Introduction of confidential enquiry into maternal deaths in Ethiopia: Implementation and methodological considerations

Patient and Public Involvement

Patients were not directly involved in the design of this study. Representative from ministry of health, heads of regional health bureaus, zonal health department, Ethiopian society of Obstetricians and Gynaecologists, Ethiopian Midwifery Association, Professional Association of Emergency Surgery Officers of Ethiopia, and people from academia were involved in the study as EthOSS Steering Committee. The steering committee reviewed the project progress, supervise and guide how the project activities should be done, approved the selection of EthOSS conditions to be investigated, and the design and data reporting forms of the project. Obstetricians, midwives, and critical care specialists were involved in the project launch and review meetings of EthOSS. The EthOSS methodology including EthOSS steering committee, case selection, reporting, data capturing and other details were previously published¹⁴. Patients and the public were not involved in dissemination plan for the study.

Study settings, design and period

This CEMD was introduced as part of the EthOSS project conducted by Haramaya University in collaboration with the University of Oxford in the United Kingdom, and Leiden University Medical Centre and University Medical Centre Groningen in The Netherlands, to adapt obstetric surveillance and confidential enquiry methodologies for use in a low-resource setting. Details pertaining to the adaptation of the methodology are described elsewhere¹⁴.

In brief, EthOSS was introduced in 13 hospitals in eastern Ethiopia to report the number of women who had sustained any of five major obstetric conditions (obstetric hemorrhage, eclampsia, uterine rupture, sepsis, and severe anemia), as well as the number of women who died, using a system of voluntary reporting midwife based at each hospital. The level of hospital ranged from several lower-level primary (district) hospitals to a tertiary (comprehensive teaching) hospital¹⁴. We used multiple methods (quantitative and qualitative) in collecting, analyzing and presenting data from this CEMD. Implementation of CEMD included the process of identification and anonymizing medical records of women who died, and establishing and training the CEMD committee, which subsequently reviewed all maternal deaths that occurred between 01 April 2021 to 31 March 2022.

Maternal death identification

All maternal deaths in participating hospitals during the study period were identified and reported (n=70) on a monthly basis by a designated midwife. We obtained permission from respective hospitals to actively search records of deaths that were filed in the medical directors’ office. In many Ethiopian hospitals, this is the location where medical records of maternal deaths are temporarily retained to ensure that women’s deaths are documented for the purposes of MDSR, and ensure that no records are changed after a death occurred. In addition, registers of the central intensive care unit and all other adult wards where deaths might have occurred were searched. The full medical records of each woman were subsequently scanned and stored on a password-protected electronic file only accessible by the project team lead and the primary investigator.

De-identification of maternal deaths

Upon receiving reports of maternal deaths in EthOSS, files for these women were scanned and deidentified centrally by the EthOSS team and stored for review by the CEMD committee. All personal identifiers, facility information, details of managing staff, and all other potentially identifiable details were removed before review by the CEMD committee. As an additional measure to ensure anonymity and confidentiality, individual CEMD committee members were not assigned to review maternal deaths that occurred in their own facilities except when they were being asked to clarify information.

Training of confidential enquiry into maternal death committee

The CEMD committee was trained by experts from Mothers and Babies: Reducing Risk through Audits and Confidential Enquiries across the UK (MBRRACE-UK), the Netherlands Audit Committee for Maternal Mortality and Morbidity, and the College of Health and Medical Sciences, Haramaya University. A one-day hands-on training on principles and the process of implementing CEMD was provided for the 24 committee members. Members included obstetrician-gynecologists, midwives, one emergency and critical care physician, a critical care nurse, one anesthesiologist, emergency surgical and obstetric officers (cadres of associate clinicians), anesthetists, and public health experts. All CEMD committee signed informed consents to maintain confidentiality of patient information.

To ensure in-depth understanding of the entire CEMD process, the committee was provided with brief orientation on how to work with the CEMD template and practiced reviewing maternal deaths and near misses under the direct supervision of experienced trainers, followed by reflections on their experiences, including comparisons to their experiences with the routine MDSR process, which many members had previously been involved in. Recommendations on how to improve the overall CEMD process were actively sought from all committee members.

The Ethiopian obstetric surveillance system confidential enquiry meetings

After maternal deaths case files were anonymized, a senior consultant obstetrician-gynecologist who was not actively in practice, but served in a managerial position and who fully supported the process, prepared case summaries for all maternal deaths. In addition to the case summaries, anonymized case files were shared with the committee members. Members of the CEMD committee then reviewed all deaths independently and recorded findings on the template prepared for this purpose, which was adapted from the MBRRACE-UK CEMD template.

Two CEMD plenary sessions were held for two days each, three months apart. Sessions were led by an independent chair and secretary. After individually reviewing the cases, each maternal death was reviewed in a plenary session to identify where improvements could be made. The plenary review of each death took an average of two hours. The CEMD committee members identified delays in care in the chain of events according to the three delays model by Thaddeus and Maine: delay one (delay in seeking care), delay two (delay in reaching an appropriate facility), and delay three (delay in receiving adequate care in the facility)¹⁶.

Statistical analysis

For quantitative outcomes, we used descriptive statistics, reporting means with standard deviations for continuous variables, and frequency and percentages for categorical variables. We calculated the maternal mortality ratio (MMR) (maternal deaths per 100, 000 live births) with its corresponding 95% confidence interval. Narrative descriptions of selected case scenarios (case vignettes) are presented alongside detailed discussion and recommendations from the CEMD committee to enable improvements to be made.

Ethical approval

The EthOSS protocol was reviewed and approved by the Institutional Health Research Ethics Review Committee (Ref No. IHRERC/024/2021) of the College of Health and Medical Sciences of Haramaya University, Ethiopia; and the University of Oxford’s Oxford Tropical Research Ethics Committee (OxTREC Reference 530-21). All methods of this study were performed in accordance with the Declaration of Helsinki. Before the actual data collection, informed consent was obtained from the medical director of each hospital. All collected data were stored anonymously in a password-protected electronic file only accessible by the project team lead and the primary investigator.

Consent statement

The CEMD committee members and data collectors were trained on major ethical issues, signed, informed consents of confidentiality and disclosed any conflicts of interest prior to participating in the review process. But, obtaining informed consent from individual woman as study participant was not applicable since this is review of case files or charts of women who died.

Role of funding source

This research was funded by the Medical Research Council (MRC) to Marian Knight as part of the 2019 Global Maternal and Neonatal Health Funding Call [grant number: MR/T037962/1]. The funders had no role in study design, data collection, data analysis, data interpretation, or writing of the manuscript.

Characteristics of study participants

During the study period, 34, 090 livebirths and 70 maternal deaths were reported to EthOSS from the 13 participating hospitals. The corresponding maternal mortality ratio was 205 (95% confidence interval 160–259) per 100 000 livebirths. A total of 59 maternal deaths were reviewed by the CEMD committee and included in this study. Deaths were excluded from review (n=11) if sufficient information is not available in the case files to enable the committee to perform a meaningful assessment, notably most commonly for deaths reported as death on arrival. Of the 59 maternal deaths reviewed, 42 (71%) were rural residents, 48 (81%) came to health facility by referral, and 39 (66%) women had no antenatal care booking. Only 13 (22%) of women who died were admitted to the intensive care unit and most deaths 47(80%) occurred during postpartum period (Table 1).

Causes of maternal deaths

As shown in details in the Figure 1 below, almost all women’s deaths were categorized as direct: 27 (46%) from hemorrhage and 15 (25%) from hypertensive disorders of pregnancy, most commonly eclampsia.

Figure 1. Causes of maternal deaths in eastern Ethiopia (n=59).

Factors contributing for maternal deaths

As indicated in detail in Figure 2 below, the CEMD committee identified a total of 102 individual factors, grouped into 16 categories, that contributed to the 59 maternal deaths we included in this report. These factors were mainly related to diagnosis, management or follow up at the facility level 29/59 (49%), maternal resuscitation and stabilization 16/59 (27%), and absence of antenatal care 10/59 (17%).

Figure 2. Distribution of factors contributing to maternal mortality in eastern Ethiopia (n=59).

Using the three-delay model, delay one, delay two and delay three were identified in 48/59 (81%), 52/59 (88%), and 54/59 (92%) of the deaths, respectively. Overall, in 55/59 (93%) of the deaths, one of the three delays was observed. In 58/59 (98%) of the deaths, improvements in care that may have made a difference in maternal outcome were identified.

Lessons learned from selected case’s vignettes

Referral problems

In several instances there were lessons related to referral and woman transfer between health facilities (an example shown in Vignette 1 below).

Vignette 1: A woman who gave birth to a stillborn baby in a nearby health center (a 10-minute drive) developed profuse vaginal bleeding postpartum. The team at the health center continued trying to manage the bleeding for 2 hours before referring her to the referral (tertiary level) hospital. Although she was in shock and unconscious upon arrival, she improved after abdominal packing, application of a non-pneumatic anti-shock garment, and receiving 4 units of blood, after which a hysterectomy was performed. She died 4 hours after the procedure in the Central Intensive Care Unit (CICU).

In this example, the CEMD committee considered that the care she received at the higher level hospital was according to locally expected standards. It was agreed that earlier transfer to the hospital could have made a difference to her outcome. The referring health center is located within a 10-minute drive and there was no shortage of ambulances. The committee also pointed out that since this woman had an intra-uterine fetal death (IUFD), according to national protocol, she should have been referred to give birth at the hospital rather than being managed at the health center. Similar difficulties related to referral or transfer between facilities were noted in the care of four other women.

Blood transfusion

Lack of adequate (compatible) blood for transfusion is one of the main contributing factors for maternal deaths in 9% of women who died, as illustrated in Figure 2 and vignette 2 below.

Vignette 2: A grand multiparous woman with a term pregnancy, who had not attended any antenatal care, presented to a tertiary hospital following vaginal bleeding for 13 hours. After giving vaginal birth to a stillborn baby, the bleeding continued. Uterotonics were given, a non-pneumatic anti-shock garment was applied, and only one unit of blood (the only available compatible blood) was transfused despite the severe blood loss, due to a lack of blood for transfusion. She went into shock and coma. No blood transfusion was available from any of the four blood banks in eastern Ethiopia and she died from severe bleeding.

Although her blood group (O-negative) may have compounded the challenge to obtain blood for transfusion, this situation emphasizes the need for availing blood for transfusion in all centers providing comprehensive emergency obstetric care. Since she gave birth in the tertiary hospital, lifesaving interventions such as aortic compression, bi-manual uterine compression and of uterine balloon catheter should have been applied to stop further bleeding and avoid a higher demand for blood transfusion. There was no documentation of such efforts in the medical records. The team involved in the care of this woman also added the point that difficulty in finding blood donors in acute settings and a lack of blood screening facility at the blood bank nearby are commonly observed major challenges.

Anesthesia-related

In 3 (5%) women, complications related to anesthesia were identified as cause for maternal deaths reviewed in Figure 1 above (an example in Vignette 3) below.

Vignette 3: A woman with a term pregnancy, who had antenatal care once during the pregnancy, presented to a primary (lower district level) hospital in labor. After pushing for 8 hours, vacuum extraction was tried but failed following which an emergency caesarean section was planned under spinal anesthesia. In the operating room, the anesthetist performed a lumbar puncture with a spinal needle, giving a regional spinal anesthetic upon the fourth attempt to reach the sub-dural space. Within five minutes of injection, the emergency surgical officer preparing for surgery noticed that the woman had a cardiorespiratory arrest. Resuscitation was tried using ambu-bag, intubation failed, and the woman died.

This case represents a typical challenge of the existing gaps in administering a spinal anesthesia in lower-level facilities in absence of a fully functional mechanical ventilator. The CEMD committee involved the health care providers who participated in the management of this case, and identified a need for improvement in technical skills around intubation and resuscitation following spinal anesthesia. To address this gap, the EthOSS project team organized two rounds of training for intensive care unit nurses and anesthetists working in the hospitals under the EthOSS consortium followed by a week of practical learning at the emergency and critical care unit of the tertiary hospital, which is better equipped in terms of staffing. The training included safe administration of anesthesia, and basic life support around anesthesia.

Indirect maternal deaths

It was also identified that undetected underlying medical conditions of a woman during pregnancy are contributing factors to maternal deaths as highlighted in vignette 4 below.

Vignette 4: A woman with a term pregnancy (primigravida) was referred from a primary hospital to a general (mid-level) hospital with complaints of shortness of breath. She was then diagnosed with severe anemia and congestive heart failure, and an intrauterine fetal death was found. She was subsequently referred to a tertiary hospital 200 km away for better medical management in Intensive Care Unit (ICU) care, but died during the ambulance transfer.

This case represents an example of situation where priority should have been given to treating the medical needs of the woman rather than her obstetric condition. Although the general hospital is staffed with internal medicine specialist doctors, these had not been consulted. It was not planned to either provide pain medication or shorten the second stage of labor (or doing a destructive birth procedure). The diagnosis of congestive heart failure was made only very late in pregnancy at the general hospital. In addition, there appeared to be no clear need to refer her without stabilizing to a distant hospital with an intrauterine fetal death. The lack of a (fully functional) intensive care unit in the primary and general hospital might also have affected her management.

Recommendations of the CEMD Committee to improve care

Based on the reviews, the CEMD committee identified a range of recommendations listed below for improvement in obstetric care and to prevent future maternal deaths.