Background: To reduce financial barriers to access, and improve access to and use of skilled maternal and newborn healthcare services, the government of Ghana, in 2003, implemented a new maternal healthcare policy that provided free maternity care services in all public and mission healthcare facilities. Although supervised delivery in Ghana has increased from 47% in 2003 to 55% in 2010, strikingly high maternal mortality ratio and low percentage of skilled attendance are still recorded in many parts of the country. Methods: We conducted qualitative research with 185 expectant and lactating mothers and 20 healthcare providers in six communities in Ghana between November 2011 and May 2012. We used Attride-Stirling’s thematic network analysis framework to analyze and present our data. Results: We found that in addition to limited and unequal distribution of skilled maternity care services, women’s experiences of intimidation in healthcare facilities, unfriendly healthcare providers, cultural insensitivity, long waiting time before care is received, limited birthing choices, poor care quality, lack of privacy at healthcare facilities, and difficulties relating to arranging suitable transportation were important health system barriers to increased and equitable access and use of services in Ghana. Conclusion: Our findings highlight how a focus on patient-side factors can conceal the fact that many health systems and maternity healthcare facilities in low-income settings such as Ghana are still chronically under-resourced and incapable of effectively providing an acceptable minimum quality of care in the event of serious obstetric complications. Efforts to encourage continued use of maternity care services, especially skilled assistance at delivery, should focus on addressing those negative attributes of the healthcare system that discourage access and use.
The qualitative data reported in this paper were extracted from within a larger, original study that the authors conducted to examine the effects of Ghana’s free maternal healthcare policy on women’s maternity care seeking experience, equity of access, and barriers to accessibility and utilization of maternal and newborn healthcare services. The design of this larger study followed a mixed methods approach; involving analysis of a nationally representative retrospective household survey data in combination with qualitative exploration using data generated from anthropological research techniques of focus group discussions (FGDs), key informant interviews (KIIs), case studies and structured field observations. While other aspects of the qualitative data from the larger study have been analysed, written up and submitted for peer-review by other journals, this paper focuses on and report findings from an aspect of the qualitative study that explored aspects of Ghana’s maternity care delivery system that discourage use of skilled care. Empirical research was conducted in Ghana during a total of 6 months between November 2011 and May 2012 in a total of 6 purposively sampled communities namely, Kuntanase, Abono, and Piase in the Bosomtwe district of the Ashanti region; and Mpaha, Sankpala and Tidrope in the Central Gonja district of the Northern region. We chose Ghana for this research not only because maternal health is seen as a ‘national emergency’, but also because Ghana presents an interesting ideal case study. It is one of only a handful of countries in Sub-Saharan Africa to have actively started implementing both universal maternity care and health insurance policies at the national level. The economic and political conditions in Ghana also make the country an interesting case study. Ghana is situated within the predominantly economically marginalized and politically unstable region of West Africa, but forms an exception. Relatively, Ghana is a small fledgling multicultural, multi-ethnic, multi-religious, and multi-party constitutional democracy, characterized by vibrant civil society activism and media pluralism. It is politically stable; a rebasing of its economy in November 2010 saw the country leap into the category of lower-middle-income countries [16]; it also recently started producing oil in commercial quantities; and is often touted as one of the most politically and economically progressive countries in the region. Although Ghana has since 2013 been experiencing economic deterioration – with some private healthcare providers including the Christian Health Association of Ghana (CHAG) cancelling their contracts with the National Health Insurance Scheme due to heavy indebtedness, thereby forcing all patients whether insured or not to resort to the old cash-and-carry system – the country is still often considered ‘an example of global good practice’ [16,18]. Despite the ‘exceptional status’ Ghana enjoys in the sub-region, maternal, neonatal and infant mortality ratios have remained persistently high. Apart from capturing a divide between a relatively destitute northern Ghana and a relatively prosperous southern Ghana, we chose the six communities to provide a diversity of social and health situations that are largely representative of the country, Ghana (Table 1). For instance to capture any differences that may exist between urban and rural communities as well as communities with at least one government or mission healthcare facility and those without, we selected Kuntanase to represent urban communities, Piase, Sankpala and Mpaha to represent rural communities with health facilities, and Abono and Tidrope to represent rural communities without any health facility. Although the six communities demonstrated variable levels of performance on maternal and newborn health indicators, access to, and use of antenatal, delivery, post-delivery and newborn care services were mostly below the national averages (Table 1). Basic characteristics of the study communities We obtained clearance from the University of Oxford Social Sciences and Humanities Inter-divisional Research Ethics Committee (Ref No.: SSD/CUREC1/11-051), and the Ghana Health Service Ethical Review Committee (Protocol ID NO: GHS-ERC 18/11/11). In addition, both informed written and verbal consent were obtained from all research participants. In all cases, consent was obtained after it was thoroughly explained to participants that their participation was entirely voluntary and that information obtained will be used for the purposes of this research only. To thank research participants, we bought biscuits and soft drinks to refresh them. The research participants were drawn from a population of pregnant women and lactating mothers from the six research communities, and from health personnel from the two regional and district health directorates. For the purposes of this research, we classified participants into ‘Women’ and ‘Healthcare Providers’. ‘Women’ here refers to women who were pregnant at the time of this research or had given birth between January 2011 and May 2012 in the six study communities. Healthcare providers comprised health professionals (doctors, nurses, midwives, healthcare managers, and health policy makers or implementers) from health facilities in the study communities, district and regional health directorates, and Ghana Health Service at the national level. The rationale for interviewing diverse actors at multiple levels was to allow for an exploration of multiple perspectives. In all, 205 participants took part in the study. Of this number, 90% (185) were pregnant women and lactating mothers. The remaining 10% (20) consisted of healthcare providers. Our strategy for recruiting research participants involved both probability and non-probability sampling procedures. For all research participants under the ‘healthcare providers’ category, a purposive sampling technique was used. This was a judgmental selection based on the participant’s knowledge of the subject of study, the participant’s role in the selected health institutions, and our evaluation and/or perception of the relevance of that role and knowledge to the research topic [19]. For the women however, a simple random sampling procedure was used. The actual sampling and recruitment procedure involved four main steps. First, we enumerated all pregnant and lactating mothers in each of the study communities using a five-item short questionnaire that we designed. The questionnaire asked whether a woman was currently pregnant or had given birth since January 2011, the name of the woman, age and house number/name. Second, after the listing was completed, we randomly selected the required number of participants from the pool of names in each study community. We predetermined the required number of participants (5% of the total enumerated population of pregnant and lactating mothers of each study community). We believe sampling 5% of the enumerated population in each community yielded sufficient numbers of participants whose views on the research topic could fairly approximate the views of the majority of women in the community. Third, the randomly chosen participants were further randomly allocated to either focus group or in-depth interview. Finally, we took the randomly selected names to the various communities wherein the research was introduced and the selection procedures thoroughly explained to each of the randomly selected women. Thereafter, the women were invited to participate in the study. Where any of the randomly selected women was not available or declined to participate in the study – and there were only 2 of such cases – we repeated the selection process to get a replacement. We acknowledge that the emphasis of qualitative research is not always on generalisation hence randomisation might not be a necessary requirement [40]. However, we used simple random sampling in this study as a pragmatic and ethical strategy to assure justice by using a fair and transparent sampling procedure that ensured that every pregnant and lactating mother in the study communities had a fair chance of taking part in the research. In fact, the idea of chance – which was embedded in our sampling procedures – helped to eliminate questions about why one woman was included and another excluded from the study. Focus group discussions and key informant interviews were the main data collection methods. We adopted focus groups partly because we believe it had the capability to reproduce women’s experiences of seeking maternity care in a normal peer-group interpersonal exchange. Six (6) focus group discussions – one in each community and involving a total of 104 pregnant women and lactating mothers – were completed. Groups consisted of 17–24 participants. This difference was mainly due to differences in the sizes of the target populations. All focus groups were held in the study communities, at venues chosen in consultation with participants and community gatekeepers. Each focus group lasted 1.30 to 2 hours, and ended when a point of saturation was reached i.e. when no new ideas and issues seemed to arise. All discussions were conducted in the local dialects – Twi in Kuntanase, Abono and Piase; Dagbani in Sankpala and Tidrope; and Gonja in Mpaha. This is not only because the literacy [written or spoken English] rates are low among the study participants, but also because we wanted to ensure that the interview language was not a barrier to effectively exploiting the full benefits of focus groups. Because our knowledge of the interview language was limited, we engaged one female research assistant from each community to facilitate the discussions. Prior to the interviews, these research assistants were trained for three days. The training took the form of classroom lectures on the objectives of the research, a question-by-question explanation of the content of the topic guides, instructions on field and interviewing procedures to be followed, and instructions on how to ask sensitive questions. To reinforce the training interviewers had received and to reduce potential bias in questioning, the training also involved mock or practice field interviewing, during which each of the research assistants interviewed 3 women using the relevant interview language. These pilot interviews were transcribed and translated into English by independent language translators. The results of the pretest suggested that all research assistants performed well in terms of asking questions in the correct way. Where any errors were detected, they were discussed with the appropriate interviewer to ensure that the same errors did not reoccur in the main interview. To further explore the research question, we conducted key informant interviews to complement the focus groups. The need for a mixed data collection technique in the social aspects of disease and health research has been widely discussed [20]. In particular, it has been argued that people may not necessarily tell the truth in any objective sense when it comes to sensitive issues such as health and disease within a group context [21]. To overcome this, we complemented our focus groups by conducting key informant interviews. Key informant interviews were also used because it was difficult to organise focus group discussions with healthcare providers. This was largely because it was extremely difficult to get a group of healthcare workers to agree on a common time period to hold FGDs. This was due to the different duty-schedules of individual healthcare workers. This was overcome more appropriately by conducting individual interviews. One major advantage of the method was its ability to address sensitive issues such as personal experiences and perceptions with regard to distribution, accessibility to, and utilization of maternity care services. In all, a total of 101 key informant interviews were completed – 81 with pregnant and lactating mothers, and 20 with selected healthcare providers. The distribution of the 81 women across the six study communities was 21, 11, 15, 12, 13, and 9 in Kuntanase, Abono, Piase, Sankpala, Mpaha, and Tidrope respectively. Interviews lasted 10 to 15 minutes. All interviews with women were conducted in Twi, Dagbani, and Gonja. Interviews with healthcare providers were however conducted in English. In all focus groups and key informant interviews, we used an open-ended thematic topic guide. The instrument was designed to ensure that similar themes were covered in each discussion or interview. The instrument however had built-in flexibility that allowed questioning to flow naturally while permitting us to pick at random and probe more on any pertinent but unexpected issues that arose during the interview process. The instruments focused primarily on exploring women’s experiences of seeking or not seeking maternity care services, issues regarding coverage, utilization and access, women’s interaction with maternal and newborn healthcare services, the barriers to access and use of services, and perceptions of in/equities in maternal health services accessibility and utilisation. In addition, a more structured questionnaire instrument that sought to collect specific socio-demographic information as well as elicit specific response (e.g. did you or will you give birth at the hospital, clinic or home?) from all women was designed and administered individually to all women. To ensure that the instrument was reliable, we engaged in a continuous review of the questions and the interview process to ensure that they were eliciting the right answers to the right questions. This proved valuable in enabling us to reframe questions, clarify and use more appropriate or easily understandable concepts as the research progressed. All discussions and interviews were tape-recorded alongside hand-written field notes. To ensure that the qualitative data analysis process was methodical and transparent, we followed the Attride-Stirling’s thematic network analysis framework [22]. Several steps were followed before the analysis proceeded to an interpretative phase in which the networks were connected into an explanatory framework consistent with the text. The first step involved transcription and reading of transcripts and field notes for overall understanding. During and after qualitative data collection, the first author and three other language specialists – Twi, Dagbani and Gonja – transcribed all tape-recorded interviews. The first author then immersed himself in all transcripts and interview notes through reading and reviewing for overall understanding and comprehension of meaning. This first step was completed with a separate summary of each transcript outlining the key points participants made in response to the questions. Once the data was reviewed and a general understanding of the scope and contexts of key experiences was attained, the interview transcripts were exported to NVivo 9 qualitative data analysis software, where the data was both deductively and inductively coded. Codes, according to Miles and Huberman [23] are labels, which are assigned to whole or segments of transcripts and interview notes to help catalogue key concepts while preserving the context in which these concepts occur. Coding provided us with a formal system to organise the data, uncovering and documenting additional links within and between concepts and experiences in the data. Data coding continued until theoretical saturation was reached. This was a point where no new concepts emerged from successive reviewing and coding of data [24]. At this stage, the code structure was deemed complete and then applied to develop and report themes. Themes simply captured something important about the data in relation to the research question, and represented some level of patterned response or meaning within the data set [25]. Finally, all the themes identified in the previous steps were assembled and a thematic chart was drawn to reflect basic themes, organising themes, and global themes (Table 2). Thematic network analysis framework (from codes to global themes) To ensure that the thematic chart reflected and supported the data, we went through the data segments related to each basic, organising, and global theme. Where necessary, refinements were made. In total, 23 codes were identified. These were grouped into 11 basic themes that were further clustered into 4 organising themes, and 2 global themes (Table 2). These global, organising and basic themes form the structure of our findings and discussion section. Where appropriate, we use verbatim quotations from interview transcripts to illustrate responses related to relevant themes.
N/A