In 2015, the overall national MMR in all Ethiopian hospitals was 149/100,000 live births. Our findings showed a substantially lower MMRs as compared to the Ethiopian Demographic and Health Survey in 2016 18. Although there are no similar studies in the country, current findings are also lower than that found in a small scale study in a specialized 19, general 26 and rural (for directly admitted women) hospital 27 in Ethiopia. Our estimate is much lower compared to findings of a systematic review of hospital-based studies in other sub-Saharan Africa, where the pooled MMR was 957 per 100,000 live births 28. It is also much lower than the findings of a Nigerian study where MMR was 2085/100,000 live births among eight included hospitals from different regions 29, but still much higher than in high-income countries 30. Relatively lower MMRs in hospitals may reflect the effectiveness of several maternal health interventions such as Maternal Death Review and Reporting system (MDSR), which is being implemented in Ethiopia. MDSR is a health reform used to continuously notify, review, analyse and respond to maternal deaths in order to take action to prevent similar deaths in the future 31.
There was a huge regional variation in the magnitude of MMRs. This is comparable with results of the Ethiopian Demographic and Health Survey 18. Regional variations in MMR were also reported in a Nigerian study and a systematic review in sub-Saharan African countries 28-29. An explanation could be huge disparity in access to healthcare among different regions of Ethiopia. Shortage of clinical skills and unavailability of a strong referral system in some regions could also play a role. Differences in the type of hospitals available in different regions could also account for regional variations 28.
Prevalence of obstetric complications among women admitted in hospitals was very high (23.4%) and surprisingly higher than the 15% WHO estimate 32. Higher prevalence of obstetric complications compared to previous findings could be a consequence of a referral system whereby complicated maternity cases are referred to hospitals. Consistent with an Eritrean study, hypertensive disorders of pregnancy were the leading cause of maternal mortality 33. Similarly, this study revealed that postpartum haemorrhage, prolonged labour and postpartum sepsis were the predominant causes of maternal mortality, supported by several small-scale studies in Ethiopia 19,34 and findings of a systematic review of literature 35.
The current study revealed a nationwide hospital-based CS rate of 20.3%. Although high, this is much lower than a CS rate of 47.6% in Dessie referral hospital, northern Ethiopia 36. It is also lower than a CS rate of 27.6% in Attat Hospital in southern Ethiopia, 37. The rate of CS in Addis Ababa city administration is much higher than the finding of a previous study, where a CS rate of 19.2% was reported 38, although the most recent study reported a comparable CS rate (38.3%) in the city 39. Our study included data from all hospitals, while previous studies used smaller samples and were conducted only in urban centers.
Huge regional variations in CS rates exist in Ethiopia. Somali, Afar and Ganbella regions, where CS rates were substantially low and access to maternal health services inadequate 18. In contrast, CS rates were considerably high in Addis Ababa, Harari and Dire Dawa, which are the urban centers of the country. Our findings were comparable with previous study reports, where Addis Ababa, Harar and Dire Dawa had the highest CS rates and Somali region the lowest 40. In line with our findings, several previous hospital-based studies revealed that CS rates were very high in these urban centers 38,40,41. These regional disparities may be attributed to better access to surgical services with overuse of the procedure and higher prevalence of elective CS in large cities of the country 6.
Type of hospital and available skills to provide CS might explain regional disparities. The majority of hospitals in pastoral regions including Somali and Afar are primary district hospitals, where emergency surgical officers perform CS. In contrast, in major regions such as Addis Ababa, Oromia, Tigray and Amhara, there are several referral teaching hospitals where highly skilled professionals, including obstetricians, perform CS. In these major regions, there is a strong referral linkage among different hospitals because there are many primary, general and referral hospitals as opposed to the pastoral regions. Furthermore, the number of hospitals available in some regions are inadequate to serve the regions’ population. For example, Gambella region has only one hospital.
A weak and statistically non-significant inverse linear association was observed between MMR and CS rate at national level, although there were significant regional disparities. Our findings in these regions are supported by a study in Turkey, where an inverse association was observed between maternal mortality and CS rates 42. This may be explained by the fact that most often CS is performed in response to emergency medical complications that saved lives of most women 43.
Nevertheless, the association between MMR and CS rate was unevenly distributed among the regional states of Ethiopia. The current study revealed that there was a direct association between MMRs and CS rates in Amhara, Benishangul Gumuz, and SNNP regions. Although Tigray region had the lowest MMR and low CS rate, there was statistically significant direct association between the two variables in this region. This may indicates that CS is not the solely option to reduce MMR. Similarly, CS rates were high both in Dire Dawa city administration and in Harari region while MMR was relatively lower in Dire Dawa than Harari region. Although the association between MMRs and CS rates in these regions was direct, this association was statistically not significant. Available evidence indicated that higher rates of CS are associated with lower MMR only when the CS rate is below 10% 17. Therefore, observing higher MMRs in these regions was not surprising because similar to the current findings, previous studies also indicated CS rates in these regions considerably exceeded 10% 36,41,44.
Findings of a previous systematic review conducted in Latin America is also consistent with our findings 10. Esteves-Pereira AP et al. 2016 reported a three-fold increased risk of maternal mortality with caesarean compared to vaginal birth in Brazil 7. In low-income countries, higher risks of death with higher CS rates may be attributed to complications with anaesthesia, lack of surgical skills and postoperative sepsis12,13. Women in low-income countries may not get effective prophylactic antibiotics and rigorous skin preparation before CS 45. This increases the probability of postoperative infection and contributes to higher maternal mortality after CS 13,45.
In the current study, the case fatality rate for women who undergone CS was very high as compared to vaginal births. This is comparable with findings of a systematic review and meta-analysis conducted in low- and middle-income countries where the risk of maternal death in women who had CS was 7.6 per 1000 women (10.9 per 1000 women in sub-Saharan Africa) 46. Although the risk of maternal death is usually higher after CS than after vaginal birth both in low 46 and high income 11,47 countries, the magnitude was greater in our study than in most previous studies. This may be because this study included only hospitals with data where complicated cases are managed while vaginal births can be provided at lower level facilities. Most mothers who experienced obstetric emergencies are referred to hospitals from lower level facilities and arrive in hospitals with advanced conditions. The majority of these women become candidates for CS, indicating higher risk of death after CS than after vaginal births. This evidence indicates a need for further in-depth review of maternal deaths after CS in order to identify where the chain of events started that led to death.11
Strengths and limitations of the study
This study has several strengths. We used national representative data of all hospitals in Ethiopia. Therefore, generalization can be applied to all regions of Ethiopia. These findings may also be useful for other low-income countries with similar demographic and socio-economic characteristics. From the commencement of the survey, experts from national and international partners were involved in data collection and management processes, which enhanced data quality. However, this study suffered from the usual limitation of a cross-sectional study in that causal associations could not be concluded. Since we calculated hospital-based CS rates based on the WHO Statement on Caesarean Section Rates, the current study did not estimate population level CS rates; hence, we are unable to generalize our findings to the general population. Although vaginal births can be attended in health centres that cannot provide CS, women who experience emergency obstetric complications are often referred to hospitals. This is a possible explanation for the high prevalence of obstetric complications and low numbers of maternal deaths after vaginal births. Maternal deaths in this study represent only the numbers, which were registered in the hospitals’ logbooks. Therefore, we were challenged while interpreting our findings due to shortage of similar literature, since the majority of previous studies were population.