8,633
Views
18
CrossRef citations to date
0
Altmetric
Original Article

Causes and contributing factors of maternal mortality in Bosaso District of Somalia. A retrospective study of 30 cases using a Verbal Autopsy approach

ORCID Icon, &
Article: 1672314 | Received 14 May 2019, Accepted 20 Sep 2019, Published online: 10 Oct 2019

ABSTRACT

Background: Somali women suffer from one of the highest maternal mortality rates in the world. Somalia characterises a specific low-income country situation with a mix of newly urbanized and nomadic culture combined with a frail health care infrastructure set in a post-conflict era. Very little is known about the effects that these contextual factors can have on maternal mortality.

Objectives: To explore and describe causes and contributing factors concerning maternal deaths in the Bosaso District, Puntland State of Somalia.

Methods: Data was collected using an adapted Verbal Autopsy tool. In 2017 30 cases of maternal deaths occurring in 2016 in the Bosaso District were reviewed. Information was assessed by three independent reviewers who classified the cause of death and the contributing factors. The Three Delay Model was employed to identify socio-cultural and economic and health system factors that may have contributed to these maternal deaths.

Results: Direct obstetric deaths accounted for 28 cases. Among these, haemorrhage was the leading cause, followed by eclampsia, sepsis and obstructed labour. Two cases were indirect obstetric deaths, caused by anaemia. All three types of delay were frequent among the studied cases. Delay in deciding to seek care was found in 25 cases, delay in reaching care in 22 cases and delay in receiving health care in 24 cases. Lack of knowledge, money, transportation, poor access and availability of adequate services, as well as substandard management by health care providers, were all underlying the delays.

Conclusion: A comprehensive intervention programme is needed in order to decrease maternal mortality among Somali women. Such a programme must include health education, improved referral systems and strategic upgrading of care services.

Responsible Editor Jennifer Stewart Williams, Umeå University, Sweden

Background

Globally, the majority (99%) of all maternal deaths take place in middle and low income countries [Citation1]. Of these, over half occur in Sub-Saharan Africa. The vast gap in maternal mortality between the high-income countries and middle- and low-income countries illustrates that a great majority of the maternal deaths that occur in low-income countries are preventable given timely and adequate intervention [Citation1]. The Millennium Development Goal 5, target set a three quarters reduction in maternal mortality for the period 1990 to 2015 [Citation2]. The global maternal mortality ratio was in fact reduced by almost half from 515 per 100,000 live births in 1990 to 303 in 2015 [Citation1]. In the 2014 annual global report by the Save the Children Fund, ranked Somalia at the very bottom out of 178 countries. It can be said therefore that Somalia is the worst place in the world to become a mother. The Maternal Mortality Rate in Somalia in 2016 was 732 maternal deaths per 100,000 live births, which is among the highest in the world, albeit this fell from 1,044 in 2012, according to the latest estimates of Somalia Health Indicators Report [Citation3]. The Total Fertility Rate in Somalia according to a UNICEF report in 2016 is 6.6 per woman, this being the third highest rate in the world [Citation3]. Accordingly, the lifetime risk of maternal death in Somalia is among the world’s highest, 1 in 22, exceeded by only Chad and Sierra Leone [Citation3].

In Somalia, a vital registration system is lacking to review maternal deaths. As a result, adequate information on the extent and underlying factors of maternal deaths is not available for use by the government or other agencies in developing policies and effective interventions. Currently, the health facilities record and classify maternal deaths solely by medical causes. This may conceal important information for understanding the underlying causes and determinants of maternal deaths. For example, information is needed on the route and actions taken by the deceased woman prior to her arrival at the health facility, which will offer clues about possible physical, socio-cultural and economic barriers that impede access to appropriate care in a timely manner. Therefore, a detailed study of cases needs to be conducted following the ‘road to death concept’ (referring to the series of events and their underlying circumstances, occurring during the care seeking process, resulting in the death of the mother if not adequate). Furthermore, it may raise awareness among health professionals about the most important factors in the facilities and in the community, which if avoided, could prevent maternal deaths and may stimulate action to address those factors.

One of the main strategies to reduce maternal mortality is providing adequate medical care in obstetric emergencies in a timely manner [Citation4]. Thaddeus and Maine,s Three Delays Model (delays in the decision to seek care, in reaching care, and in receiving care) was developed more than 20 years ago for evaluating the circumstances surrounding access to adequate emergency obstetric care [Citation5]. A number of studies have used the model to identify barriers and potential points of intervention to access appropriate emergency obstetric care [Citation6].

This study uses the Verbal Autopsy (VA) technique, which is a widely used approach due to the substantial benefits it offers as a tool in countries such as Somalia, where there is no system of vital registration and the practice of the medical certification of death is non-existent [Citation7]. The VA methodology was introduced in the 1980s and has become a well-utilized and well-validated method in settings lacking other routines for determining cause of deaths [Citation8]. VA is now recommended by WHO as the method of choice in such situations and has published well-validated guidelines [Citation9]. Recently, the VA tool has been further developed into the InterVA-4 version, which provides an increased opportunity to obtain information about emergency obstetric care [Citation10].

The aim of this study is to identify and describe causes contributing to maternal deaths in Bosaso District, Puntland State of Somalia and to further explore the socio-cultural and economic and health service factors in connection with those deaths.

Methods

Study setting

The study was performed in Bosaso District, Administrative Divisions of Bari region with an estimated population of 719,512 according to population estimation survey in 2014 of Somalia. It is situated in the North-eastern part of the country on the coast of the Gulf of Aden. Bosaso city is the densest city in Puntland State of Somalia. The estimated total population uptake for Bosaso health facilities was 448,009 in the year 2017 according to Puntland Ministry of Health report. The city is home to the only public district hospital in the region which serves as a referral hospital for all of the four maternal health centres (MCH) that provide basic obstetric care. The hospital, also receives a considerable number of patients from all the other districts in the region.

Study design and subjects

The study was designed as a cross-sectional assessment of maternal deaths that could be retrieved using records at a defined set of health facilities during a defined period of time. The target population consisted of women who died while pregnant, during delivery, or up to 42 days after delivery between the time 1 March 2016 and 30 March 2017 and who resided within Bosaso District during their last pregnancy until death. The death must have occurred in a health facility (hospital, health centre, health posts), in the community or en route to a health facility.

This resulted in the identification of thirty cases of maternal mortality according to the World Health Organisation International Classification of Diseases version 10 (WHO ICD 10), from the health records in the hospital, three MCH and five traditional birth attendants (TBA).

Data collection

Data were collected in 2017 by a VA approach regarding 30 cases of maternal deaths during the previous year in Bosaso District. The study used data from records from the mentioned types of health facilities, in addition to VA interviews, which were conducted with the communities; Families and relatives (husband, sisters, mother in law) neighbours, TBAs, community health workers (CHW) and others who had knowledge of the particular cases were interviewed. A recall period between one to 30 weeks was applied, this being well within the recommended range of not more than five years [Citation11]. One main interviewer was used at all times to increase the reliability of the collected information. The ‘road to death’ concept (referring to the series of events and their underlying circumstances, occurring during the care seeking process, resulting in the death of the mother if not adequate) was followed in a quest to generate adequate information [Citation12].

A pre-validated Verbal Autopsy Questionnaire (VAQ) was used that contains sections of structured questions to probe for specific signs and symptoms and the background characteristics of the deceased. It also included a section for open ended questions to probe for relatives’ account of events around the woman’s illness and death and symptoms developed during pregnancy, labour, and within six weeks after delivery [Citation13].

A classification form developed by WHO guidelines for Maternal Death Review was used for our VA [Citation12]. This tool was used to review and assign the medical cause of death and the contributing factors. The form consists of two main aspects of maternal death: the first is used for the cause of death (either direct or indirect cause of maternal death) and the second aspect is a checklist, which looks at the contributing aspect of the death.

Classification by reviewers

All 30 cases were reviewed by three obstetricians using the standard classification form. Information on medical cause of death and the contributing factors was added. The reviews were based on the information collected through the VAQ and the deceased’s health facility records (case notes). The medical cause of death assigned to each case was accepted when two of the reviewers agreed upon a main cause of death and if the three reviewers did not agree then the cause of death was reported as ‘unknown’.

Data analysis

The R statistical packages was used for descriptive analysis of the quantitative data with regard to causes and contributing factors. Qualitative ínformation was analysed using Open Code software and was coded and allocated according to different categories in the ‘Three Phase Delay’ Model. Finally, the categories were put into a framework for the analysis.

Results

shows the socioeconomic and demographic characteristics of the women including pregnancy and birth outcome. The average age of the deceased women was 31.1 years (SD =  8.4). The youngest among the deceased was 15 years while the oldest was 50 year. Of the 30 cases, only seven (23.3%) had attended formal schooling of which three had completed secondary school. The majority of the women 23 (76.7%) had no education at all or had only attended basic Islamic school (Madrasah). In contrast, 23 of the husbands (76.7) had some formal education. All of the women were married and most of them were their husband’s only wife. More than half of the women (56.7%) were housewives. The husband’s work varied across a wide range of occupations.

Table 1. Sociodemographic characteristics and pregnancy details of 30 cases of maternal death in Bosaso District, Puntland State, Somalia.

Pregnancy and birth characteristics of the cases

One third of the women made no antenatal care visits. The twenty women who attended prenatal care had on average two antenatal care visits during their last pregnancy (range = 0–5, SD = 2). Fifteen of the women (50.0%) had seven or more pregnancies. The average number of pregnancies was 7.1 (range =  1–17, SD =  4.3).

Outcome of the pregnancy

The majority (18) of the mothers died in their home. Looking at the time of death, 26 (86.6%) died after delivery and more than half of those deaths occurred between 1 and 42 days after delivery. Nineteen (73.1%) of the women who died postpartum, delivered at home assisted by a TBA. There were three multiple pregnancies (twins) among the deliveries. However, none of them was diagnosed before delivery. Two of these mothers made more than one antenatal care visits. There were only 11 live births, four of which were from three twin pregnancies. Four women died before they had delivered while the child died as well as the mother in the remaining 18 cases.

Medical causes of direct and indirect obstetric deaths

presents direct and indirect deaths, according to the causes determined by the three reviewers. Twenty-eight (93.3%) of the cases were classified as direct obstetric deaths. Haemorrhage was the most common main cause of death in this group, accounting for (43.3%) deaths. In addition, a further five cases had haemorrhage as an underlying cause of death and if added to the 13 cases where haemorrhage was the main cause of death, this resulted in 18 cases (not in the table). In three cases, the main cause of death was sepsis and in another three cases sepsis was determined as the underlining cause of death. (not in the table)

Table 2. Causes of maternal death among 30 women in Bosaso District, Puntland State, Somalia.

Healthcare seeking process

All of the cases sought healthcare from different providers (). There were 16 cases (53.3%) for whom the first place to seek care was from a TBA. Only three cases sought care at only one provider. Reaching a medical facility did not always mean that a woman received the treatment necessary to save her life. In this study 14 of the cases visited as many as three healthcare providers. This indicates that 27 (90.0%) of the women presented to centres where they did not receive the required services. Ambulance service was not available for the majority of these women (92.6%). Hence, the woman and her relatives had to arrange their own means of transportation by hiring a vehicle to take them to the facilities to which they had been referred (This information is not included in the table).

Table 3. Sequence of health care seeking among 30 cases of maternal death in Bosaso District, Puntland State, Somalia.

Funds to purchase medical care

The average expenditure on medical care was 79.6 USD ranging between zero to 358 USD (This information is not in the table). Regardless of the blood being donated by relatives or outside people, this service costs patients 40 USD covering the price of the blood bag container and the testing of the blood group.

Contributing factors

shows frequency of contributing factors as per reviewer. Three independent reviewers identified the most frequent factors that caused maternal deaths. Delay in the decision making process and getting to see professional health staff instead of TBAs or CHWs or even pharmacy staff as well as, the lack of knowledge of treatment possibilities were the most common contributing factors.

Table 4. Ranking order of contributing factors among 30 cases of maternal death as per reviewer in Bosaso District, Puntland State, Somalia.

Table 5. Determined cause, reported symptoms and experienced delays among 30 verbally autopsied cases of maternal death in Bosaso District, Puntland State, Somalia.

Verbal autopsy and Three Delays Model

Finally, the Three Delays Phases Model was applied for the analysis of the information collected from the informants. A delayed decision to seek medical care was indicated for 25 of the cases. Twenty-two of the women had delays in reaching an appropriate obstetric care facility once the decision to seek care was made. Even when an appropriate obstetric care facility was reached, 24 out of the 30 cases had not received the care services they needed, according to the reviewers. Looking at the phase of delay, twelve of the 30 cases had all three delays; 21 cases experienced two phases of delays and two cases experienced only one type of delay. A summary of the characteristics of all of the delays experienced among the 30 cases is provided in .

Delay in deciding to seek health care (phase 1)

The main factor behind this type of delay, which was classified as a ‘patient factor’ identified by this study, was an underestimation of signs, symptoms, and severity of the problem. Lack of knowledge about danger signs and warnings during pregnancy and failure to seek prompt treatment from health facility were the causes of delay in deciding to seek health care. Family members such as mothers and sisters were often involved in the decision-making process. In Case 13, a sister narrated:

‘During pregnancy she complained about breathlessness at the eighth month of pregnancy, her body was swollen too. Therefore, she had to measure her blood pressure and it was found to be high. She had no medication for the diagnosed hypertension due to the difficulty in finding money for the medicines. However, we occasionally bought medicines from the local pharmacy when we could afford, and we got some medicine from MCH and also from another health centre. Eventually, she delivered a dead baby at home, and after delivery she lost a large amount of blood, I was constantly changing her bed. The bleeding subsided in the next six days. She was very weak and became pale, she had fever and headache, and her swelling increased. The family applied traditional remedies to treat her. Health providers visiting the village examined her and gave her injections and told us to take her hospital. Therefore, she told me to go to the market and sell her earrings. I sold them for $50 worth of shillings but soon after I arrived she died at home. It was on her ninth day after delivery’.

Delay in reaching obstetric care facility (phase 2)

Transportation difficulties and seeking care at more than two medical facilities, were contributing factors in failing to access quality emergency obstetric care. The families experienced difficulty in collecting enough money to go to another health facility to access care. One of the interviewed mothers narrated:

‘She felt that the foetus did not move for a day and she went to the local pharmacy, where she was advised to have an ultrasound. She visited the ultrasound clinic, but the clinic had closed at the time due to the daybreak at noon. She returned home and in the afternoon the labour started, The TBA was called for … . .It turned out to be a multiple pregnancy after one baby was delivered the next baby did not follow, so I walked in the heat of the sun to the MCH, but the staff told me to bring her to the clinic. … … When I came back home with a rented car she already had lost a lot of blood and died soon after’.

Delay in actually receiving health care after reaching the hospital (phase 3)

Inadequate health services was one of the factors in this study. This includes lack of blood as in Cases 5 and 9 and incompetence of the available staff as shown in Case 2: Placenta was retained after delivery in the health centre; however no manual removal of placenta was performed, and no timely referral was made at the health centre. The woman’s neighbour narrated:

‘After she experienced labour pain she decided to go to the health centre. Her daughter and I accompanied her to the centre two hours later when her contraction became more intense. A friend with a car helped us to take her there and her aunt was left home to babysit. After the nurse had examined her she told us that her cervix was dilated 4 cm. So she stayed to deliver there. Finally, she delivered a baby girl but the placenta was not coming out. Immediately she told the nurse that in her previous pregnancy the placenta was also retained first and then manually removed. However, this was not attempted. Consequently, she was bleeding heavily. The midwife gave her an injection to stop the bleeding. After two hours when the dawn broke she was transferred to the hospital. After 20 minutes in the hospital the doctors told us that she was dead’.

Other inadequate health services factors included missed diagnoses such as anaemia (Case 11), tumour (Case 16), and multiple pregnancies, which were not identified during their ANC visits. Another factor was mismanagement of intra and postpartum haemorrhage by a TBA delaying timely referral (Cases 18 and 19). Case 10 who made no antenatal care visits but instead visited the pharmacy after complications, where medicine was given but no referral was made. Case 11 made five ante-natal care visits but had to queue for seven consecutive days at the health centre because of a delay in receiving a registration for vouchers. Delay in accessing health services may have been experienced as a result of perceived poor quality of care by the families: some families testified about health staff’s mistreatment towards escorts in addition to limited privacy and lack of personal care at the delivery facility.

Some patient factors were also identified. For instance, refusal to accept care (Case 3, 14 and 20) because of fear of caesarean section. Furthermore, patient factors often contributed to discharge from a health facility without complete treatment. One woman (Case 20) went home from the hospital after 10 days because of financial constraints and another woman (Case 7) made the same decision, because of childcare responsibilities. However, in some cases, the health staff made the decision to discharge the woman without complete treatment (Case 6, 11 and 17) giving the women medicine to take home.

Discussion

In Somalia, the lifetime risk for a woman to die because of pregnancy is estimated to a staggering 5%, which makes maternal mortality a major contributor to the Burden of Disease among adult women in this country. Moreover, maternal deaths often have detrimental effects on family health and wellbeing, in a low-income country context where child-rearing responsibilities rest heavily on the woman in the household. This detailed VA-based study of 30 cases of maternal deaths in Bosaso District, which exhibited the typical direct and indirect obstetric causes of death, has broadened understanding of the socio-cultural and health systems-related factors behind maternal mortality.

The majority of the deaths were direct obstetric deaths due to haemorrhage, followed by eclampsia, obstructed labour and sepsis. This is consistent with findings of other studies including a WHO global estimate [Citation1, Citation6, Citation14Citation16]. One study from Mozambique and some other African countries identified sepsis as the second leading direct cause of death and HIV/AIDS was the most common indirect cause of death [Citation17].

The main finding in this study was that all three phases of delay were frequent, often in parallel.

Delay in decision to seek health care (first delay)

A combination of factors contributed to the first delay, which lead the pregnant women not to seek prompt treatment from a health facility with capacity to manage pregnancy and delivery-related complications. Some mothers did nothing in this situation, while others sought treatment from TBAs (as the first place to seek care which is in accordance with a study from Gambia [Citation18]) or Village Health Workers (VHWs), while other purchased medicine from the pharmacy or sought treatment by spiritual healers.

Obstetric emergency among the studied cases, occurred mostly during the intrapartum or immediately postpartum. In this study 19 of the women delivered at home increasing the risk of maternal mortality from delays.

Traditional birth attendants played a major role in this study as health service providers. Delay in referral from a TBA to a health facility with emergency obstetric care and skilled birth attendants can result in a high risk of maternal death [Citation19]. For example, the six cases in this study where sepsis was either direct or underlining cause of death, were all delivered at home assisted by a TBA. It has been shown in other studies that sepsis usually follows haemorrhage in maternal deaths [Citation12] and previous studies have shown that facility-based delivery can reduce the risk of maternal death [Citation15]. It has also been shown in a study of maternal near miss cases in Brazil, that delay in accessing emergency obstetric care increases the severity of the complications [Citation6].

Delay in reaching health care once decided (second delay)

Contributing factors for not reaching an appropriate health facility were often transport issues or financial constraints. Other factors were lack of doctors on duty, causing the women to move from one facility to the next in desperation for getting adequate service. Further delay in reaching care may have been caused by the perceived poor quality of care by the families. Some families in this study reported experiences of health personnel’s mistreatment towards escorts. Moreover, they mentioned limited privacy and lack of personal care at the delivery facility.

Delay in receiving prompt and adequate care once in health facility (third delay)

Once the woman had reached the hospital, in some cases she did not receive prompt and adequate treatment due to both patient and health worker factors. A previous systematic review of third delay studies [Citation20] has shown that women are at risk of dying even if they arrive in the health facility without any complications [Citation20Citation22]. Other studies reviewing mortality audits also found health service factors to be common contributing factors [Citation20Citation22] Our finding of inadequate health service factor is consistent with their results. A recent study performed in Bosaso health facilities assessing the essential newborn care supports the finding of this study of poor quality of obstetric care [Citation23].

In the cultural setting represented by Somalia, women are expected to take full responsibility for childcare duties and all of the domestic chores, regardless of other circumstances. In addition, many of them often have the obligation to help supporting the family through poorly paid jobs or by selling items in the market, even during pregnancy. Almost half of the women in the study worked for income. Moreover, almost all of their jobs were in the informal/labour sector. Hard work and responsibilities coupled with financial constraints during pregnancy might have been a reason for women overlooking their health status and not utilizing facility based obstetric services such as ultrasound.

Methodological considerations

A strength of this study was that the VAQ was piloted to identify issues that could improve validity. A further strength was that agreement regarding cause of death between the three reviewers was 100%. However, a limitation was the absence of information regarding women who survived their delivery experience, which might have given more clarity to the actual prevalence of risk factors of maternal emergencies. Another limitation was the small sample size because of resource constraints, meaning that the results cannot be generalized. The study should, therefore, be regarded as mainly exploratory.

We observed that family members or people in the community felt more comfortable about talking if the interviewer was accompanied by someone they knew. For that reason, the investigator was accompanied by a health staff person (nurse or midwife) from the nearby MCH, or by a TBA or a VHW. However following introductions the accompanying person left so that the interviewer could proceed with the interview. One of the challenges was that the tracing and retrieval of health facility patient records was at times difficult. Some records were missing, incomplete and lacked clarity regarding key information.

Conclusion

Haemorrhage, eclampsia, sepsis and obstructed labour were important causes of direct obstetric deaths. Anaemia was a risk factor behind the indirect obstetric deaths. Patient factors such as delaying the decision to seek care when obstetric complications arise and health service factors were identified as common contributors to maternal death.

There should be national policies or protocols put in place to track and review all maternal deaths. Furthermore, Health Demographic Surveillance Sites utilising VA techniques could be initiated to monitor and audit cases of maternal mortality. Institutional delivery is the key to reducing high maternal mortality and pregnant women need to access adequate delivery facilities as soon as labour starts.

Ethics and consent

East Africa University Institutional Review Board (EAUIRB) approved the study. The Director of Bosaso General Hospital, Directors of the three MCHs and the TBAs permitted access to the health facility patient records. Individuals or families approached for participation in the study were first fully briefed on the purpose of the study and as well as their roles. It was also explained to them in Somali the right of refusal or withdrawal from interviews. The participants then signed a consent form with their name in writing or by their thumbprint.

Paper context

There is a lack of systematic studies exploring the ‘road to death’ of maternal deaths in Somalia, a country without vital registration. This study analysed the background factors and determinants of maternal mortality using an adapted VA approach based on the ‘Three Delay Model’. Delays were often identified in the maternal deaths resulting from sociocultural, economical and health service factors obstructing access to adequate and prompt health care. This study adds important new knowledge about contextual factors behind maternal mortality in Somalia.

Acknowledgments

We would like to thank the communities, families, relatives and neighbours of the deceased women, TBAs, CHWs and others in Bosaso District, who provided us with knowledge of the cases in this study. We also want to thank Somali-Swedish Research Cooperation for Health (SSRCH) program for coordinating and funding the study and specially East Africa University, which contributed with funding to the collaboration.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was supported by SSRCH program in particular through the collaboration, East Africa University funded the study. Alliance for Health Policy and Systems Research provided additional funding for the preparation of the manuscript through the collaboration.

Notes on contributors

Jamila Ahmed Aden

JA conceptualized, drafted, and led the writing of the manuscript. PO, HJ provided overall conceptual guidance for the study and PO closely worked with JA to develop the article. All authors have contributed to the writing and reviewed and approved the final manuscript.

References

  • WHO, UNICEF, UNFPA. World Bank Group, and United Nations population division trends in maternal mortality: 1990 to 2015 Geneva: World Health Organization. Popul Dev Rev. 2016;42:726.
  • United Nations. Dept. of General Assembly Affairs Conference Services, Services C. United Nations correspondence manual: a guide to the drafting, processing, and dispatch of official United Nations communications. Geneva: United Nations Publications; 2000.
  • United Nations Children’s Fund. Women and children in Somalia: a situation analysis. Geneva: United Nations Children’s Fund; 2016.
  • Paxton A, Maine D, Freedman L, et al. The evidence for emergency obstetric care. Int J Gynecol Obstet. 2005;88:181–10.
  • Thaddeus S, Maine D. Too far to walk: maternal mortality in context. Soc Sci Med. 1994;38:1091–1110.
  • Pacagnella RC, Cecatti JG, Parpinelli MA, et al. Delays in receiving obstetric care and poor maternal outcomes: results from a national multicentre cross-sectional study. BMC Pregnancy Childbirth. 2014;14:159.
  • Hussain-Alkhateeb L, Fottrell E, Petzold M, et al. Local perceptions of causes of death in rural South Africa: a comparison of perceived and verbal autopsy causes of death. Glob Health Action. 2015;8:28302.
  • Bauni E, Ndila C, Mochamah G, et al. Validating physician-certified verbal autopsy and probabilistic modeling (InterVA) approaches to verbal autopsy interpretation using hospital causes of adult deaths. Popul Health Metr. 2011;9:49.
  • Nichols EK, Byass P, Chandramohan D, et al. The WHO 2016 verbal autopsy instrument: an international standard suitable for automated analysis by InterVA, InSilicoVA, and Tariff 2.0. PLoS Med. 2018;15:e1002486.
  • Byass P, Chandramohan D, Clark SJ, et al. Strengthening standardised interpretation of verbal autopsy data: the new InterVA-4 tool. Glob Health Action. 2012;5:19281.
  • Campbell O, Ronsmans C. Programme WHOMH and SM. Verbal autopsies for maternal deaths: World Health Organization workshop; 1994 Jan 10–13; London; 1995 [cited 2019 Jul 5]. Available from: https://apps.who.int/iris/handle/10665/61029
  • Graham W. Safe motherhood needs assessment: maternal death review. Geneva: World Health Organization; 1996.
  • Ronsmans C, Etard J-F, Walraven G. Verbal autopsies: learning from reviewing deaths in the community. Beyond Numbers. 2004:43.
  • Pasha O, McClure EM, Saleem S, et al. A prospective cause of death classification system for maternal deaths in low and middle-income countries: results from the global network maternal newborn health registry. BJOG. 2018;125:1137–1143.
  • Say L, Chou D, Gemmill A, et al. Global causes of maternal death: a WHO systematic analysis. Lancet Glob Health. 2014;2:e323–e333.
  • Sageer R, Kongnyuy E, Adebimpe WO, et al. Causes and contributory factors of maternal mortality: evidence from maternal and perinatal death surveillance and response in Ogun state, Southwest Nigeria. BMC Pregnancy Childbirth. 2019;19:63.
  • Chavane LA, Bailey P, Loquiha O, et al. Maternal death and delays in accessing emergency obstetric care in Mozambique. BMC Pregnancy Childbirth. 2018;18:71.
  • Cham M, Sundby J, Vangen S. Maternal mortality in the rural Gambia, a qualitative study on access to emergency obstetric care | | t [Internet]. [cited 2019 Jul 5]. Available from: https://reproductive-health-journal.biomedcentral.com/articles/10.1186/1742-4755-2-3
  • D’Ambruoso L. Global health post-2015: the case for universal health equity. Glob Health Action. 2013;6:19661.
  • Knight HE, Self A, Kennedy SH. Why are women dying when they reach hospital on time? A systematic review of the ‘third delay’. PLoS One. 2013;8:e63846.
  • Baharuddin M, Amelia D, Suhowatsky S, et al. Maternal death reviews: a retrospective case series of 90 hospital-based maternal deaths in 11 hospitals in Indonesia. Int J Gynecol Obstet. 2019;144:59–64.
  • Merali HS, Lipsitz S, Hevelone N, et al. Audit-identified avoidable factors in maternal and perinatal deaths in low resource settings: a systematic review. BMC Pregnancy Childbirth. 2014;14:280.
  • Amsalu R, Morris CN, Chukwumalu K, et al. Essential newborn care practice at four primary health facilities in conflict affected areas of Bossaso, Somalia: a cross-sectional study. Confl Health. 2019;13:27.