Background

Most maternal deaths occur during the intrapartum and immediate postpartum period from preventable or treatable causes such as hemorrhage, eclampsia and sepsis [1]. The majority of these women die in low-income countries, and several countries in sub-Saharan Africa have very high maternal mortality ratios (MMR), often double the estimated global MMR of 400 per 100,000 live births [2].

Although global and regional population based numbers of maternal deaths remain high, numbers at the facility level are relatively low. It has been shown that reviewing critical incidents, including obstetric audits at facility level, may improve quality of facility based care [3] and functioning of the broader health system [4–6].

Until recently, a variety of criteria was used to define severe maternal morbidity or maternal near miss (MNM) [4]. In 2011, the World Health Organization (WHO) developed a systematic MNM approach. The aim was to facilitate comparisons between different studies, and between countries and regions [7], in order to address shortcomings in quality of obstetric care and improve pregnancy outcome [8]. Within this approach, an MNM refers to ‘a woman who almost died but survived a complication during pregnancy, childbirth, or within 42 days after termination of pregnancy’ [9, 10], and organ failure based criteria reflect critical illness and allow for identification of MNM. A number of studies have applied these criteria in low-income countries [8, 11–14].

In Rwanda, the MMR has been reduced by roughly two-thirds from 750 in 2005 to 210 per 100 000 live births in 2015 [15, 16]. This may have resulted from strong government commitment combined with a well-organized community based health program, performance based financing, innovative community health insurance, and SMS-based alert systems among other policy changes [17, 18]. Nevertheless, additional progress is needed, and evaluation of MNM may be of increasing use in light of this mortality reduction. This study aimed to assess MNM characteristics and process indicators related to quality of care at a district hospital in rural Rwanda, by applying the WHO MNM approach.

Methods

This was a prospective cohort study among pregnant women who were admitted for delivery or pregnancy related complications, and who sustained severe acute maternal morbidity at Ruhengeri maternity ward in Musanze district, Rwanda, between June 2013 and December 2014. Musanze is located in northern Rwanda, at 94 km from the capital Kigali, and has a population of 368 267 inhabitants with a total fertility rate of 4.6 births per woman [16].

Ruhengeri hospital acts as a provincial referral hospital for high-risk obstetric cases from health centers and other district hospitals in the northern province. Annually, 3500 deliveries are conducted at the facility. During the study period, medical staff consisted of one specialist obstetrician, four medical officers, two intern doctors and 18 midwives. Patients with critical conditions were admitted into the intensive care unit (ICU), where vital signs were monitored hourly, and treatment with vasoactive drugs and mechanical ventilation could be provided. Blood for transfusion was supplied by the regional blood bank located next to the hospital.

Eligibility for the study was not restricted by gestational age. Therefore, women who had had an abortion or ectopic pregnancy and met the WHO MNM criteria were also included [7, 10]. Women who developed complications more than 42 days after termination of pregnancy were not eligible. On a daily basis, we captured relevant data for every woman who presented with severe acute maternal morbidity or died during admission, by using available medical records. Severe Maternal Outcome (SMO) was defined as MNMs and maternal deaths combined.

Two trained research assistants screened possible participants and the principal investigator (RK) verified suitability for study inclusion. Possible participants included patients who had severe maternal complications at baseline, received critical interventions or were admitted into the ICU, according to WHO guidelines [10]. We tried as much as possible to apply the WHO MNM criteria but had to make several modifications due to lack of laboratory tests and management options (Table 1). For every case, information was collected regarding socio-demographic characteristics, gestational age, maternal and neonatal outcome, as well as process indicators regarding prevention and treatment of the main causes of MNM and death: giving magnesium sulfate for treatment of eclampsia, oxytocin for prevention and treatment of postpartum hemorrhage, and antibiotics for prophylaxis (during cesarean section) and treatment of sepsis.

Table 1 WHO near miss criteria and modifications for Ruhengeri hospital
Full size table

Data were entered into Microsoft Excel, and subsequently entered into STATA version 12 for analysis. Outcomes were given in proportions with 95 % confidence intervals (CIs). The results are presented in accordance with the WHO MNM approach.

Results

During the 19-month study period, there were 3979 deliveries, in which 3827 singletons and 152 twins pairs were born. Among the 4131 neonates, there were 3994 live births, and 137 stillbirths. Three hundred thirty two women sustained potentially life-threatening conditions and 99 had SMO: 86 near misses and 13 maternal deaths. The cesarean section rate for Ruhengeri hospital was 34.9 %.

Cases of morbidity were classified by underlying causes and type of organ system dysfunction (Tables 2 and 3). The commonest conditions contributing to MNM were severe hemorrhage (n = 49; 57 %), but with MI 5.8 % (95 % CI 1.5–15.7). Severe hypertensive disorders was the second main cause of morbidity (n = 27; 31.4 %) with MI 13 % (95 % CI 4.1–31.1). Eclampsia (n = 4/13; 30.7 %) was the leading cause of maternal mortality, while sepsis/peritonitis following cesarean section (n = 2/6; 33.3 %) had the highest mortality index. Other common contributing factors among MNM cases were lack of health insurance (n = 23; 26.7 %), anemia (n = 21; 24.4 %) and previous cesarean section (n = 12; 14 %).

Table 2 Underlying causes and contributory factors of potentially life-threatening conditions (PLTC) and severe maternal outcomesa
Full size table
Table 3 Frequencies of maternal near miss and maternal death by type of organ system dysfunction
Full size table

Most common organ dysfunctions among women with MNM conditions were cardiovascular dysfunction (n = 56; 65.1 %) followed by multiple organ dysfunction (n = 35; 40.7 %). The mortality index was highest for hepatic dysfunction, (33.3 %). The MNM/mortality ratio was highest for cardiovascular dysfunction (8:1), followed by renal and neurologic dysfunction (5:1) (Table 3).

Cesarean section was performed in many women with potentially life threatening conditions (n = 125/332; 37.7 %), MNM (n = 37/86; 43 %) and maternal deaths (n = 3/13; 23.1 %). Neonatal outcomes (preterm births, stillbirths and perinatal deaths) were worst among maternal deaths and comparable between women with potentially life threatening conditions and maternal near misses (Table 4).

Table 4 End of pregnancy and birth outcomes in our study population
Full size table

The SMO ratio was 24.8 per 1000 live births (95 % CI 20.3–30.1), and the MNM ratio 21.5 per 1000 live births (95 % CI 17.3–26.5). The maternal mortality ratio (MMR) was 325 (95 % CI 181–543) per 100 000 live births, the mortality index (MI = maternal deaths/MNM+ maternal deaths) 13.1 % (95 % CI 7.3–21.9), and the maternal near miss mortality ratio (MNM: maternal deaths) 7:1 (Table 5). In 332 women, 234 critical interventions were carried out, including 108 blood transfusions (32.5 % of women) and 71 laparotomies (21.4 % of women). The ICU admission rate was 1.4 % (61/4353) among all women, whereas ICU admission rate among women with SMO was 28.3 % (28/99) and proportion of maternal deaths occurred without ICU admission was 46 % (6/13) (Table 5). Of 134 women with severe hemorrhage, 52 (38,8 %) were transfused five or more units of blood.

Table 5 Severe maternal outcomes, near miss indicators and facility-related indicators, including 95 % confidence intervals
Full size table

Among women with SMO, oxytocin was given in 3862 women (93.5 %) for prevention of postpartum hemorrhage, and only 97 of 134 women with severe PPH (72.4 %) received oxytocin as treatment. Almost all women with eclampsia (52/53; 98.1 %) received magnesium sulphate. Intra-operative prophylactic antibiotics were administered to 1409 out of 1442 women (97.7 %) who delivered by cesarean section. Only 18 of 38 women (47.4 %) with sepsis received parenteral antibiotics (Table 6).

Table 6 Use of interventions for prevention and treatment of major obstetric complications
Full size table

Discussion

This is the first study of SMO at a district hospital in rural Rwanda. The high prevalence of MNM in our study lies within the wide range of ratios reported in studies from other low-income countries, which used similar near miss criteria [11, 19, 20]. The high MNMR of 21.5 per 1000 live births in our study may be explained by: i) different interpretations of the criteria [12–14], ii) delayed referral of SMO cases (77.9 % compared to 20.9–64.4 % in the other studies) or iii) the high proportion of women without medical insurance which may have led to delay in seeking or providing care. The higher mortality index for MNM of 13.1 % in our and similar studies from Africa [8, 13, 14] compared to middle-income countries [5], demonstrates that a larger number of critically ill women in low-resource settings die from such severe complications.

The main direct causes of SMO were obstetric hemorrhage and hypertensive disorders, comparable to other studies in low-resource countries [11–14]. Mortality indices for these conditions were lower than in other studies, probably due to wide availability of blood for transfusion and magnesium sulfate in our setting. Particular attention is needed for conditions with a high mortality index, in particular post-cesarean sepsis and peritonitis, which were previously reported to also be a problem at a tertiary hospital in Rwanda [14]. The relative inexperience from medical officers working in district hospitals (generally recent medical school graduates) may compound this problem [21], combined with improper use of antibiotics and poor sterilization procedures [22]. These findings are extremely worrying in light of the very high cesarean section rate. This rate calls for rigorous audit of cesarean section indications [23].

Two-thirds of SMO cases were received in critical condition from other facilities or home. This may indicate difficulties in the referral system or detection of pregnant women with life threatening complications outside health facilities due to delays in seeking or reaching care [19]. Women who were referred constituted the large majority of all women with SMO. It also addresses quality of care in lower-level health facilities. When analyzing access to care, it is essential to separate near miss cases on arrival from those developing in the hospital. This would be unexpectedly high given the government’s commitment towards improving maternal and neonatal health care through use of its existing programs [17, 18] to reduce first and second delays [24]. This highlights a number of potential interventions, such as educating pregnant women and their caretakers on obstetric danger signs [21] and training health care workers at lower level facilities in emergency obstetric care [25].

MNM developed in one quarter of women with potentially life threatening conditions, with high MNMRs due to cardiovascular, renal and neurologic dysfunction. Suboptimal use of some evidence-based practices explains these SMO cases. The use of uterotonic drugs to prevent PPH was high; however, less than three out of four women who developed PPH received oxytocin as a treatment. Similarly, use of prophylactic antibiotics for cesarean delivery was high; however, less than half of women with sepsis received parenteral treatment with antibiotics. The high patient load at the facility might have a negative impact on quality of care.

The low number of women with SMO admitted into ICU is similar to what other authors in low resource settings found [25, 26]. It is easy to under-estimate severe morbidity in the absence of sophisticated laboratory diagnostics and shortage of nurse-midwives and clinicians to identify clinical signs of deteriorating patients [12, 13]. In sub-Saharan African countries, studies have shown maternal morbidity and mortality can be significantly reduced by improving maternal health care when health workers themselves use audit to identify and analyze deficiencies and apply the findings to improve their obstetric care practices [3, 27]. However, our setting is more focused on treating conditions instead of looking further on how to avoid these in the future. This type of audit will be initiated at Ruhengeri hospital as it is paramount to follow-up actions and maternal health outcomes after an audit [28].

Our study is the first study to be conducted in a rural setting in Rwanda to assess SMO using the new WHO MNM criteria. The advantage of this approach is its standardized methodology, which may allow for a comparison of health facilities and systems. Using this approach led to the integration of regular assessment of the quality of maternal health care in the hospital. In terms of limitations, the lack of follow up after discharge may lead to under reporting of actual MNM and maternal deaths. In addition, the quality of medical records was sometimes poor. We could not apply all WHO criteria due to limited resources at our facility. Some women had severe complications, which did not fulfill any clinical criteria, while laboratory criteria could not be tested for an important part. This may have caused additional under estimation of MNM. Autopsies were not performed and the underlying causes of death were only based on clinical records. Future longitudinal studies should be conducted to assess maternal morbidity and mortality using the WHO approach nationwide. Interviewing women with MNM may help to elucidate barriers to receiving hospital care.

Conclusions

In conclusions, MNM is common at district level in Rwanda. Our study highlights some shortfalls in clinical practice and the referral system, improvements of which could lead to further reductions in maternal mortality and morbidity. We recommend intergrated continous quality of care interventions by use of audit at all levels including primary care.