Table 1 Summary of the 17 Articles

Abodunrin et al., 2010 [30]

Nigeria (urban and rural communities in Ilorin, the capital of Kwara State)

To assess factors that determine referral practices of Traditional Birth Attendants (TBAs).

Descriptive quantitative survey: pre-tested semi-structured questionnaire

162 Registered TBAs, mean age = 46 years, 89.5% females, 71.6% married, 92% Islam, 64.2% had at least primary school education, 85.8% were part-time TBAs

▪ Identified timely and appropriate referrals among TBAs with more than one re-training (69.2%) and TBAs who have ever been visited by a supervisor (45%). Timely and appropriate referral was explained as referring women with high-risk pregnancies such as previous stillbirths, bleeding in previous or current pregnancies, multiple pregnancies, abnormal lie and not interfering with them. It also comprised immediate referral of women who had complications during labour management, such as bleeding during labour, prolonged labour, tiredness or loss of strength, seizures and retained placenta.

▪ Inappropriate referral comprised three conditions: delayed referral irrespective of the reason, wrong referral and non-referral. Delayed referral was defined as “not referring immediately any identified high-risk pregnancy and complicated labour.”

▪ Wrong referrals were those to any place other than a modern health facility.

▪ A significant relationship was found between initial source of skills acquisition, re-training, supervision and prompt/appropriate referral of high-risk pregnancies.

▪ Young and unmarried TBAs with higher education had a higher tendency of appropriate and timely referral.

▪ Most TBAs who started through inheritance usually refer late or not.

Afari et al., 2014 [31]

Ghana (4 health posts, 6 health centres and one district hospital in the Assin North Municipality)

To describe whether healthcare workers (HCW) identified systemic challenges and the significance of local engagement in developing strategies to enhance emergency maternity referral related processes.

Qualitative study: semi-structured interviews

18 HCWs (1 doctor, 2 emergency room nurses, 3 medical assistants, 4 community health

officers, 8 midwives)

▪ Gaps in existing referral protocols-signal function recognition for referral, stabilising patients, initiating referrals, transportation arrangement for referral:

“Sometimes they hire commercial vehicles and sometimes too they use the motorbike. If there is no commercial vehicle at the station, they will beg someone to use their motorbike to convey them to the nearest health center or hospital, and then maybe somebody’s private car. The person might sacrifice.” CHO, Health Post

▪ Few facilities adhered to national referral protocols. Poor referral documentation and lack of communication between sending and receiving facilities were reported, although national referral protocols existed:

“Apart from that [one] guy (HCW) who calls, the others don’t call so you’ll be here and such a case comes in. And […] with no…nobody accompanying… it’s really a challenge. Because if you know […] somebody is coming with eclampsia… you know you’re supposed to prepare first so that you receive [appropriately].” Emergency Nurse, District Hospital

“Somebody who is fitting (or convulsing), a pregnant woman who is fitting… somebody (HCW) needs to accompany. But this is someone who is coming with relatives. They don’t know they have to turn the head to the side, [or] the person can aspirate saliva and any other thing[s].” Nurse, District Hospital

▪ HCWs recommendations: standardising implementation of the referral protocol, enhancing transportation, ensuring dependable data reporting and management systems, actively engagement of community and offering continuous training for health staff.

Akaba & Ekele, 2018 [32]

Nigeria (from either a primary, secondary, tertiary or private health facility to University of Abuja Teaching Hospital, Gwagwalada)

To determine maternal and fetal outcomes of emergency maternal referrals and reasons for these referrals.

Prospective longitudinal study-November 2015 to March 2016, data from case-notes, cross-checked with referral documentation when available.

All women requiring EmONC and referred from primary, secondary, tertiary or private health facility to University of Abuja Teaching Hospital, aged between 20 to 44.

▪ Nine cases (7.3%) were transported by ambulance.

▪ There was 8.9% emergency referral fatality rate (11 maternal deaths/123 maternal referrals).

▪ 7/11 maternal deaths occurred among women referred from secondary health facilities.

▪ Poor emergency maternity referrals and fetal outcomes were reported: 14 (11.5%) fresh stillbirths and six (4.9%) macerated stillbirths) due to late presentation.

Awoonor-Williams et al., 2015 [1]

Ghana (Upper East Region)

Maternity referral audit to strengthen the referral system for pregnant women and newborns in northern Ghana.

Quantitative, two-cycle prospective audit in early 2011and late 2011; questionnaire, 32 facilities in all-16 facilities, 12 health centres, 3 district and 1 regional hospital

223 referred women and their newborns (223 in each of the two cycles)

▪ Observed enhanced referral facilitative mechanisms-increased use of ambulances/vehicles for referrals (48 to 63%); higher usage of referral forms (66 to 77%); alerting receiving facilities through phone calls (38 to 65%); increment in feedback from receiving facilities (58 to 70%); all 6 women referred twice in the 2nd cycle were accompanied by health staff.

Carnahan et al., 2016 [33]

Tanzania (14 government facilities providing maternal health care, urban Dar es Salaam.

To investigate healthcare providers regarding prevention and management of postpartum haemorrhage (PPH).

Quantitative cross-sectional survey, questionnaire

102/115 (88.7%) nurses with midwifery training, 9 (7.8%) nurses without midwifery training, 4(3.5%) doctors/medical/ clinical officers from 10 dispensaries (60.9%), 2 hospitals (18.3%), 1 health centres (20.9%).

104/115 (90.4%) females, 71 (62.8%) with more than 6-year experience.

▪ All 14 facilities had referred 42.6% of women within the past three months.

▪ Forty-nine (42.6%) providers had referred at least one woman in the three months preceding the survey.

▪ 67.8% of 115 providers indicated consultation and referral communication systems are in place.

▪ 65.2% of 115 providers reported establishment of maternal referral transport system.

Elmusharaf et al., 2017 [34]

South Sudan (Renk County, Upper Nile State)

To ascertain patterns and contributory factors of pregnant women’s pathways from the onset of labour or complications until arriving in suitable health facilities.

Qualitative, Critical Incident Technique (CIT), Stakeholder Interviews.

28 key informants (2 from local government, 4 from county health department, 14 healthcare providers, 2 NGO employees, 3 Faith-Based Organisation (FBO) employees, 3 community religious leaders).

▪ Identified four referral pathways-late referral, zigzagging referral, multiple referrals and bypassing non-functioning facilities.

▪ Women who directly went to appropriate health facilities and by-passed non-functioning facilities survived.

▪ Competencies of healthcare providers and functionality of the initial point of care determined the pathway to further care.

▪ Trained midwives were found to be competent but TBAs were not.

Goodman et al., 2017 [35]

Ghana (Ridge Regional Hospital (RRH), Accra)

To describe maternal referrals received in Ridge Regional Hospital (RRH) and explore the timeliness with which women enter CEmOC.

10-week prospective cohort time-sequence information at arrival and from records and logbooks period from 9 to 9-2012 to 11-11-2012.

1082 women with pregnancy complications, 15–46 years, 0–8 parity, 24–49 weeks gestation age.

▪ Long waiting time upon arriving in receiving facility-40 min on average.

▪ The most distant referral facilities were 50 km from RRH.

▪ Gaps were identified in how maternal vital signs and labour assessments were recorded with 25 of 90 referrals found to be inappropriate. The most common reason for referral was for fetal-pelvic disproportion, however, fundal height was less than 40 cm, which does not support this diagnosis.

Kyei-Onanjiri et al., 2018 [36]

Ghana (120 health facilities across Upper East region)

To investigate the availability of emergency maternal care interventions in Upper East region.

Quantitative cross-sectional survey, questionnaire.

120 health facilities (9 public and private hospitals, 17 clinics, 41 health centres, 52 Community-based Health Planning and Service

s (CHPS) centres, 1 maternity home).

▪ 94% health facilities were having standardised or printed referral forms for maternal referrals.

▪ 83% had a standard maternal referral procedure.

▪ 64% had shortwave radio/telephone for referral communication.

▪ 56% of facilities without shortwave radio/telephone could not access one within minutes in instances where it is needed.

▪ Most facilities always had a trained health provider.

▪ 73% had a midwife or doctor either on call or present at all times.

Mirkuzie et al., 2016 [37]

Ethiopia (10 public health centres with similar staff profile and providing EmONC, Addis Ababa)

To assess the proportion of maternal referrals resulting from premature rupture of membranes and investigate its correctness and management in Health centres.

Sequential explanatory mixed methods, routine retrospective data from birth and intrapartum referral logbooks and interviews, focused interview guide.

2820 women with maternal complications; 10 senior midwives

▪ All HC with high referral rates had many SBAs per caseload.

▪ 77.8% of the referred women had a spontaneous labour could have been wrongly referred because they were not in labour when they were referred.

▪ Some health centres observed women for about 8 h before referral initiation:

“… when we get mothers saying that their water has broken, after we evaluate them, they will be admitted to our health center and observed for about eight hours. If there is no spontaneous labour in eight hours we refer them to hospital after giving them a loading dose of Ampicillin… in the referral slip we write how long the mothers had been observed in our health center.” (An informant from HC I).

Mselle & Kohi, 2016 [38]

Tanzania (Comprehensive Community Based Rehabilitation, a private, non-governmental organisation in Dar es Salaam)

To use women’s narratives to demonstrate the challenges leading to failure in accessing adequate maternal care in a timely manner

Qualitative, narrative research, semi-structured interview guide

16 women with obstetric fistula, aged between19 and 43 years, 82% rural dwellers, all unemployed, 88% had no or primary education

▪ Delay in making referral decision was reported:

“… it took 4 days at the village health facility, I could not give birth and then I was referred to the big hospital” (Divorced, aged 29, Kibakwe, Dodoma).

“…In the health facility, I spend the night until morning … I had pains, the day passed, I slept again until morning again, and it was when a decision was made to transfer me to another hospital. They said it was because I had urine retention. On the third day is when I was transported to a big hospital” (Divorced, aged 20, Mlandizi-Pwani).

Nuamah et al., 2016 [39]

Ghana (Antenatal clinics, Amansie West District in the Ashanti region)

To evaluate the role of socio-economic factors, perception and transport availability in fulfilling maternity referrals

Quantitative cross-sectional study, questionnaire

720 confirmed pregnant women from 5 sub-districts, 65.5% cohabitating, 28.8% married, 49.6% JHS/Middle School, 17% No formal education

▪ > 90% reported meeting staff in the receiving facility always.

▪ 76.6% disclosed that health staff in the receiving facility solved their problems.

▪ Most women were referred once and were not referred further.

▪ Commercial cars (88.2%) are more often used for referral than ambulances (6.6%).

Nwameme et al., 2014 [40]

Ghana (ante-natal care (ANC) clinics in Ga East district, Greater Accra region)

To examines the situation faced by women when they need emergency maternity care

Mixed Method, Questionnaire, In-depth Interview guide, Referral and facility review checklist

390 women attending ANC antenatal care clinic attendees and in-depth interviews with principal health care personnel, 17–46 aged women, 92% married, 44.6% Unorthodox Christians, 35.6% Orthodox Christians, 52.1% Junior High School, 12.8% no education, 76.2% traders, 43.6% parity 1, 29.7% parity 2

▪ Out of 17 women referred in their current pregnancies, none of them was sent by Ambulance, ten had public transport whilst seven made their own transport arrangement.

▪ Of the 17, fourteen got to the referral centre within 24 h, two within 48 h and one woman got there after 10 days.

▪ 15 had referral letters, but only 1 was accompanied by staff.

▪ Only one hospital had information computerized for easy access.

▪ Referring health facilities hardly received feedback:

“We don’t receive any feedback from the hospitals. At least it would help us understand what we could have done better.” (In-depth Interview, Medical Officer)

▪ During maternity emergencies, they contact referral centres by mobile phone to find out if beds are available:

“There are hindrances between the two hospitals, no beds, no doctors…all these contribute to the delays.” (In-depth Interview, Nursing Administrator).

Okafor et al., 2015 [8]

Nigeria (Semino Hospital and Maternity (SHM), Enugu State)

To audit childbirth emergency referrals by trained TBAs

Quantitative, retrospective, case records retrieved and data extracted with case record forms

205 women with childbirth emergencies, 41.5% rural dwellers, 58.5% urban dwellers, 90.2% married, 58.5% unemployed, 56.1% nullipara

▪ 155 (75.6%) of the women were delayed for more than 12 h before referral.

▪ 75.6% (155/205) arrived walking unsupported prior to admission whilst 24.4% (50/205) could not walk on admission.

Shimoda et al., 2015 [41]

Tanzania (urban, one regional referral hospital and one health centre in Dar es Salaam city)

To describe how midwives monitor and manage childbirth in order to achieve early consulting and timely referral to obstetricians

Qualitative, semi-structured interviews

11 midwives, 12.5 average year experience, 6 in the regional referral hospital, 5 in the health centre during the day, 4 with certificate, 4 with diploma, 2 with bachelor’s degree, 1 with master’s degree, 30–80 daily average births for their wards.

▪ Intrapartum management and monitoring/examination to arrive at referral decision consisted of 3 phases: 1) initial encountering, 2) monitoring, and 3) acting that finally resulted in referral.

▪ Prompt referral upon identifying signal function beyond the facility’s capacity:

“When she put in the catheter, we saw some blood starting to pass. That is the sign of obstructed labor. That’s why I decided to refer immediately.” (F)

▪ In instances where mother and fetus conditions are worsening, midwives decide earlier without taking time to confirm labour.

Strand et al., 2009 [42]

Angola (3 peripheral birth units-Cazenga, Palanca and Sambizanga)

To assess the efficacy of the newly established network of peripheral birth units and their linkage to hospitals.

Two-phase quantitative survey, review of maternal records

157 referred women for 1st and 92 for 2nd phase, 24.1 mean age, 36% < 20 years, 43% primiparae, 32% ≥4 previous births in 1st phase

▪ 157 deaths (17.8% case fatality rate) occurred among traced referrals in the first phase, no maternal death in the second phase.

▪ Redacted proportion of referred women who were left without medical evaluation/treatment observed from the women’s records (45% in the first phase of the study to 27% in the second phase (p = 0.007).

Tayler-Smith et al., 2013 [43]

Burundi (rural district, Kabezi)

To describe Medecins sans Frontieres’ communication and ambulance service, examine relationship between referral time and adverse outcome, explore effect of referral service on coverage of complications and caesarean sections.

Cross-sectional study, retrospective analysis

1478 ambulance call-outs/referrals.

▪ Median referral time (time between call-out to the ambulance returning with the patient at CURGO) was 78 min.

▪ One maternal death occurred among referred women but it was not possible to evaluate the linkage between death and referral time.

▪ 3-h referral duration or higher was associated with increased risk of early neonatal mortality-15% as compared with 9%

Windsma et al., 2017 [44]

Ethiopia (20 health centres in the Eastern Gurage Zone)

To assess BEmONC, knowledge of high-risk pregnancies and referral capacity in health centres

Cross-sectional survey

37 healthcare providers (18 heads of health centres, 14 midwives, 3 nurses, 1 health officer, 1 other), 45 months average of experience among heads with 27.5 median age, median age of 24 years for others with 24 months average experience.

▪ Most staff used their own mobile phones for referral correspondence-only 5 facilities (26.3%) had a working landline and 1 (5.3%) facility had a mobile phone.

▪There were 5 ambulances for the Eastern Gurage Zone population: Two stationed in health centres and 3 in District Health Offices.

▪ Distance to the referral Butajira General Hospital used by all health centres was 16.5 km on average.

▪ There is a need to train staff of the health centres in the identification of signal functions and BEmONC.