A health system framework for perinatal care in South African district hospitals: a Delphi technique

Background The majority of perinatal deaths occurring in low- and middle- income countries are preventable. South Africa is a middle-income country with consistently high perinatal mortality rates and most factors contributing to preventable deaths are linked to the functioning of the health system. Particularly of concern in South Africa is the high perinatal mortality in district hospitals, where most births occur and where intrapartum care is provided to women of low and intermediate risk. Therefore, it is crucial to strengthen the health system for perinatal care in district hospitals. There is currently no consolidated documented framework outlining contextual health system domains and indicators that are key to providing effective perinatal care in district hospitals. The purpose of this study was to derive key health system domains and indicators necessary to measure the performance of the health system for perinatal care in South African district hospitals. Methods The Delphi technique was used in collecting data from a panel with experts drawn from disciplines connected with the functioning of the health system for perinatal care in South Africa. The study enrolled thirteen experts from whom data on key health system domains and indicators for perinatal care were derived. The project reference group gave guidance to the development of the framework and ascertained its relevance to the South African setting. Results The Facility Based Health System Framework for Perinatal Care comprising domains and indicators necessary to measure health system performance in South African district hospitals was derived from data. The broad structure of the proposed framework aligns with the WHO Health Systems Framework. Each critical building block has detailed domains and indicators that illuminate essential facility-level and programme-specific elements that require attention for strengthening the health system for perinatal care. Conclusion The proposed framework can enable district hospital management teams to identify gaps in the health system for perinatal care, which need to be strengthened in order to alleviate the burden of perinatal deaths in district hospitals.


Background
High perinatal mortality remains a global public health concern. The majority of perinatal deaths occurring in low-and middle-income countries (LMIC) are preventable and most are intrapartum-related [1][2][3][4]. Although patient factors do contribute to intrapartum-related deaths, evidence shows that most factors contributing to preventable mortality are directly linked to the functioning of health systems [1,3]. Specifically, in low resource settings, 66.7, 11.9 and 7.1% of preventable intrapartumrelated deaths have factors relating to substandard care by a health worker, administrative factors, and referral factors respectively [1]. Consequently, there is a great need to strengthen the health system for perinatal care in LMIC in order to save lives [5].
South Africa is a middle-income country with consistently high perinatal mortality rates even among babies who are expected to have a good chance at survival [4]. The high prevalence of intrapartum-related deaths for babies who weigh at least 2500 g at birth clearly shows weaknesses in perinatal health care in South Africa [4,6]. Essentially, in 2008 and 2009, approximately 35% of fresh stillbirths in South Africa weighed at least 2500 g, and in the 2010/2011 period, the reported intrapartumrelated mortality for babies in this weight category was 6.37 per 1000 births [7,8]. In one province in South Africa, the analysis of preventable perinatal deaths following an intrapartum event indicated that early neonatal deaths were more likely to have a healthy mother [9]. Such deaths, which cannot necessarily be predicted during antenatal care, and are not prevented during intrapartum care, reflect a need to focus on potential underlying causes; and these can be better understood by exploring the context in which these deaths occur.
South Africa's public healthcare delivery system is organized into levels of care, providing perinatal care health services for low risk to high risk pregnancies and births [10,11]. Low risk births are designated to be midwife-led, in clinics and community health centres (CHCs), which provide basic emergency obstetric care. Clinics and CHCs refer intermediate risk births to district hospitals which provide comprehensive emergency obstetric care [11]. High risk births are designated to be doctor-led in regional, tertiary and national central hospitals [11]. Overall, the healthcare delivery system has been designed to meet the healthcare needs of women during pregnancy, around the time of birth and after birth.
Perinatal mortality rates in South African district, regional, tertiary and national central hospitals were 30.5, 42.7, 55.3 and 64.3 per 1000 total births respectively in the 2012/2013 period [6]. The perinatal mortality in regional, tertiary and national central hospitals is presumably from high risk pregnancies and births with complications [11].
Although district hospitals have a lower perinatal mortality rate, the absolute number of perinatal deaths occurring in district hospitals is higher compared to other levels of care [6]. The high perinatal mortality in district hospitals is of concern as deaths are occurring in intermediate risk pregnancies [11]. The high number of deaths in district hospitals could be partially attributed to the quality of care, which as measured by the perinatal care index, is poorer [8,12]. A survey conducted in 2012 indicated that the ability of district hospitals to provide lifesaving services is poor, implying that safe intrapartum care is not consistently available in district hospitals [13]. The inefficiencies in district hospitals could be resulting from a lack of resources and competencies to deliver comprehensive intrapartum care services [10,13]. District hospitals account for approximately 41% of births in the public health sector [8]. Therefore, there is a great need to focus health system strengthening strategies on district hospitals in order to reduce perinatal mortality in babies born to intermediate risk mothers, toward improving perinatal outcomes.
The environment in which perinatal care is provided must be better understood before implementing health system strengthening strategies [14]. There is currently no consolidated documented framework outlining contextual health system domains and indicators that are key in providing effective perinatal care in South African district hospitals. Consequently, district hospital management teams are constrained in making evidencebased decisions in prioritizing areas of intervention for strengthening the health system for perinatal care. Hence, this study focused on deriving key health system domains and indicators, necessary to measure the performance of the health system for perinatal care in South African district hospitals. Understanding key contextual health system domains and indicators will in turn enable district hospital management teams to identify gaps in the health system for perinatal care, which need to be strengthened in order to alleviate the burden of perinatal deaths in district hospitals.

Aim of the study
The aim of the study was to develop a health system framework for perinatal care in South African district hospitals.

Study design
The Delphi technique was used in collecting data from a panel of experts, comprising disciplines connected with the functioning of the health system for perinatal care in South Africa. The Delphi technique was implemented as it allowed an iterative process of data collection and analysis until consensus was reached on key health system domains and indicators for perinatal care [15].

Study setting
Experts were based in two South African provinces, namely KwaZulu-Natal and Gauteng, but had a national perspective. The project reference group were based in two provinces in South Africa, namely KwaZulu-Natal and Western Cape, and had a national perspective. Indepth interviews were conducted in the experts' workplaces, and consensus was reached through electronic communication.

Sampling strategy and sample size
There were two groups of study participants: a group of experts and a project reference group.

Selection of experts
Thirteen experts, from whom data on key domains and indicators of a health system for perinatal care were derived, were purposively sampled to participate in the study. Experts were selected based on their expertise in perinatal care and/or on sound knowledge of the functioning of the health system for perinatal care in South Africa. A sample size of 10 is the recommended minimum for the Delphi technique [16]. The sample size of 13 was implemented to ensure maximum variation [17]. Table 1 presents background information of experts as well as the criteria that influenced their purposive selection.

Selection of the project reference group
Four experts in disciplines related to the functioning of health system for perinatal care were requested to be part of the project reference group, which provided guidance to the development of the framework and ascertained its relevance to the South African setting. The group comprised one obstetrician and one paediatrician, who were respectively responsible for governance of the obstetric and paediatric services in one South African province. The third member was an advanced midwife with postgraduate training in public health and with previous experience in governing perinatal care issues at provincial level. The three members of the project reference group who had experience of governing perinatal care issues at provincial level had a nuanced understanding of health systems performance for optimal perinatal care, as policies for implementation are governed at provincial level in South Africa. The fourth member was a public health specialist who had a strong publication record in health systems research and was recruited to particularly contribute to the components and domains of the health system framework.

Data collection and analysis
Data were collected and analysed in three rounds as described below.

Data collection and analysisfirst round
In the first round of data collection, the principal investigator conducted face-to-face in-depth interviews with experts. A data collection tool, comprising open-ended questions, was developed by the principal investigator and a co-investigator who had solid experience in qualitative research and health systems research. Questions were formulated to elicit health system factors with either positive or negative influence on perinatal outcomes. The developed data collection tool was used to interview first two experts. Data obtained were transcribed and analysed by both the principal investigator and the same co-investigator. The data collection tool was modified based on the nature, quality and depth of data obtained in the preliminary interviews. Specifically, the enhanced data collection tool covered questions specific to each health system building block. An experienced transcriber was hired to transcribe recordings of the subsequent 11 expert interviews, and the analysis was done by the principal investigator and the same coinvestigator.
Interview data was analysed using the directed content analysis approach, to identify codes and themes [18]. To ensure simplicity and comprehensiveness, the World Health Organisation (WHO) Health System Framework served as an organising framework for sorting codes and identifying major themes [19]. One member of the project reference group advised on the most correct formulation of domains and indicators based on South African guidelines for maternity care, and on the alignment between domains and building blocks [11,19].

Data collection and analysissecond round
A list of all extracted domains and indicators, categorized by building blocks, was sent to twelve experts, as the thirteenth expert was lost to follow-up. Experts were requested to rate the impact of each indicator on perinatal care and perinatal outcomes. Out of twelve experts who were requested to provide ratings, one expert declined further participation in the study. Therefore, ratings were received from eleven experts. The rating scale ranged from zero to five with the connotation presented in Table 2.
Inclusion of domains and indicators in the health system framework for perinatal care was determined by assessing the percentage of experts who gave a rating of four or five, stipulating that an indicator had either large or profound impact on perinatal care and perinatal outcomes. Frequencies and percentages of the ratings were calculated by the principal investigator using Stata version 13. The guidelines informing inclusion of indicators are presented in Table 3.
The results were discussed in a face-to-face meeting with all members of the project reference group who advised on relevance of the domains and indicators. In addition, the project reference group were provided with an opportunity to suggest additional domains and indicators, which they felt had been omitted through the previous round of data collection.

Data collection and analysisthird round
All indicators that were considered indeterminate in the second round, as well as additional indicators suggested by the project reference group were sent to eleven experts in the third round. At this stage, experts were requested to make a binary choice, and indicate whether they agree to include each indicator in the framework or not. All indicators that were deemed necessary to be included by at least 70% of experts were incorporated in the framework. Subsequently, all domains and indicators on which there was consensus were included in the framework. Table 4 presents the number of domains and indicators that were included in the framework in both the second and third rounds of data collection.

Results
The domains and indicators for health system performance for perinatal care derived from the three rounds of the Delphi technique are presented in Tables 5, 6

Leadership
Leadership is defined as the role of both senior hospital managers as well as labour and neonatal ward managers  in ensuring that the health system for perinatal care functions optimally. Leadership domains and indicators of health system performance, which by consensus were considered to have a major impact on perinatal mortality, are presented in Table 5.
Most indicators in the leadership building block are responsibilities of operational managers at ward level. However, the impact of senior hospital managers on perinatal outcomes was emphasised. Apart from ensuring availability and implementation of policies, both hospital and ward level operational managers are responsible to oversee overall functioning of the health system for optimal perinatal care. In recognition of the role of senior hospital managers in perinatal care, the Department of Health authorities have included 'ensuring improved perinatal outcomes' in the hospital managers' key results areas (KRA). In addition to ensuring optimal functioning of the health system for perinatal care, hospital managers should engage with communities concerning potential risk factors for perinatal deaths.
"….and if they [the CEOs] have picked up problems, they must communicate with the community; and say 'Guys, you come in late, that's why we've got so many perinatal deaths'. So that is the responsibility of the manager." .
"If you ignore the community, then maybe you must start forgetting about the whole thing [improved perinatal outcomes]. Honestly." [KI-03].

Health information system
The Health Information System building block focuses on routine data management processes as well as the use of data for decision-making. The processes include capturing, sorting, analysing and sharing knowledge amongst stakeholders. Domains and indicators of health information system at ward level and at district hospital level are presented in Table 6.
Experts highlighted that the value health care professionals attach to data has great influence on the quality of the data entered in source documents. Health care professionals are primarily trained to provide clinical care. As a result, the importance of health systems related data might not be apparent, more especially to newly qualified professionals. Ensuring that all health care professionals appreciate the importance of health systems related data, and its impact in improving quality of care and facilitating proper decision-making, was emphasised.
"If you are talking about a new nurse, a new midwife, data is the last thing on her mind…. Even myself as a new nurse back then, I didn't appreciate data." [KI-01].
"And the nurse has to know how important the data are, because without data we don't know where we're going." [KI-04].

Health workforce
In this study, health workforce refers to health care professionals who provide perinatal care. Several domains and indicators were identified, relating to the availability and competence of the health workforce, and are presented in Table 7.
The skill mix ratio, the number of health care professionals, and the availability of an anchor or a dedicated doctor are the domains identified to ensure the availability of an adequate and diverse health workforce to provide perinatal care. Clinical mentorship, training, drills and staff rotation are domains identified that reflect   Some doctors migrate to regional hospitals as they feel they get limited exposure with respect to the scope of clinical services they offer in district hospitals. The challenge of high staff turnover, more especially in rural district hospitals, has a potential to leave district hospitals with less competent staff. In addition, it may affect the quality of service delivery, as health care professionals become overwhelmed by high workloads.
"On a whole, yes they are competent, but that competent staff member, as I am speaking, she is handing in her resignation letter… The hospitals send them [clinical staff] for training, but they are only committed to stay for that certain period that they have signed in the contract. After that, they go. It's now become a culture, which, if one advanced midwife   In addition, teamwork amongst health care professionals was specified as an important domain for the performance of the health workforce. The main challenge mentioned was the lack of collaboration between doctors and nurses, particularly in conducting joint ESMOE and HBB drills, as well as in joint clinical decision-making.
"And the doctors sometimes don't listen to the nurses because they say, 'No, you're a nurse, you are supposed to listen to me. It's not [about] me listening to you.' I think that is wrong. So part of teamwork is that this patient that you're looking after, both of you are equally responsible. So if I see something as a nurse, I should be able to say, 'Doctor, this patient has got this, I think we should do this'. The doctor must not see that as a threat. And the same thing with a nurse, she should be able to listen to the doctor and respond appropriately. So what you see sometimes is that because the person is a doctor, he comes in as an intern, he's just got one year out of school and a nurse has been there for thirty years, who's got experience. All of a sudden, because he is a doctor, the name is 'doctor'; she [the nurse] must listen to this guy with one-year experience. So to me something is very wrong." [ KI 12].

Medical resources and infrastructure
The medical resources and infrastructure building block focuses on the availability of essential medicines, medical equipment, transport services and infrastructure. Basically, it measures the "hardware" of the health system for perinatal care [20]. The identified domains and indicators that must be monitored to ensure optimal perinatal care are presented in Table 8 below.
Experts strongly emphasised the consistent adequate supply of medicines for perinatal care. Reportedly, district hospitals have shortages of functional medical equipment, exacerbated by a tiresome procurement and replacement process. However, one expert believed that intrapartum care medical equipment is mainly important for mothers who experience complications during labour. However, health care professionals tend to over-rely on medical equipment for low-risk births in district hospitals.
"Again, going back to the fact that you can labour at home and deliver a very good nice bouncing baby. That's how we were made. We were made to deliver properly. But [when] in labour, then it's [important] to monitor [the labour]. That's why we have tools to monitor the labour; the partogram was made to identify a woman who struggles to follow the normal route." [KI-06].

Health financing
Health financing mainly refers to the role played by operational managers at ward level in ensuring that the wards that offer perinatal care have adequate resources. South African district hospitals offer free health services to mothers and neonates, and government financing is the only stream of funds for healthcare services [21]. Domains and indicators for health financing for perinatal care are presented in Table 9.
Apparently, hospital managers do not have full control on the total funds allocated to their hospitals. Funds are directly allocated to the entire hospital and the distribution of funds among the diverse services offered by the hospital is at the discretion of senior hospital managers.

Service delivery
Service delivery in this study refers to a process through which health care professionals, health programmes and policies are coordinated and implemented, to provide optimal perinatal care with the aim to minimized perinatal deaths. Key domains and indicators of service delivery are presented in Table 10.
The variety of domains and indicators relating to service delivery show that it is not enough for the labour and neonatal wards to offer the essential packages of care, but the manner and process of providing care is important. Of note is the involvement of senior hospital managers in ensuring optimal delivery of perinatal care services. Senior hospital managers are expected to be involved in the perinatal audit process; this also reflects their invaluable influence on perinatal outcomes.
"The position you are holding, the responsibility [for accountability] that is on you, makes people comply, without even saying anything, just by being there….. It gives [credibility to] the meeting. But the CEOs were not participating that much. Again, the decisions, it [availability of the CEOs in perinatal review meetings] also creates an opportunity for thorough decisionmaking. We will not need to have another forum meeting because the person who needs to make a decision is there. Actually, that person [the CEO] is the chairperson of the cash flow, if we need money to do something. Obviously, when we put a motivation, its going to be approved. You have already agreed with him in the meeting. It makes a lot of difference." [KI-02].
Essentially, the essence of strengthening health systems is to improve the quantity and quality of service delivery. Any weaknesses in the other building blocks of the health system eventually disadvantage service delivery. Thus, it is important that every actor in the health system play their role fully and continually, in order to alleviate the burden of perinatal deaths in South African district hospitals.

Discussion
The key contribution of this study has been to provide the Facility Based Health System Framework for Perinatal Care (FBHSF-PC) comprising domains and indicators that are necessary to measure health system performance for perinatal care in South African district hospitals. Most health system frameworks that have been developed to date are applicable at national level, and have not been translated for implementation at facility level [19,[22][23][24][25]. However, managers at facility level have a strong influence on health system functioning and on how policies are implemented and hence experienced by health care professionals and the communities they serve [26]. Moreover, the majority of existing health system frameworks are not designed for programme-specific health system strengthening interventions [19,22,27]. Further, no consensus exists on what are programme-specific health system strengthening strategies to deliver optimum programmespecific interventions [23,28]. Therefore, a framework to guide facility-level and programme-specific health system strengthening interventions is crucial.
Health facilities are at the frontline of service delivery, therefore their functionality has direct influence on health outcomes, which are the best measure of health system performance [29]. Since health facilities are the most peripheral organisational units within a decentralised health system, developments at facility level should accrue improvements at national level. Substantial progress has been noted globally in the delivery of health related interventions [19]. However, performance indicators aggregated at national level hide peripheral variation in performance, and therefore hide poor outcomes for the most vulnerable groups [19]. Focusing health system strengthening interventions at facility level will enable increased access to quality health services even for the most vulnerable groups in communities. Subsequently, the mandate of health systems to deliver efficient quality services to all people would be achieved [5,19].
Programme-specific health system strengthening interventions have been shown capable to improve targeted health outcomes [23,30]. For instance, application of health system strengthening interventions pertaining to maternal, neonatal and child health in several African countries had definite improvements on maternal and newborn health outcomes [23,30]. Remarkably, programme-specific health system strengthening have a great potential to strengthen the overall health system as interventions are implemented within a broader health system [23,31]. Since the functionality of a health system is evident in its capacity to deliver quality services for priority health issues, it is important to focus interventions to strengthening health systems for perinatal care, as maternal and newborn care are global health priority issues [5,27,32].
The broad structure of the FBHSF-PC aligns with the WHO Health Systems Framework which was used as an organising schema [19]. The alignment of FBHSF-PC to the WHO Health Systems Framework ensured that all critical health system building blocks previously identified as contributing to strengthening health systems are incorporated [19,29]. The detailed domains and indicators within each critical building block of the FBHSF-PC illuminate essential facility level and programme-specific elements requiring attention for optimal perinatal care. The FBHSF-PC has four additional components that are not featured in the WHO Health Systems Framework. These are: a) community engagement; b) teamwork amongst health care professionals; c) staff attitude; and d) the role of health facility managers in service delivery. Essentially, consistent engagement with communities with regard to health promotion and disease prevention is considered an effective and sustainable way to ensure improved health outcomes [22]. Hence, the Framework for Strengthening Health Systems to Improve Maternal, Neonatal and Child Health Outcomes and the Ouagadougou Declaration on Primary Health Care and Health Systems in Africa reflected community engagement as a major priority in the functioning of a health system [22,23]. Effective teamwork amongst health care professionals is key for ensuring effective and safe patient care, and particularly for perinatal care, as cooperation from multiple disciplines is necessary to provide effective quality care [11,33,34]. Poor staff attitude is one of the barriers to health systems providing equitable and respectful health care, and is mainly experienced in public sector health facilities [30,35,36]. In South Africa, poor attitude of health care professionals towards mothers has been experienced routinely [10,26,37]. Hence it has been considered a key factor for maternal health in South Africa [25]. It is therefore necessary for managers at facility level to monitor staff attitude in order to ensure that mothers are treated with respect. The role of senior facility managers in the FBHSF-PC is not constrained to governance and supervision, as is the case with WHO Health Systems Framework and the World Bank Control Knobs Framework [19,24]. It includes involvement in the perinatal audit process which has a direct impact on quality of services offered [38]. The four additional components included in the FBHSF-PC are crucial, as they strengthen the functioning of the health system at facility level.
The FBHSF-PC aligns well with the essential packages of care and the Guidelines for Maternity Care published in 2015 by the Ministry of Health in South Africa [11]. Therefore, application of the FBHSF-PC would be a first attempt to integrate health system strengthening with best practice clinical guidelines shown to be effective in the management of maternal and newborn health [39]. Therefore, the FBHSF-PC can become an effective and relevant mechanism to guide facility and unit managers in managing perinatal care in South African district hospitals.
The FBHSF-PC has been particularly designed to strengthen the health system for perinatal care for births occurring within a health facility. Thus, it does not address antenatal care challenges and patient related factors that are mainly associated with pre-admission perinatal deaths. Approximately 1.6% of women present with absent foetal heart on admission in South Africa [40]. Thus, implementation of the FBHSF-PC will ensure improvement of the performance of a health system in a majority of women who have live foetuses on admission.
The FBHSF-PC was validated during the course of its development. Specifically, face validity was assured through participation of the project reference group in ascertaining relevance of domains and indicators to the South African setting [41]. In addition, content validity was ensured by giving members of the project reference group an opportunity to suggest additional domains and indicators that were omitted by experts [41].
Through a Delphi technique, the FBHSF-PC identifies domains and key indicators to measure health system performance for perinatal care. The framework requires implementation for further refinement, which will involve an analysis of measurement challenges encountered. The FBHSF-PC does not measure the effect of critical building blocks and of overall health system performance on perinatal outcomes. Since the essence of strengthening health systems performance is to ensure improved health outcomes, there is a need to translate the FBHSF-PC to a model that will measure the impact of the identified building blocks on overall performance of health system, and in turn on perinatal outcomes [19,29]. That is, criterion validity of the FBSHF-PC needs to be determined [41]. In addition, since a health system is a set of interconnected elements that function cooperatively, interactions between critical building blocks must be defined [29]. Thus, research on the application of the FBHSF-PC must be conducted to ascertain its utility and to develop a model that will assist managers to recognise areas of intervention in the health system that will yield greatest improvements in perinatal outcomes.

Conclusions
Programme-specific health system strengthening interventions applied at facility level can yield optimum health outcomes. The proposed framework can empower district hospital management teams to identify gaps in the health system for perinatal care that need to be strengthened in order to alleviate the burden of perinatal deaths in district hospitals.