Evaluation of Maternal Death Surveillance and Response System in Dewachefa District, Oromia Zone, Amhara Region, Ethiopia, 2018

Maternal death surveillance and response (MDSR) is the “litmus test” of the health system that provides evidence for accomplishment, and provides information in real time and allows improvement towards catching all maternal mortalities. The aim of study was to evaluate maternal death surveillance and response system in Dewachefa. A cross sectional study design was conducted in two health centers, ve health post, district health oce and from these facilities 32 health workers were included. Data were collected through focal person, health worker and health extension worker interview by using checklist. Collected data were entered into Epi data version 3.1. These data were exported to statistical package for social science for analysis. Analyzed data were presented in the form of text, table and gures.


Introduction
Maternal Death Surveillance and Response (MDSR) is a system of uninterrupted investigation that connects health information to quality enrichment from grass root level to countrywide, and that measures a program's ability to answer to women's health requirements, mainly throughout and after pregnancy and birth (1)(2)(3). MDSR also provides evidence for accomplishment, links activities to results, makes maternal death visible at all levels, informs communities & health workers, increases country ownership of data, provides information in real time and allows improvement towards catching all maternal mortalities (1,4,5).
Routine identi cation, noti cation, quanti cation, and determination of causes and avoidability of all maternal deaths are the function of MDSR and that will prevent future deaths (6, 7). In 2012, the World Health Organization (WHO) and partners introduced the Maternal Death Surveillance and Response (MDSR) approach as a new method to maternal death review (8,9).
Maternal death is the death of a woman while pregnant or within 42 days of the termination of pregnancy, irrespective of the duration and site of the pregnancy, from any cause related to or aggravated by the pregnancy or its management but not from accidental or incidental causes (1,10,11) A suspected maternal death is the death of any woman while pregnant or within 42 days of the termination of pregnancy, irrespective of causes (10,11).
Probable maternal death is death of a woman of reproductive age (between 15-49 years of age) (11).
Globally, an estimated 287 000 women die each year as a result of pregnancy and childbirth, and about 99% of them were in developing countries (8). In sub-Saharan Africa, the life time risk of maternal mortality is more than 47 times greater than for those in the developed countries (12,13).
In Ethiopia, about 13,000 women died from pregnancy related conditions in 2013, making the country maternal death ratio 420/100,000 live births and providing approximately 4% to the total maternal deaths (7). The place of death revealed that the high percentage of deaths occurred in facilities, with 72% deaths taking place in hospitals or health centers, although this is about to be an object of better facility-based reporting. Deaths in travel accounted for 13% of deaths, while14% of deaths were occurred at home (14).
Even where resources were limited, almost all maternal deaths are preventable and could be eliminated but, it needs the right kind of information for decision making (16). Observed evidence is very vital in the combat alongside maternal death where exact context established interventions are important to meaningfully reduced maternal mortality (17). However, the correct burden of maternal deaths has been unidenti ed and unavailable, and assessment of maternal mortality needs a large countrywide survey which is unaffordable for countries (18). conducting MDSR evaluation at local level is the rst step to recognize where action is required (16). It is likely to establish a structure to measure the burden of maternal mortality (12).
Most countries with high maternal death have weak public registration systems (5). Therefore, numerous maternal deaths and the explanations overdue these deaths remain unregistered and unreported, mainly when women die at home (19).
If accurate information is available, preventing maternal deaths can be achievable to provide targeted actions (1). MDSR makes each maternal death a noti able event, and con rms that communities and facilities report and respond to end preventable maternal deaths (19). Despite the importance of MDSR in producing up to date information for decision making, evidence on the process and implementation is lacking (20).
Dewachefa district launched MDSR in 2013, but the implementation of MDSR was not evaluated yet (21).
The aim of this study was to evaluate MDSR system and to design locally relevant interventions based on the study ndings for government and other stakeholders in Dewachefa district.

Study design and period
Cross sectional study design was conducted from January 1-30/ 2018 in Dewachefa district, North-East of Ethiopia.

Study setting
Dewachefa district is located about 325 Kilometers from Addis Ababa (the capital city of Ethiopia) and 555 Kilometers from Bahir Dar (the capital city of Amhara region) at an altitude of 1623 meter to 2570 meter above sea level. The district is bordered by Majetie district in the South, South Wello in the West, Dewi Harewa in the Northwest, Artuma fursi district in the Northeast and the East. The area of the district is 782.22 square kilometer with the total population 151645. The district is governmentally alienated into 26 kebeles. There are 7 health centers (health center is the primary health care units that provides preventive and curative health care services) and 26 health posts (health post is the subset of health centers that provides preventive health care services at the community level) in the district (21).

Source population
The source population for this study were all governmental health facilities, and health extension workers (who are certi ed by government and working at community levels or health post) and health professionals (public health o cers, Nurses and midwives) who have been working in governmental health facilities in Dewachefa district.

Study population
Dewachefa health o ce, Dulcha and Weldi health centers, and Teref, Tochie, Weldi, Gerbi and Kelo health posts, and health extension workers and health professions in this health facilities were study population. Maternal, youth and Public health emergency o cers who are responsible for MDSR data collection from lower levels and they compile, analyze and propose response plan. Public health o cers, nurses and midwives who are responsible for identi cation, noti cation, reviewing health center maternal deaths and developing action plan. Health extension workers who are responsible for identi cation, noti cation and conducting verbal autopsy for community maternal deaths were included in the study. All (26 weekly reports) for 2017 were also reviewed by 3 eld Epidemiology residents in all health facilities. Those health extension workers and health professionals who were not present during data collection due to annual leave, medical and social problem were excluded.

Sample size determination and sampling technique
We assumed that 79.7% of health workers used the data that were generated by the surveillance system (5). We have used the single population formula to calculate the sample size. Where: n= Sample size [where population> 10,000] Z= Normal deviation at the desired con dence interval. In this case it was taken at 95%, Z value at 95% is 1.96 P= Proportion of the population with the desired characteristic (p=79.7%) d= Degree of precision; was taken to be 5%.
We have used Population Correction formula because the total number of study population was 32. Where: nf= The desired sample size for population <10,000 n= the calculated sample size N= the total population (N= 32) There was not the need of sampling for 32. So, the total sample size that was included in this study was 32.
Dewachefa district was purposively selected because the surveillance evaluation was not conducted before and more than half of Zonal maternal death was occurred in this district. Among the total health centers in this district, 2(25%) of the district health centers were included in the study. These health centers were selected by lottery method of simple random sampling. From these health centers; all health posts were involved in this evaluation. Equally, all health professionals in the health centers and all health extension workers in the health posts were invited to be included in the study.

Data collection tools
Standard data collection tools were prepared from US Centers for Disease Control and Prevention guidelines for surveillance system evaluation in 2013 (1) and National guideline for MDSR (9,11). The tool consisted of two separated Questionnaires (i.e. for health extension workers and health professionals) and checklists for woreda (the middle administrative structure that owned its budget) health o ce, health centers and health posts.
Both questionnaires of health extension workers and health professionals consisted of sociodemographic, knowledge and attitude related questions toward MDSR.
Pre-testing of data collection tools Collection tools were checked in Artuma Fursi district, because the district provided a similar setting with the area under study. The orders of the questions in the questionnaire was changed so that the questions followed a logical sequence that made meaning to the participants.

Data collection techniques
Structured self-administered technique was used to collect data from health workers and health extension workers to assess their sociodemographic, knowledge and attitude towards the surveillance attributes and operations of the MDSR system (1,9).. A checklist, using MDSR guidelines on surveillance system evaluation was used to assess the stability of the system and completeness of retrospective record review of the report of 31weeks (9,11). MDSR identi cation, noti cation and weekly and casebased report formats were reviewed to check for data quality, completeness (If all components of the report and noti cation form were lled then we assigned 1 point and, for incomplete variable, we gave 0 point.) and timeliness of the system. In addition to these, identi cation, noti cation, reviewing and response plan for maternal death were assessed.

Data analysis
Epidata versions 3.1 was used to enter collected data. Entered data were exported into statistical package for social science (SPSS) version 20 for data clearness and analysis. The knowledge related questions were coded on a true/false basis. A correct answer was assigned 1 point and an incorrect/unknown answer was assigned 0 point. The total knowledge was computed in SPSS. The median of knowledge was determined. Then variable was categorized as knowledgeable and non-knowledgeable based on the median. Frequency mean, median and proportion were computed. The ndings of this evaluation were stated in the form of text and tables.

Attributes of the surveillance system
Simplicity is an easiness of a surveillance system as both its structure and implementation while quiet meeting their aims. This attribute was evaluated by assessing the training status of the implementers and determining the experience of the implementers of ever lling MDSR form.
Acceptability is the preparedness of persons and organizations to take part in a surveillance system. Health care workers were asked whether they were ready to remain participating in the MDSR. As well as completeness and timeliness were also be assessed as a substitution of acceptability.
Usefulness is ability to use maternal mortality data to implement changes that leads to maternal care and mortality reduction Sensitivity; The sensitivity of a surveillance system can be evaluated by the percentage of cases identi ed by the surveillance system. This attribute was considered by asking main respondents the number of maternal deaths that were picked by the MDSR system, through verifying whether maternal deaths were correctly classi ed Flexibility is the ease with which system can integrate another disease or event with little or no additional resources Stability is the consistency and availability of the system. Consistency is the capability to collect, manage and provide data properly without failure. Availability is the capacity of a surveillance system to be functional when it is required. Stability of the MDSR was evaluated by examining for consistency in reporting, availability of communication apparatus and other material resources needed for the surveillance system.
Data quality The quality of data is prejudiced by the clearness of surveillance forms, the quality of training and the observation of persons who complete the maternal death noti cation forms and the amount of care that is practiced in handling the surveillance data. A review of these structures of a surveillance system provided an unintended measure of the quality of data. Maternal death noti cation form (MDNF) were revised to check for completeness of the noti cation forms.
Timeliness states as the speed at which data is communicated between different levels in the surveillance system. It was measured by checking whether MDNF are completed within seven days of a maternal death and are then sent to the district health o ce within 14 days of the maternal days as restricted.
Operational de nitions Satisfactory Knowledge: Those respondents who score median or above median score of MDSR related knowledge questions Positive attitude: Those respondents who score above median score of attitude assessing questions toward MDSR.

Result
Socio demographic characteristic of Health professions and Health extension workers Twenty-two health professionals and 10 health extension workers were participated in this evaluation. More than three-fourth,17 (77.3.1%) of health professionals were male whereas all health extension workers were female. The median age of health professionals was 28. Regarding to professional background of health workers, 12 (54.5%) of them were nurses. The mean years of experience of health professionals and health extension workers were 8 and 7.3 years respectively (Table 1). Thirteen (59.1%) of health professionals had unsatisfactory knowledge on MDSR (Table 2). Three-fth, 6(60%) of the health extension workers did not de ne the probable case de nition of maternal death. Regarding to the overall knowledge of MDSR, 6(60%) of health extension workers had unsatisfactory knowledge (Table 3). Surveillance attributes Simplicity Fourteen (63.6%) of health professionals had negative attitude toward simplicity of the surveillance system whereas 6(60%) of health extension workers had positive attitude towards simplicity of the surveillance system. All health centers surveillance focal person and woreda Public health emergency management (PHEM) focal person agree the simplicity of the system (Table 4).

Flexibility
Thirteen (59.1%) and 5(50%) of health professionals and health extension workers believe that system adopts to the user improvement demands respectively (Table 4).

Acceptability
Thirteen (59.1%) of health professionals had positive attitude toward the acceptability of the system. Among the total visited health posts, 4(80%) health posts were not reported suspected maternal deaths.
Verbal autopsy was not done by all visited health posts. All visited health centers focal persons accepted the system. All visited Health facility and woreda health o ce sent weekly report to the next level by hard copy. Reported data were not analyzed at woreda and visited health facilities. Rapid response team was not functional at woreda and visited health facilities. Action for response was not developed in all visited health facilities and woreda health o ce (Table 4).

Predictive positive value
Sixteen (72.7%) of health professionals and 4(40%) health extension workers believe that death reported in this system are actually maternal deaths respectively. Positive predictive value was not calculated because 12 suspected maternal death were not identi ed and noti ed by health centers (Table 4).

Sensitivity
Half, 11 (50%) of health professionals and 7(70%) health extension workers had positive attitude toward the sensitivity of the system. In 2017, three maternal deaths were identi ed and noti ed to Woreda PHEM unit whereas 5 maternal deaths were reported through health management information system. The sensitivity of this surveillance was 3/5(60%) ( Table 4).

Representativeness
More than half, 12 (54.5%) of health professionals and 7(70%) health extension workers perceived that maternal death report represents the situation in the facility/community. Twelve suspected maternal death were not noti ed from the community in 2017. The representativeness of the system in MNCH unit was 5/17(29.4%). In PHEM unit its representativeness was 3/17 17.6% (Table 4).

Timeliness
More than half,13 (54.5%) of the health professionals agree with MDSR data is always ready when we need it for planning purposes whereas 7(70%) of health extension workers did not agree for the timeliness of MDSR data. In 2017, all maternal deaths were noti ed after 8 day of death. The case-based report forms of 4 maternal deaths were sent to the woreda after 1 month of death. One case-based report format was reported from health facility to woreda after 3 months (Table 4).

Completeness
The revised weekly reports were 31 for all visited health facilities and woreda health o ce. Among the total revised weekly report form, the report forms left black were 7(22.6%). The report completeness of the woreda was 19/26(73.1%) .
Cost Twelve (54.5%) of health agree with the system is not costly as compared to the current bene ts we gain from it. The mean annual nancial cost of the system could be 55859 Ethiopian birrs in each health facilities and woreda health o ce (Table 4).

Usefulness
Twelve (54.5%) of health professionals said current system have an ability to estimate maternal death in the facility/community, show the trend of maternal death in the facility/community, the progress and effect of preventive and control methods applied against maternal death and indicate major causes of maternal deaths in the health facility/community (Table 4).  Table 2).
All visited health posts were not using revised PHEM tools for maternal death surveillance and response whereas all visited health centers and woreda health o ce were using revised PHEM tools.

Discussion
The overall knowledge of health professionals and health extension workers on MDSR were 40.9% and 40% respectively. This is lower than the study conducted in Zimbabwe 50% (5). This difference might be due to training provision so that much training could be given to health care providers of the participants of previous study. Another justi cation might be due to internet access increment. Now, this internet access might increase reading habit which may impose to increase knowledge of health professionals and health extension workers on MDSR.
The average completeness of weekly report form of the woreda was 77.4%. This is consistent with the study conducted Zimbabwe 79% (22) and in Mutare district, Zimbabwe76% (5).
The timeliness of this surveillance system was poor because all reports of maternal deaths were sent after 8 days of maternal death from health facility to woreda o ce. In all Health facilities, the case based form of maternal death was reported after 2 months which is greatly discrepancy with the standard FMOH report which is scheduled with 48 hours and another study conducted in Ethiopia (23).
The possible justi cation might be due to fear of taking responsibility of maternal death by health professionals and health extension workers (4,23).
Maternal death surveillance and response system was not representative because 12 suspected maternal death were not reported from health post/ community. This is greatly contradicted with the FMOH and the study conducted in Cambodia (24). MDSR which recommends all suspected maternal death report from community/health post (9). This might be due to unavailability of noti cation form in health posts, and weak supervision and feedback system.
The sensitivity of this surveillance system was 17.6%. This is discrepancy with the study conducted in Cameron 42.9% (2). This might be due to unavailability of maternal identi cation and noti cation form at the health posts, absence of feedback and supervision system.
This system is unstable because all visited health posts were not using revised PHEM tools for maternal death surveillance and response, and there was not established feedback and regular supervision system. This is contradicted with the FMOH target which recommends all facilities have revised PHEM tools (9) and the study conducted in Zimbabwe (5). This might due to weak coordination between woreda PHEM and MNCH units.
Woreda health o ce PHEM and MNCH unit had delegated o cers. The visited health centers had PHEM and MNCH focal persons. This is consistent with the target of FMOH 100% (9)]. The proportion of woreda PHEM and MNCH unit o cer, health center focal person training was 100%. This is consistent with the target of FMOH 100% (9).

Conclusion And Recommendation
The overall knowledge of Health professionals and health extension workers on MDSR was fair. The surveillance system is not sensitive, timeliness and representative. The System is not sustainable/ not standardized.
More work should be needed to improve the sensitivity, representativeness, timeliness and sustainable of the surveillance system.

Limitation
This study cannot show factors associated with surveillance evaluation of MDSR because it is descriptive data analysis.