Barriers to formal health care seeking during pregnancy, childbirth and postnatal period: A qualitative study in Siaya County in rural Kenya

Background: There is broad agreement that antenatal care (ANC) interventions, skilled attendance at birth and management of complications arising after delivery are key strategies that can tackle the high burden of maternal mortality in sub-Saharan Africa. In Kenya, utilisation rate of these services has remained low despite a government policy on free maternal care. The present study sought to understand what factors are leading to the low healthcare seeking during pregnancy, child birth and postnatal period in Siaya County in Kenya. Methods: Six Focus Group Discussions were conducted with 50 women attending ANC in 6 public primary healthcare facilities. Participants were drawn from a sample of 200 women who were eligible participants in a Conditional Cash Transfer project aimed at increasing utilization of healthcare services during pregnancy and postnatal period. Interviews were conducted at the health facilities, recorded, transcribed and analysed using thematic analysis. Results: Multiple factors beyond the commonly reported distance to health facility and lack of transportation and finances explained the low utilization of services. Emergent themes included a lack of understanding of the role of ANC beyond the treatment of regular ailments. Women with no complicated pregnancies therefore missed or went in late for the visits. A missed health visit contributed to future missed visits, not just for ANC but also for facility delivery and postnatal care. The underlying cause of this relationship was a fear of reprimand from the health staff and denial of care. The negative attitude of the health workers explained the pervasive fear expressed by the participants, as well as being on its own a reason for not making the visits. The effect was not just on the woman with the negative experience, but spiraled and affected the decision of other women and their social networks. Conclusions: The complexity of the barriers to healthcare visits implies that narrow focused solutions are unlikely to succeed. Instead, there should broad-based solutions that focus on the entire continuum of maternal care with a special focus on ANC. There is an urgent need to shift the negative attitude of healthcare workers towards their clients. This study was part of a broader pilot study investigating the role of conditional cash transfers (CCTs) in motivating health facility visits for ANC, delivery and postnatal checks in Siaya County. The results of this qualitative phase was meant to inform the design and delivery of an intervention to tackle the identified barriers to health visits in the County, which is currently underway [39]. Participants were drawn from pregnant women attending ANC at public primary health care facilities in Boro Division in Siaya County. Boro is one of the four Divisions in Siaya, and was purposively selected for the study by recommendation of the County Health Committee because it exhibited the lowest service utilisation rates by pregnant women in Siaya County. Recruitment was done in six healthcare facilities, from where we drew a random sample of 10 women per facility, giving a total sample size of 60 women. Enrolment in the facilities was carried out by incentivized health staff with a research staff supervising the process. The enrolment was continuous and ceased when the target sample size for each facility had been accomplished. Criteria for enrolment were that the woman had a pregnancy of 6 months or less, was a long-term resident of Boro Division (6 months and above) and had no plans to relocate in the next 12 months. Due to the nature of the intervention study whereby cash was being transferred to participants through mobile phone wallets (M-pesa), an additional enrolment criterion was for the women to have access to a mobile phone. The phone could belong to them or to a member of their household or anyone else they trust. There is very high mobile phone penetration in Kenya, and this criterion was easily met by all the women who were approached as potential participants. There were no refusals to participate in the study. The qualitative data was collected using focus group discussions (FGDs) that took place in August 2015. Out of the 60 recruited women, 50 participated in the FGDs. In total, there were six FGDs, each one comprising of between 7 and 10 participants. Those who did not participate had either moved from the study location, miscarried or were untraceable. The FGDs were administered in the local language (Luo) by two local researchers who are familiar with the setting and moderated by the Field Coordinator for the study (ASO). The design of the data collection tools (FGD guides) was coordinated by the Principal Investigator (CAO) who is also conversant with the local sociocultural context and language; with further input from the research advisory team. CAO trained the local researchers on how to use the FGD guides and led the piloting of the tools. All the FGDs took place at the health facilities where the women were enrolled, which tended to be their nearest health facility and the one they would seek most of their care from. The six health facilities were Segere, Kadenge, Kaluo, Karuoth, Boro and Nyathi. The health facilities were the only neutral and confidential venues available in the village where the women felt comfortable gathering in. The healthcare staff facilitated venue arrangements for the discussions but were not present during the interviews. Refreshments were provided to participants before commencement of the discussions which lasted for about 1 h. Participants also received a transport reimbursement of Ksh. 300 (USD 3). Prior to the FGDs, the moderators reiterated the details of the study and the aim of the FGDs as well as what taking part in the FGDs meant for the women individually and as a community. This was done to ensure that the women fully understood what they consented to. They were then asked to sign consent forms if they were happy to proceed with the FGDs. They all opted for oral consent. The moderators therefore read out the contents of the consent form and the participants verbally agreed to the contents and to take part in the study. The FGDs were conducted using a loosely structured question guide (Additional file 1) that had been piloted with 5 pregnant women in Sigomre health centre, a health facility in Siaya county but outside Boro division where the study was being conducted. The piloting was done to ensure that the final FGD guides had clarity and sensitivity to cultural nuances about health and personal life issues. All the FDGs were audio-recorded with the respondents’ consent. These were complemented with notes taking on key issues that emerged during the discussions. At the end of each FGD, one of the moderators summed up the key issues discussed in the sessions as part of data validation and verified with the respondents their take on the proceedings as part of data quality assurance. The moderators also sought participants’ permission to contact them at a later date in case there was need for clarification of some issues they raised or additional information during data analyses. All the participants consented to follow up contact and also to be involved in verification of the findings. After each FGD, the moderators held debriefing sessions to reflect on the process, share their observations and identify ways of improving the process in the subsequent FGDs. Data coding and analysis was carried out by the CAO, under the guidance of a research advisor (JAO). The analytical approach used for this paper was largely transcript and notes based thematic analysis [23, 24]. Data analysis started with familiarization through listening to the audio tapes several times followed by verbatim transcription of each FGD audio tape. As Krueger and Casey [23] have noted, it is important to listen to someone’s story and understand it well before telling it, because that is what analyses of focus group data entails. The transcripts were then re-read repeatedly, noting down any points that stood out from the initial readings. The transcripts were then subjected to coding, initially assigning descriptor statements to text that presented particular meanings. This was done for each of the FGD transcripts and then compared across the transcripts. The process entailed breaking down the data and reassembling it together through constant comparison [24] of transcripts and notes to tease out underlying meanings. Emerging themes were identified, checked to see if they corresponded with the extracted data linked to the codes attached to them, and if they fitted the entire data from the transcripts and the notes. After retrieving all the emerging themes, they were then redefined and renamed to develop clear story lines from the data. These themes and related issues were also compared with data from literature to check for similarities and differences. For example, barriers to ANC attendance reported by the present participants such as distance to facilities and financial constraints were checked against reports from related research from other low resource settings.