Background
Maternal mortality is a measure of a woman’s risk of dying during pregnancy, in childbirth or during the 42 days following delivery. It is a tragedy, as no woman should die giving birth [
1]. In 2013, the number of maternal deaths worldwide was estimated at 292,982 [
1]. In addition, there are nearly 6 million perinatal deaths (stillbirths and early neonatal deaths) that occur each year worldwide. It is in the poor countries of Africa and Asia that the risk of these deaths is highest [
2].
The causes of maternal deaths are generally known [
1]. Antepartum and post-partum haemorrhage, obstructed labor, severe pre-eclampsia or eclampsia, complications related to abortion, uterine rupture and postpartum sepsis are the direct obstetric complications (DOC) that account for more than 80% of maternal deaths [
1,
3]. Complications from preterm births, intrapartum-related disorders or birth asphyxia, and infections are the main causes of perinatal deaths in several sub-Saharan African countries [
4].
The global strategy for women’s, children’s, and adolescents’ health (2016–2030) is aligned with the Sustainable Development Goals (SDG) [
5]. SDG 3.1 and 3.2 are devoted to maternal and child mortality. The targets associated with these indicators were not achieved in 2015 in the framework of the Millennium Development Goals. To achieve by 2030 the targets of SDG 3.1 – reduce global maternal mortality to fewer than 70 deaths per 100,000 live births – and SDG 3.2 – decrease newborn mortality to no more than 12 deaths per 1000 live births in all countries [
6]– it is necessary to improve coverage and utilization of evidence-based interventions.
The effectiveness of emergency obstetric and neonatal care (EmONC) in reducing these mortality rates is proven [
7‐
10]. The nine interventions constituting the EmONC services package are called EmONC signal functions. They were chosen based on their effectiveness in addressing the major causes of maternal mortality and most causes of perinatal mortality (fetal distress and respiratory distress) [
10]. The availability and use of EmONC can reduce maternal mortality by 85% and perinatal mortality by more than 75%, which makes it the most effective service package for directly improving maternal and neonatal prognoses [
11‐
14]. Nevertheless, while the effectiveness of this healthcare package is known and proven, its use and quality remain matters for concern, especially within developing countries. In Africa, where maternal and perinatal mortalities are the highest, less than 30% of women who present with obstetric complications are admitted to healthcare facilities providing EmONC [
15‐
21].
In the Democratic Republic of the Congo (DRC), where it is estimated that 693 maternal deaths occur for every 100,000 live births [
22] and 72 perinatal deaths occur for every 1000 births per year [
3], we know little about the availability, use and quality of EmONC. We know that more than 70% of women give birth with skilled attendance [
23]. We also know that EmONC functions should be integrated into the package of maternal-child health (MCH) care. Thus, according to national guidelines, Basic emergency obstetric and neonatal care (BEmONC) should be offered by all health facilities independent of their level in the health system, while Comprehensive emergency obstetric and neonatal care (CEmONC) constitutes a package reserved for reference facilities like reference hospitals and hospital centers [
24], but at both the national and subnational levels, there is little information about the availability, use, and quality of these services.
In fact, the maternal mortality ratio (MMR) at the national level, evaluated every 5 years, gives no information about disparities between provinces or between urban and rural areas [
25,
26]. The data provided by the national sanitary information system (NSIS), in addition to being frequently inconsistent, give little information about the MMR (due to underreporting of maternal deaths) or use of MCH services. They are also too lacking in completeness to be used to track progress toward reduction of maternal and perinatal mortality [
27]. In terms of monitoring progress in the reduction of maternal and perinatal mortality, EmONC has the advantage of being evaluable even in small units –like the health zone (HZ) – and at short intervals, which permits the capture, even at the subnational level, of those processes which can help avoid these deaths [
10]. Indicators of EmONC are key to orienting the strategies that are put in place to effectively reduce maternal and perinatal mortalities [
10].
In the city of Lubumbashi, the second most populous in DRC, the availability, use and quality of EmONC services are not known [
27]. This study was initiated with the aim of assessing these elements in Lubumbashi.
Discussion
The EmONC functions are a service package proven to have an impact on the reduction of maternal and perinatal mortality [
3,
10‐
13,
34]. In Lubumbashi, availability of EmONC is limited. This study found one health care facility providing the nine signal functions of comprehensive EmONC for 918,819 inhabitants. Apart from the tertiary hospital (Hospital Sendwe), no other facility provided all seven signal functions of BEmONC. This falls short of WHO’s recommended minimum, which indicates that there should be five EmONC facilities and at least one facility providing CEmONC for a population of this size [
10].
What explains this low EmONC distribution despite this being a mostly urban health district [
29] is that most health facilities do not perform vacuum extractions, since some providers consider them dangerous. If signal function 6, assisted vaginal delivery using vacuum (ventouse) or forceps extraction, was not taken into account in our assessment of EmONC availability, our results would show five facilities providing CEmONC for 500,000 inhabitants, which meets the WHO recommendation. In this study, we observed that several HCs that were willing to perform vacuum extraction instead carried out Cesarean sections, even though it was not within their scope of practice. The fact that certain health care facilities –particularly private ones– that are considered HCs, that is, the lowest level of service delivery, perform Caesareans poses a problem, on the one hand, in tracking the availability and quality of EmONC functions, and on the other hand, in terms of management of information available in the health care system. In fact, with the disengagement of the state from the financing of health care, the private sector has registered more and more facilities, to the point that more than 60% of health care facilities in Lubumbashi are private [
30]. The weak regulation of health care provision and the medicalization of services at primary health care centers has engendered an overlap of technical services, and competition between the primary and secondary and between the secondary and tertiary levels. The results of the present study show that an evaluation of EmONC based on the presumption that HC offer only BEmONC while only GRH offer CEmONC is insufficient to determine exactly which EmONC functions are provided. In the context of a health care system dominated by the private sector and little regulated, as is the case in Lubumbashi [
30], in the context of evaluating availability, use, and quality of EmONC, all the EmONC signal functions – especially cesarean, which is often performed at primary-care facilities for the economic survival of personnel – should be systematically researched in private-sector facilities that handle deliveries, regardless of their status in the health care system hierarchy (HC, HoC, or GRH).
This observation implies the reconsideration of the role of primary and secondary facilities vis-à-vis the tertiary GRHs in terms of collaboration, complementarity, and management of information in the HZ. However, in an environment where caesarean section is less available, assisted vaginal delivery becomes all the more important. This is the case in the city of Lubumbashi, where dystocia, certain forms of which can be managed using the ventouse, represents 40% of all delivery complications.
When assessing the availability of supplies and equipment for the management of obstetric complications, we found that 28.3% of the facilities had ventouse devices (Table
1). However, the fact that these were not being used despite there being an indication for it, raises the issue of staff training and of the decisions that the maternity staff make when it comes to the choice of procedures. The reluctance of health care staff to use assisted vaginal delivery for fear of maternal and neonatal complications is unjustified. It reveals a lack of provider training and experience, both of which are determinants of the outcome of vacuum- or forceps-assisted delivery [
16,
19]. In Lubumbashi, this lack of training is one of the main obstacles to the introduction of vacuum extraction by ventouse as the tool of choice for instrument-assisted delivery. This is due to the lack of attention given to this technique in training of doctors, midwives, and nurses, both in their schooling and on the job, and suggests that programs to train providers in vacuum extraction by ventouse should be developed in the context of the internship or on the job. Professional organizations should be mobilized to ensure this training, and content should be based in part on hands-on simulation. Adoption of vacuum extraction by ventouse as a first-line intervention should be encouraged only after a minimum standard of training has been attained.
The observation that more health facilities offer signal functions 8–9 (blood transfusions and caesarean sections) than either BEmONC or CEmONC supports the finding that some services are preferred due to their financial returns [
35,
36]. Similar observations have been reported by Lawn et al. in West Africa [
3]. However, the fact that not using the ventouse or forceps prevents a facility from being considered as offering EmONC services raises the question of the applicability of this function. In the case of certain forms of dystocia, for example, the use of forceps or the ventouse is one alternative. But more often than not, in facilities where it is possible to do a cesarean, staff prefer this over extraction by ventouse. In this case, for the same indication, staff have the choice between the cesarean and the ventouse. This choice automatically affects the provision of EmONC functions insofar as the two will not be used simultaneously. This consideration suggests that if extraction by ventouse, forceps, and cesarean are, in certain circumstances, alternatives, then one might well consider that EmONC is available in Lubumbashi, but one alternative, cesarean section, is preferred over the others.
Thus, staff training and the regulation of how obstetric and neonatal care is provided are both essential features which, if improved, would help ensure that women and newborns obtain health interventions in the most cost-effective way.
Regarding the use of EmONC, we noted that Lubumbashi is dotted with many health facilities, but only a small proportion of facility births (7.6%) take place in health facilities providing CEmONC, while 92.4% take place in facilities that offer some, but not all, EmONC signal functions. In Table
3, more than 90.0% of health facilities claimed to have referred complicated cases to a higher-level facility. While this information might seem to imply that obstetric complications are appropriately managed within the health system, information is still lacking concerning the proportion of cases referred relative to the number of patients admitted to the facility, and concerning the time it takes the staff within the lower-level facility to refer their complicated cases. Very often, the referral is delayed due to false optimism and indecisiveness within these lower-level health facilities, or due to unavailable or inappropriate transportation [
20]. Even though the facilities report that they refer, in this context, this cannot be understood to imply all the necessary conditions (transport, accompaniment by a health care provider and a referral note); referral is rather a decision informing the family to go to such and such health facility, sometimes without precise information. Women are often not accompanied and there is often no transport. Given that she is not accompanied by a staff member from the referring facility, the decision to go or not to the indicated facility depends on the woman or her family. It is therefore not guaranteed that the woman who is referred to
x facility will arrive there. Going there or not depends on her ability to find transport and pay the presumed costs of care, and on her perception of the quality of care in the facility.
As indicated by WHO [
10], all women who had obstetric complications delivered in facilities that offered at least one EmONC signal function during the 3 months that preceded our survey. However, this indicator is not very appropriate, since admission to a health facility following a DOC does not necessarily guarantee satisfactory care. The word “satisfactory” here refers to the principle that care is respectful, but also that the time between the arrival of a woman in a facility and her treatment is short. Such timeliness is only possible if skilled staff is present, and if the appropriate supplies and equipment are available. However, since several health facilities in Lubumbashi do not have the necessary emergency kits for the management of complicated cases, and rely on the families of women to obtain these [
37], it is clear that the delay is inevitably long. This delay could contribute to increased rates of maternal mortality or intrapartum mortality [
10].
The low incidence of Caesarean sections relative to the number of expected births reported in this study is another indicator that EmONC is not frequently used by providers to care for women who present with complications. While this incidence is expected to range from 5 to 15% [
10], we found a rate of 3% among the women cared for in the surveyed health care facilities. In the context of Lubumbashi, this low rate of cesareans can be explained in several ways; the most common explanation put forward by women is the high cost [
35,
36]. From another point of view, taking into account the proportion of facility births (≈95%) [
36] and the medicalization of first-line facilities in Lubumbashi [
30], this low rate could be explained by the fact that many women for whom cesareans are indicated delivered in facilities not authorized to provide them, but no information about them (women) was mentioned in the official documents of the health facilities [
36,
38].
The quality of care for women who gave birth in the health facility that offered CEmONC was not good. The reported maternal mortality due to DOC (3.9%) was higher than the acceptable level (≤1%). This reflects the lack of supplies, drugs and equipment, as well as the lack of skilled staff. We observed, for example, that out of all health facilities, only the THs possessed the necessary supplies or equipment for the management of obstetric complications. It was indeed in one of these facilities that all of the EmONC functions were offered. We would therefore expect that EmONC would be of a good quality. However, there is a need to distinguish between availability, sufficiency and the quality of supplies and equipment [
3]. At Sendwe Hospital, even if the staff are supposedly skilled, given that supplies such as surgical kits, blood, and intravenous fluids must often be sought by the family of the parturient, the time needed to acquire them contributes to prolonging the considerable delay linked to the overwhelm of the first-line facilities (the so-called third delay), to the means of transport used to reach the health facility (second delay), and the decision, by the woman and/or her family, to seek care (first delay) [
31,
35].
When we look through the causes of reported maternal deaths (Fig.
3) and the interval between admission and death (Fig.
6), we can see that these delays are generally the basis for fatal outcomes. Indeed, according to Filippi [
39], none of these complications would be fatal (in a 12-h timeframe) if they were referred in time, and if the receiving facility was sufficiently equipped and organized to deal with any kind of emergency. Thus, even when the CEmONC functions were provided at Sendwe, we do not know how long after her arrival a woman was cared for.
These considerations are equally valid in explaining intrapartum mortality. We have noted that the number of reported intrapartum deaths and maternal deaths by DOC was higher in the CEmONC facility than in those which provided only some EmONC functions. This difference is explained by the characteristics of referrals to CEmONC facilities. These facilities are often considered a last resort – even for the other GRHs – when complications cannot be managed at the first level of resort. It is thus obvious that the number of deaths would be higher – though still under the acceptable threshold─ in these facilities than in other facilities.
IOCs accounted for only 24.4% of all maternal deaths, but non-intrapartum deaths represented more than half of all perinatal deaths (61.0%) reported in this study. The proportion of non-intrapartum deaths was higher in Sendwe (74.1%) than in facilities that offered only partial EmONC (50.9%) (Table
5). This situation implies that, EmONC alone is not enough; even facilities providing EmONC need to improve the quality of care given to newborns who have problems requiring interventions other than those featured in the EmONC signal functions [
3,
38,
40]. Only urgent neonatal complications are considered in the EmONC signal functions, with others not being taken directly into account [
10]. For example, although prematurity and low birth weight are the leading causes of neonatal and perinatal death, these cannot be prevented or appropriately managed by the health procedures included in the EmONC functions. The primary aim of the EmONC functions is to manage respiratory distress. Although we know that prematurity can be secondary to complications such as eclampsia, severe pre-eclampsia or placenta previa, neither magnesium sulphate, oxytocin nor caesarean sections can prevent neonatal death associated with this complication. Other procedures not accounted for by the EmONC functions can address neonatal mortality from non-emergency causes. According to Lee et al. [
7,
8], thermal protection of the newborn, the use of oxygen therapy and nasogastric tubes for feeding as well as other emergency care associated with the prevention and treatment of severe sepsis of the newborn could prevent over 80.0% of perinatal deaths. In this study, we focused only on EmONC functions; it would be good to take other services into account when assessing the availability, use and quality of neonatal care in general, rather than solely relying on the EmONC signal functions, which focus on neonatal emergency care (resuscitation). Anecdotally, we can report that these other procedures were not always available in the facilities surveyed, despite there being strong indications for their use during the 3 months period covered by our study.
Our observations concerning EmONC availability, use and quality in Lubumbashi are similar to those made in urban or urban–rural centers in several African countries [
16‐
21]. In all these settings, between 2009 and 2015, signal functions 1 and 2 were the most often performed (70–95%) by health facilities. By contrast, removal of retained products, assisted vaginal delivery, cesarean, and transfusion were the least often performed (3–30%) [
17]. Pattinson observed in 12 districts in South Africa that although the provision of EmONC signal functions was higher (≥60%) than that reported in Lubumbashi and in other African settings, vaginal delivery assisted by ventouse was also the function that was least offered both by the health centers and the GRHs (≈60%) [
17]. In this study the authors also observed that only 48% of GRHs had provided all the signal functions. In surveys carried out in Madagascar, Mali, Ethiopia, Malawi, Zambia, Uganda, and other African countries, as in Lubumbashi, the proportion of facilities that had provided all these functions did not exceed 30% (10–30%); similarly, the proportion of deliveries in EmONC facilities did not exceed this percentage (10–30%) [
16‐
21,
41]. The proportion of obstetrical needs met and the rates of cesarean sections reported in these studies were also weak and variable, ranging, for the obstetrical needs met, from 9.6% in Madagascar [
41] to 48% in South Africa [
17], and for cesarean sections, from 1.5 to 9.0%, respectively, in the same countries [
17,
41]. The rate of cesarean section observed in our study is also similar to those reported by Chu [
20] in Lubutu and Masisi in DRC (≈3.0%), Bo in Sierra Léone and in Kabezi in Burundi, where it remains generally lower than the minimum recommended by the WHO [
10].
This study, which involved staff interviews and the analysis of archival data, has some limitations. It is possible that members of staff forgot to mention a particular procedure which was carried out during the period of study. This could have changed our classification of the facilities, which was based on whether or not they had provided EmONC functions. It was to limit this recall bias that we chose to record the provision of EmONC functions during a relatively short 3 months window.
The relationship between EmONC signal functions and maternal indications is likely to have affected the proportion of health care facilities that accomplished a specific EmONC function. For example, concerning the reasons that justified the lack of provision of EmONC, 25.3% of the maternity ward managers acknowledged not having had indications to offer it. This remains debatable. First of all, if functions 1 and 3–9 are offered in the context of complications, function 2 is systematically offered preventively to all women delivering. Not having administered uterotonics on the pretext of lack of indication shows a lack of information due to the fact that these providers have not been trained about AMTSL – even if they did not want to say so. This is also the case for facilities that conducted caesareans, but where the administration of antibiotics was not mentioned among the functions offered (Additional files
1 and
3).
Secondly, for certain types of dystocias, cesarean and vacuum-assisted delivery are two management alternatives; however, the cesarean is preferred by providers. Thirdly, most functions specific to certain complications can also be indispensable for others [
10]. Blood transfusion is a fundamental function in case of hemorrhage due to placental retention, uterine rupture, complications of abortion, or during a cesarean, the management of extra uterine pregnancy and also the severe anemias observed in poor countries [
34], and parenteral antibiotics are similarly offered as infection prevention in case of uterine rupture, after a surgical intervention, complications of abortion, and post-partum hemorrhage after placental retention. This is not the case for magnesium sulfate, which is very specific for eclampsia. These relationships raise the issue of adjustment or not of EmONC functions according to complications.
With regards to the quality of the data collected, only one registry existed within the HCs, and this contained information about the facility’s main activities. In this registry, several columns were added and held information about maternal complications and maternal-foetal outcomes. In other health facilities, complications relating to abortions were sometimes recorded in a general care registry, interspersed with other care procedures. Newborns practically never had their own file, the information instead being written into the mother’s file, which otherwise contained data concerning her stay. The data presented here was extracted from these documents, held by the health facilities. It is therefore possible that we underestimated the use of the EmONC services due to poor recordkeeping. Nevertheless, we attempted to reduce this uncertainty by first liaising with the maternity ward manager so that he or she could confirm the validity of the records, and also by triangulating independent sources of data management tools: i) maternity registers filled out by the maternity ward manager and her colleagues; ii) admission registers that manage the data from the maternity unit using files entered before admission and discharge of the women, on the basis of which the NSIS reports are developed iii) and the prevention of mother-to-child transmission (PMTCT) of HIV registry [
3]. In fact, the health care facilities included in this study are among those in which we have integrated PMTCT activities since 2004. An HIV screening registry of parturients was maintained independently of the maternity register by the facility’s point person for this activity. Given the high quality of these records, this was considered a reference to verify the completeness of the data.
All the reference facilities were included in the study, so the extrapolation applies only to health centers and hospital centers. Given that the extrapolation was performed on the basis of the data from surveyed facilities, it is possible that it doesn’t reflect the situation of HCs and hospital centers that were not surveyed if their characteristics – urban or urban–rural location and private or public sector – were different from those of surveyed facilities. However, to reduce the effect of this bias, we selected facilities while ensuring that the sample was proportional to their location.