Review of maternal death audits in refugee camps in UNHCR East and Horn of Africa and Great Lakes Region, 2017–2019
Abstract
Objectives
To review the quality of maternal death audits and identify factors contributing to refugee maternal deaths in the East and Horn of Africa.
Methods
Maternal death audits submitted to The UN Refugee Agency (UNHCR) from 2017 to 2019 in 43 refugee camps in eight countries were analyzed for completeness, obstetric history, cause of death, and contributing factors.
Results
A total of 191 refugee maternal death audits were retrieved. The mean age of the deceased was 28 years (range, 15–45 years), and 13% were adolescents and 17% were of advanced maternal age. Most patients (55%) were grand multigravida (≥5 pregnancies). The majority (86%) attended antenatal care visits, with 51% attending four or more visits. Among women who delivered (n = 140), 91% were facility-based deliveries. Most (68%) deaths occurred postpartum. Obstetric hemorrhage (49%) was the leading direct cause of death (with 77 cases of postpartum hemorrhage), followed by hypertensive disorder (19%) and infection (15%). Delays in care were identified in 185 (97%) cases. Delays in receiving care were more prevalent (81%) than in seeking (61%) and reaching (26%) care.
Conclusion
Factors contributing to delays in receiving care highlight the capacity gaps in provision of emergency obstetric care, including management of postpartum hemorrhage, requiring urgent additional investments. Audit findings also show the need for attention and action towards family planning, contraception, and adolescent sexual and reproductive health services.
1 INTRODUCTION
The East and Horn of Africa and Great Lakes (EHAGL) region is home to the majority of forcibly displaced persons in sub-Saharan Africa. In 2021, the region hosted approximately 5 million refugees and over 12 million internally displaced people.1 Refugees in the EHAGL are hosted by countries that face a high burden of maternal mortality themselves.2 Although most host countries have policies on maternal death surveillance and response (MDSR), many still face significant obstacles relating to reporting, reviewing, and follow-up of maternal deaths.3 Evidence shows MDSR is still surrounded by misconceptions about confidentiality and accountability, limited community engagement, and absence of policies and guidelines at all health levels.4 Health workers may feel blamed, and audits are sometimes seen as a punitive measure. Additionally, already overburdened healthcare systems face competing priorities when it comes to the implementation and follow-up of MDSR.4-7 Similar obstacles were found across humanitarian settings, associated with context-specific factors such as health worker security and moral distress.8
The UN Refugee Agency (UNHCR) works with national governments and partners to promote and provide international protection and humanitarian assistance to refugees, including enabling access to health services. In line with the Global Compact on Refugees, UNHCR advocates for inclusion of refugees in national health services and implements programs to support services to meet the needs of both refugees and host communities.9 Strengthening maternal health services is a UNHCR priority and is supported by the collection and analysis of robust data.
UNHCR encourages routine reporting and auditing of all maternal deaths in refugee settings in line with national approaches and in conjunction with ministries of health and partners. In 2007, UNHCR developed and introduced a maternal death audit tool, based on World Health Organization (WHO) guidance, which documents and facilitates review of demographic characteristics, obstetric history, cause of death, and contributing factors.10, 11 Audit reports are compiled and examined yearly with aggregated analyses, and recommendations are fed back to UNHCR country operations and partners.
A previous study of UNHCR maternal death review reports in 2008 to 2010 analyzed data from 10 countries, including six in the EHAGL region, finding timing and causes of death similar to reported global trends and identifying delays in care at health facilities that contributed to preventable deaths.11 There have been no other publications focused on MDSR in refugee operations in EHAGL in the past decade, and no studies of national MDSR systems that examined implementation in health facilities serving refugee populations. This study was therefore undertaken to assess the quality of maternal death audits, analyze the causes and circumstances surrounding maternal deaths, and elucidate bottlenecks in the availability, access to, and provision of quality maternal care in refugee settings in the EHAGL region.
2 MATERIALS AND METHODS
2.1 Study setting
UNHCR conducted a retrospective review of 3 years of maternal death audit data from refugee operations in the EHAGL region. Maternal death audit reports were requested from UNHCR operations in Burundi, Djibouti, Eritrea, Ethiopia, Kenya, Rwanda, Somalia, South Sudan, Sudan, Tanzania, and Uganda.
2.2 Study population
The WHO definition of maternal death was used.12 All maternal death audits completed by UNHCR and its partners in the EHAGL region from January 1, 2017 to December 31, 2019, were reviewed. This included audits among refugee populations and among host country nationals utilizing UNHCR-supported health services.
2.3 Data collection and analysis
As per UNHCR standards, maternal death audit forms are stored at the facility where deaths are recorded and are digitized and submitted as Microsoft Word or PDF files to the national authorities and the UNHCR public health personnel of that respective area/camp/settlement.
Three UNHCR EHAGL staff compiled the audit reports received from eight country operations and entered data into Microsoft Excel (2019) for analysis using descriptive statistics and qualitative content analysis methods. All three analysts had a clinical background in maternal and newborn care and humanitarian experience.
To assess the quality and completeness of the forms, the analysts examined missing outcomes, the time of reporting and presence/involvement of community members in the death review. To assess the causes and circumstances surrounding maternal deaths, the characteristics of women who died (demographics, gravidity, parity, stage of pregnancy at time of death) as well as the mode of delivery were noted. To elucidate bottlenecks in the availability, access, and provision of quality maternal care, analysts assessed the number of antenatal care (ANC) visits, place of delivery, skilled attendance at delivery, birth outcomes (live birth or stillbirth), status of women at admission (stable or critical, based on the clinical history, and vital signs that were reflected in the audit report) and cause of death. To classify the cause of death, the International Classification of Diseases, Tenth Revision-Maternal Mortality (ICD-MM) classification was used.13 Analysts reviewed the cause of death identified in the audit alongside the case history; if the history did not match the cause of death identified in the audit report or was deemed unlikely by the reviewer, the cause of death was adjusted. For deaths caused by postpartum hemorrhage (PPH), deaths were further classified using the four Ts: tone, trauma, tissue, and thrombin.14
The “Three-Delays” model was used to classify factors contributing to maternal deaths.15 Analysis was conducted separately by year and for refugee and national deaths. Findings are presented for all maternal death audit reports reviewed, and for the subset of audit reports for deaths caused by PPH.
2.4 Ethical clearance
The data analyzed within the scope of this research was deidentified before entering the findings into the database. The data are part of routinely collected data for health surveillance and quality purposes. The Johns Hopkins Bloomberg School of Public Health institutional review board reviewed the study plans and determined that the analysis does not qualify as human subjects research and thus did not require institutional review board oversight.
3 RESULTS
A total of 191 refugee maternal death audit reports were collected from 43 camps/settlements in eight countries (Table 1). An increase in reports was seen over time (47 in 2017, 65 in 2018, and 79 in 2019). In addition, 19 reports were received of host country nationals: three in Kenya, three in South Sudan, and 13 in Uganda.
| Study locations | No. of maternal death audits | ||
|---|---|---|---|
| Country | No. of camps | Camp names | Total number of maternal death audits received (no. of maternal refugee death audits) |
| Burundi | 1 | Bwagiriza | 1 (1) |
| Djibouti | 2 | Al Sabieh, Holl-Holl | 2 (2) |
| Ethiopia | 6 | Bambasi, Nguenwiel, Hilewyn, Isore, Melkadida, Kule | 9 (9) |
| Kenya | 6 | Hagadera, Turkana West, Ifo 1, Ifo 2, Hamey, Kakuma | 72 (69) |
| Rwanda | 4 | Gihembe, Mahama, Kigeme, Kiziba | 7 (7) |
| South Sudan | 9 | Ajoung Thok, Doro, Pamir, Kaya, Batil, Gendrossa, Yabus, Yida, Gorem | 21 (19) |
| Tanzania | 3 | Mtendeli, Nyarugusa, Ntuda | 14 (14) |
| Uganda | 12 | Rwamanja, Kyangwali, Adjumani, Kyaka II, Nakivale, Imvepi, Kiryandongo, Bidibidi, Palabek, Orugchinga, Palorinya, Rhinocamp | 83 (70) |
| Total | 43 | - | 210 (191) |
3.1 Audit process
Audit formats varied from minutes of a maternal death review meeting to national ministry of health audit forms to UNHCR audit forms. Most audits were conducted within 1 week (149 [71.0%]), but 25 (11.9%) of the 210 refugee and national maternal death audits were conducted more than 1 month after the death. In 133 (63.3%) of the 210 cases, a family or community member participated in the audit process. Data regarding the timing of the audit was missing in 13 (6.2%) reports and information about family/community presence was missing in 32 of 210 (15.2%) reports (Table 2).
| Refugees (n = 191) | Nationals (n = 19) | |
|---|---|---|
| Demographics | Mean (range) | Mean (range) |
| Age, years | 28 (15–45) | 27 (17–39) |
| No. of pregnancies | 5 (1–17) | 5 (1–11) |
| Parity (number of prior deliveries) | 4 (0–13) | 3 (0–8) |
| No. of ANC visits | 3.5 (0–8) | 2.3 (0–5) |
| Age categories, n (%) | ||
| Adolescents (aged 10–19 years) | 24 (12.6) | 4 (21.1) |
| 20–34 years | 119 (62.3) | 11 (57.9) |
| 35+ years | 33 (17.3) | 4 (21.1) |
| Not documented | 15 (7.8) | 0 (0) |
| Gravidity, n (%) | ||
| Primigravida | 36 (18.8) | 5 (26.3) |
| Multigravida2-4 | 46 (24.1) | 5 (26.3) |
| Grand multigravida5-9 | 87 (45.5) | 7 (36.8) |
| Great grand multigravida (≥10) | 18 (9.4) | 2 (10.5) |
| Not documented | 4 (2.1) | 0 (0) |
| Parity at time of ANC, n (%) | ||
| Nullipara | 28 (14.7) | 5 (26.3) |
| Primipara | 18 (9.4) | 3 (15.8) |
| Multipara (2–4 deliveries) | 53 (27.7) | 4 (21.1) |
| Grand multipara5-9 | 83 (43.5) | 7 (36.8) |
| Great grand multipara (≥10) | 7 (3.7) | 0 (0) |
| Not documented | 2 (1.0) | 0 (0) |
| ANC visits, n (%) | ||
| No attendance | 9 (4.7) | 5 (26.3) |
| 1–3 visits | 67 (35.1) | 9 (47.4) |
| ≥4 visits | 98 (51.3) | 5 (26.3) |
| Not documented | 17 (8.9) | 0 (0) |
| Delivered at time of death, n (%) | ||
| Did deliver | 140 (73.3%) | 14 (73.7%) |
| Did not deliver | 51 (26.7%) | 5 (26.3%) |
| Audit process data | ||
| Timing of maternal death audit from date of death, n (%) | ||
| <1 week | 137 (71.7) | 12 (63.2) |
| 1 week-1 month | 20 (10.4) | 3 (15.8) |
| More than 1 month | 22 (11.5) | 3 (15.8) |
| Not documented | 12 (6.3) | 1 (5.3) |
| Family or community member present at audit, n (%) | ||
| Yes | 121 (63.4) | 12 (63.2) |
| No | 42 (22.0) | 3 (15.8) |
| Unknown | 28 (14.7) | 4 (21.1) |
- Abbreviation: ANC, antenatal care.
3.2 Maternal deaths among refugee women
The mean age of refugee women/girls with deaths audited was 28 years (range, 15–45 years) (Table 2). Adolescents (aged 10–19 years) made up 12.6% of the pregnancies (24 of 191), while 17.3% (33 of 191) were of advanced maternal age at time of death (≥35 years). There was high gravidity and parity, with an average of five pregnancies (range, 1–17) and four births (range, 0–13) prior to the current pregnancies. A total of 91 of 191 (47.2%) refugee women had a high-risk pregnancy, with more than five deliveries previously. In total, 165 of 191 (86.4%) refugee women had attended ANC visits at least once, of whom 98 (51.3%) had attended at least four times. Nine of 191 refugee women (4.7%) never attended ANC and, for 17 (8.9%), ANC attendance was not documented. Most women (140 of 191) had delivered at time of death (73.3%).
In one case, the mode of delivery was not documented, for the others, 74 of 140 had a vaginal delivery (52.9%), while 65 of 140 (46.4%) delivered by cesarean section. A total of 93 of 140 (66.4%) deliveries resulted in a live birth, while 40 (28.6%) resulted in a stillbirth; for seven of the 140 deliveries (5.0%), the birth outcome was not documented. Among those who died during or after delivery, 127 of the 140 deliveries (90.7%) occurred at a health facility. Of the facility deliveries, 38 of the 127 (29.9%) had documented partograph use. In 72 (56.7%) of the 127 reports, information on partograph use was not recorded (Table 3).
| Delivery data | Refugees, n (%) | Nationals, n (%) |
|---|---|---|
| Mode of delivery | n = 140 | n = 14 |
| Vaginal birth | 74 (52.9) | 10 (71.4) |
| Cesarean section | 65 (46.4) | 4 (28.6) |
| Not documented | 1 (0.7) | 0 (0) |
| Attended by skilled health personnel | ||
| Skilled | 127 (90.7) | 11 (78.6) |
| Not attended by skilled personnel | 13 (9.3) | 3 (21.4) |
| Not documented | 0 (0) | 0 (0) |
| Birth outcome | ||
| Live birth | 93 (66.4) | 8 (57.1) |
| Stillbirtha | 40 (28.6) | 6 (42.9) |
| Not documented | 7 (5.0) | 0 (0) |
| Partograph use among those who had skilled delivery, n (%) | n = 127 | n = 11 |
| Yes | 38 (29.9) | 1 (9.1) |
| No | 17 (13.4) | 4 (36.4) |
| Not documented | 72 (56.7) | 6 (54.5) |
| Admission data | ||
| Admitted at time of death, n (%) | n = 191 | n = 19 |
| Yes | 178 (93.2) | 19 (100) |
| No | 13 (6.8) | 0 (0) |
| Clinical status at time of admission, n (%)b | n = 182 | n = 19 |
| Stable | 91 (50.0) | 6 (31.6) |
| Critical | 91 (50.0) | 13 (68.4) |
| Stage of pregnancy at time of admission, n (%)b | n = 182 | n = 19 |
| Antepartum | 79 (43.4) | 5 (26.3%) |
| Intrapartum | 64 (35.2) | 0 (0) |
| Postpartum | 39 (21.4) | 14 (73.7) |
| Stage of pregnancy, location and cause of death data, n (%) | ||
| Stage of pregnancy at time of death | n = 191 | n = 19 |
| Antepartum | 39 (20.4) | 5 (26.3) |
| Intrapartum | 12 (6.3) | 0 (0.0) |
| Postpartum | 140 (73.3) | 14 (73.7) |
| Postpartum (delivered at facility), n | 127 | 11 |
| Postpartum (delivered at home), n | 13 | 3 |
| Location of death, n (%)c | n = 191 | n = 19 |
| In the community (including TBA/traditional medicine practitioner /on the way to facility), n (%) | 13 (6.8%) | 0 (0%) |
| At the facility | 178 (93.2) | 15 (78.9) |
| NGO-managed facility, n | 98 | 4 |
| Government-run facility, n | 80 | 11 |
| Not documented, n (%) | 0 (0) | 4 (26.3) |
| Direct cause of death, n (%)d | n = 191 | n = 19 |
| Pregnancy with abortive outcome | 5 (2.6) | 1 (5.3) |
| Hypertensive disorder | 37 (19.4) | 4 (21.1) |
| Obstetric hemorrhage | 94 (49.2) | 12 (63.2) |
| Pregnancy related infection | 29 (15.2) | 2 (10.5) |
| Other obstetric complication | 26 (13.6) | 0 (0) |
| Unanticipated complications of management | 6 (3.1) | 0 (0) |
- Abbreviations: NGO, nongovernmental organization; TBA, traditional birth attendant.
- a For stillbirth, no distinction was made between fresh and macerated stillbirth, as this is often misclassified in the field because of a lack of capacity and means to monitor the status of the fetus intrauterine.
- b Four mothers were admitted in an earlier stage (antenatal) but were discharged and died at home afterwards.
- c Note some sought care for their specific problem but were discharged and died outside the facility.
- d Total of percentage is more than 100% as some deaths had multiple causes (six deaths in which more than one cause was identified).
The majority of the refugee women deaths occurred at health facilities (178 of 191 [93.2%]). However, an additional four women were admitted prior in their pregnancy but were discharged and died there. At the time of admission, 79 of 182 women (43.4%) were antepartum and 64 of 182 (35.2%) were in labor. Of those admitted in the facility, half (91 of 182) were in a critical condition and half (91 of 182) were stable on admission to the first facility. For those admitted, no difference in distribution was found for certain subgroups (adolescents or high parity) relating to clinical status on arrival (Table 3).
A total of 178 (93.2%) of the 191 refugee deaths occurred after admission at the first facility, while 13 (6.8%) of the 191 deaths occurred in the community or on the way to the first facility, including deaths of the four women who were discharged. Of those who died in a facility, 98 of the 178 women (55.1%) died in a nongovernmental organization–managed facility and 80 of the 178 (44.9%) in a government-run facility. Of all deaths, most women were postpartum at time of death (140 of 191 [73.3%]), followed by antepartum (39 of 191 [20.4%]) and intrapartum (12 of 191 [6.3%]). Obstetric hemorrhage was the leading direct cause of death (94 of 191 [49.2%]), followed by hypertensive disorder (37 of 191 [19.4%]) and pregnancy-related infection (29 of 191 [15.2%]). Some audits identified more than one cause of death (e.g., eclampsia and PPH) (Table 3).
3.3 Maternal death among host country national women
The mean age of host country national women/girls was 27 years (range, 17–39 years), four of 19 (21.2%) were adolescents. High-risk pregnancies were also visible, with seven of 19 (36.8%) women having more than five previous deliveries. Five of 19 (26.3%) women did not attend ANC while five of 19 (26.3%) had completed at least four visits (Table 2). The three leading causes of death among host country nationals were obstetric hemorrhage (12 of 19 [63.2%]), followed by hypertensive disorder (four of 19 [21.1%]) and pregnancy-related infection (two of 19 [10.5%]) (Table 3).
3.4 Delays among refugee women
With the use of the Three-Delays model, in 185 (96.9%) of the 191 audits of refugee women, a delay in care was identified. In 117 of 191 reports (61.3%), a significant first delay was identified (i.e., in the decision to seek care), in 49 of 191 reports (25.6%) the second delay played a role (i.e., in reaching the healthcare facility), and in 154 of 191 reports (80.6%) a third delay (i.e., in receiving care at the facility) played a significant role. In-depth analysis reflects poor health-seeking behavior, preference for traditional medicine/home care, and reluctance to seek care as first delay factors. The second delay factors included lack of transport, long distances to the clinic, poor road conditions, and weak linkage between community structures and health clinics. Finally, under the third delay, inadequately trained and poorly motivated staff, lack of resources, and an inadequate referral system were factors (Table 4).
| First delay: delay in decision to seek care (n = 117) | |
| Poor health-seeking behavior | Not recognizing danger signs of obstetric complications |
| Late arrival for delivery care | |
| Preference for traditional medicine/home care | Preference of home delivery |
| Traditional healers first consulted | |
| Home treatment with (herbal) medicine | |
| Not willing to seek care | Unsafe abortion (fear of repercussions) |
| Refusal to consent to care | |
| Fear of cesarean section if seeking care at facility | |
| Self-discharge against medical advice | |
| Second delay: delay in reaching care (n = 49) | |
| Lack of transport | Nonfunctional/unavailable ambulances for community transport |
| Lack of any transport means | |
| Long distance to clinic | Distance from health clinic |
| Poor road condition | Poor road condition (especially during rainy season) |
| Lack of communication means | Lack of communication means/structures to alert the health facility |
| Third delay: delay in receiving adequate care (n = 154) | |
| Lack of resources | Lack of human resources |
| Lack of equipment | |
| Lack of medicines | |
| Inadequately trained and poorly motivated staff | Misdiagnosis/treatment |
| Lack of trained human resources available in the clinic | |
| Poor family involvement (communication) | |
| Poor staff attitude | |
| Poor monitoring (e.g. incomplete/no partograph or poor postoperative monitoring) | |
| Failure to recognize high-risk pregnancies | |
| Poor documentation | |
| Inadequate referral system | Comprehensive Emergency Obstetric and Newborn Care facility not fully operational |
| Comprehensive Emergency Obstetric and Newborn Care facility hard to reach because of road condition, inadequate transport, long distance, lack of ambulances | |
| Poor communication between primary and secondary care facilities | |
3.5 Characteristics of refugee women with maternal deaths caused by PPH
A total of 77 of 191 refugee women (40.3%) died of PPH (Table 5). Women who died of PPH were on average 29 years old (range, 16–40 years) and had an average gravidity of six (range, 1–17) and a parity at the time of ANC of five (range, 0–11). Sixteen (20.8%) of 77 women were of advanced maternal age and six of 77 (7.8%) were adolescent. Among women who died of PPH, the majority (55 of 77 [71.4%]) were grand or great-grand multigravida, while of women who died of other causes the majority (60 of 114 [52.7%]) were primigravida or multigravida. Similarly, high parities were found among the women who died of PPH, 47 of 77 (61%) had five or more previous deliveries, compared with 42 of 114 (36.8%) women who died of other causes. Of women who died of PPH, the mode of delivery was cesarean section in 32 of 77 (41.6%) deliveries and vaginal delivery in 44 of 77 (57.1%).
| Refugee women who died of PPH (n = 77) | Refugee women who died of other causes (n = 114) | |
|---|---|---|
| Demographics | Mean (range) | Mean (range) |
| Age | 29 (16–40) | 27 (15–45) |
| Gravidity | 6 (1–17) | 4 (1–14) |
| Parity | 5 (0–11) | 4 (0–13) |
| Age categories, n (%) | ||
| Adolescents (10–19 years) | 6 (7.8) | 19 (16.7) |
| 20–34 years | 52 (67.5) | 63 (55.3) |
| Advanced maternal age pregnancy (≥35 years) | 16 (20.8) | 21 (18.4) |
| Not documented | 3 (3.9) | 11 (9.6) |
| Gravidity, n (%) | ||
| Primigravida | 8 (10.4) | 28 (24.6) |
| Multigravida2-4 | 14 (18.2) | 32 (28.1) |
| Grand multigravida5-9 | 43 (55.8) | 44 (38.6) |
| Great grand multigravida (≥10) | 12 (15.6) | 6 (5.3) |
| Not documented | 0 (0) | 4 (3.5) |
| Parity at time of ANC, n (%) | ||
| Nullipara | 5 (6.5) | 24 (21.) |
| Primipara | 5 (6.5) | 13 (11.4) |
| Multipara2-4 | 20 (26.0) | 33 (28.9) |
| Grand multipara5-9 | 42 (54.5) | 40 (35.0) |
| Great grand multipara (≥10) | 5 (6.5) | 2 (1.8) |
| Not documented | 0 (0) | 2 (1.8) |
| Mode of delivery, n (%) | ||
| Cesarean section | 32 (41.6) | 34 (29.8) |
| Vaginal delivery | 44 (57.1) | 31 (27.2) |
| Did not deliver | N/A | 49 (43) |
| Not documented | 1 (1.3) | 0 (0) |
- Abbreviations: ANC, antenatal care; N/A, not applicable; PPH, postpartum hemorrhage.
In 17 of 77 cases the direct cause of the PPH could not be identified because of a lack of documentation, clinical follow-up, or death occurring outside the facility. For the other 60 cases, the leading direct cause of PPH was atonic uterus, 27 (45%) of which were after cesarean section. Atonic uterus was followed by trauma as a direct cause (lacerations, hematoma, inversion, and rupture) with 22 (36.7%) of the 60 cases reporting lacerations or rupture; among these 22, 13 (59.1%) were grand or great-grand multipara. In 11 of the 60 cases (18.3%), a retained placenta or clot was identified as the direct cause of PPH. No cases of thrombin issues/coagulopathy were identified but it is possible some of the PPHs were additionally complicated by this (Table 6).
| Total (n = 60)a | Percentage | |
|---|---|---|
| Atonic uterus | 27 | 45 |
| After vaginal delivery | 10 | 16.7 |
| After cesarean -section | 17 | 28.3 |
| Trauma (lacerations, hematoma, inversion, rupture) | 22 | 36.7 |
| Rupture | 16 | 26.7 |
| Lacerations | 8 | 13.3 |
| Tissue (retained tissue, invasive placenta) | 11 | 18.3 |
| Thrombin (coagulopathy) | 0 | 0 |
- a In 17 cases, the cause of death could not be identified and were therefore excluded. PPH, postpartum hemorrhage.
4 DISCUSSION
In the current study, the authors examined the characteristics of, and contributing factors to, maternal deaths in refugee camps and settlements in the EHAGL region. The authors saw a notable proportion of adolescents and high gravidity/parity among the deceased and noted that most deaths occurred postpartum. Many women had attended at least one ANC visit and among those who delivered, the vast majority delivered in the health facility. Obstetric hemorrhage, specifically PPH, was identified as the most common cause of death. Avoidable factors contributing to the death were mainly related to delays in receiving care, followed by seeking care and reaching care.
4.1 Quality and completeness of audits
There was an increase in audit reports over time, which does not necessarily reflect an increase in the number of maternal deaths but may be because of improved reporting and coverage of maternal death audits. The increase was largely attributed to reports from Uganda. There were marked differences in reporting of maternal mortality between countries with similar size refugee populations that reflect suboptimal surveillance in some settings. The authors noted a lack of completeness of the audits with regards to demographic data and contributing factors and acknowledged that most likely not all maternal deaths were captured and not all deaths were audited. The difficulty in capturing data in humanitarian settings is known.16 However, this study, along with the Hynes et al. study, is the only recent study that examines this number of audits from refugee settings.11 Factors that contribute to audits not being completed include deaths occurring in distant referral facilities, with limited ability to retrieve this information. The incompleteness of some audits could be related to late reporting of deaths, a lack of MDSR institutionalization including reluctance of health providers to engage in audits, incomplete clinical record-keeping, or lack of community involvement.5, 7, 8, 17 To effectively monitor reproductive health programs and ensure the best way to serve the needs of refugees, health program managers should continue to focus on the quality of MDSR in refugee settings.8, 10, 18 A revised audit form was developed to improve the audit process in refugee settings and facilitate collection of quality data.19, 20 Stillbirths and neonatal deaths are also common in these settings, and many are preventable. Where resources allow and when the coverage of perinatal death reporting is high, audits of perinatal deaths are also able to highlight factors contributing to adverse maternal and perinatal outcomes.21 However, in settings with incomplete maternal and perinatal death reporting the focus should be on capturing all maternal and perinatal deaths and auditing all maternal deaths.
4.2 Risk factors
This study highlights the high parity among maternal deaths and a considerable number of deaths in adolescents. Higher order births are less likely to be covered by critical maternal and child health interventions, and complications during pregnancy and childbirth are the leading cause of death for 15- to 19-year-old girls globally.22 The high parity and early pregnancy require a focus on comprehensive SRH, which includes family planning and contraceptive services but also on overcoming social and cultural obstacles through multisectoral enablers including promoting gender equality and empowerment, reducing child marriage, and encouraging higher educational attainment in girls.23
4.3 Delays and quality of care
Compared with an analysis of refugee maternal death audits from 2008 to 2010, the authors saw an increase in ANC attendance among the deceased, with 55% attending at least four visits versus 33% in the previous study.11 This may reflect efforts made in the past decade to improve ANC coverage or that previous studies included a wider range of countries. Efforts have also been made to improve the coverage of facility-based deliveries. Indeed, an increased proportion of deliveries are now occurring in facilities, but this study demonstrates that required care was not always available.11 In our study the third delay was the most prominent. While the authors acknowledge the importance of the third delay, half of the patients arrived in a critical condition, meaning the first and second delay still require significant attention to ensure women and their families decide to seek care and are able to reach appropriate care.
Partograph use was not well documented, indicating either poor record keeping or underuse of the partograph. Barriers to partograph use have been observed in low and in low−/middle-income countries.24 Additional focus should be placed on postoperative monitoring as 46% of patients had a cesarean section.
4.4 Cause of death
The current study showed that the most reported direct cause of death was PPH. This is similar to regional estimates,25 yet with adequate and timely interventions, the morbidity and mortality associated with PPH can be minimized, regardless of the cause.14 The known risk factors of PPH, such as grand multiparity, need to be similarly addressed.14
5 CONCLUSION
Reducing maternal and neonatal mortality is central to the Sustainable Development Goals. Significant improvements have been made in access to maternal and neonatal care services in refugee settings. However, the significant contribution of the third delay to maternal deaths highlights the capacity gaps in provision of emergency obstetric care among healthcare providers, whether nongovernmental organizations or national services. While evidence-based guidance is available, implementation is often far behind.
Despite improvements, MDSR in refugee settings still has challenges with underreporting and audits of variable quality. UNHCR is strengthening its capacity at the country level and has developed updated guidance on maternal death auditing.20 Many of the same factors that contribute to maternal deaths also affect fetal and neonatal outcomes. Strengthening the reporting of all stillbirths and neonatal deaths should be a priority in refugee settings in sub-Saharan Africa. Achieving the global target of less than 70 maternal deaths per 100 000 live births by 2030 requires a mortality reduction more than double the rate achieved between 2000 and 2015. As an immediate measure there is a crucial need to improve quality and coverage of comprehensive SRH services, particularly comprehensive emergency obstetric care, which includes management of PPH; these services should be inclusive of refugee hosting areas.26, 27 Sustained progress in reducing maternal deaths will require progress on health system and multisector enablers and an equity-driven approach encompassing, inter alia, universal health coverage, poverty reduction, improved nutrition, and gender equality.19
AUTHOR CONTRIBUTIONS
A.B., C.S., and E.M. conceived the idea of the paper. All authors were involved in the planning of the article. T.B. and E.M. created the database for the analysis and analyzed the data. T.B. and H.H. prepared the first draft of the manuscript and all authors were involved in the review of the paper.
ACKNOWLEDGMENTS
We want to acknowledge F. Jones and D. Yiweza (UNHCR Regional Bureau for East and Horn of Africa and Great Lakes) for their contribution in data collection and compilation, as well as the UNHCR public health officers and health partners of the respective country operations for their support in the audit process and compilation of the reports.
CONFLICT OF INTEREST
There are no conflicts of interest to report.
Open Research
DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available because of privacy or ethical restrictions.
