“These are not good things for other people to know”: How rural Tanzanian women’s experiences of pregnancy loss and early neonatal death may impact survey data quality
Introduction
The day after being interviewed about her own stillbirth, Hawa found our project interviewer at the mud-walled home of her friend Zuwena in rural Tanzania. Whispering, Hawa explained that she hoped we might take her younger sister Imani to the district hospital, as it was too far to walk and eroded crevasses in the mud roads rendered them impassable by bicycle. “You have transport, and I trust you,” she said. “Please, can you help? I am afraid Imani will die.” Imani was nearly three months pregnant but no one but her husband knew until that morning, when she had suddenly begun hemorrhaging and experiencing abdominal pains. Her husband was away farming, so Imani had gone to her sister Hawa hoping that because Hawa had participated in our maternal health study, she might be able to convince us to help her. Imani lay bleeding on a straw mat on her home’s mud floor. Hawa wrapped khangas (printed cotton cloth) around Imani’s waist to absorb the blood and helped her into our truck. Concerned, our local study driver asked Imani what was wrong. Imani hesitated, then stammered BP (“blood pressure”), the local name for hypertension. Later, the driver asked why she was bleeding. Imani repeated that she had BP, which Hawa corroborated. Why would Imani confide in the interviewer yet hide the truth from our driver? (During follow-up interviews, we learned that Imani had suffered a ruptured fallopian tube from an ectopic pregnancy, her life saved by an emergency salpingectomy.)
Administration of household surveys is the methodological cornerstone of population health research, permitting measurement of events using statistical inference from a sample to a population. Standardized questionnaires offer a fairly reliable, quick, and comparable method for collecting data from many subjects (Stone & Campbell, 1984); however, the trustworthiness of survey findings may be compromised when measuring sensitive events. Concerned about data quality, demographers have critiqued survey design and administration (Stone & Campbell, 1984), and a few have considered methodological challenges in measuring sensitive events including induced abortion (Hammerslough, 1986, Udry et al., 1996). However, surveys generally do not treat pregnancy and pregnancy loss as sensitive events, or consider how perceptions about other sensitive events such as induced abortion could impact disclosure of pregnancy and pregnancy loss. Imani’s strategic disclosure and concealment of her pregnancy and complications in the opening vignette suggest otherwise, but demographic studies have largely ignored the sociocultural context of reproductive events, and few data exist on how accurately surveys capture these events (Erviti, Castro, & Collado, 2004).
Measuring stillbirths—late fetal deaths—is increasingly viewed as critical to reduce the world’s 6 million annual perinatal deaths (Lawn et al., 2005, Lawn et al., 2009, Thatte et al., 2009). Better neonatal death and stillbirth data are necessary to track mortality trends and to inform selection of appropriate interventions, as causes vary by time of death (Darmstadt et al., 2005, Lawn et al., 2009). Yet stillbirths often go uncounted: stillbirths are not routinely reported to the World Health Organization (WHO) and are rarely counted systematically in vital registration systems and population-based surveys (Lawn et al., 2005, Stanton et al., 2006). Current estimates of 3.2–3.3 million annual stillbirths globally are conservative, as stillbirths are thought to be underreported by at least 20–40% in both survey and vital registration stillbirth data used to model these estimates (Lawn et al., 2009, Stanton et al., 2006, WHO, 2006).
Stillbirth is a constructed concept of Western colloquial origin describing late-term fetal deaths; although a stillbirth universally indicates a baby born dead, multiple definitions of stillbirth exist, complicating measurement (Lawn et al., 2009). Distinctions between miscarriage and stillbirth use variable benchmarks in gestational age, birthweight, and body length. In high-income countries, stillbirth rates often include some proportion of early fetal deaths, which the WHO defines as fetuses ≥500 g at birth, or at least 22 completed weeks’ gestation (ICD-10, 1993). Because few countries collect complete or reliable data on early fetal deaths, for international comparability, the WHO recommends reporting only late fetal deaths—fetuses ≥1000 g at birth (if birthweight is available), or at least 28 weeks of gestation or 35 cm body length. If early fetal deaths are included in global estimates, assuming similar rates as in countries with available data, approximately 6 million stillbirths occur annually (Goldenberg, McClure, & Belizan, 2009).
Social and cultural factors also contribute to the hiddenness of perinatal losses (Lawn et al., 2009, WHO, 2005). Stillbirths and neonatal deaths at birth may be difficult to differentiate, particularly if no skilled birth attendant is present (Stanton et al., 2006), gestational age at death may be unknown, and vital status at birth may be assessed and/or documented inaccurately, whether unintentionally or deliberately to facilitate coping or obtain benefits reserved for mothers of a live-born infant (Macfarlane and Mugford, 2000, WHO, 2006).
Especially where vital registration coverage of stillbirth is poor, improved methodologies to measure pregnancy loss and early neonatal deaths are needed to bridge data gaps, especially at the community level (Espeut, 2002, Thatte et al., 2009). Studies to understand local concepts, meanings, and social consequences of pregnancy loss are prerequisite for the development of such methodologies, as context may influence stillbirth enumeration in mortality measurement exercises.
Little social science research has investigated the meaning and impact of pregnancy loss, particularly in low-income countries (Cecil, 1996, Erviti et al., 2004, Layne, 2003). Despite pregnancy and pregnancy loss not being regarded as sensitive events, with pregnancy losses even being called “non-events” in high-fertility, high mortality settings because they are not mourned publicly (Scheper-Hughes, 1992, Stewart and Stotland, 1993), sociocultural issues surrounding pregnancy loss likely contribute to underreporting.
Almost universally, pregnancy is concealed for some or all of gestation. In numerous African and Asian countries, pregnancy concealment shields women from social risks such as gossip, sorcery, and spirit possession, to which women are thought more vulnerable during pregnancy (Chapman, 2005, Savage, 1996, Stokes et al., 2008).
Women’s low social status in many low-resource settings limits their sexual negotiating power and access to education and health care, rendering them vulnerable to victimization, blame, and stigma, especially in reproductive matters. Although induced abortions are often financed and/or coerced by intimate partners or parents, women suffer stigmatizing consequences. This stigma creates a shroud of reproductive secrecy that may contribute to underreporting of unintentional loss. Where childlessness is stigmatized (Inhorn and van Balen, 2003, Kielmann, 1998, Miall, 1986), women may be blamed for poor reproductive outcomes (Savage, 1996, Stewart and Stotland, 1993). Risks and consequences of pregnancy loss are also related to women’s vulnerability to emotional, physical, and sexual intimate partner violence: a WHO study found that half of women in African sites reported experiencing intimate partner violence (Garcia-Moreno, Jansen, Ellsberg, Heise, & Watts, 2006). Studies using Demographic and Health Survey (DHS) data have found associations—likely through multiple causal pathways—between spousal violence and increased risk of single and recurrent pregnancy loss (Cameroon) and unwanted pregnancy, induced abortion, and miscarriage (Bangladesh) (Alio et al., 2009, Silverman et al., 2007). Risk of abuse may also increase because of a pregnancy loss. The most vulnerable women appear at greatest risk of pregnancy loss and associated stigma and abuse.
Data from rural India revealed severe underreporting of negative reproductive events in surveys, yet demonstrated that prospective pregnancy identification and community mobilization could improve ascertainment of these events. Compared to a retrospective 1-year pregnancy history survey at baseline, a neonatal health study in Uttar Pradesh documented significantly higher stillbirth (56 versus 27.2/1000 births) and neonatal death rates (84.2 versus 54.2/1000 live births) among the intervention group at endline using prospective pregnancy surveillance (Darmstadt, unpublished data), enhanced by strong community engagement, mobilization and empowerment efforts (Kumar et al., 2008).
These data and the social science literature suggest that sociocultural factors, as well as data collection methodologies, may impact disclosure and measurement of stillbirths in many low-resource settings. To explore and identify these factors, we conducted reproductive narratives in rural southern Tanzania with 50 women who had experienced recent reproductive loss. We describe how pregnancy losses are defined and socially managed. We then identify issues that complicate enumeration in population-based surveys, including differences between biomedical and local categories; abortion-associated stigma that inhibits disclosure of spontaneous losses; social control mechanisms that suppress discourse, grief, and mourning; and methodological considerations that affect disclosure of losses. These findings inform recommendations to improve stillbirth data quality using culturally appropriate measurement strategies.
Section snippets
Methods
We conducted 95 open-ended interviews with 50 women in 7 rural villages of Mtwara and Lindi Regions, Tanzania, who had suffered infertility (inability to conceive for ≥2 years prior to the study despite efforts to conceive; N = 10), miscarriage (spontaneous loss of a pregnancy before 28 completed weeks of gestation; N = 10), stillbirth (a baby born dead at or after 28 completed weeks of gestation; N = 20), or early neonatal death (a baby born alive but who died during the first week of life; N
Results
The 50 respondents, broadly representative of the study area (Armstrong Schellenberg et al., 2008), were predominantly married, Muslim women of Makonde or Mwera ethnicity, with limited formal education (Table 1). Approximately one-quarter had polygamous marriages, and more than one-third had lost multiple pregnancies and/or children.
Discussion
These study findings demonstrate how qualitative research can inform quantitative methodologies to enumerate losses in childbearing (Table 4). The findings reveal that biomedical terms and categories used in surveys may not measure the intended outcome. Surveys designed to measure “miscarriage,” “stillbirth” and “neonatal death” reflect a reliance on constructed Western biomedical categories that do not map neatly onto local categories. That numerous respondents reported one biomedical outcome
Conclusions
The collection of consistent, internationally comparable data on stillbirth and early neonatal deaths is complicated by concealment of pregnancy and pregnancy loss, perceived social risks and mechanisms of social control that discourage discourse about loss, and incommensurability between local and biomedical categories of loss. Particularly in settings like rural Tanzania where home birth is common and “immature” losses are hidden, rates of stillbirths and early neonatal deaths may be much
Acknowledgments
We thank Dr. Hassan Mshinda, director of the Ifakara Health Institute (IHI), and the social science team and support staff of the Intermittent Preventive Treatment of Malaria in Pregnancy (IPTi) Project at IHI for their field support. Funding for the original research was provided to RA Haws through a J. William Fulbright Foundation Fellowship and endowed funds at Johns Hopkins, including the Wright and Bodian Scholarships. Funding for the writing of this manuscript was provided through an
References (53)
- et al.
Spousal violence and potentially preventable single and recurrent spontaneous fetal loss in an African setting: cross-sectional study
Lancet
(2009) - et al.
Evidence-based, cost-effective interventions: how many newborn babies can we save?
Lancet
(2005) - et al.
Prevalence of intimate partner violence: findings from the WHO multi-country study on women’s health and domestic violence
Lancet
(2006) - et al.
Commentary: reducing the world’s stillbirths
BMC Pregnancy and Childbirth
(2009) - et al.
Effect of community-based behaviour change management on neonatal mortality in Shivgarh, Uttar Pradesh, India: a cluster-randomised controlled trial
Lancet
(2008) The pregnancy that doesn’t stay: the practice and perception of abortion by Ekiti Yoruba women
Soc Sci Med
(1996)Nyo dua lili – 30 days confinement: traditions and changed childbearing beliefs and practices among Hmong women in Australia
Midwifery
(2000)- et al.
Adolescent girls, illegal abortions and "sugar-daddies" in Dar es Salaam: vulnerable victims and active social agents
Social Science & Medicine
(2001) - et al.
Stillbirth rates: delivering estimates in 190 countries
Lancet
(2006) - et al.
Health and survival of young children in southern Tanzania
BMC Public Health
(2008)
Investigating induced abortion in developing countries: methods and problems
Studies in Family Planning
Obstetric and perinatal events: the accuracy of maternal report
Clin Pediatr
The (Re) Negotiation of illness diagnoses and responsibility for child death in rural Mali
Medical Anthropology Quarterly
The anthropology of pregnancy loss: Comparative studies in miscarriage, stillbirth, and neonatal death
Embodying risk, defining danger in safe motherhood
Diagnostic accuracy of verbal autopsies in ascertaining the causes of stillbirths and neonatal deaths in rural Ghana
Paediatr Perinat Epidemiol
Strategies used by low-income Mexican women to deal with miscarriage and "spontaneous" abortion
Qualitative Health Research
A validation of birth and pregnancy histories in Matlab, Bangladesh
Do women forget their births? A study of maternity histories in a rural area of Senegal (Niakhar)
Popul Bull UN
Can fertility be estimated from current pregnancy data?
Pop Stud
Correcting survey-based contraceptive failure rates for abortion underreporting (dissertation)
Survey questions for the measurement of induced abortion
Studies in Family Planning
International statistical classification of diseases and related health problems
Infertility around the globe: New thinking on childlessness, gender, and reproductive technologies
Underreporting of induced and spontaneous abortion in the United States: an analysis of the 2002 national survey of family growth
Stud Fam Plann
Barren ground: contesting identities of infertile women in Pemba
Cited by (99)
“Athra,” a maternal illness in the rural Punjabi-Pakistan, as a source of stigma
2023, Asian Journal of Social ScienceExploring the impact of healthcare workers communication with women who have experienced stillbirth in Malawi, Tanzania and Zambia. A grounded theory study
2023, Women and BirthCitation Excerpt :Communication around stillbirth is influenced by cultural attitudes and norms, which determines how stillbirth is accepted and discussed in society [16]. In many countries, the death of a child before birth is associated with curses and spirits, thus its disclosure its kept in the private sphere to avoid further harm and embarrassment for the woman [17]. Lack of public acknowledgment of these stillborn babies means parenthood is often not recognised, and grief is likely to be disenfranchised, resulting in stigma, isolation, and more suffering for the couple [15].
(In)visibilising pregnancy loss in Southern Malawi
2024, Culture, Health and SexualityDeterminants of stillbirths in sub-Saharan Africa: A systematic review
2024, BJOG: An International Journal of Obstetrics and Gynaecology