Introduction

An important target of Sustainable Development Goals (SDGs), which urges the international community to reduce maternal mortality rate (MMR) to less than 70 per 100,000 live births by 2030, [1] is still far from reach. Despite huge international efforts to reduce its incidence, maternal mortality remains one of the most important public health problems. Even though the global maternal mortality rate fell from 442 to 211 per 100,000 live births within the past twenty years [2], the number of maternal deaths is still high in most developing countries. Sub-Saharan Africa countries bear the highest burden of maternal mortality contributing to 66% of the total maternal deaths reported globally [2].

The high burden of adverse maternal events in some resource-poor areas of the world is a clear manifestation of the inequalities in health service access, and highlights the gap between the rich and poor. Four largely preventable obstetric complications (bleeding, infections, preeclampsia/eclampsia, and unsafe abortion) account for 80% of all maternal deaths [3]. Most maternal deaths are avoidable since there are well-established healthcare solutions to prevent and manage obstetric complications [4]. Yet, majority of maternal deaths that occur in developing countries are largely preventable [5,6].

Information on the determinants of MNM and mortality is crucial for general assessment of resource requirements, appropriate policy formulation, as well as prioritization and effective implementation of interventions. Previously a number of distant (socio-economic and demographic characteristics of women) and intermediate (obstetric history, preexisting medical conditions, and reproductive health characteristics) factors were revealed as determinants of maternal mortality and morbidity [7–10].

The WHO MNM tool [11] recommends investigation of all cases of MNM and maternal death to gain better understanding of the circumstances surrounding maternal death and severe maternal morbidity. This principle is based on the assumption that all maternal deaths experience at least one life threatening condition (organ dysfunction), and therefore are similar to cases of MNM in every aspect except for the ultimate outcome of interest. Even though there are several studies in the literature that focus on the determinants of either maternal mortality [12,13] or MNM, [7,8,14] there is still scarcity of research focusing on potential factors associated with severe maternal outcome (SMO).

Eritrea is one of the sub-Saharan countries with high rate of maternal mortality. Despite this fact, studies on the subject of maternal morbidity and mortality are virtually inexistent in the country. Hence, there is lack of knowledge regarding factors associated with SMO in Eritrea. The aim of this study was, therefore, to assess determinants of SMO among mothers admitted to Keren Regional Referral Hospital.

Methods

Results

During the 3-year period reviewed, 701 (SMO) cases were identified (662 cases of MNM and 39 maternal deaths). We also included 1,402 women who delivered without any notable complications during the same period to be used as controls in this study.

Socio-demographic, obstetric, prenatal and perinatal characteristics of study participants

There was no significant difference between mean age of SMO cases (27.6) and controls (28.2). Larger proportion of women with SMO did not attend antenatal care services compared to controls (p<0.001). Moreover, compared to the control group, cases of SMO tended to have higher parity, have premature delivery, experience abortion, be referred from lower facilities, and reside 30 kilometers or further from the hospital, all of which were statistically significant (Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Distribution of socio-demographic, obstetric, prenatal and perinatal characteristics of women admitted to Keren Regional Referral Hospital.

https://doi.org/10.1371/journal.pone.0299663.t001

Past obstetric and medical characteristics of study participants

The percentage of women with SMO who had history of anemia, caesarean section, and stillbirth was significantly higher than controls. Higher proportion of SMO cases had preexisting medical conditions compared to controls as well; this difference, however, did not reach statistical significance. Slightly higher percentage of women in the control group also had history of abortion compared to the SMO group, although the difference proved to be statistically insignificant (Table 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Distribution of past obstetric and medical characteristics of women admitted to Keren Regional Referral Hospital.

https://doi.org/10.1371/journal.pone.0299663.t002

Determinants of severe maternal outcome

In the bivariate binary logistic regression analysis, we assessed the association of 12 distant and proximate variables with maternal outcome. The following variables were significantly associated with SMO: parity of four or more, not attending ANC, lower gestational age, caesarean section in the current pregnancy, referral from lower level facilities, longer distance from the hospital, history of anemia, previous caesarean section and history of stillbirth. All variables considered in the bivariate analysis were significant at p = 0.25, and therefore, were eligible to be include in the multivariate binary logistic regression analysis.

In the final multivariate logistic regression model, the following variables remained independently associated with SMO: not attending ANC follow up (AOR: 4.54; CI: 3.15–6.54), caesarean section in the current pregnancy (AOR: 3.70; CI: 2.44–5.17), referral from lower level facilities (AOR: 11.71; CI: 9.01–15.23), residing more than 30 kilometers away from the hospital (AOR: 2.93; CI: 2.25–3.80), history of anemia (AOR: 2.38; CI: 1.84–3.07), and previous caesarean section (AOR: 3.33; CI: 2.06–5.38) (Table 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Bivariate and multivariate logistic regression analysis of factors associated with severe maternal outcome among women admitted to Keren Regional Referral Hospital.

https://doi.org/10.1371/journal.pone.0299663.t003

Discussion

In the present study, ANC follow up, caesarean section in the current pregnancy, referral from lower level facilities, proximity of residence to hospital, history of anemia, and history of caesarean section were the major determinants of SMO.

In our study, we found that women who did not attend ANC were more likely to experience SMO. Studies from Ethiopia [7,8,14,17–20] and Brazil [9,10] also reported similar findings. There is overwhelming evidence regarding the effectiveness of ANC in preventing undesirable outcomes of pregnancy [21]. ANC provides opportunity to early detect and treat risk factors such as iron deficiency anemia and hypertensive disorders of pregnancy that can increase the odds of complication during late stage of pregnancy and delivery. Moreover, women who attend ANC tend to have better knowledge of obstetric danger signs [22] and are more likely to give birth at health facilities [23], and therefore have better chances of avoiding the occurrence of first and second delays. This suggests that a substantial number of SMO incidents can be avoided through improvements in ANC services.

Our findings further revealed that Women who underwent caesarean section in the current pregnancy had higher odds of SMO compared to women who had spontaneous vaginal delivery. This result was in line with similar studies from Brazil [24,25] that reported a three-fold increase in the risk of SMO among women who undertook caesarean section. Caesarean section is associated with several risk factors including thromboembolism, postoperative adhesion, hysterectomy, incisional hernia, and wound infections (especially if performed without administering prophylactic antibiotics) [26]. In addition, caesarean section might cause severe hemorrhage that can be threatening to the mother’s life, and may require blood transfusion [27]. However, these findings should be interpreted with caution. Most of the time, women who require caesarean section are those with higher risk pregnancies or severe medical conditions. Hence, it is difficult to distinguish problems caused by the caesarean section itself from the problems caused by the conditions that require it. It has been reported that annually millions of unnecessary caesarean section are performed throughout the world [28]. Therefore, ensuring caesarean section rates stay within the recommended range (15%) [29] through promoting evidence-based practices is crucial.

This study also revealed that women who previously had caesarean section were at higher risk of developing SMO compared to women who had normal delivery in their preceding pregnancies. This finding was in agreement with previous researches from Ethiopia [8,14,19]. Studies have shown that women who had multiple caesarean sections were more likely to have problems with later pregnancies. Previous caesarean section has been recognized as a strong risk factor of uterine rupture [30]. Caesarean section leaves a scar in the uterus that weakens the elastic property of the myometrium, making the uterus easily ruptured. Moreover, the risk of placenta accrete, a potentially life threatening condition, dramatically increases with each caesarean section performed [27,31]. Previous caesarean section is also a strong risk factor for placenta previa, abruption placenta, and hysterectomy [31]. Obstetricians have different opinions on the relative advantages of vaginal birth and caesarean section following a caesarean delivery. However, in accordance with WHO recommendation, the rate of caesarean section should be restrained within the ideal range of 5–15% [29] through efforts to reduce caesarean sections performed for reasons other than medical necessity.

Consistent with previous studies from Ethiopia, the current study also found that women who had history of anemia were more prone to SMO event [7,14]. Anemia increases the mother’s risk to complications of pregnancy including placenta previa, abruption placenta, postpartum hemorrhage, shock, uterine sub-involution, and caesarean delivery morbidity [32]. Anemia also contributes to SMO through its negative impact on immune functions. It has been documented that anemia increases susceptibility or severity of infections [33]. Early detection and treatment of anemia through quality and accessible antenatal care services is imperative to reduce the effect of anemia on adverse maternal outcome.

Among all factors, referral from lower-level health facilities had the strongest link to SMO in this study. Women referred from lower level facilities had more than twelve-fold higher odds of ending up with SMO compared to self-referred women. The scientific literature has documented a number of challenges in the emergency referral system that compromise the chances of survival in women referred due to obstetric complications. Distance from CEmOC facility and poor quality roads are the main challenges faced by referring facilities in transporting women with obstetric complication in most developing countries [34]. Most peripheral health facilities do not have readily available ambulance and will often have to rely on unreliable sources of transportation such as private vehicles [34–36]. Poor communication between the referring and receiving health facilities is another factor that negatively affects the effectiveness of referral system. In most instances, referring facilities fail to inform staff members of the recipient hospital in advance when referring a woman with obstetric complication [35,37]. As a result, valuable time is often lost to make necessary preparations once the woman reaches the hospital. Clinical skill of health workers in the referring facilities could also have profound effect on the efficiency of referral system [37]. Health workers may fail to recognize danger signs on time or may refer the woman without proper resuscitation and stabilization [35]. Effectiveness of the referral system can be enhanced by improving the transportation systems, raising technical skills of health workers, and establishing standard referral protocol [35,38,39].

This study also identified distance from the hospital as a risk factor for SMO. Accordingly, women residing more than 30 kilometers away from the hospital were found to be at higher odds of developing SMO. Similarly, two studies done in Ethiopia [17,20] reported that mothers who travelled 10 Kilometers or more to reach the hospital were twice as likely to develop SMO. In their classic three delays model [40], Maine and Thaddeus have clearly outlined how distance from health facility can cause both first and second delays. Long distances not only hinder reaching healthcare but also discourage women from seeking it in the first place. Women who reside in remote villages often wait until their illness gets serious before they decide to seek care, hopping that it will resolve by itself. The effect of distance is even more catastrophic when coupled with lack of transportation and poor quality roads. In addition, in most developing countries including Eritrea, specialized obstetric facilities are concentrated in the urban centers [40]. This means that women who reside in remote villages will have difficulty to reach the closest health facility, leave alone a CEmOC facility. Ensuring equitable distribution of specialist obstetric care facilities and enhancing maternity waiting home services are the most feasible solutions to mitigate the negative impact of long distance on maternal health outcome.

In the present study, we could not establish significant association between preexisting medical conditions and SMO. However, previous studies from Brazil [24] and Ethiopia [7,8,19] reported that women with preexisting medical conditions were more likely to have SMO. The proportion of women with medical comorbidities was very low in both SMO and control groups in this study. Therefore, the fact that our study was underpowered for this particular variable might be the most plausible explanation for the observed inconsistency.

Conclusion

In this study, ANC follow up, caesarean section in the current pregnancy, referral from lower level facilities, proximity of residence to hospital, history of anemia, and history of caesarean section were found to be independently associated with SMO. These findings imply that improving access to antenatal care services will have a remarkable impact in preventing SMO. Quality ANC service is critical for early identification and correction of obstetric complications in general and iron deficiency anemia in particular. Health authorities should also be committed to reduce the rate of caesarean sections performed without sound medical indications. Moreover, improving transportation facilities, establishing robust referral protocol as well as ensuring equitable distribution of CEmOC facilities can play crucial role in reducing the incidence of SMO in Keren Hospital.

Supporting information

Acknowledgments

I would like to thank the research and ethics committee of ministry of health Anseba region for granting me permission to conduct this study. I would also like to extend my gratitude to staff members of Keren Hospital who participated in the process of data collection.