Background: The provision of quality obstetric care in health facilities is central to reducing maternal mortality, but simply increasing childbirth in facilities not enough, with evidence that many facilities in sub-Saharan Africa do not fulfil even basic requirements for safe childbirth care. There is ongoing debate on whether to recommend a policy of birth in hospitals, where staffing and capacity may be better, over lower level facilities, which are closer to women’s homes and more accessible. Little is known about the quality of childbirth care in Liberia, where facility births have increased in recent decades, but maternal mortality remains among the highest in the world. We will analyse quality in terms of readiness for emergency care and referral, staffing, and volume of births. Methods: We assessed the readiness of the Liberian health system to provide safe care during childbirth use using three data sources: Demographic and Health Surveys (DHS), Service Availability and Readiness Assessments (SARA), and the Health Management Information System (HMIS). We estimated trends in the percentage of births by location and population caesarean-section coverage from 3 DHS surveys (2007, 2013 and 2019–20). We examined readiness for safe childbirth care among all Liberian health facilities by analysing reported emergency obstetric and neonatal care signal functions (EmONC) and staffing from SARA 2018, and linking with volume of births reported in HMIS 2019. Results: The percentage of births in facilities increased from 37 to 80% between 2004 and 2017, while the caesarean section rate increased from 3.3 to 5.0%. 18% of facilities could carry out basic EmONC signal functions, and 8% could provide blood transfusion and caesarean section. Overall, 63% of facility births were in places without full basic emergency readiness. 60% of facilities could not make emergency referrals, and 54% had fewer than one birth every two days. Conclusions: The increase in proportions of facility births over time occurred because women gave birth in lower-level facilities. However, most facilities are very low volume, and cannot provide safe EmONC, even at the basic level. This presents the health system with a serious challenge for assuring safe, good-quality childbirth services.
Liberia had a population of approximately five million in 2021 and a total fertility rate of 4.1 children per woman [23]. The Liberian health system, weakened by civil war, was further damaged by the 2014–16 Ebola epidemic [24]. Maternal mortality remains high, at an estimated 661 per 100,000 live births in 2017, a decrease of 26% from 2000 [5]. The estimated neonatal mortality in 2020 was 31 per 1000 live births, a decrease of 35% from 2000 [25]. The three levels of facility which provide childbirth care in Liberia are clinics, health centres and hospital. Clinics, the lowest level of facility, are expected to provide routine antenatal, labour, and postnatal care, and deal with certain obstetric emergencies, and should be staffed by one nurse and one midwife. Health centres are expected to provide BEmONC and should be staffed with at least two physician assistants, four midwives and one nurse. Finally, hospitals (the highest level) are intended to provide Comprehensive Emergency Obstetric and Neonatal Care (CEmONC), including caesarean section and blood transfusion, and be staffed a doctor, three physician assistants, six midwives and ten nurses. We used three data sources: the DHS surveys (2007, 2013 and 2019–20), the Liberia Service Availability and Readiness Assessment (SARA) (2018) and the Liberia Health Management Information System (HMIS) 2019. The DHS surveys were used to estimate coverage of facility births by type of facility and caesarean section rates among all live births in the 5 years preceding the survey. The total sample size was 19,106 live births across the three surveys. To assess staffing and EmONC signal functions, we used data from the SARA 2018 which included all health facilities in Liberia in December 2017 and January 2018. We removed duplicate assessments and those with incomplete data on childbirth services, and included facilities in our analysis only if they reported offering childbirth services. To examine volume of births and caesarean sections, we used data from facilities reporting at least one live birth in 2019 to the Health Management Information System (HMIS). To enable us to jointly assess volume of births and signal functions, we matched facilities in the HMIS and SARA datasets on name, region and district. We estimated the percentage of births in public hospitals, public health centres, public clinics/other public, private facilities, home, and other/missing for each DHS, using the mid-point of the 5-year recall period for each survey (2004 for 2007, 2010 for 2013 & 2017 for 2019–2020). Data on private facilities by level were not available in the DHS. Full details of the birth location indicator definitions are given in the Additional file 1: Table S1. We also report the percentage of births by caesarean section for each DHS, examining each singleton birth and for the neonate who was born last in each multiple birth. Readiness to deliver EmONC was assessed through the availability of signal functions [22]. We defined facilities as having readiness to provide BEmONC-1 if they reported having carried out six of the seven signal interventions for management of basic obstetric emergencies at least once in the 12 months preceding the 2018 SARA visit, and if the necessary drugs or equipment were observed in the facility. The six signal functions were parenteral administration of antibiotics, parenteral administration of oxytocin, parenteral administration of anticonvulsants, manual removal of placenta, removal of retained products and neonatal resuscitation (full details Additional file 1: Table S2). We removed the requirement for assisted vaginal delivery as part of BEmONC because this is so rarely provided. Facilities were defined as having CEmONC-1 readiness if they could provide the six basic signal functions and the two comprehensive functions, caesarean section and blood transfusion. Readiness to make an emergency referral was assessed through availability of vehicles and telephones reported in SARA 2018. A facility was defined to have readiness for emergency referral if it had either a) a functional ambulance or other vehicle stationed at the facility, or b) access to an ambulance or other vehicle stationed at another facility, and a functioning telephone (landline or mobile) supported by the facility. Skilled birth attendants (SBAs) were defined as doctors, physician assistants, midwives and nurses. Total numbers of SBAs, and numbers by cadre, were calculated from the numbers of each cadre reported working at the facility in the 2018 SARA, irrespective of their full- or part-time status. Physician assistants were grouped with doctors for analyses by cadre due to low numbers. The number of live births per facility reported to HMIS in 2019 were grouped in five categories, as suggested by Kruk et al. [9]: < 53 a year (or 500 a year [9]. Five hundred births a year is an internationally used threshold, for example in the UK and the US [26, 27]. Facilities which reported at least one caesarean section were grouped into four volume categories for caesareans: < 25 a year, 25–99 a year, 100–199 a year and ≥ 200 a year. We calculated the percentage of facility births and caesarean sections occurring in facilities with different levels of EmONC or volume categories directly, by summing the number of live births and caesarean sections reported by each facility in the HMIS 2019 for each EmONC readiness and volume category. All analysis was carried out in Stata version 17.0.