Variation of caesarean section rates in Sub-Saharan Africa: A literature review

Ensuring quality access to caesarean section (CS) care is a key millennium development strategy and the next sustainable development goal [1] to reduce maternal and infant mortality. As with any surgical procedure, a CS involves risks and complications, which means that it should be performed in an approved way and should not be performed excessively. Although the optimum CS rate at population level is diffi cult to assess, the World Health Organisation (WHO) recommends that the national CS should not exceed 10 15% [2]; however, in many countries, the CS is rising [3].


Introduction
Ensuring quality access to caesarean section (CS) care is a key millennium development strategy and the next sustainable development goal [1] to reduce maternal and infant mortality. As with any surgical procedure, a CS involves risks and complications, which means that it should be performed in an approved way and should not be performed excessively. Although the optimum CS rate at population level is diffi cult to assess, the World Health Organisation (WHO) recommends that the national CS should not exceed 10 -15% [2]; however, in many countries, the CS is rising [3].
Studies on the relationship between the CS rate and maternal and perinatal mortality and morbidity have concluded contradictory results [4]. In Latin American hospitals, for example, the increase in the CS rate above 10-15% was associated with more premature deliveries and an increase in neonatal mortality [5]. However, in Sub-Saharan Africa, where the average CS rate was 8.8%, the risk of maternal death was lower in hospitals with a higher elective CS rate [6]. Studies on CS have reported that the 10-15% rate represented the optimum result as regards maternal and neonatal mortality [7]. However, these studies were limited by data that was sometimes incomplete. These ecological studies did not consider the individual risks because they were based on data aggregated by different countries.
The goal of our analysis is to document, based on the selection of population studies, the variability in CS rates based on studies conducted in Sub-Saharan Africa.

Material and Methods
The data were selected according to the selection: The type of study on CS, the target population, data analysis using the following PubMed search equations for a period of 7 years, from 2010 to 2017. Inclusion criteria selected articles with frequency of CS in sub-Saharan Africa from 2010 to 2017.

Results
These data were obtained after research on PubMed using key words. 570 citations were found, 544 articles were excluded and 26 studies retained ( and concludes that access to CS is still diffi cult in Benin, and that diagnostic errors and delays are frequent [14]. The study by Makanya V et al. shows an average CS rate of 42.4% and concludes that use of the Robson classifi cation helps reduce the CS rate [15]. The study by Muti M et al. presents an average CS rate of 12.5% and concludes that high blood pressure during pregnancy affects 5 to 8% of pregnant women and that this increases maternal and perinatal mortality and morbidity [16]. The study by Rukewe A et al. shows an average CS rate of 31.1% and concludes that maternal risk increases with general anaesthesia [17]. The study by Ugwa E et al. shows the average CS rate of 17.1% and concludes that there is an absence of signifi cant correlation between the CS rate and perinatal mortality [18] (Graph 2).
The forest plot above shows an overall CS proportion of 19%, with a confi dence interval of 95% [14% -24%]. The I 2 index is equal to 99.92%, suggesting a very high level of heterogeneity. This heterogeneity is associated with the studies that are included in the analysis. The different studies show variable CS rates depending on the indications for a CS, the CS being performed in a rural or urban setting, the population studied, and on whether the CS is an emergency or programmed ( Table  2).
The only variable explaining some of the heterogeneity observed is the impact factor. As the graph below shows (graph 2), the studies published in higher impact journals tend to show a lower CS rate than the articles published in lower impact journals (Graph 2).
This graph shows the link between the impact factor of the journal and the reported proportion of CS. The purpose of this graph is to identify the source of the difference in CS rates reported in the different articles analysed, and we can conclude that the only predictive factor for the difference in the rates reported in the various articles is the impact factor.

Discussion
The objective of this study is to document the variability in CS rates in Sub-Saharan Africa.
The average CS rate of 19% reported in this analysis corresponds with the overall estimate described in the literature (WHO). In 1985, a group of WHO experts concluded that there was "no justifi cation for any region in the world to have a CS rate higher than 10-15%" [34]. This declaration was based on evidence that was limited, but available at that time, on the CS rate (CS) observed in North European countries, which had the lowest maternal and perinatal mortality rate. Over time, this fi gure came to be considered by the international community as the "optimum" CS rate. Since then, these rates have risen in developed and developing countries alike, sometimes reaching very high rates as, for example, in Brazil and the United States [4,35,36]. Our study shows that this CS rate varies depending on the population studied and especially on access to healthcare [21]. Experts and public health offi cials have expressed concern over the unprecedented increase in CS sections and the consequences. However, the validity of the historical declaration of 1985 has sometimes been challenged in light of data accumulated over three decades on the basis of major improvements in obstetric practices, and which could justify a higher CS rate according to certain publications and recommendations [4,37]. Worldwide concern over this uncontrolled increase in CS rate is not unjustifi ed.
In Sub-Saharan Africa, a CS increases the risk of maternal and neonatal morbidity and mortality [16]. Although a CS is an effective technique for preventing maternal and perinatal mortality when performed properly, this technique is not without risk, and is associated with short and long-term consequences [4,38]. Although the CS rate for a population with limited healthcare services measures that population's access to obstetric care, this rate must be effective to save lives. The CS rate has been used as an assessment measure for governments, policymakers and public healthcare professionals when seeking to assess progress in the fi eld of maternal and infant health, the monitoring of obstetric emergencies, and the use of different resources [2]. However, determining the optimum CS rate within a population is not an easy task. Several studies have addressed this issue by analysing the link between the type of delivery and the maternal and infant outcome at population level [7,35]. However, these analyses used different methodologies and led to different interpretations.

Conclusion
The caesarean rate is the one that gives the best outcome as regards foetal-maternal benefi t. This rate can vary depending on the various obstetric problems in the different countries.
The articles published in higher impact journals tend to report a lower caesarean rate than the articles published in lower impact journals. Graph 1: Twenty-six studies were analysed according to the selection criteria.
Graph 2: Caesarean frequency in 26 studies. , and if positive, positive link between the variables (when variable X is higher, variable Y is higher)) 2. Se = standard error = measurement error for each measurement point on the regression line 3. Zval = critical value for this variable 4.Pval = p-value associated with this critical value (if the zval is suffi ciently high, the p-value will be signifi cant). In this case, only the impact factor has a p-value below 0.05. 5. ci.lb and ci.up = 95% confi dence interval lower bound and upper bound for the estimator (point 1). Note the link between the confi dence interval and the p-value: if the confi dence interval contains 0, the P-value is not signifi cant, but if it does not contain 0, the p-value is signifi cant.