Bregje De Kok


Loss in childbearing (maternal mortality; induced and spontaneous abortions; perinatal mortality) is an important problem in sub-Saharan Africa (SSA), specifically, in Malawi. Regardless of sustained global initiatives, maternal mortality (MM) and pregnancy loss are unacceptably common. In Malawi, life-time risk of maternal death is 1 in 18 (UK: 1 in 8200). Delays in seeking and receiving care are important contributors to the burden. They are not fully explained by practical (e.g. costs, distance) and cognitive (lack of knowledge) barriers. We need more insight into how local rationales affect use and provision of care, in particular interpretations of responsibility, blame and women’s entitlement to care. For instance, in Malawi, some interpret obstructed labour as signalling a woman’s infidelity, who is kept at home until she confesses. Practitioners’ moral judgements matter too; they seem to underpin substandard maternal care in SSA. In various African contexts, practitioners have been found to verbally and physically abuse especially ‘deviant’ clients (those not attending antenatal care, teenage mothers).

Using an innovative analytic approach –discourse analysis (DA)- this project will document how communities and practitioners interpret loss of the (unborn) child and mother and care provided; how these interpretations assign responsibility, blame, and entitlement to care and how they may affect use and provision of care. Interpretations will be elicited in interviews and focus groups. Acknowledging that accounts are never neutral, DA will be used to examine statements’ functions (e.g. justifying actions, allocating blame). Observations elucidate how interpretations of blame and entitlement to care may affect provider- client interactions. This study is theoretically and methodologically innovative.  Going beyond the usual assessment of the content of accounts of loss and maternity care, it examines how such accounts produce moral judgements, evoke local interpretations of human rights and how judgements may affect maternity care.  It includes providers and men, whose views have been neglected. Moreover, use of DA elucidates methodological issues affecting the validity of maternal death audits, increasingly used in resource-poor settings to infer cause of maternal death from practitioners’ and relatives’ accounts. However, they neglect how interpersonal issues at stake (responsibility, blame) for all parties concerned colour these accounts.

The Research Idea

Addressing loss in childbearing (maternal and perinatal deaths; induced and spontaneous abortions) in sub-Saharan Africa (SSA) requires a social rather than techno-medical approach (Penn-Kekana et al. 2007). Delays in seeking and obtaining quality maternity care contribute greatly to the problem (Ensor & Cooper, 2004). Using an original analytic framework –discourse analysis- this project examines accounts of loss in the context of normative expectations and moral judgements which may delay access to quality care. In African settings, married people ought to bear children (Inhorn & van Balen, 2002). Husbands, relatives, and providers may hold women responsible for pregnancy outcomes (Behague et al., 2008; Cecil, 1996), but women often lack the power to seek care (Filippi et al., 2006). Access to maternal care is considered a human right (ICPD, 1994), but local interpretations of rights underpin people’s behaviours (Petchesky, 1994). This project’s innovative thesis is that we should examine how community members’ and providers’ accounts  of loss manage interpersonal concerns (e.g. maintaining a positive identity, minimizing responsibility) and evoke interpretations of women’s entitlement to care. This provides novel insights into interpersonal  barriers to care; these insights can inform development of social interventions which address such barriers and increase maternal health strategies’ effectiveness  by tailoring them to local, gendered realities.    

The Focus

Loss in childbearing in SSA is a significant social and public health problem, acknowledged by Millennium Development Goal 4 (child mortality) and 5 (maternal health). Maternal mortality ratios remain unacceptably high in SSA (Hogan et al., 2010), especially in Malawi where life-time risk of maternal death is one in 18 (WHO, 2005). For every maternal death, an estimated 30 women suffer significant reproductive morbidities, like fistula, often co-occurring with (future) pregnancy loss (Nhlema Simwaka et al., 2005).  Perinatal deaths  (stillbirths and deaths in the first week of life) are five times as common in developing than developed regions (WHO, 2007). In SSA, 92% of induced abortions are unsafe and they are a main cause of maternal mortality and morbidity in SSA (Sedgh et al., 2008). All these forms of loss have significant economic, social and psychological implications for surviving women and their families (de Kok et al., 2010).


Maternal, neonatal and child health (MNCH) policies focus on increasing access to institutional deliveries and emergency obstetric care (Campbell & Graham, 2006). Their success has been limited especially in terms of reducing maternal mortality  in SSA (Hogan et al., 2010). Attempts  to reduce delays in seeking and receiving quality care focus on supply-side determinants (human resources, supplies) and demand-side barriers, mainly of a practical (users fees, transport) and cognitive (knowledge of danger signs) nature (Ensor & Cooper, 2004). However, information and access to free, nearby care does not guarantee use of services (Chapman, 2003). Fewer studies have examined cultural barriers including preferences, attitudes and norms (see next section), and they have influenced  MNCH policy less  (Ensor & Cooper, 2004).

Much MNCH policy and research treats potential users of maternal health services as rational decision-makers who will seek care if given enough information and practical barriers are removed. On the supply side, providers are (implicitly) assumed to provide adequate care if properly trained, paid and resourced. However, health providers are robots nor angels (Blaauw et al., 2006). Women who encounter pregnancy complications and loss are embedded in morally saturated relationships with husbands, relatives and providers who have their own expectations, interests, and judgements concerning their reproduction (de Kok et al., 2010). Qualitative social science studies illuminate such matters, but MNCH initiatives have relied on quantitative, epidemiological evidence (Ensor & Cooper, 2004).

Theory & Evidence Base

Ethnographic evidence shows that in resource-poor settings, pregnancies and deliveries are not always considered as justifying hospitalisation and expenses (Jeffery& Jeffery, 2010; Thaddeus & Maine, 1994). Thus, actions taken following complications  appear related to ideas about a woman’s entitlement to care, which in turn will relate to notions of responsibility and blame. Perinatal loss breaches the normative requirement that married people bear children (de Kok et al., 2010). Women are indeed blamed and stigmatised for induced and spontaneous abortions and perinatal deaths by partners, relatives, community members and even practitioners (Haws et al., 2010; van der Sijpt, 2010). Such moral judgements can affect care sought. Women who have undergone an abortion, illegal in Malawi, may delay seeking help for complications, fearing others’ moral judgements (Koster, 2003). Further, in Malawi, obstructed labour is sometimes interpreted as a sign of a woman’s infidelity, and no professional help will be sought until she confesses (McCoy et al., 2004).

Judgements of responsibility, blame, and entitlement to care also seem to affect practitioners’ behaviour. Abortion providers  (Koster, 2003) and traditional birth attendants  (TBA) (Bisika, 2008) have been found to delay referral for complications for fear of being held accountable. Death of the baby or mother potentially threaten practitioners status of competent professional (Mander, 1994), making practitioners likely to manage their accountability in their interpretations of, and responses to, loss. Furthermore, substandard interpersonal care is common in SSA (Behague et al., 2008;  Jewkes et al., 1998; McCoy et al., 2004).  Some nurses neglect and abuse patients verbally and even physically. Malpractices, which may well affect service utilisation, appear triggered by lack of resources but also by moral judgements: ‘deviant’ patients (e.g. those not attending antenatal care, single or teenage mothers) appear common targets (op cit.).


This small scale study pilots new ways of examining barriers to quality maternal care in Malawi, by documenting:

      1. How community members and providers interpret loss of the (unborn) baby or mother.
      2. How community members and providers account for maternity care sought and provided (e.g. in terms of quality of care).
      3. How accounts of loss and care a) manage interpersonal concerns, such as responsibility and blame, and b) construct women’s entitlement to care.
      4. Whether and how interpersonal concerns (e.g. responsibility and blame) and interpretations of entitlement to care affect use and provision of maternal care.

Accounts will be elicited in:

  • 34 semi-structured interviews with women who experienced pregnancy loss (8); their husbands and relatives (8); husbands and relatives of women who died during pregnancy (8); TBAs and biomedical practitioners (10).
  • 6 focus groups (FGs) with adult community members, TBAs and biomedical practitioners

Fieldwork (10 weeks) takes place in one town and one village in southern Malawi. I speak Chichewa at intermediate level, but will use an interpreter in interviews with non-English speakers. An RA will conduct three FGs in Chichewa. Participants will be purposively sampled (according to e.g. age, socio-economic class) and recruited through hospitals, health surveillance assistants, and snowballing, facilitated by my extensive network in Malawi.

For the analysis,  I will use one form of Discourse Analysis (DA) (Potter, 2010) which identifies descriptions interpersonal effects (e.g. justifying actions, minimizing responsibility) and has been applied to translated and interpreter-mediated interviews before (d’Hondt, 2002; de Kok, 2009). DA is particularly appropriate because it analyzes ‘talk’ and accounts, prime mechanisms for managing normative expectations and moral judgements (Bergmann, 1998), which ethnographies show become relevant when pregnancy goes awry. DA acknowledges that descriptions are never neutral, but attend to actors’ agency and responsibility (Potter & Wetherell, 1987). DA provides the tools to understand how interpersonal concerns are managed in and colour accounts of loss. This requires more attention, also considering increased use of ‘verbal autopsies’ (WHO, 2007b), which assess incidence and causes of MM based on relatives’ accounts of loss.

In order to obtain insight into how interpretations may affect maternity care, the RA and I will observe provider-patient interactions in 2 hospitals and 2 TBA clinics.

If participants consent, FGs, interviews and provider-patient interactions will be recorded and transcribed verbatim and if necessary translated into English, with samples being verified by a second translator.

 Software like NVivo is not suitable for DA, but will be used for coding observations recorded in field-notes deductively (codes derived from research questions and literature, e.g. interpersonal aspects of care, blaming) and inductively (emerging from analysis).

 Considering the topic’s sensitivity, ethical concerns will be taken particularly seriously (e.g. by ensuring total anonymity, letting participants choose interview time and location) and continuously reviewed, following professional guidelines (http://www.britsoc.co.uk/equality/Statement+Ethical+Practice.htm).

Methodological and ethical matters arising will be discussed in an advisory team.


Substantive and methodological insights will be disseminated through two conference presentations; two papers to be submitted to high impact, interdisciplinary journals (e.g. Social Science and Medicine); two policy briefs for human rights advocates and MNCH policy-makers; dedicated project website highlighting key findings.


This project synthesizes insights from public health, medical sociology, anthropology and  psychology. Its deeply interdisciplinary approach is essential because loss in childbearing is a medical and social problem for which we need new solutions. These require new conceptualisations and investigations, and interdisciplinarity will facilitate the required thinking outside the box’. This study critiques assumptions underpinning MNCH approaches and brings a new mode of enquiry (DA) into an area dominated by quantitative public health approaches and, to a lesser extent, anthropology. This generates new insights into interpersonal barriers to quality maternal care, valuable for development of social interventions which address the real world tragedy of loss in childbearing.

As outlined in my proposal, factors contributing to loss in childbearing (maternal and perinatal deaths, miscarriages) include poor quality maternity care, under-utilisation of care and ill-functioning health systems (Lawn et al., 2011). This study seeks to illuminate health system functioning and the behaviours of users and providers of maternity care. Its objectives are to examine a) community members’ and providers’ interpretations of accountability and blame for pregnancy complications and loss and women’sentitlement to care and b) how these interpretations may affect provision and use of maternity care.

Achievements & Challenges

In 13 weeks of fieldwork, two Malawian researchers and I generated a substantial data-set: 58 semi-structured interviews and 15 focus groups with health providers; women with a loss experience; their relatives; relatives of women who died during pregnancy; community members. Pursuing our intended focus on induced abortions, a highly sensitive issue, proved too time-consuming. Ethnographic observations of maternity care were conducted in two rural and one semi-urban health centre and in two referral hospitals, mostly by myself due to limited availability of a qualified Malawian researcher. However, the Malawian researchers will contribute to interpretation of my detailed fieldnotes. I was able to interview 6 Traditional Birth Attendants but because they were banned from practicing, I could not observe their care.

All data have been transcribed in detail and translated, with a selection checked by myself and a professional translator. The large volume of data necessitated a thematic analysis prior to a more detailed discourse analysis, which is in progress. I presented initial findings at 5 public health and social science conferences/seminars (Global Women’s Health conference, Development Studies seminar, Uni. of Malawi; BSA Human Reproduction conference; Social Determinants of Health conference, Queen Margaret Uni.; ISRF). I plan to submit three papers to high impact interdisciplinary journals (e.g. Social Science & Medicine; Health Policy & Planning). The British Journal of Obstetrics & Gynaecology is interested in a reflection on the risk of providers neglecting the loss experience, based on my ISRF Bulletin essay.

  1. ‘Holding women accountable for losing their baby.’ This paper will discuss the symbolic (fines), behavioural (e.g. slapping) and verbal (e.g. scolding) mechanisms which providers, traditional authorities and relatives use to hold women accountable for ‘risky’ health behaviours (e.g. home birth), pregnancy complications and loss. Interpretations of accountability may inform providers’ and relatives’ disrespectful care, and this may affect women’s use of care. Rather than merely condemning providers’ behaviours, we should situate them in the wider policy context, which currently holds providers accountable for the number of institutional deliveries and maternal and perinatal deaths, but not quality of care.
  2. ‘Providers’ accountability for maternal and perinatal health and quality of care’. Discursive analysis illuminates how community-members implicitly address providers’ accountability for the quality of care and loss of a baby or mother. However, certain provider behaviours (e.g. not providing information) and power differentials limit users’ ability to hold them explicitly and officially accountable.
  3. ‘Inequalities and entitlements to care: Communities’ perspectives’ This paper discusses community members’ accounts of inequalities in the provision of care. Respondents often portray inequalities as logical by invoking obligations and entitlements associated with certain categories of people (e.g. the uneducated). For instance, they suggest that as a health professional and relative, one is obliged to prioritize giving care to one’s relatives, who are entitled to jump the queue. Thus, socially shared notions facilitate problematic institutional practices.

Finally, I am currently drafting a report for health practitioners and policy makers in Malawi, highlighting the psycho-social consequences of loss and quality issues in care.