Improving emergency obstetric referral systems in low and middle income countries: A qualitative study in a tertiary health facility in Ghana

Background: Timely access to emergency obstetric care is crucial in preventing mortalities associated with pregnancy and childbirth. The referral of patients from lower levels of care to higher levels has been identified as an integral component of the health care delivery system in Ghana. To this effect, in 2012, the National Referral Policy and Guidelines was developed by the Ministry of Health (MOH) to help improve standard procedures and reduce delays which affect access to emergency care. Nonetheless, ensuring timely access to care during referral of obstetric emergencies has been problematic. The study aimed to identify barriers associated with the referral of emergency obstetric cases to the leading national referral centre. It specifically examines the lived experiences of patients, healthcare providers and relatives of patients on the referral system. Methods: Korle Bu Teaching Hospital, Accra was used as a case study in 2016.The qualitative method was used and in-depth interviews were conducted with 89 respondents: healthcare providers [n = 34];patients [n = 31] and relatives of patients [n = 24] using semi-structured interview guides. Purposive sampling techniques were used in selecting healthcare providers and patients and convenience sampling techniques were used in selecting relatives of patients. Results: The study identified a range of barriers encountered in the referral process and broadly fall under the major themes: referral transportation system, referrer-receiver communication barriers, inadequate infrastructure and supplies and insufficient health personnel. Some highlights of the problem included inadequate use of ambulance services, poor management of patients during transit, lack of professional escort, unannounced emergency referrals, lack of adequate information and feedback and limited supply of beds, drugs and blood. These findings have implications on type II and III of the three delays model. Conclusions: Initiatives to improve the transportation system for the referral of obstetric emergencies are vital in ensuring patients’ safety during transfer. Communication between referring and receiving facilities should be enhanced. A strong collaboration is needed between teaching hospitals and other stakeholders in the referral chain to foster good referral practices and healthcare delivery. Concurrently, supply side barriers at referred facilities including ensuring sufficient provision for bed, blood, drugs, and personnel must be addressed. Korle Bu Teaching Hospital was established in 1923 and located in the capital city of Ghana, Accra. It is currently the third largest hospital in Africa and the leading national referral centre in Ghana. The hospital currently has a 2000 bed capacity and 17 departments/units serving an average of 1500 patients a day. The Obstetrics and Gynaecology department serves mainly as a referral centre for the southern part of the nation, which has a population of over 10 million. The Obstetric unit of the Obstetrics and Gynaecology department has two labour wards and 275 beds for obstetrics cases. The unit receives referral cases from other lower level facilities both public and private within and outside Accra. Often, more than half of all the referred cases are of emergency nature. The unit records several pregnancy and delivery complications in a year with hypertensive diseases mostly at the top. The unit has in time past had more than 80% of referral cases resulting in maternal deaths each year [16]. A wide range of services ranging from minor through to major procedures are conducted by health professionals in the unit. The unit has two operating theatres and these theatres have a five-bed recovery ward. The study employed the qualitative method and used phenomenology. The social constructivist paradigm was used as a guide for the study. This paradigm acknowledges the subjective and multiple nature of reality and posits a very close contact with the participants being studied [17]. Assumptions made under this paradigm suggest that individuals develop subjective meanings of their experiences as a way of seeking understanding of the world in which they live and work. Social constructivists however state that these subjective meanings are not imprinted on individuals but are rather co-constructed with others through interaction [18]. This presents an opportunity for several individuals to share or describe their experience; an approach in qualitative inquiry referred to as phenomenology [17]. This approach was used to understand the lived experiences of providers, patients and relatives of patients on the challenges associated with accessing emergency care at the hospital. An interview guide was developed to elicit views from respondents. The study population included all women who were referred under emergency conditions to the obstetric unit of the hospital, all relatives of emergency referred patients and all healthcare personnel in that unit. Study participants were selected using purposive and convenience sampling techniques. Key informants such as Doctors, midwives and nurses at the Obstetrics unit of the Obstetrics and Gynaecology department of the hospital were purposively selected for In-Depth Interviews (IDIs) because they have the knowledge in the management of obstetric cases in the facility. Data was collected from 34 health providers comprising 16 medical doctors, 13 nurses and 5 midwives. Again, IDIs were conducted with 31 purposively selected patients. The health staff at the unit were contacted to help in the selection of women who were referred from other facilities (public or private) under any form of emergency (those requiring immediate attention) within and outside Accra. However, participation in the study was guided by health professionals in selecting those who are fit primarily and willing to take part in the study. These included in-patients and discharged patients. Self-referred women and those who reported with no emergency issues were excluded from the study. Convenience sampling technique was used to select 24 relatives of patients for IDIs comprising 17 husbands, 4 sisters, 1 in-law and 2 mothers. Relatives of patients were contacted during visiting hours and those who agreed to take part in the study were interviewed at a time and place of their preference. In all, a total sample size of 89 was determined using a saturation approach [19]. Eighty-nine (89) face-to-face IDIs were conducted with the help of semi-structured interview guides which were pre-tested before their final adoption. This exercise was undertaken after an approval was granted by the Scientific and Technical Committee and Institutional Review Board of the hospital. Questions listed in the guides were open ended and had the principal aim of eliciting responses from healthcare providers, patients and their relatives on the referral of emergency cases to the hospital and the challenges associated with it. Views on how to ensure effective referral and management of emergency cases to the hospital were also sought from all participants. All IDIs were conducted by two trained nursing students. Interviews were conducted in English and two local languages (Twi and Ga). The interview guides were not translated into the local languages but the research assistants were taken through rigorous training to accurately translate the interview guides into the two local languages to ensure consistency. An arrangement was made to present the study to all healthcare providers and a date was agreed on to commence the data collection. The purpose of the study was again explained to all participants before the interviews. Participants read and signed the consent forms accordingly before each interview. For participants who could not read, the research assistants read the content of the forms after which they signed or thumb printed. Healthcare providers were visited in the wards and in their offices within the unit for the interviews. Interviews with patients were also in the wards while some relatives were also interviewed in the wards during the visiting hours. Other relatives however chose a place outside the wards for the interviews. Responses from participants were recorded using a digital voice recorder. However, responses from participants who did not consent to audio recording were handwritten. An opportunity was however given to respondents to crosscheck what was recorded after which they signed to validate the information given. The study was aided by two different sets of semi-structured interview guides (see Additional file 1); one for healthcare providers and the other for patients and their relatives. Participants were assured of confidentiality and anonymity with regards to responses given. They were also reminded that participation was voluntary and thus they could freely withdraw from the study when they deem it necessary. All interviews with participants lasted between 20 and 35 min. This study made use of both primary and secondary data. Primary data was sourced from in-depth interviews with healthcare providers, patients and their relatives. Secondary data was obtained from review of policy and guidelines on referrals and annual statistical report of the obstetric unit of KBTH. Information on obstetric emergency cases and maternal death records were sourced from the unit’s annual statistical records. Analysis of data involved various phases of thematic analysis [20]. All voice recordings from interviews were transcribed verbatim. Responses in the local languages (Twi and Ga) were translated into English with the help of experts prior to analysis. To ensure accuracy, recordings in Twi and Ga were compared with the English versions for accurate translations while listening to the original voice recordings. Initial ideas were given codes while reading the data. Handwritten responses were also typed and saved in Microsoft word documents. All transcripts were stored in softcopy formats in password secured folders on a password-protected laptop computer. Transcripts were entered into Microsoft excel to help in identifying common themes regarding barriers to effective referral and management of emergency obstetric cases to the teaching hospital. Codes were collated and repeated patterns identified across the data set were identified as themes using the referral policy and guidelines and the three-delay model as a guide. Four (4) global themes were developed through a review of the various themes. The next step involved the refinement of the global themes which led to the development of sub-themes. Finally, compelling extracts of the data were used to back analysis in relation to the objectives of the study.