I should have started earlier, but I was not feeling ill! Perceptions of Kalenjin women on antenatal care and its implications on initial access and differentials in patterns of antenatal care utilization in rural Uasin Gishu County Kenya

Aim The aim of this study was to explore how Kalenjin women in rural Uasing Gishu County in Kenya perceive antenatal care and how their perceptions impede or motivate earlier access and continuous use of antenatal care services. Methods A study was conducted among 188 pregnant and post-natal mothers seeking care in 23 rural public health facilities. Gestational age at the initial antenatal care booking was established from their medical cards. Further researcher-administered questionnaire with closed and open-ended questions was used. Key informant interviews with traditional birth attendants (n = 6) and maternal and child health nursing officers (n = 6) were also conducted for triangulation. Descriptive statistics were applied using SPSS programme. The interviews of women who gave consent to be audio recorded (n = 52) were transcribed and thematically analysed using MAXQDA program, based on Andersen and Newman’s (1973) behavioural model of health services utilization. Results The mean gestational age at booking initial biomedical care was 23.36 weeks. Only 18 patients (10%) booked before 13 weeks and 45% made four or more visits. The main reasons given for early booking were: illness in index pregnancy (42%) checking the foetus position and monitoring foetus progress (7%). The main reasons given for late booking were: no reason (31%), was not feeling sick (16%), fear or shame due to unexpected pregnancy (13%). Almost half of the respondents (44%) used both biomedical and traditional antenatal care services. Main reasons for visiting traditional care were to: check foetus position and reposition it (63%), collect medicinal herbs (31%), relief discomforts through massage (18%). Conclusion Early antenatal care booking is meant for women with unpleasant physical signs and symptoms. Later ANC is meant to check foetus position and reposition it to cephalic presentation and monitor its progress and this is only possible if the foetus is large. Data for this study were collected as part of broader research investigating the social cultural context of nutrition in pregnancy and the utilization of nutrition intervention services in rural Uasin Gishu County in western part of Kenya. Maternal nutrition interventions in Kenya are offered free of charge in all government hospitals as part of routine ANC services [28]. Women’s utilization of ANC at health facilities plays a crucial role in uptake of these interventions. Exploring factors that influence ANC attendance was therefore a key objective of this research. For this reason, women who had at least one prior ANC visit in a health facility during the current pregnancy or had delivered a baby within one month were recruited for the study in order to elicit their experiences with nutrition interventions during their previous appointments. Uasin Gishu is one of the 47 counties of Kenya and it covers a total area of 3,345.2 km2 with a total estimated population of 1,023,656 [13]. Most settlements are rural (64.1%). The climatic conditions and soil type in this region are generally favourable for a wide range of livestock and crop production with an average rural land holding of 5 ha, hence the County is commonly known as the “country’s food basket” [29]. However, despite the food surplus in the county, maternal malnutrition indicators of Uasin Gishu are worse than the national norm, particularly with respect to stunting; statistics indicate 31.2% of children in Uasin Gishu are stunted compared to 26.0% nationally [8]. To establish the social cultural context of maternal nutrition in the county, Uasin Gishu County was thus purposively selected. The Kalenjin are the predominant ethnic population in Uasin Gishu County. This ethnic group is composed of smaller sub-ethnic groups (the Kipsigis, Nandi, Tugen, Keiyo, Marakwet, Pokot, Sabaot and the Terik) that share a common dialect and similar cultural traits. The Nandi occupies the largest settlement in Uasin Gishu County, followed by the Keiyo. There are 171 health facilities in the county, of which 90 are government owned and offer maternal care services free of charge [13]. Most of the facilities are concentrated in the county headquarters (Eldoret town). Uasin Gishu county is administratively divided into six sub-counties namely: Turbo, Soy, Moiben, Kapseret, Kesses and Burned forest. Each sub-county has a sub-county hospital equipped with one medical doctor, nurses, clinicians, a delivery room and maternity wards. However, these sub-county hospitals do not provide maternal services for high-risk women and so refer such cases to the county hospital, which is only one. These sub-county hospitals are the largest facilities in the rural areas and thus serve as referral centres within each sub-county. There are also other health centres (headed by a clinical officer) and dispensaries (headed by nursing officer) which offer ANC services for normal pregnancies but they are not equipped to attend deliveries. Study subjects were selected from the six sub-county hospitals. Pregnant women attending ANC between March and June 2017 were enrolled. Only Kalenjin women, who had at least one prior visit to an ANC during the current pregnancy or post-natal care within one month, were included. The number of women seeking care in the previous 6 months was determined by reviewing maternal-care registration records. This was used to estimate the number of women who would be attending the clinic during the period when the study was to be implemented. As per the hospital records, approximately 60–240 women seek maternal care per month in each of the six sub-county hospitals. Thus on average, a total 795 women were seen per month in these hospitals. Systematic sampling technique was used to select study participants where by every second woman who met the inclusion criteria was recruited until the minimum desired sample size of 188 was attained. This selection criterion excluded the following women: non-Kalenjin, pregnant and visiting ANC for the first time, unable or unwilling to participate. A researcher-administered questionnaire with closed and open-ended questions (S1 Doc) was chosen in order to provide room for probing, clarity of questions and enable participants to express their views on the topic in order to generate rich detailed insight information [30]. The questionnaire was developed after a literature review and discussions with nursing officers in charge of Maternal and Child Health (MCH). The topics covered in the questionnaire included: demographics, reasons for early ANC booking, reasons for late ANC booking, whether the interviewee had ever used TBA services for the current pregnancy, the gestational age and the number of times they visited TBA care, the nature of TBA care and their opinions on both TBA and ANC services. The data-collection exercise was coordinated and conducted by the first author with the help of four research assistants who were fluent in both local Kalenjin dialects, English and Swahili languages, with social science research experience. The research assistants were properly trained in the research instruments, language translation and they participated in the pilot study and review of research instruments after piloting to ensure consistency and inter-researcher reliability. Data were collected in the local language or Swahili depending on the preference of the respondent. If the respondent consented, her responses were noted and recorded and later transcribed verbatim and translated into English for analysis. If a respondent objected to being recorded, detailed notes were taken. Individual face-to-face interviews were preferred because they give the opportunity to observe respondents’ facial expressions and body language, particularly important for correct interpretation of the answers [31]. The individual interviews were conducted in a quiet private room at the health facilities to avoid distractions, ensure privacy and anonymity of the responses and enhance crystal-clear recordings [32]. Each woman was interviewed once and the interviews lasted for 30–45 minutes. The reliability of the findings was ensured through triangulation using cross-checking questions, observation, key informant interviews with TBAs who were also herbalists (n = 6) and nurses offering ANC (n = 6). The health workers were interviewed at their place of work whereas TBAs/herbalists were interviewed either in their homes or at the market centres where they sell their herbal medicines. The TBAs were identified by snowball and convenience sampling through respondents who gave birth at home and who took herbal remedies during pregnancy. Information gathered from key informants was further employed to explore meanings and enrich the responses obtained from the interviews with pregnant women. Research approval was obtained from the National Commission for Science, Technology and Innovation (NACOSTI/P/15/2335/5353; 2-Apr-2015) (S2 Doc). NACOSTI is a state corporation with the overall mandate to review and regulate the quality of science in the country and approve research studies. NACOSTI offers the researcher approval to conduct research activities in the community only if the study does not require further clearance from an ethical institutional review board. In this case, the study was not recommended for further ethical review. For this study additional permission were obtained from the Uasin-Gishu County Director of Health, the County Commissioner and the County Director of Education (S2 Doc). Institutional approval was obtained from the Moi University (Kenya) and the Vrije Universiteit Amsterdam (Netherlands) (S2 Doc), to undertake this research. Study participants were provided with information about the study before any consent to participate was sought. The participants were also informed of their right to abstain from participating in the study, or to withdraw from it at any time, without reprisal if they felt uncomfortable to continue with the study. Measures to ensure confidentiality of information was also provided. All respondents provided written informed consent to participate in the study. Informed consent from adolescents below 18 years was guided by Fisher et al’s (2003) recommendation that unlike younger adolescents, those over 16 can make informed decisions as well as adults [33]. Ruiz-Canela et al 2013, further recommends that “If adolescents are mature enough to understand the purpose of the proposed study and the involvement requested, then they are mature enough to consent” [34]. However in this case, the consent forms were read out to the adolescents in the presence of a legal guardian/parent and informed assent was sought from minors, while legal guardians/parents gave written informed consent. Statistical data were coded and analysed using SPSS software (version 23) to establish frequencies of descriptive statistics and the results were tabulated. The recordings were transcribed verbatim and translated into English with each participant being identified with a pseudonym, and these are used as narratives in the results section. With the help of MAXQDA 12.3.2 software, both notes and voices were further coded into themes and sub-themes based on Andersen and Newman’s behavioural model of health services utilization [35] as the initial coding guide. The Andersen and Newman model used in this study specifically focuses on individual determinants of general patterns in the use of health services. The underlying model assumes that use of health services is dependent on: (1) the predisposition of the individual to use services; (2) enabling factors (individual’s ability to secure services); and (3) need for care (individual’s illness level). Predisposing factors refer to the propensity to use health services and this can be predicted by individual characteristics that pre-date the onset of specific episodes of illness. These factors include demographics (e.g. age, sex, marital status and past illness), social-structural characteristics (e.g. education, occupation, family size, ethnicity, religion and residential mobility) and attitudinal-beliefs (e.g. values concerning health and illness, attitudes towards health services, and knowledge about medical care, physicians, and disease) of individuals, which influence people’s attitudes towards illness and care. According to Andersen and Newman predisposing variables as such are not considered to be a direct reason for using health services but do result in differences in inclination towards using them. Enabling factors: Enabling variables refer to means that make health service resources available to the individual patient and may emanate from the family or community. Enabling factors include family resources, such as income, level of health insurance coverage, or other source of third-party payment, whether the individual has a regular source of care, the nature of that regular source of care, and the accessibility of the source. In addition, the community resources in the area in which the family lives can affect the use of services, e.g. the number of health facilities and personnel in a community, the price of health services, region of the country and the rural or urban nature of the community. These variables might be linked to use because of local norms concerning how medicine should be practised or overriding community values that influence the behaviour of an individual living in the community. Illness level: Individuals or their family must perceive illness or the likelihood of its occurrence for them to make use of health services. Illness level represents the most immediate cause of using health services. Measures of perceived illness include number of disability days (days during which individuals are unable to do what they usually do, be that work, go to school, take care of the house, or play with other children), symptoms experienced in a given time period, and a self-report of general state of health (e.g. excellent, good, fair, or poor). In addition, evaluated illness measures determine the need for care. These measures are attempts to get at the actual illness that the patient is experiencing and the clinically judged severity of that illness. The symptoms reported by the individuals can also be weighed by physicians regarding the probability of need for care. According to this model, therefore, patients must perceive a need for care, they have to respond to this need and the patient’s environment must enable the search for care. Fig 1 presents each of these components of the model and the variables that operationalize them as established by Andersen and Newman. This model thus actively facilitated the exploration of respondents’ perception on ANC, integrating these perceptions into predisposing, enabling and need for care contexts to producing one behavioural outcome regarding maternal ANC seeking behaviour. Newly emerging codes in the transcripts were inductively added to the framework’s variables that correspond to the findings of this study to build our model of personal factors influencing ANC-seeking behaviour as indicated in Fig 2. When new codes or themes were added to the framework, all data were re-scrutinized several times to obtain a sense of the whole. Researchers with different backgrounds provided input to the analysis to increase its validity [36].