A lost cause? Extending verbal autopsy to investigate biomedical and socio-cultural causes of maternal death in Burkina Faso and Indonesia
Introduction
When a woman dies in pregnancy or childbirth, her death is likely to have been preventable with prompt and good quality treatment (Ronsmans & Graham, 2006), she is likely to have been poor and living in a rural location (CSDH, 2008) and she is likely to have lived in a less developed country (WHO, UNICEF, UNFPA, & World Bank, 2007). Being poor and pregnant in a poor country seems to confer a cumulative disadvantage that is expressed with sharp severity during an obstetric emergency. In addition to the 536,000 maternal deaths thought to occur globally and annually (WHO et al., 2007), 15–20 million more women survive but suffer long-term disabilities (Ashford, 2002).
Maternal deaths are classified according to medical cause, or in the absence of accurate information, on the state of pregnancy as proxy for cause. Research has revealed that most deaths occur during or immediately after delivery (Ronsmans & Graham, 2006) from complications related to the physiological states of pregnancy, delivery and the postpartum period. These ‘direct’ causes of maternal mortality — haemorrhage, hypertensive disorders, sepsis, abortion and obstructed labour — are the leading biomedical causes of maternal death (Khan, Wojdyla, Say, Gulmezoglu, & Van Look, 2006).
Intrapartum care (delivered ideally in primary health centres) to treat the main causes of death at the time they occur is thought to be the most effective means to reduce avoidable maternal mortality and morbidity (The Lancet, 2006). Two inter-related interventions are established in international policy discourse in this regard These are: deliveries with skilled birth attendants (SBA) equipped with the necessary drugs, supplies and transport (Graham, Bell, & Bullough, 2001) to manage normal deliveries, identify, stabilise and refer complications (WHO, 2004); and secondly, a package of life-saving care delivered in hospitals (including manual removal of placenta, blood transfusion and Caesarean section) to treat the direct causes of maternal mortality, referred to collectively as emergency obstetric care (EmOC) (Paxton, Maine, Freedman, Fry, & Lobis, 2005). Ensuring the availability, quality and use of EmOC is a core part of the strategy, and a sophisticated monitoring system exists (WHO, UNFPA, UNICEF & AMDD 2009).
SBA and EmOC represent coherent, global approaches to maternal mortality reduction and have been applied and evaluated in many settings (AMDD, 2008, FCI, 2009; IMMPACT, 2010). Ensuring good quality delivery care for maternal mortality reduction is also enshrined in international health and development targets, most notably the Millennium Development Goals (MDGs) (UN, 2009a). And, more recently, has been reframed in terms of social justice and human rights (UN, 2009b).
Despite knowledge of the necessary care, no change has been detected in the global burden of maternal mortality since international measurement began in the 1980s (WHO and UNICEF, 1996, WHO et al., 2001, WHO et al., 2004; and WHO et al., 2007). Despite recent reductions via modelled estimates (Hogan et al., 2010), empirically measuring the extent of the problem is notoriously difficult. In lieu of robust civil registration, large household surveys are required to identify sufficient numbers of deaths. These typically incur high costs, technical difficulties, and are thought to underestimate the magnitude of the problem (Graham, Ahmed, Stanton, AbouZahr, & Campbell, 2008).
If maternal deaths are systematically underestimated, it follows that maternal health planning, informed by incomplete data, may fail to account for disadvantaged groups (those systematically missed by civil registration and surveys), and policy, planning and services may become, inadvertently, configured to maintain inequities, by data failing to capture, and services failing to respond to, the burden among the poorest.
This article describes a Verbal Autopsy (VA) survey. VA is a health surveillance technique used to quantify levels and causes of mortality in settings where vital registration is poor or absent and/or where people die outside hospitals (WHO, 2007a). Initially used to investigate child deaths (Bang et al., 1992) and latterly with a focus on maternal mortality (Bartlett et al., 2002), VA surveys involve a structured interview with final caregiver(s) on medical signs and symptoms of their relative prior to death. This information is interpreted (by physicians or through algorithms) to conclude probable biomedical cause(s). When this process is undertaken for a sufficient number of representative cases, cause-specific mortality profiles that reflect disease burdens in populations can be generated.
More recently, the scope of VA has broadened to capitalise on the perspectives of family (and community) members directly involved with the end of a woman’s life. In this sense, VA is being used regardless of place of death, where the quality of medical reporting is low, to investigate the perspectives of those directly involved in acute care situations, on causes of death and delays, as well as to ascertain burdens of disease (Allotey & Reidpath, 2001; Barnes-Josiah et al., 1998, Castro et al., 2000, Parkar et al., 2009).
Today, over 45 countries in Africa, Asia and Latin America use VA as an interim solution to lack of routine data, for monitoring and public health planning (Baiden et al., 2007, Fottrell and Byass, 2010). The widespread use of the technique has led to the international prioritisation of its standardisation, to facilitate cross-country comparisons and international monitoring (WHO, 2007a).
In this paper, VA is referred to as a ‘health surveillance technique’. We acknowledge that, from a sociological perspective, the term ‘surveillance’ is not neutral and has negative connotations of mistrust and control over individuals (Foucault, 1973). From a public health perspective, however, the term is more practical in nature, with suspicion and control directed towards infectious agents, diseases and disorders rather than towards human beings. We adopt the public health interpretation in this paper, but acknowledge its otherwise contentious nature.
By virtue of the deaths it investigates, VA provides an opportunity to gather information from, and thus configure services for, those whom life-saving obstetric care fails to reach. The study aims were to 1) extend the standard VA (interview and analytical) schedules to gather and present additional information on experiences of emergency care-seeking, from the perspectives of family members who experienced some or all of the relevant events; 2) identify biomedical and socio-cultural factors relevant to outcomes, and; 3) present an illustrative analysis that considers the utility of the extension for routine health planning and surveillance.
Burkina Faso is one of the poorest countries in the world (UNDP, 2008). 46% of the population live in poverty, and of those 92% live in rural areas (Nikiéma, Haddad, & Potvin, 2008). 77% of the total population live in rural areas (INSD, 2008) and income is predominately generated through subsistence farming. The country has high total fertility (6.2 children per woman) (INSD, 2008), low life expectancy (52 years) (Bodart, Servais, Mohamed, & Schmidt-Ehry, 2001) and low literacy (34% and 24% of men and women are literate respectively) (INSD, 2008). Maternal mortality estimates vary. Estimates indicate ‘high’ and ‘very high’ levels, between 484 (INSD & Macro, 2000) and 700 (WHO et al., 2007) maternal deaths per 100,000 live births, with the most recent census recording an national level of 307 (INSD, 2008), possibly indicative of an improvement. Otherwise, home deliveries are common, 61% of all deliveries occur in the home (69% in rural areas) and 44% are not attended by trained health providers (INSD & Macro, 2004).
In 1994, under policies of decentralisation, regional hospitals assumed financial and operational responsibility for public healthcare service organisation and delivery (Sommerfeld, Sanon, Kouyaté, & Sauerborn, 2002). Healthcare has since become almost entirely financed through fee-for-services (Ridde, 2003) and health insurers have been seen to proliferate (Sommerfeld et al., 2002).
The study reported here was undertaken in Ouargaye district in November 2007. Ouargaye has a remote and rural population of approximately 214,000, one district hospital and 19 health centres or CSPSs (Centres de Santé et de Promotion Sociale). The district has 2% literacy, 18% institutional delivery and 1% family planning (modern methods) (Hounton et al., 2008).
In Indonesia, 48% of the population live in rural areas (BPS & Macro, 2008) and 52% on less than US$2 per day (World Bank, 2006). Most employment is in agriculture, sales and services (BPS & Macro, 2008). Total fertility is low (2.6 children per woman) and literacy and life expectancy are high (91%/87% literacy and 68/72 years for males and females respectively (BPS & Macro, 2008)). It is impossible to say with precision what the level of maternal mortality is in Indonesia, the literature suggests a range from around 200 (BPS & Macro, 2008) to over 400 maternal deaths per 100,000 live births (WHO et al., 2007). Less than a third of women in the study districts deliver with professional assistance, and the majority of the remainder do so with traditional birth attendants (TBAs) at home (Achadi et al., 2007).
To address high levels of maternal mortality and home deliveries, the Indonesian village midwife programme, launched in 1989, aims to post a trained midwife in every rural village to attend deliveries and refer complications (Shankar et al., 2008). Despite two decades of the programme, insufficient numbers of midwives, limited practical training, low public salaries and the privatisation of public services have led to differential coverage, quality and outcomes (Ensor et al., 2008, Ronsmans et al., 2009).
Decentralisation of public services was introduced in Indonesia in 1999. A national social safety net and three health-insurance-for-the-poor schemes have been in operation since 1998 (Ensor et al., 2008, Kristiansen and Santoso, 2006).
The study was undertaken in January 2008, in the Serang and Pandeglang districts of Banten Province, on Java. The districts are rural and remote respectively and have a combined population of 3 million. They are served by two public and three private hospitals and 66 health centres (Achadi et al., 2007).
Section snippets
Case selection
Cases were selected from two previous surveys: in Burkina Faso, from a household census conducted in 2006 that recorded, and conducted VAs on, 385 pregnancy-related deaths that had occurred between 2002 and 2006 (Bell et al., 2008); and in Indonesia, from a village-based informant survey conducted in 2006 that identified, and conducted VAs on, 474 maternal deaths that occurred between 2005 and 2006 (Ronsmans et al., 2009). 70 deaths in Ouargaye and 104 from Serang and Pandeglang were selected.
Profile of women and respondents
In both settings, over 80% of women were aged between 20 and 39 years at death and virtually all were married. Education and literacy differed widely; 64% of the Indonesian cases had primary education and 93% were literate, whereas 97% of the Burkinabé cases had no education and 4% were literate. In both settings, approximately half the women had from one to three previous live deliveries. 31% and 48% died at home, 59% and 46% in a health centre or hospital and 10% and 6% en route to a facility
Implications for health planning and surveillance
Health planning based on the results presented in Table 2, Table 3 would be likely to invest resources for the detection and management of obstetric sepsis and haemorrhage, through the strengthening of clinical competencies, supplies and equipment, possibly coupled with monitoring for coverage, quality and equitable provision. Standardised reporting of cause is also congruent with the International Classification of Diseases, Revision 10 (ICD-10), an elaborate classification system of
Conclusion
Pregnant women who die outside health facilities without proper registration are likely to be poor, marginalised and excluded from access to healthcare. Without an understanding of their situations and needs, the problems they face cannot be identified nor addressed. In the acute situations that we investigated, families’ preferences for traditional medicine and therapies, combined with unaffordable emergency services of limited availability, led to the severe worsening of delivery
Acknowledgments
This work was undertaken as part of Immpact, funded by the Bill & Melinda Gates Foundation, the UK Department for International Development, the European Commission and USAID. Immpact is an international research programme which also provides technical assistance through its affiliate organisation, Ipact. The funders have no responsibility for the information provided or views expressed in this paper. The views expressed herein are solely those of the authors. The authors would like to
References (76)
- et al.
Establishing the causes of childhood mortality in Ghana: the ‘spirit child’
Social Science & Medicine
(2001) - et al.
The “three delays” as a framework for examining maternal mortality in Haiti
Social Science & Medicine
(1998) - et al.
Maternal mortality among Afghan refugees in Pakistan, 1999–2000
The Lancet
(2002) - et al.
Maternal mortality for 181 countries, 1980-2008: a systematic analysis of progress towards Millennium Development Goal 5
The Lancet
(2010) - et al.
WHO analysis of causes of maternal death: a systematic review
Lancet
(2006) - et al.
Surviving decentralisation?: impacts of regional autonomy on health service provision in Indonesia
Health Policy
(2006) Human rights, the millennium development goals, and the future of development cooperation
World Development
(2007)- et al.
Women bargaining to seek healthcare: norms, domestic practices and implications in rural Burkina Faso
World Development
(2008) - et al.
The evidence for emergency obstetric care
International Journal of Gynecology & Obstetrics
(2005) - et al.
Maternal mortality: who, when, where, and why
Lancet
(2006)