Health system readiness to manage maternal death data and avail evidence for decision-making through the Maternal Death Surveillance System in Ethiopia, 2020

Globally, maternal mortality continues to be a significant public health concern. In 2017, maternal mortality ranged from 462 per 100 000 live births in the least-developed countries to 11 per 100 000 in high-income countries. The report also indicated that low and middle-income countries in Sub-Saharan Africa and Southern Asia make up 86% of global maternal deaths [1, 2]. Based on the Sustainable Development Goals (SDGs) goal for 2030, all countries should reduce MMR by two-thirds of their 2010 baseline level. The average global target is an MMR of less than 70/100 000 live births by 2030. However, accurate information on how many women died, where they died, and why they died is essential yet currently inadequate [3].

Ethiopia has one of the highest Maternal Mortality Rates (MMR) globally, with an estimated maternal mortality ratio of 412 maternal deaths per 100,000 live births [4]. On the path to ending preventable maternal mortality, the WHO guide "Beyond the numbers" was launched in 2004 to encourage not just estimating the number of maternal deaths to know the magnitude of the problem but also to understanding why and where the women died to be able to do something about it [5‐10].

Following the WHO recommendation, the Ministry of Health (MOH) – in Ethiopia established the Maternal Death Review and Response (MDSR), a system to count and investigate all maternal deaths to generate and avail real-time information for decision-making in 2013. Since 2014, the MDSR has been integrated with the national Public Health Emergency Management (PHEM) system. The MDSR system is expected to avail reliable and quality data and information in real-time on the burden, causes, and contributing factors of maternal deaths and preventability of fatalities in the nation, thereby assisting with timely responses for improving maternal in Ethiopia [11, 12]. To date, several promising achievements have been gained. However, one of the challenges in eliminating preventable maternal death in Ethiopia is the absence of information that shows the magnitude and causes of maternal deaths to assist with decision-making regarding the responses [8, 9]. Therefore, this study was conducted to determine the MDSR system implementation status and describe the health system's readiness for generating and availing good quality data and evidence for the decision-making process in Ethiopia.

This national MDSR system evaluation covers the three components of the Maternal Death Surveillance and Response system in Ethiopia's community and health facility health system structure. The scope of this national MDSR system evaluation is shown below (Fig. 1).

This system evaluation addresses the below-indicated evaluation questions.

Ethiopia is the second-most populous nation in Africa, with more than 110 million people (CSA 2015), located in the Horn of Africa. A household's average size is 4.6 people. The population's pyramidal age structure has remained predominantly youthful [10, 13]. The total fertility rate is 4.6 children per woman (2.3 in urban areas and 5.2 in rural areas). The Maternal Mortality Ratio is estimated at 412 per 100,000 live births [3].

According to the health sector transformation plan (HSTP), Ethiopian health services are being restructured into a three-tier system; primary, secondary, and tertiary level of care. The primary level of care includes a primary hospital, health center, and health post. The Primary Health Care Unit comprises a health center (HC) and five satellite health posts (HPs). Availability, accessibility, equity, efficiency, and quality of health services depend on the infrastructure's distribution, functionality, and quality [14]. This evaluation was conducted at the national level and included nine regional states of the nation: Tigray, Afar, Amhara, Oromia, Somali, Benishangul Gumuz, the Southern People's Nations and Nationalities (SNNP), Gambella, and Harari, as well as two city administrations: Addis Ababa and Dire Dawa. This evaluation was conducted from September 2019 to April 2020.

A cross-sectional study assessed selected health facilities (hospitals and health centers) and health posts in all regions and city administrations. A modified version of the WHO framework for evaluating communicable disease surveillance systems and monitoring and evaluation framework for the MDSR system for Ethiopia were used to design and evaluate the system [12, 15].

This study considers the clustering effect units assuming the hierarchical nature of study units. All regions were selected as primary sampling units. Then, the required number of health centers and hospitals was determined at each stratum using stratified random sampling techniques using the types of facilities as stratifying variables. The sample size for this study based on the parameters is computed using \({n}_{0i}\) and \({N}_{i}\) as: \({n}_{i}=\frac{{n}_{0i}*{N}_{i}}{{n}_{oi}+{N}_{i}-1}\), \(i=\mathrm{1,2}\) denoting the sample of health center size or hospital size determined. Thus, the sample size for each stratum is \({n}_{Health centers}=552\) \({n}_{Hospitals}=77\) And the total sample size is computed as \(n={n}_{Health centers}+{n}_{Hospitals}=629\)

The final sample size represents 15% of health centers and 30% of hospitals. Once the total sample was determined, sample allocation was made based on the probability proportional to their size (PPS) for each region using the \(\frac{n}{N}=\frac{{n}_{r}}{{N}_{r}}\), formula. Then the sample allocated for each region was computed with \({n}_{r}=\frac{n}{N}*{N}_{r}\).

During the computation of required samples for regions, the sample size for Dire Dawa and Harari regions for hospitals was equivalent to zero but rounded up to 1 to ensure representation of the regional hospitals, and the final sample size reached 631.

Multistage sampling techniques were used to select health facilities for the study. Available health facilities in the nation were considered during the selection process, and the final list of health facilities for the study was selected using the lottery method. Then one health post under the chosen health center is included in the study.

Standardized quantitative assessment tools designed using ODK with integrated data quality assurance features using android-based mobile tablets were used for data collection. The refinement of data collection tools and procedures was done based on observations from field-level pre-testing. Field epidemiology residents mobilized from universities, national MDSR Technical Working Group (TWG) members, and MDSR officers from regional health bureaus were participated during the data collection and as field team coordinators.

All data collectors were trained for three days on general data collection guidelines and the basic components of the Maternal Death Surveillance and Response System. Each data collection team's implementation was tracked by the central data manager, who also assigned a central survey coordinator. Field supervisors verified the collected data for accuracy and completeness at the field level and then transferred it to a secure central server in the national PHEM data management center.

Data cleaning was carried out during and after the completion of the data collection time and before the final data analysis and information generation process. Data were exported to Stata 14, and descriptive statistics were disaggregated by type of facility.

A total of 400 (77.1%) visited health facilities implementing the MDSR system during the visit. Of all implementing health facilities, 332 (74.6%) and 68 (91.9%) were health centers and hospitals respectively. Furthermore, a total of 264 (71.5%) health posts were implementing the MDSR system during the time of the visit. The geographical distribution of the MDSR system implementing health facilities shows significant variation among different regions in the nation. System implementation at health facilities in the Somali and Afar regions is low. A substantial difference in community-level MDSR system implementation status was also observed among regions in the nation (Table 2, Fig. 3).

Based on this evaluation findings, 279 (69.8%) of MDSR-implementing health facilities have assigned MDSR focal persons to coordinate MDSR-related activities in their institutions. Out of the assigned experts at health facilities, 153 (54.6%) were trained on the MDSR system at least once in the last 24 months. Additionally, 101 (38.5%) HEWs reported that they are trained on MDSR, and out of the trained, 40 (56.3%) received on-the-job orientation or training on community-level MDSR systems in the last 24 months.

Of the total MDSR system-implementing health facilities, 162 (40.5%) and 169 (42.3%) had case definitions and maternal death notification formats, respectively. Besides, experts are assigned to routinely review registration books and client charts for unreported maternal deaths at 224 (56.0%) facilities. Furthermore, about three-fourths (75.5%) of MDSR implementing facilities (both health centers and hospitals) and 314 (94.6%) health centers reported the existence of established mechanisms for identification and formal notification of maternal deaths occurring in their health facilities and from their community respectively (Table 3).

Of the total implementing health facilities, only 37 (11.6%) and 14 (20.6%) MDSR-implementing health centers and hospitals have optimal readiness for conducting maternal death identification and notification process. Among system-implementing health posts, 39 (14.0%) have locally translated community case definitions for maternal death. The vast majority, 231 (87.5%), of the health posts reported the availability of established mechanisms for receiving maternal death reports from the community. Besides, a total of 203 (76.9%) MDSR implementing health posts reported the existence of an established system to screen death of women of reproductive age group occurred in their catchment community (Table 3).

301 (75.1%) and 390 (97.5%) implementing health facilities that archive maternal death reports and have a means of communication for maternal death data reporting to the next level, respectively. Regional variation is also observed in archiving maternal death reports. None of the MDSR system-implementing health facilities in the Afar region had archived maternal death reports during the visit (Table 3).

268 (67.0%) MDSR-implementing health facilities (both health centers and hospitals) assigned experts to oversee the investigation of suspected maternal deaths in their institution, and 158 (39.5%) health centers assigned responsible experts (surveillance focal person) to monitor the investigation status of reported maternal deaths by HEWs from their catchment areas. Additionally, 349 (87.2%) health facilities (both health centers and hospitals) had a maternal death review committee in their respective institutions. Only 42 (12.4%) of the review committee held more than or equal to six death review coordination meetings during 2011 EFY (2019 GC). Among them, 275 (78.7%) are chaired by CEOs or medical directors, followed by surveillance focal persons 54 (15.3%). In addition, only 6 (1.5%) health facility respondents claimed that they were engaged in routine analysis and information generation of maternal death data in their health facilities at least once.

Furthermore, only 26 (6.5%) and 45 (11.3%) facilities reported the availability of finance for MDSR system implementation and functional computers for surveillance in their health facility, which varied across regions. Of all the MDSR system-implementing health facilities, 23 (6.9%) health centers and 18 (26.5%) hospitals have optimal readiness to perform the overall maternal death investigation and review process at the national level. Regarding community-level maternal death investigation status, only 28 (10.6%) health posts had locally translated verbal autopsy format to investigate maternal deaths (Table 4).

This system evaluation profiled the MDSR system performance status at the health facility and community level across all regions and city administrations of Ethiopia. The evidence indicated that the system is not well-equipped to handle the very objective of the surveillance, which was to eliminate preventable maternal death by obtaining and using the information on each maternal death to guide public health action. Even though there may be a long list of surrounding factors for this suboptimal achievement, the lower-level-level commitment of experts assigned to coordinate MDSR-related activities at the health facilities and health facilities' leadership commitment takes the greater portion. Additionally, findings on the community-level MDSR system implementation status uncover a gap in community MDSR systems in capturing and investigating community death. Even though most of the implementing health facilities established a committee in their respective institutions, a considerable number of the committees were not functional, as evidenced by the absence of routinely scheduled meetings to discuss maternal death as per the guidance. Moreover, routine data analysis was low, violating the surveillance's goal and cannot proceed beyond counting deaths.

Even though this system evaluation shows the national MDSR is not working as expected, this system evaluation showed that the national-level MDSR system implementation in Ethiopia is partially satisfactory and encouraging. However, gaps remain, and upgrading the current system offers a unique opportunity to implement the necessary structural changes and exploit the system to its potential.

Specific recommendations categorized under WHO building blocks are provided below.