Variation in maternal mortality in Sidama National Regional State, southern Ethiopia: A population based cross sectional household survey

PLoS ONE, Volume 18, No. 3 March, Year 2023

Ukuhunyushwa

Introduction Maternal mortality studies conducted at national level do not provide information needed for planning and monitoring health programs at lower administrative levels. The aim of this study was to measure maternal mortality, identify risk factors and district level variations in Sidama National Regional State, southern Ethiopia. Methods A cross sectional population-based survey was carried in households where women reported pregnancy and birth outcomes in the past five years. The study was conducted in the Sidama National Regional State, southern Ethiopia, from July 2019 to May 2020. Multistage cluster sampling technique was employed. The outcome variable of the study was maternal mortality. Complex sample logistic regression analysis was applied to assess variables independently associated with maternal mortality. Results We registered 10602 live births (LB) and 48 maternal deaths yielding the overall maternal mortality ratio (MMR) of 419; 95% CI: 260–577 per 100,000 LB. Aroresa district had the highest MMR with 1142 (95% CI: 693–1591) per 100,000 LB. Leading causes of death were haemorrhage 21 (41%) and eclampsia 10 (27%). Thirty (59%) mothers died during labour or within 24 hours after delivery, 25 (47%) died at home and 17 (38%) at health facility. Mothers who did not have formal education had higher risk of maternal death (AOR: 4.4; 95% CI: 1.7–11.0). The risk of maternal death was higher in districts with low midwife to population ratio (AOR: 2.9; 95% CI: 1.0–8.9). Conclusion The high maternal mortality with district level variations in Sidama Region highlights the importance of improving obstetric care and employing targeted interventions in areas with high mortality rates. Due attention should be given to improving access to female education. Additional midwives have to be trained and deployed to improve maternal health services and consequently save the life of mothers. We used a cross sectional study design employing population-based survey in households that reported pregnancy and birth outcomes in the past five years (July 2014-June 2019). The study was conducted in six woredas (districts): Aleta Chuko, Aleta Wondo, Aroresa, Daela, Hawassa Zuriya and Wondogenet of Sidama National Regional State, southern Ethiopia from July 2019 to May 2020. Sidama National Regional State is one of the 11 regional states in Ethiopia. The region had a population of 4.3 million people in 2020 [20] and administratively divided into 30 rural districts, 6 town administrations and 536 rural kebeles (the smallest administrative structure with average population of 5000). Under the kebele, there are local structures known as limatbudin (administrative unit organized by 40–50 neighbouring households). The region has 18 hospitals (13 primary, 4 general and 1 tertiary), 137 health centres and 553 health posts operated by the government [21]. In the region, there are also 4 hospitals (1 general and 3 primary), 21 speciality and higher clinics, 131 medium clinics and 79 primary clinics run by private owners. The health centres provide basic emergency obstetric and new born care (BEmONC) whereas hospitals are responsible for comprehensive obstetric and new born care (CEmONC) in addition to the BEmONC [22]. All women who experienced pregnancy and birth outcomes in the past five years in Sidama National Regional State were the source population. Women residing in sampled households and who had pregnancy and birth outcomes (live births, stillbirths and neonatal deaths) in the past five years preceding the survey were the study population. Fig 1 shows the sampling strategy of the study. We followed multistage cluster sampling technique to select the study population. Probability sampling technique: the gold standard technique recommended to observe reliable findings (precision) was employed at each sampling stage [23]. In first stage, we listed all the 30 rural districts of the region with unique identification code. Then, we selected 6 districts (20% of the districts) by simple random sampling. At the second stage, we listed all the kebeles in the 6 districts and randomly selected 40 kebeles proportional to the size of the kebeles in the districts. We employed complex sampling technique and used seed number (245987) in statistical package for social science (SPSS) to generate the sample of kebeles. In third stage, we listed all the limatbudins for each of the selected kebele and randomly selected 6 limatbudins from each kebele; altogether 240 limatbudins from the 40 kebeles. To identify a mother who experienced pregnancy and pregnancy outcomes in the past five years, we visited all the households in the selected limatbudins and listed all the households that reported births in the past five years. Finally, we selected 37 households from each limatbudin; which amounts 8880 households in total from 240 limatbudins. Maternal mortality was the outcome measurement of the study. Explanatory variables were: educational level of mother, educational level of husband, road type used to reach the nearest health facility, distance to the nearest health centre, distance to the nearest hospital, occupation of household head, number of births given in past five years, family size, wealth index, hospital to population ratio, health centre to population ratio, doctor to population ratio and midwife to population ratio. The geographic locations of the households, the nearest health centres and the nearest hospitals were mapped with a global positioning system (GPS) receiver by data collectors who visited all the sampled households during data collection. Traveling time by walking to the nearest hospital was assessed by the data collectors based on reports from the respondents. Data on number of hospitals, health centres, doctors and midwives of the sampled and other districts of the region was obtained from Sidama National Regional Health Bureau, Human Resource Department (unpublished). Wealth index was created using 15 household asset variables [18] broadly categorized in five groups: assets owned (radio, mobile phone and motorbike), livestock owned (cattle, horse or mule or donkey and sheep or goat), housing characterstics and utilities (flooring materials, roofing materials, number of rooms used for sleeping, source of drinking water, type of toilet facilities, access to electricity and use of kerosene lamp), cash crop grown and ownership of horse or mule used for transportation. Household utilities and asset variables used for household wealth index creation are presented in S1 Table. Type of road to the nearest health facility was obtained from the report of participant interview. Maternal death. A death of a woman while pregnant or within 42 days of 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; International classification of diseases and related health problems (ICD-10) [24]. Late maternal death. A death of a woman from direct or indirect obstetric causes, more than 42 days but less than one year after termination of pregnancy [24]. Comprehensive maternal death. A grouping that combines both early and late maternal deaths (ICD-11) [25]. Maternal mortality ratio (MMR). Is the number of maternal deaths during a given time period per 100,000 live births during the same time period. Verbal autopsy for maternal health. A method of finding out the medical causes of death and ascertaining factors that may have contributed to the death in women who died outside of a medical facility. The VA consists of interviewing people who know about the events leading to the death such as family members, neighbours and traditional birth attendants [26]. The data was collected from households that reported pregnancy and pregnancy outcomes in the past five years. In a household which did not have maternal death, a mother was interviewed about her pregnancy experiences and household characteristics using interviewer administered questionnaire. When a mother was absent during the initial visit, the data collectors revisited the household the next day. The data was collected by diploma level teachers recruited from each kebele. In a household where maternal death occurred, we interviewed a father or any adult knowledgeable about the death of a mother. The data was obtained through administering VA questions adapted from the WHO manual for maternal death [27]. Two public health officers who were familiar with the language and culture of study area independently conducted the VA interview. The VA interviewers determined the cause of death using pre-coded options of major causes of maternal deaths: bleeding (haemorrhage), fever (sepsis), convulsion (hypertension), prolonged or obstructed labour and including the option of other causes [24]. The questionnaire was developed after reviewing similar studies. Initially, the questionnaire was prepared in English, translated into local language (Sidaamu Afoo) and then back translated to English by another individual. VA interview questions were adapted from the World Health Organization (WHO) VA guideline [27]. We used the WHO ICD-10 guideline for the ascertainment of causes of maternal deaths [24]. Inter-rater agreement between the two VA interviewers while ascertaining the cause of maternal deaths was assessed by kappa statistic. We used the Landis and Koch inter-rater reliability classification to interpret the kappa coefficient: < = 0.4: poor to fair; >0.4-< = 0.6: moderate agreement, >0.6–0.8: substantial agreement and >0.8-high agreement [28]. The computed Kappa statistics test result was Kappa = 0.75 (95% CI: (0.62–0.87) which indicates substantial agreement between the two VA interviewers. Internal consistency of the variables used for wealth index creations was determined using Cronbach’s Alpha reliability statistics which was found 0.54 and the sampling adequacy was assessed by Kaiser-Meyer-Olkin test with test result of 0.64. The data collectors, the supervisors and VA interviewers were given training by the principal investigator. Key terms and concepts were translated into local terms during the training. The questionnaire was pretested in one district not included in the survey. The supervisors followed the data collectors, checked consistency and completeness of the questionnaire on daily basis. The data was double entered and validated using EpiData version 3.1 software (EpiData Association 2000–2021, Denmark). Sample size estimation for the survey was determined based on the following assumptions: MMR of 412/100,000 LB, crude birth rate of 32 per 1000 population and average household size of 4.6 [3]. With the assumption of a MMR of 412 per 100,000 LB, we used design effect of 2 (as the study employed multistage cluster sampling method) and 0.14% precision level to obtain the number of LB needed for this study. The estimated sample was 15879 LB. We wanted to estimate maternal mortality within 0.14 percent point of the true value with 95% confidence. From a population of 100,000 people and assumed crude birth rate of 32 per 1000 people, we would have (32/1000*100000) 3200 LB per year (16000 LB in 5 years). Hence, we expected to observe 66 maternal deaths over five years among 16000 LB with 95% confidence interval of MMR; 412 (324–524) per 100,000 LB [29]. We assumed that two LB would occur in one household over a five-year period [18] and hence 8000 households would be visited to get the 16000 LB. By considering 10% non-response, the final households estimated for the survey were 8800 households. We used OpenEpi software to calculate the sample size (Source Epidemiologic Statistics for Public Health version 3.01, www.OpenEpi.com) [29]. We used Stata version 15 for data analysis (Stata Corp., LLC. College Station, Texas, USA). This study used data obtained through multistage cluster sampling design [30, 31]. To account for the sampling design, we employed complex survey data analysis method with sampling weight adjusted for non-response [30, 32]. The sampling weight was employed to correct for unequal probability of selection so that to produce meaningful estimates which correspond to the population of interest [33]. This study had four sampling units: district, kebele, limatbudin and household. In primary sampling unit, we applied similar sampling weight since the districts were selected with equal probability of selection. However, the kebeles, limatbudin and households were selected with different selection probability at their respective levels and hence we computed the sampling weight for each of them that differ according to their sampling probability. We computed sampling weight adjusted for non-response by using three steps stated below [32]. We initially calculated the sampling weight for each sampling unit. The sampling weight was computed as the inverse of selection probability. Secondly, we adjusted for non-response for each sampling unit. Nonresponse was calculated as the inverse of response rate. Finally, we calculated sampling weight adjusted for non-response by multiplying the inverse of sampling probability (inverse of inclusion probability) with the inverse of response rate at each sampling unit [32]. We also estimated finite population correction (FPC) factor for each sampling unit to adjust for variance estimators as the survey data was sampled from finite population without replacement [34]. The FPC was calculated using the following formula where N is population and n is sample: Principal component analysis (PCA) was computed to create wealth index [35]. We categorized the wealth index using the first principal component with eigenvalue of 2.3 that explained 15.2% of the total variance. We used geographic coordinates of households, the nearest health centres and hospitals to calculate distance between them. We calculated straight-line distances using proximity analysis “generate near table function” in ArcGIS 10.4.1 [36] and exported the data to Stata 15 for further analysis. Walking time to the nearest hospital according to the participants’ report was also used. We did descriptive statistics like mean, proportions and ratios. Chi-square test was computed to test the association between the outcome variable and potential explanatory variables. Complex sample logistic regression analysis was used to measure the effect of explanatory variables with maternal mortality. We carried out both weighted and non-weighted analysis, but reported only weighted analysis. The ethical approval for this study was obtained from institutional review board of Hawassa University College of Medicine and Health Sciences (IRB/015/11) and Regional Ethical Committee of Western Norway (2018/2389/REK vest). Support letter to respective district (woreda) health offices was obtained from Sidama National Regional State Health Bureau (formerly known Sidama Zone Health Department). Letter of permission to respective kebeles was sought from each woreda health office. Informed written (thumb print and signed) consent was obtained from the study participant before interview. Participant identifiers were anonymized during data entry and analysis to maintain confidentiality of the participants.

I-AI Digest

Study Justification:

This study aimed to address the lack of information on maternal mortality at lower administrative levels in Ethiopia. National-level studies do not provide the necessary data for planning and monitoring health programs in specific regions. By conducting a population-based survey in the Sidama National Regional State, the study aimed to measure maternal mortality, identify risk factors, and assess district-level variations. This information is crucial for improving obstetric care and implementing targeted interventions in areas with high mortality rates.

Highlights:

– The study registered 10,602 live births and 48 maternal deaths, resulting in an overall maternal mortality ratio (MMR) of 419 per 100,000 live births.
– Aroresa district had the highest MMR with 1,142 per 100,000 live births.
– The leading causes of maternal death were hemorrhage (41%) and eclampsia (27%).
– The majority of maternal deaths occurred during labor or within 24 hours after delivery, with 47% occurring at home and 38% at health facilities.
– Mothers without formal education had a higher risk of maternal death.
– Districts with low midwife to population ratios had a higher risk of maternal death.

Recommendations:

– Improve obstetric care and access to maternal health services in areas with high maternal mortality rates, particularly in Aroresa district.
– Focus on targeted interventions for preventing and managing hemorrhage and eclampsia, the leading causes of maternal death.
– Increase access to female education to reduce the risk of maternal death.
– Train and deploy additional midwives to improve maternal health services.

Key Role Players:

– Sidama National Regional Health Bureau: Responsible for coordinating and implementing interventions to improve maternal health services.
– District Health Offices: Responsible for implementing interventions at the district level and ensuring access to obstetric care.
– Health Centers and Hospitals: Provide emergency obstetric and newborn care and comprehensive obstetric care, respectively.
– Ministry of Education: Responsible for improving access to female education.

Cost Items for Planning Recommendations:

– Training and deployment of additional midwives.
– Improvement of obstetric care facilities and equipment.
– Education programs to increase access to female education.
– Monitoring and evaluation of interventions.
– Awareness campaigns and community engagement initiatives.

Amandla Obufakazi

The strength of evidence for this abstract is 8 out of 10.
The evidence in the abstract is strong because it provides detailed information about the study design, population, and methods used. The study employed a cross-sectional population-based survey, which is a robust method for measuring maternal mortality. The sample size estimation was based on appropriate assumptions and the study used complex survey data analysis to account for the sampling design. The abstract also presents key findings, such as the overall maternal mortality ratio and district-level variations. The conclusion suggests actionable steps to improve obstetric care and access to education. To improve the evidence, the abstract could include more information about the limitations of the study, such as potential biases or confounding factors. Additionally, providing information about the statistical analyses used and the significance of the findings would further strengthen the evidence.

Abstract

We used a cross sectional study design employing population-based survey in households that reported pregnancy and birth outcomes in the past five years (July 2014-June 2019). The study was conducted in six woredas (districts): Aleta Chuko, Aleta Wondo, Aroresa, Daela, Hawassa Zuriya and Wondogenet of Sidama National Regional State, southern Ethiopia from July 2019 to May 2020. Sidama National Regional State is one of the 11 regional states in Ethiopia. The region had a population of 4.3 million people in 2020 [20] and administratively divided into 30 rural districts, 6 town administrations and 536 rural kebeles (the smallest administrative structure with average population of 5000). Under the kebele, there are local structures known as limatbudin (administrative unit organized by 40–50 neighbouring households). The region has 18 hospitals (13 primary, 4 general and 1 tertiary), 137 health centres and 553 health posts operated by the government [21]. In the region, there are also 4 hospitals (1 general and 3 primary), 21 speciality and higher clinics, 131 medium clinics and 79 primary clinics run by private owners. The health centres provide basic emergency obstetric and new born care (BEmONC) whereas hospitals are responsible for comprehensive obstetric and new born care (CEmONC) in addition to the BEmONC [22]. All women who experienced pregnancy and birth outcomes in the past five years in Sidama National Regional State were the source population. Women residing in sampled households and who had pregnancy and birth outcomes (live births, stillbirths and neonatal deaths) in the past five years preceding the survey were the study population. Fig 1 shows the sampling strategy of the study. We followed multistage cluster sampling technique to select the study population. Probability sampling technique: the gold standard technique recommended to observe reliable findings (precision) was employed at each sampling stage [23]. In first stage, we listed all the 30 rural districts of the region with unique identification code. Then, we selected 6 districts (20% of the districts) by simple random sampling. At the second stage, we listed all the kebeles in the 6 districts and randomly selected 40 kebeles proportional to the size of the kebeles in the districts. We employed complex sampling technique and used seed number (245987) in statistical package for social science (SPSS) to generate the sample of kebeles. In third stage, we listed all the limatbudins for each of the selected kebele and randomly selected 6 limatbudins from each kebele; altogether 240 limatbudins from the 40 kebeles. To identify a mother who experienced pregnancy and pregnancy outcomes in the past five years, we visited all the households in the selected limatbudins and listed all the households that reported births in the past five years. Finally, we selected 37 households from each limatbudin; which amounts 8880 households in total from 240 limatbudins. Maternal mortality was the outcome measurement of the study. Explanatory variables were: educational level of mother, educational level of husband, road type used to reach the nearest health facility, distance to the nearest health centre, distance to the nearest hospital, occupation of household head, number of births given in past five years, family size, wealth index, hospital to population ratio, health centre to population ratio, doctor to population ratio and midwife to population ratio. The geographic locations of the households, the nearest health centres and the nearest hospitals were mapped with a global positioning system (GPS) receiver by data collectors who visited all the sampled households during data collection. Traveling time by walking to the nearest hospital was assessed by the data collectors based on reports from the respondents. Data on number of hospitals, health centres, doctors and midwives of the sampled and other districts of the region was obtained from Sidama National Regional Health Bureau, Human Resource Department (unpublished). Wealth index was created using 15 household asset variables [18] broadly categorized in five groups: assets owned (radio, mobile phone and motorbike), livestock owned (cattle, horse or mule or donkey and sheep or goat), housing characterstics and utilities (flooring materials, roofing materials, number of rooms used for sleeping, source of drinking water, type of toilet facilities, access to electricity and use of kerosene lamp), cash crop grown and ownership of horse or mule used for transportation. Household utilities and asset variables used for household wealth index creation are presented in S1 Table. Type of road to the nearest health facility was obtained from the report of participant interview. Maternal death. A death of a woman while pregnant or within 42 days of 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; International classification of diseases and related health problems (ICD-10) [24]. Late maternal death. A death of a woman from direct or indirect obstetric causes, more than 42 days but less than one year after termination of pregnancy [24]. Comprehensive maternal death. A grouping that combines both early and late maternal deaths (ICD-11) [25]. Maternal mortality ratio (MMR). Is the number of maternal deaths during a given time period per 100,000 live births during the same time period. Verbal autopsy for maternal health. A method of finding out the medical causes of death and ascertaining factors that may have contributed to the death in women who died outside of a medical facility. The VA consists of interviewing people who know about the events leading to the death such as family members, neighbours and traditional birth attendants [26]. The data was collected from households that reported pregnancy and pregnancy outcomes in the past five years. In a household which did not have maternal death, a mother was interviewed about her pregnancy experiences and household characteristics using interviewer administered questionnaire. When a mother was absent during the initial visit, the data collectors revisited the household the next day. The data was collected by diploma level teachers recruited from each kebele. In a household where maternal death occurred, we interviewed a father or any adult knowledgeable about the death of a mother. The data was obtained through administering VA questions adapted from the WHO manual for maternal death [27]. Two public health officers who were familiar with the language and culture of study area independently conducted the VA interview. The VA interviewers determined the cause of death using pre-coded options of major causes of maternal deaths: bleeding (haemorrhage), fever (sepsis), convulsion (hypertension), prolonged or obstructed labour and including the option of other causes [24]. The questionnaire was developed after reviewing similar studies. Initially, the questionnaire was prepared in English, translated into local language (Sidaamu Afoo) and then back translated to English by another individual. VA interview questions were adapted from the World Health Organization (WHO) VA guideline [27]. We used the WHO ICD-10 guideline for the ascertainment of causes of maternal deaths [24]. Inter-rater agreement between the two VA interviewers while ascertaining the cause of maternal deaths was assessed by kappa statistic. We used the Landis and Koch inter-rater reliability classification to interpret the kappa coefficient: 0.4-0.6–0.8: substantial agreement and >0.8-high agreement [28]. The computed Kappa statistics test result was Kappa = 0.75 (95% CI: (0.62–0.87) which indicates substantial agreement between the two VA interviewers. Internal consistency of the variables used for wealth index creations was determined using Cronbach’s Alpha reliability statistics which was found 0.54 and the sampling adequacy was assessed by Kaiser-Meyer-Olkin test with test result of 0.64. The data collectors, the supervisors and VA interviewers were given training by the principal investigator. Key terms and concepts were translated into local terms during the training. The questionnaire was pretested in one district not included in the survey. The supervisors followed the data collectors, checked consistency and completeness of the questionnaire on daily basis. The data was double entered and validated using EpiData version 3.1 software (EpiData Association 2000–2021, Denmark). Sample size estimation for the survey was determined based on the following assumptions: MMR of 412/100,000 LB, crude birth rate of 32 per 1000 population and average household size of 4.6 [3]. With the assumption of a MMR of 412 per 100,000 LB, we used design effect of 2 (as the study employed multistage cluster sampling method) and 0.14% precision level to obtain the number of LB needed for this study. The estimated sample was 15879 LB. We wanted to estimate maternal mortality within 0.14 percent point of the true value with 95% confidence. From a population of 100,000 people and assumed crude birth rate of 32 per 1000 people, we would have (32/1000*100000) 3200 LB per year (16000 LB in 5 years). Hence, we expected to observe 66 maternal deaths over five years among 16000 LB with 95% confidence interval of MMR; 412 (324–524) per 100,000 LB [29]. We assumed that two LB would occur in one household over a five-year period [18] and hence 8000 households would be visited to get the 16000 LB. By considering 10% non-response, the final households estimated for the survey were 8800 households. We used OpenEpi software to calculate the sample size (Source Epidemiologic Statistics for Public Health version 3.01, www.OpenEpi.com) [29]. We used Stata version 15 for data analysis (Stata Corp., LLC. College Station, Texas, USA). This study used data obtained through multistage cluster sampling design [30, 31]. To account for the sampling design, we employed complex survey data analysis method with sampling weight adjusted for non-response [30, 32]. The sampling weight was employed to correct for unequal probability of selection so that to produce meaningful estimates which correspond to the population of interest [33]. This study had four sampling units: district, kebele, limatbudin and household. In primary sampling unit, we applied similar sampling weight since the districts were selected with equal probability of selection. However, the kebeles, limatbudin and households were selected with different selection probability at their respective levels and hence we computed the sampling weight for each of them that differ according to their sampling probability. We computed sampling weight adjusted for non-response by using three steps stated below [32]. We initially calculated the sampling weight for each sampling unit. The sampling weight was computed as the inverse of selection probability. Secondly, we adjusted for non-response for each sampling unit. Nonresponse was calculated as the inverse of response rate. Finally, we calculated sampling weight adjusted for non-response by multiplying the inverse of sampling probability (inverse of inclusion probability) with the inverse of response rate at each sampling unit [32]. We also estimated finite population correction (FPC) factor for each sampling unit to adjust for variance estimators as the survey data was sampled from finite population without replacement [34]. The FPC was calculated using the following formula where N is population and n is sample: Principal component analysis (PCA) was computed to create wealth index [35]. We categorized the wealth index using the first principal component with eigenvalue of 2.3 that explained 15.2% of the total variance. We used geographic coordinates of households, the nearest health centres and hospitals to calculate distance between them. We calculated straight-line distances using proximity analysis “generate near table function” in ArcGIS 10.4.1 [36] and exported the data to Stata 15 for further analysis. Walking time to the nearest hospital according to the participants’ report was also used. We did descriptive statistics like mean, proportions and ratios. Chi-square test was computed to test the association between the outcome variable and potential explanatory variables. Complex sample logistic regression analysis was used to measure the effect of explanatory variables with maternal mortality. We carried out both weighted and non-weighted analysis, but reported only weighted analysis. The ethical approval for this study was obtained from institutional review board of Hawassa University College of Medicine and Health Sciences (IRB/015/11) and Regional Ethical Committee of Western Norway (2018/2389/REK vest). Support letter to respective district (woreda) health offices was obtained from Sidama National Regional State Health Bureau (formerly known Sidama Zone Health Department). Letter of permission to respective kebeles was sought from each woreda health office. Informed written (thumb print and signed) consent was obtained from the study participant before interview. Participant identifiers were anonymized during data entry and analysis to maintain confidentiality of the participants.

Izinto zokwakha

Emisha

Based on the provided information, here are some potential innovations that can be used to improve access to maternal health:

1. Telemedicine: Implementing telemedicine services can help improve access to maternal health by allowing pregnant women to consult with healthcare providers remotely. This can be particularly beneficial for women in remote or underserved areas who may have limited access to healthcare facilities.

2. Mobile health (mHealth) applications: Developing mobile applications that provide information and resources related to maternal health can help educate and empower pregnant women. These apps can provide guidance on prenatal care, nutrition, and warning signs during pregnancy, as well as connect women with healthcare providers and support networks.

3. Community health workers: Training and deploying community health workers can help bridge the gap between healthcare facilities and pregnant women in rural or underserved areas. These workers can provide basic prenatal care, education, and referrals to healthcare facilities when necessary.

4. Transportation services: Improving transportation infrastructure and providing transportation services specifically for pregnant women can help overcome geographical barriers to accessing maternal health services. This can include ambulances, mobile clinics, or partnerships with local transportation providers.

5. Maternal waiting homes: Establishing maternal waiting homes near healthcare facilities can provide a safe and comfortable place for pregnant women to stay during the final weeks of pregnancy. This can ensure that women have timely access to skilled birth attendants and emergency obstetric care when needed.

6. Financial incentives: Implementing financial incentives, such as conditional cash transfers or vouchers, can help incentivize pregnant women to seek and utilize maternal health services. This can help overcome financial barriers and increase access to quality care.

7. Health education and awareness campaigns: Conducting targeted health education and awareness campaigns can help improve knowledge and understanding of maternal health issues among pregnant women and their communities. This can help reduce cultural and social barriers to accessing maternal health services.

8. Strengthening healthcare infrastructure: Investing in and strengthening healthcare infrastructure, including the availability of skilled healthcare providers, medical equipment, and essential supplies, can help ensure that pregnant women have access to quality maternal health services.

It is important to note that the implementation of these innovations should be context-specific and tailored to the local needs and resources of the community.

AI Innovations Description

Based on the information provided, the following recommendation can be developed into an innovation to improve access to maternal health:

1. Strengthen Obstetric Care: Given the high maternal mortality rate and district-level variations, it is crucial to improve obstetric care services. This can be achieved by enhancing the capacity of healthcare facilities to provide comprehensive obstetric and newborn care (CEmONC) and basic emergency obstetric and newborn care (BEmONC). This includes ensuring the availability of skilled healthcare providers, necessary medical equipment, and essential drugs.

2. Targeted Interventions in High Mortality Areas: To address the district-level variations in maternal mortality, targeted interventions should be implemented in areas with high mortality rates. This can involve deploying additional healthcare providers, such as midwives, to these areas to improve access to maternal health services. Additionally, efforts should be made to improve transportation infrastructure and referral systems to ensure timely access to emergency obstetric care.

3. Improve Access to Female Education: The study identified a higher risk of maternal death among mothers without formal education. To address this, it is important to prioritize and invest in female education. This can be done by implementing policies and programs that promote girls’ education, provide scholarships, and create supportive environments for girls to attend and complete school.

4. Training and Deployment of Midwives: The study highlighted the importance of increasing the midwife to population ratio in districts with high maternal mortality rates. To achieve this, it is necessary to train and deploy more midwives to these areas. This can be done through targeted recruitment and training programs that focus on increasing the number of midwives in underserved regions.

5. Strengthen Data Collection and Monitoring: To effectively address maternal health challenges, it is essential to have accurate and up-to-date data. This can be achieved by strengthening data collection systems and monitoring mechanisms. Regular population-based surveys, like the one conducted in this study, can provide valuable information on maternal mortality and risk factors. By regularly collecting and analyzing data, policymakers and healthcare providers can make informed decisions and track progress towards improving maternal health.

Overall, the innovation should focus on improving access to quality obstetric care, targeting interventions in high mortality areas, promoting female education, increasing the number of midwives, and strengthening data collection and monitoring systems. By implementing these recommendations, access to maternal health can be improved, leading to a reduction in maternal mortality rates.

AI Innovations Methodology

Based on the information provided, here are some potential recommendations to improve access to maternal health:

1. Increase access to female education: The study found that mothers without formal education had a higher risk of maternal death. Therefore, improving access to education for women can empower them with knowledge and skills to make informed decisions about their health and seek appropriate maternal healthcare.

2. Train and deploy additional midwives: The study identified a higher risk of maternal death in districts with low midwife to population ratio. Increasing the number of midwives and ensuring their deployment in areas with high mortality rates can improve access to skilled birth attendants and essential obstetric care.

3. Improve obstetric care: The study highlighted the importance of improving obstetric care in areas with high maternal mortality rates. This can be achieved by strengthening healthcare facilities, ensuring availability of essential equipment and supplies, and providing continuous training and support to healthcare providers.

4. Targeted interventions in high-risk areas: The study identified district-level variations in maternal mortality. Implementing targeted interventions in areas with high mortality rates can help address specific challenges and barriers to accessing maternal healthcare.

To simulate the impact of these recommendations on improving access to maternal health, a methodology could include the following steps:

1. Define the indicators: Identify key indicators that reflect access to maternal health, such as maternal mortality ratio, proportion of births attended by skilled birth attendants, and distance to the nearest healthcare facility.

2. Collect baseline data: Gather data on the current status of the indicators in the study area. This can be done through surveys, interviews, and data analysis of existing health records.

3. Define the intervention scenarios: Develop different scenarios based on the recommendations, such as increasing the number of midwives, improving obstetric care facilities, and implementing targeted interventions in high-risk areas.

4. Simulate the impact: Use statistical modeling techniques to simulate the impact of each intervention scenario on the selected indicators. This can involve analyzing the data collected in step 2 and applying appropriate statistical methods, such as regression analysis or mathematical modeling.

5. Compare the results: Compare the simulated outcomes of each intervention scenario to the baseline data to assess the potential impact on improving access to maternal health. This can help identify the most effective interventions and prioritize their implementation.

6. Refine and iterate: Based on the results, refine the intervention scenarios and repeat the simulation process to further optimize the recommendations and assess their potential impact.

By following this methodology, policymakers and healthcare providers can gain insights into the potential impact of different interventions on improving access to maternal health and make informed decisions on resource allocation and program planning.

Ababhali & Co-Ababhali

Statistics:

Authors: 4

Identifiers:

DOI: 10.1371/journal.pone.0272110

Research Areas:

Community Interventions, Environmental, Health System and Policy, Maternal Access, Maternal and Child Health, Noncommunicable Diseases, Quality of Care, Sexual and Reproductive Health, Social Determinants, Workforce

Study Countries:

Ethiopia

Study Design:

Cross Sectional Study, Grounded Theory

Study Approach:

Quantitative

Participants Gender:

Female

Yabelana ngalokhu: