the UNDP Human Development Index [7]. Eighty four percent of its population live in rural areas [8] and 39% live below the poverty line of USD 1.25 per day [7]. In 2009, life expectancy at birth was 53 years for males and 56 for females [9] and the under five mortality was 104 per 1,000 live births (ibid.) although, encouragingly, data released by the Ethiopian Ministry of Health in 2011 places under five mortality at 88 per 1,000 live births [10].

Healthcare services in Ethiopia are biased towards urban areas [11]. There are two public sector hospital beds per 10,000 of the population and only 1,806 physicians working in the country [9]. The healthcare system is overstretched and under resourced, and services targeted at women are thought to have particularly low coverage [12]. Maternal morbidity and morality rates in Ethiopia remain amongst the highest in the world. Each year, approximately 22,000 women die in Ethiopia due to complications during pregnancy or childbirth [13] and only 10% of births are attended by a skilled attendant in a health facility, whilst the remainder occur at home [10].

The Ethiopian government is committed to achieving the Millennium Development Goals, particularly Goals 4 (reduction in child mortality) and 5 [14, 15]. They have increased the number of trained midwives and implemented initiatives such as the health extension programme (the training and deployment of 35,000 paid female community health workers to be based at village level) [10, 16, 17]. Despite such interventions, it is well documented that existing maternal newborn and child health (MNCH) services continue to be underutilised by mothers in developing countries [18, 19]. It is clear that demand and delivery does not always equate with service uptake.

In seeking to improve maternal health services and encourage uptake, it is imperative to understand factors that influence care-seeking behaviour in a given context. Several studies have explored maternal care-seeking behaviour in Ethiopia. A number review data collected in the Ethiopia Demographic Health Surveys [11, 20, 21] but the main body of work builds on the Safe Motherhood Survey conducted in 2004–2005. This was designed to explore community values and practices surrounding pregnancy, childbirth and the postpartum and neonatal period, the influences of those values and practices on health-seeking behaviour, and the barriers and enablers to seeking and utilising health services. The project led to the in-depth analysis of four regions, Tigray, Oromiya, Southern Nations Nationalities and Peoples and Amhara [22]. Building upon this (unpublished) research, our study sought to identify and assess factors that influence home versus health facility delivery in South Wollo, a largely rural zone in Amhara. We conducted a rapid ethnographic assessment of maternal care-seeking behaviour in the target area [23]. It addressed local beliefs and practices associated with pregnancy and childbirth and aimed to determine the socio-cultural factors contributing to the low uptake of maternal health services in South Wollo.

Our study highlights that whilst a large proportion of women are actively engaged with health services, most often their interaction does not extend to delivery, and the majority of women give birth at home. In the context of improving MNCH in Ethiopia, it is important to understand this aspect of care-seeking behaviour. Previous studies [11, 12, 22, 24] have cited logistical issues including distance to clinic, lack of transport and associated cost as barriers precluding access to health services. Whilst we recognise these as important factors in care-seeking behaviour, in this paper we address a more focused question: why do women who access health facilities and utilise MNCH services at other times, not deliver at health facilities?

Methods

The study was conducted in line with prevailing ethical principles to protect the rights and welfare of all participants. The study was supported by the South Wollo zonal health office that granted permission for the research to be conducted in their area.

Research Team

The research team consisted of an international medical anthropologist, the primary investigator, and an Ethiopian health professional, the research assistant. Fieldwork support and logistical planning was provided by the organisation that commissioned the study.

Study Site

Data collection was carried out in January 2011 in five woredas (districts) in South Wollo: Kelela, Genete, Sayint, Mekane-selam and Wogedi. These were chosen for pragmatic and logistical reasons of access. The research team visited the main health centre in each woreda, plus a health post in two kebeles (neighbourhoods, the smallest administrative unit in Ethiopia). Due to time constraints, only one kebele was visited in Wogedi.

Participants and Recruitment

The woreda and zonal health offices, main health centre and a health post in each kebele were contacted in advance. A Health Extension Worker (HEW) was appointed in each kebele to make prior contact with mothers to seek informal permission for the research team to visit. From this pool of potential interviewees, mothers were purposively selected for interview by the research team. Pregnant and recently delivered mothers (primiparous, muliparous and grand-multiparous who had given birth within the previous 7 months) were recruited as participants, with an equal distribution between home and health-facility deliveries to ensure the full spectrum of influences determining location of childbirth were captured.

Data Collection

A rapid literature review was undertaken at the start of the research period, focusing on maternal and child health in Ethiopia and maternal health-seeking behaviour in developing contexts. This highlighted key issues upon which the research team based their design of a semi-structured topic guide that included a broad spectrum of research questions and probes (see “Appendix”). Interviews had seven sections: background; general healthcare; maternal healthcare; pregnancy and ante-natal; childbirth and immediate newborn care; post-natal; general conclusions. Specific questions and probes were reviewed and refined during the research period in light of themes arising. Although the direction of each interview was determined by the interviewee and largely focused on issues they self-prioritised (rather than on what the research team may have presupposed to be important), the key topics were addressed in each interview and therefore allowed generalisation of themes across participants. Questions focused on the current, or most recent pregnancy, although mothers were encouraged to compare this to previous pregnancies.

The research questions were translated and back translated between English and Amharic. All interviews were conducted by the English-speaking primary investigator with the research assistant translating consecutively between English and Amharic. Each interview lasted for approximately 1 h. Audio recordings were not made. This helped foster a sense of trust and privacy and encouraged mothers to speak more candidly than may otherwise have been possible. The primary investigator and research assistant made extensive notes during each interview, and large sections of narrative were transcribed ad verbatim.

Interviews were conducted at the mother’s home or nearest health post, and were held in as much privacy as possible. At the start of each interview, it was made clear to the interviewee that their participation was optional and voluntary and would not affect any future referral or medical service required or received. As the majority of interviewees were illiterate, the study’s consent form was read and explained in detail. Informed consent was given by signature or thumbprint of all participating mothers.

Data Analysis

At the conclusion of each day, the research team compiled and transcribed their interview notes, including the sections of narrative that were translated and back translated. Preliminary analysis was conducted in-country throughout the research process. Using an inductive approach, initial findings were discussed throughout the fieldwork and at its conclusion in two round-table focus groups comprising the research team, project support staff and key stakeholders in Dessie and Addis Ababa.

The primary researcher was responsible for the complete thematic analysis of the interviews using grounded theory [25–28]. Dominant themes were identified through the systematic sorting of data, labelling ideas and phenomena as they appeared and reappeared. Coding and analysis was iterative and by hand. The emerging trends were analysed according to the research objective using the critical-interpretive approach of medical anthropology [29, 30].

Methodological Limitations

This study was carried out in a challenging research environment. It was set in a difficult to access area of rural Ethiopia, and was conducted with limited time, budget and manpower. Throughout, we sought to mitigate or minimise the impact of these constraints by employing a methodology carefully designed to be pragmatic and by deploying resources efficiently.

Inevitably, some limitations remained. We elected not to use audio recording to enable the interviewees to speak more openly. It was therefore not possible to produce full transcriptions with translation and back translation of interviews. Risks associated with misinterpretation are inherent in consecutive translation, but a number of strategies were used to improve accuracy. In translating between English and Amharic, the researchers planned translation and interpretation styles in advance and decided how to best capture colloquialisms, abstractions, idiomatic expressions and jargon. We used short units of speech and careful phraseology that was refined during the finalisation of the interview question framework. During the interviews, the research team validated sections of narrative that were transcribed ad verbatim and certain responses were reiterated to the interviewee for clarification and confirmation. Narrative sections were translated and back translated by the research assistant after the interview was concluded. These were later checked by a bi-lingual member of the project’s support team.

Transcriptions of the interviews were made from extensive shorthand notes taken throughout the interview by the primary investigator. The transcripts were cross referenced with the research assistant’s notes, and any areas of digression highlighted and discussed. That the research team had full visibility of the growing data and were able to query potential anomalies throughout the study, served to mitigate the risk of errors in the translation or transcription process.

The final coding and thematic analysis was conducted by the primary researcher. Most qualitative studies employ two or more researchers to code data in order to reduce bias, but this was not feasible in our study. We sought to limit the potential for bias by conducting ongoing preliminary analysis during the fieldwork and by sharing data and results with the project team and stakeholders for their review and input at several stages. Some view a single analyst as preferable in certain settings [31].

It is possible that interviewees expressed what they perceived to be appropriate or socially desired responses. This is a risk in most interview-based qualitative research, but was not seen to be a major limitation, however, as we conducted informal, private interviews that were not recorded, the interviewees did not know the research team, and the semi-structure interview format allowed questions to be asked in multiple ways and responses triangulated.

Although relatively small, the sample size resulted in saturation of findings. This acted to lessen the impact of convenience sampling. The results are likely representative of the population in South Wollo, but are not generalisable and cannot be extrapolated to a wider Ethiopian context.

Our results are broadly corroborated by other literature (as discussed) and the study identifies important determinants influencing maternal care-seeking behaviour and the location of childbirth in South Wollo. Further to this research, key themes may be developed to form the basis of a more rigorous study into operationalisable strategies to improve maternal health services in the area.

Results

Forty-six interviews with mothers were completed. This was the maximum number possible in the allotted 2-week fieldwork period and was sufficient to achieve saturation of thematic findings. Thirty mothers interviewed had recently delivered and 16 were pregnant (5 with their first child).

Health Facility Attendance

Forty-two mothers interviewed had attended a health facility at least once during their current or most recent pregnancy. Of the four mothers who had not attended a health facility, three had attended during a previous pregnancy but did not feel compelled to do so in their current or most recent pregnancy. Only one mother, who had recently delivered her seventh child, had never attended a health facility.

Location of Delivery

Of the 30 delivered cases, 14 had given birth at a health facility, 14 at home, 1 at a health post and 1 on the roadside as she journeyed to a health centre. Only 3 of the 14 mothers who had delivered at a health facility had planned to do so. The remaining 11 were compelled to attend due to long or complicated labour.

Of the 16 pregnant cases, 3 claimed not to have considered the location of birth or had no preference, 5 intended to deliver at home, and 8 were considering attending a health centre. In July 2011, a project assistant followed up all 16 cases that had given birth in the interim period. Seven mothers had given birth at home and 9 at a health facility, 6 through choice, and 3 due to emergency complications.

Five dominant themes influencing the location of delivery were identified.

Perceptions of a Normal Delivery

A ‘normal delivery’ was perceived to be short (around 4 h), easy and at home, surrounded by a large number of both male and female relatives and neighbours, some of whom would play a practical role in physically supporting the mother as she laboured. During home deliveries, most mothers called for a traditional birth attendant (ylimidawalaj). Usually unpaid, birth attendants were seen as a positive presence providing experience rather than skilled specialist assistance. Labouring mothers were able to move freely around the house, and all delivered in a kneeling position (memberkek). This was regarded as the normal and dignified position in which to give birth.

If a delivery was normal, it was not considered to warrant medical intervention. Normal delivery did not require preparations or advanced planning and there was a sense that it was a routine activity women must endure. Despite this, interviewees perceived childbirth to be dangerous and there was an underlying sense of concern attached to labour and confinement. Women conveyed practical advice during a pregnancy, but labour and childbirth were never discussed, so as not to frighten an expectant mother.

We never talk about labour, there is a habit that you don’t inform a pregnant woman so you don’t scare her. For me, I would have felt stressed if I had heard about labour, what it was, so it was better for me that nobody spoke of it.

Motivations Encouraging Health Facility Delivery

The key motivation to attending a health facility for delivery was the length and nature of labour. The majority of mothers did not consider giving birth at a health facility unless labour was protracted or complicated. The most often cited conditions that prompted seeking assistance were excessive bleeding, breech birth or retained placenta.

Generally, treatment received at health facilities was regarded as positive and there was confidence in medical services provided. Mothers spoke favourably of injections to quicken labour and stop heavy bleeding, and of nurses’ ability to ‘sew if there is a tear’. Among the majority of mother participants, even those who had given birth at home, delivering in the presence of trained health professionals was seen to be beneficial as it minimised risks known to be associated with home births.

If you are at home, nobody knows there is a problem and you might die, but if you follow what they [health professionals] say then your labour won’t take long and it will be clean.

If a health facility was in close proximity, mothers suggested they were more likely to consider using its services. In addition, a number of younger mothers preferred to give birth at a health facility to avoid the pressure of delivering in front of relatives who, they thought, may judge the progression of labour and their behaviour.

Deterrents Preventing Health Facility Deliveries

Several mothers explained it was their ‘custom’ or ‘habit’ to give birth at home. This ‘custom’ precluded them from attending health facilities during childbirth, although they accepted medical intervention at other times. A health facility was perceived as a place of illness, and as ‘normal’ labour was not conducive for ‘treatment’, attendance at a health facility was not considered. Even mothers who sought preventative health measures (such as contraception and vaccinations) found it difficult to commit to health facility delivery until labour was prolonged or complicated.

I always give birth at home because it is easy. If somebody is not ill, then why go to the health centre? I take vaccinations when I am not ill because it prevents against future illness for me and my child, and so I am happy with that. But I didn’t think to give birth at the health centre, I never considered going for delivery.

Planning to attend a health facility or making preparations to do so in advance were rare (although in subsistence living, having the capability and capacity to plan is unusual). Most often, the latest pregnancy was considered independently from previous deliveries. A mother who had given birth at a health facility once would not necessarily plan to deliver there again, and may only attend if the next labour was also problematic.

The main factors deterring mothers from health facility delivery were those that contrasted most markedly with home births. Health facilities did not permit relatives or neighbours to accompany a mother into the delivery unit. Consequently, many mothers perceived they were ‘alone’ during labour, despite the presence of attending health professionals. The position of delivery was also key. Rather than allowing mothers to deliver in a kneeling position, health facilities instructed them to lie down, often with their legs in stirrups. Such physical exposure was deemed by mothers to be highly problematic. They also discussed their dislike of internal physical examinations, and the fact that the delivery units were so bright.

They say that at the health centre you must come for any pain, and for delivery too it is the best place. But I don’t want to give birth there, we have no habit from before, and I am somehow afraid to go. It is hard to get there and nobody knows me, whilst at home people know me during labour. We have no experience of going to a health centre from this rural area. We don’t want to show our bodies to people who don’t know us.

Distance from home to health facility and lack of transport are known obstacles that individuals overcome to receive care during illness episodes, or for routine or preventative health measures. During labour, however, the same logistical factors may assume greater significance than at other times and deter mothers from attending health facilities. To access a facility, most mothers would walk, ride a horse or donkey, or be carried on a stretcher, necessitating social rather than economic costs. Being carried on a homemade stretcher by male relatives and neighbours did not, in itself, prevent attendance, but was heavily imbued with negative connotations. Mothers perceived this to be a public display that labour was not normal, that she was in difficulty and was being taken to a health facility as if she was ill. Many expressed a sense of shame at having to attend a health facility rather than giving birth normally at home.

After giving birth at the health centre I was carried home on a stretcher. I was not happy with that. It made me feel as though I was ill. They walked very slowly and it took around forty minutes to get home. There were lots of people with the stretcher, including my husband and his friends.

For many, the possibility of onwards referral and lack of immediate treatment was a general deterrent against health facility attendance. It was not uncommon for staff at health facilities to lack the skills and resources to manage the late presentation of severe obstetric complications for which onwards referral becomes necessary. This contributed to a negative image of health facilities, especially in cases where the mother or child did not survive. In contrast to many interviewees who spoke favourably of health facility deliveries, a number discussed their dissatisfaction with services received and their lack of confidence in health professionals dealing with their case. Several mothers who presented during early stage labour were sent away with instructions to return when they were in true labour, and then stayed at home for their delivery.

In terms of economic cost, only those associated with onward referral (from a local health facility to district hospital) were discussed as problematic. Transport from home to local health facility was usually free or incurred minimal cost, and whilst the majority of mothers did not know in advance that delivering at a health facility was free, the presumed expenditure was not found to deter their attendance.

Decision-Making Processes

In South Wollo, decision-making processes are dominated by men and the male household head is usually responsible for making the final decision. If a long or complicated labour warranted health facility attendance, the decision to take the labouring woman was made by her husband (or father) in consultation with relatives and neighbours present. Such collective decision-making ensured that those assembled were willing to help transport the woman to the health facility (often by making and carrying the stretcher).

In several cases, mothers reported family members ignoring their requests to take them to a health facility during labour. In the face of their relatives’ opposition, the women were unable to assert themselves and were not in a position to compel others to take action on their behalf.

My family and neighbours were with me when I was in labour. It was coming worse and worse, and I asked them many times to take me to the health centre, but they didn’t, they thought I might give birth at home if I waited. I had little choice, I had to listen to them and wait at home. I thought let me die.

Level of Knowledge and Health Education

Forty-two mothers in our sample had attended a health facility at least once during their current or most recent pregnancy, yet only 3 recently delivered mothers had planned to give birth there. There was a feeling amongst mothers who had delivered at home, that advice they had received from health professionals (such as preparing a clean sharp blade for cutting the umbilical cord) encouraged them to have home births, and even validated their actions. The majority of mothers concluded they had never been explicitly instructed to deliver at a health facility. In contrast, mothers who had received take-home educative materials reported feeling better prepared for childbirth and as a consequence, more likely to consider health facility delivery. A large proportion of interviewees called for more information, education and ‘sensitisation’ about ‘the process from conception to delivery’ to be made available through both formal and informal channels.

Discussion

Women in South Wollo acknowledge that MNCH and access to related services have improved over time. Perceptions of risk are changing and in the community, there is a feeling that health facilities now provide a safety net by offering assistance during prolonged or complicated labour. Cases where the birth goes smoothly and the women are satisfied with their care, build confidence about the quality of health services offered. Nigussie et al. [12] found that utilisation of safe delivery services was significantly higher among women who had previously experienced obstetric complications. But only turning to health facilities in emergencies, often as a last resort, is problematic and contributes to high levels of maternal and neonatal deaths. It is also reinforces the view that facility attendance is for direct medical intervention. Presenting after the event, for a check up after a long and difficult home labour, for example, is rarely considered necessary if mother and child do not require immediate assistance.

The dominant perception of a ‘normal’ birth precludes many mothers from engaging with health facilities for routine deliveries. A paradoxical situation has developed in which increasing numbers of women who have the ability to access health facilities and utilise other MNCH services, still do not give birth there. Nigussie et al. [12] suggest that prenatal visits to a health facility are a strong predictor of safe delivery service utilisation. This was not reflected in our study. The situation in South Wollo underlines that having services in place does not automatically result in mothers who are able to utilise them doing so. Mothers, as health service consumers, are selecting specific MNCH services rather than complying with the whole continuum-of-care package.

For an MNCH intervention to be successful, to be accepted and utilised by the population it is intended to serve, the community must regard it as necessary, appropriate and relevant. For some, these attributes are not apparent in health facility attendance for routine delivery. Yet, the perception of a normal birth is slowly changing and many mothers speak positively about the benefits of health facility delivery, even if they have not experienced it themselves. Individual care-seeking is not always predictable and despite apparent choices, a mother’s actions (or lack of action) is often pragmatic rather than considered, for as Woldemicael et al. [11] contend, women with lower autonomy are less likely to seek, or able to seek, healthcare during delivery. In Ethiopia, issues of maternal health must be interpreted against a backdrop of gender inequality. One aspect of empowering women within the wider community is giving mothers the right and ability to determine their own health-related actions, particularly concerning pregnancy and childbirth [32–34].

The health system and its services need to be structured in a way that encourages mothers’ positive engagement. This involves managing levels of expectation (on the side of mothers and healthcare professionals alike). As Warren [35] concludes, ‘the location of childbirth involves a balance between retaining control of the process and outcome, and securing a safe delivery’. Because expectant mothers receive scant information from their community beyond the practical advice given during pregnancy, ante- and post-natal education provided by trained professionals at health centre or health post level is the only source of health education available to many mothers. However, the opportunity to provide and reinforce positive health education during ante-natal visits is not currently being maximised by health facilities.

Our study emphasises that it is unrealistic to expect mothers to comply with health facility deliveries, if basic features of the service (such as the position of delivery and the barring of relatives) are not socially acceptable. Altering the environment of health facility delivery units to better facilitate the needs of mothers is one example of a strategic measure likely to encourage attendance. This could include maximising opportunities to familiarise women with facilities and services during the ante-natal period; permitting relations to accompany the labouring woman into the delivery unit; and allowing the woman to choose her delivery position. The importance of environment and personal interaction in the context of health facility deliveries has also been emphasised in other studies [36].

Conclusion

In South Wollo, motivations to deliver at home are reinforced by deterrents to health facility delivery. Identifying and analysing the socio-cultural determinants that influence the location of delivery goes some way towards explaining why women who access health facilities and utilise MNCH services at other times, do not necessarily give birth at health facilities.

This has implications for service provision. To increase the utilisation of facilities during childbirth, it is not sufficient to only channel resources into the expansion of services and access. Key socio-cultural factors need to be harnessed as positive drivers and their impact as barriers minimised. There is great potential to improve MNCH in Ethiopia, but to achieve this it is crucial to focus on the context of service delivery. Socio-cultural determinants should be at the forefront of MNCH policy and programming, and actively incorporated in ways that encourage the utilisation of health facilities, even for routine deliveries.