Access to facility delivery and caesarean section in north-central Liberia: a cross-sectional community-based study.

OBJECTIVE: Rural north-central Liberia has one of the world's highest maternal mortality ratios. We studied health facility birthing service utilisation and the motives of women seeking or not seeking facility-based care in north-central Liberia.
DESIGN: Cross-sectional community-based structured interviews and health facility medical record review.
SETTING: A regional hospital and the surrounding communities in rural north-central Liberia. PARTICIPANTS: A convenience sample of 307 women between 15 and 49 years participated in structured interviews. 1031 deliveries performed in the regional hospital were included in the record review. PRIMARY OUTCOMES: Delivery within a health facility and caesarean delivery rates were used as indicators of direct utilisation of care and as markers of availability of maternal health services.
RESULTS: Of 280 interview respondents with a prior childbirth, only 47 (16.8%) delivered their last child in a health facility. Women who did not use formal services cited cost, sudden labour and family tradition or religion as their principal reasons for home delivery. At the regional hospital, the caesarean delivery rate was 35.5%.
CONCLUSIONS: There is an enormous unmet need for maternal health services in north-central Liberia. Greater outreach and referral services as well as community-based education among women, family members and traditional midwives are vital to improve the timely utilisation of care.

Evaluate the utilisation rate of key maternal health services in north-central Liberia.

Assess the barriers to seeking facility-based maternal health services.

There is a massive unmet need for maternal health services in rural Liberia, including facility-based delivery and caesarean section.

Significant barriers to access included cost, transport and family tradition of home delivery with a traditional midwife.

Local facility-based and community-based assessments of maternal health services are critical in order to accurately measure community needs and inform policy and programmatic interventions.

Design strengths included the use of mixed facility- and community-based methodologies to assess local utilisation of maternal health services in a highly underserved and under-researched region.

Study limitations included non-randomised sampling of participants in community interviews and limited quality of comprehensive data in hospital records.

Introduction

In 2008, an estimated 342 900 women died in childbirth worldwide.1 A vast global disparity exists, with more than 99% of maternal deaths taking place in low-income and middle-income countries. The maternal mortality ratio (MMR) represents the number of maternal deaths per 100 000 live births. MMR has decreased in Asia and Latin America in the past two decades, but has remained near-constant in sub-Saharan Africa.1–3 According to the WHO, most maternal deaths could be prevented with ready access to adequate obstetric services.4 5 Millennium Development Goal 5 calls for a three-quarters reduction in MMR globally by 2015. Strides have been made toward this target; however, large-scale international programmes have not reduced maternal mortality significantly in areas of the world that bear the greatest burden.1 6 7

Programmes to reduce maternal mortality must consider the local context of maternal mortality and access to maternal health services if they are to be effective.7–9 Recent efforts to improve the measurement of maternal mortality globally have yielded more accurate and complete data.1 Still, there is a paucity of primary community information in many countries with the highest MMR, such as Liberia.1 3 10 11 In addition, a high degree of variability in MMR and access to maternal care exists within communities.12 13

In areas with high maternal mortality and poor information systems where data on maternal mortality are not available, process indicators related to this outcome can be highly informative to evaluate the unmet need for maternal healthcare. Process indicators for maternal mortality include the proportion of women who deliver in a health facility under skilled attendance and the rate of caesarean delivery.14–16 These indicators are easier to capture partly because maternal death is difficult to ascertain, especially in rural areas, or may occur in small catchment areas with low numbers over a short evaluation timespan.17

Liberia is among the poorest countries in the world; it has 1 of the 10 highest national MMR of any nation.18 Hogan et al1 estimated MMR in Liberia to be 859 (uncertainty interval, 547–1287) maternal deaths per 100 000 live births in 2008. By comparison, the MMR in Ghana, among the lowest in the region, was 409 (248–633), and in neighbour Sierra Leone, it was 1033 (635–1627), the region's highest. The wide uncertainty intervals are reflective of the relative lack of data captured by poor health information systems in Liberia, which is still recovering from a 14-year civil war that ended in 2003. Health infrastructure and human resources were decimated during that span. In 2009, the country was estimated to have a deficit of 10 000 healthcare workers.19 The 2007 Liberia Demographic and Health Survey confirms a corresponding gap in access to essential services, citing only 46% of women in Liberia deliver under skilled attendance in health facilities.12 In addition, the national caesarean delivery rate in 2007 was 3.5%, and <1.0% in rural parts of the country.12 20 As Liberia rebuilds its health infrastructure and targets maternal mortality, community-level research is needed to appropriately direct maternal health interventions.

The objective of this study was to characterise utilisation of maternal health services, including facility delivery and caesarean delivery, in rural north-central Liberia and describe the motives of women seeking or not seeking facility-based care.

Methods

We conducted a cross-sectional study employing a structured interview in a community setting, as well as a review of medical records at the local hospital. The individual structured interview comprised of questions on the behaviour and attitudes of women regarding maternal healthcare. Eight communities were chosen based on the presence of existing outreach programmes for antenatal education and childhood immunisations for a regional hospital in Nimba County, located in north-central Liberia. All communities were within 15 km of the hospital. More than 95% of the county's 462 000 residents live in a rural setting.12

All women between the ages of 15–49 years were eligible, consonant with age sampling of the 2007 Liberia Demographic and Health Survey.12 Women were approached for participation in their homes or in other public spaces such as community meeting places. All households and public locations in each village were visited once, resulting in a convenience sample given absenteeism. Community health workers were trained to conduct 25 min standardised interviews. Questions in English were designed in collaboration with clinicians, public health technicians and community health workers in Liberia. Interviews were conducted orally in English or Mano, a language spoken in the region, as per a participant's preference. The interviewers were trained to record responses including multiple answers given to open-ended questions. Responses were coded by a member of the research team. Verbal informed consent was obtained from all participants. Women received a small non-monetary compensation (value≈US$2) for their time.

The review of medical records involved a retrospective evaluation of all deliveries at a regional hospital in Nimba County using general medical and operating room records. Data were extracted on method of delivery, incidence of stillbirth and maternal and fetal indications for caesarean delivery. Extraction was performed by a hospital data officer and a medical student.

Statistical analysis

Participant responses were analysed for frequencies, means and SDs as indicated. Participant characteristics were correlated with their interview responses. Nulliparous women were excluded from analysis of questions related to prior deliveries. Respondents with missing data were dropped from all analyses. Fisher's exact test was used to evaluate categorical variables and Student's t test was used for continuous variables. All analyses were performed with SAS V.9.1.3 (SAS Institute, Cary, North Carolina, USA). Our primary outcomes of interest were facility delivery rate assessed in the community and caesarean delivery rate in a regional hospital.

Data were kept anonymous at all times by removing all identifiers. Ethical approval was granted by the Vanderbilt University Institutional Review Board and the Medical Director of the Liberian hospital.

Results

Between May 2009 and August 2009, 316 women were approached to participate in community interviews. Of these 307 (97.2%) met eligibility criteria, consented and completed the interview. There was 1 (0.3%) documented refusal. Participants had a mean age of 30.5 (SD + 7.5; table 1). The mean parity was 4.8 (SD + 3.0). More than half (n=174, 57.1%) of women reported having no formal education. There were 6 (2.0%) nulliparous women and 21 (6.8%) who chose not to answer questions regarding the site of delivery in the immediate prior pregnancy. Of the remaining 280 women, 210 (75.0%) reported delivering their previous child at home with a traditional midwife (TM) and 23 (8.2%) at home without a TM. Interviewers then asked the open-ended question, “Why did you choose to deliver your child in this place?” for which multiple responses were accepted. Responses for those delivering at home included, ‘cost’ or ‘difficulty with transport’ (31.5%), ‘quick onset of labour’ (30.6%), ‘family tradition’ or ‘religion’ (16.9%), ‘TM is safe’ (16.9%) and ‘conflict’ or ‘flooding’ (3.7%). Women who reported delivering in a hospital responded, ‘labour complications at home’ (57.8%), ‘hospital nearby’ (13.3%), ‘antenatal referral’ (11.1%) ‘safety’ (11.1%) and ‘previous caesarean delivery’ (6.7%).

Table 1

Interview responses of women (n=307)* interviewed in their homes in rural northern Liberia in 2010

Characteristic	Value
Age,* mean years (SD)	30.5 (7.5)
Parity,* mean no. (SD)	4.8 (3.0)
Education level, no. (%)
No formal education	174 (57.1)
Some primary	114 (37.4)
Completed primary or greater	17 (5.6)
Relationship status, no. (%)
Married	223 (72.6)
Unmarried	84 (27.4)
Lost a child during prior pregnancy, no. (%)
Yes	113 (37.0)
No	192 (63.0)
Site of delivery in immediate prior pregnancy, no. (%)†
Hospital	47 (16.8)
Home with traditional midwife (TM)	210 (75.0)
Home without TM	23 (8.2)
Why did you choose to deliver your child in this place?†
Home, no. (%)
‘Cost’ or ‘Difficulty with transport’	69 (31.5)
‘Quick onset of labour’	67 (30.6)
‘Family tradition’ or ‘Religion’	37 (16.9)
‘TM is safe’	37 (16.9)
‘Conflict’ or ‘Flooding’	8 (3.7)
Hospital, no. (%)
‘Labour complications at home’	26 (57.8)
‘Hospital nearby’	6 (13.3)
‘Antenatal referral’	5 (11.1)
‘Hospital is safer’	5 (11.1)
‘Previous caesarean delivery’	3 (6.7)

*Totals may not add to 307 due to missing values. 280 respondents reported age. 305 respondents reported parity.

†N=280 excluding nulliparous women.

In women who reported delivering at home, and those who reported delivering in a hospital, there was no statistical evidence to suggest a difference in age, parity, relationship status or whether a woman reporting losing a child during a prior pregnancy (table 2). Women who reported delivering at home were less likely to have education at or beyond the primary level, a result that had borderline statistical significance (p=0.057).

Table 2

Comparison of characteristics of rural Liberian women who reported delivering their last child at home or in hospital, excluding nulliparous women (n=280)

Characteristic	Home	Hospital	p Value
Age, mean years (SD)	30.9 (7.3)	29.6 (8.3)	0.29
Parity, mean no. (SD)	4.9 (2.9)	4.5 (3.2)	0.32
Education level, no. (%)
None	139 (59.9)	20 (43.5)	0.057
Some primary	84 (36.2)	22 (47.8)
Completed primary or greater	9 (3.9)	4 (8.7)
Relationship status, no. (%)
Married	168 (72.1)	33 (70.2)	0.86
Unmarried	65 (27.9)	14 (29.8)
Lost a child during prior pregnancy, no. (%)
Yes	86 (36.9)	18 (38.3)	0.87
No	147 (63.1)	29 (61.7)

In a 1-year period between April 2008 and March 2009, 949 live births occurred in the study hospital. In the same period, 82 stillbirths were recorded, 8% of total births (n=1031). The rate of caesarean delivery at the hospital was 35.5% (n=337). The stillbirth rate in the caesarean subset was 12.4%. Common indications for caesarean delivery included cephalopelvic disproportion (50%), previous caesarean delivery (12.7%), antepartum haemorrhage due to placenta previa or placental abruption (10.2%), eclampsia (8.5%), breech presentation (5.0%) and ruptured uterus (4.2%).

Discussion

In structured interviews conducted in villages proximate to a hospital in Nimba County, Liberia, only one in six women reported delivering their last child in a hospital, that is, 83.2% of women reported their last delivery to have been at home. Of women delivering at a regional hospital, 35.5% were by caesarean section and 8% were stillborn. These metrics indicate severe underutilisation of timely, supervised, institutional birthing services among women in north-central Liberia. Participants reported financial and transportation barriers to seeking care, as well as cultural traditions including the practice of delivery under the care of a TM. (Traditional midwife is the term used in Liberia, analogous to the traditional birth attendant in other nations.) Further, more than half of women who delivered their last child in a hospital did do so only after experiencing complications in labour at home. This was validated by the hospital's high caesarean section rate. These findings of underutilisation of the formal health sector are similar to those of a 2011 study in the same region of Liberia that highlighted the impact of geographic and financial barriers, perceptions of quality in the formal and informal healthcare systems, and prevailing care by informal health workers, including TM, on the healthcare utilisation.21

The rates of hospital delivery in our study were significantly lower than the national (46.3%) and county (32.6%) statistics reported in 2007 for skilled attendance at delivery, a related metric used by the WHO and Liberia Ministry of Health to assess women's access to safe maternal health services.4 12 Skilled attendance differs from facility delivery in the requirement of supervision of childbirth by a physician, nurse, trained midwife or physician's assistant.11 This is deemed vital to the early recognition of life-threatening intrapartum complications, and their effective management.5 14 While hospital delivery is a pre-requisite for skilled attendance in Liberia where home deliveries are rarely supervised by a professional in the above cadres, it is not necessarily true that all hospital deliveries have the appropriate personnel present. Therefore, our measured rates of facility delivery potentially overestimate the true figure for safe delivery under skilled attendance among the participants.

The definition for skilled attendants notably excludes TM because they do not have access to resources for medical or surgical management of labour.4 Women under TM care who have conditions such as dystocia, eclampsia or bleeding are exceedingly vulnerable. In addition, TM have limited training in the recognition of complications, and in the study community there were no formal mechanisms for timely referral if complications arose.10

When we compared women who delivered their last child at home to those who delivered in a hospital, only education differed; home-delivered women were more likely to report having no formal education. Previous studies have demonstrated the alarming inequalities in access to maternal healthcare based on education levels, as well as wealth, rural venue and religion.7 13 22

In our review of hospital medical and operating room records we found that caesarean delivery accounted for more than a third of all deliveries, though there were relatively few births at the facility in the study period. The hospital serves a region of approximately 250 000 people where nearly 10 000 births were expected annually.12 In addition it is the only facility within a 35 km radius with capacity to provide caesarean delivery service. While smaller health facilities in the region equipped to offer assisted vaginal delivery may provide services in a portion of births in the catchment area, the delivery number at the study site is still very low and reflects the trend towards home delivery found in interviews.

In addition, the high rate of caesarean delivery may stem from referral of complicated cases from other facilities and late presentation after attempted home delivery.23 24 Prolonged labour, antepartum haemorrhage and ruptured uterus were the common antecedents to caesarean delivery.25 Unfortunately, we were not able to ascertain the duration of labour, location where labour was initiated and delays in accessing care. Prior research in the region using the ‘three-delays’ model found significant delays at all three stages—deciding to seek care, reaching a health facility and receiving needed care at the facility.10 23 These prior studies similarly highlighted the importance of family and TM influence on timely access to care.

Caesarean delivery is cited as a life-saving component of Comprehensive Emergency Obstetric Care (EmOC).16 Criteria for Comprehensive EmOC include 24 h caesarean delivery and blood transfusion availability.26 Notably, these standards were not met by this regional hospital. Typical obstacles to the delivery of Comprehensive EmOC in the region include human resource shortages, intermittent electricity supplied exclusively by generator, and scarcity of donated blood that often necessitates donations from the patient's family at the time of the operation.

Limitations of our study included non-randomised sampling of participants in community interviews. In fact, convenience recruitment in proximate communities with active hospital outreach programmes in maternal and child health may have over-represented ‘health-seekers’ and women with access to hospital services, or skewed the sample population in unknown ways by excluding women who were not at home when interviewers visited. Unfortunately we do not have data from communities without hospital outreach programmes for comparison. In addition, refusal to participate among women approached was not documented consistently by survey administrators so we are unable to accurately report on this figure. Finally, the hospital-based portion of our study was limited by poor medical record keeping and organisation, and missing data, particularly pertaining to ‘history of present illness’, a key timeline necessary to evaluate the dynamic causes of birth complications in the context of home and facility delivery.

In summary, the combination of community and facility data enabled us to identify deficiencies in the utilisation of maternal health services in north-central Liberia. Many women in need of such services did not access them, even if care was available locally. Additionally, there was a high rate of serious complications among women presenting to the hospital for caesarean delivery. Greater outreach and referral services as well as education among women and TM are needed to improve timely utilisation of care. Our study also highlights the importance of performing local assessments of maternal health services in order to accurately measure community needs that otherwise may be underestimated by national surveys. These geographically focused evaluations are critical to inform policy and programmatic interventions at the community level.