Study area
The study took place in Ondo State, Nigeria, between May and September 2016. According to the most recent population estimate, around 4,671,700 people were living in the Ondo State in 2016 [17]. The state is located in the Southwestern region of Nigeria and populated mainly by the Yoruba speaking people of Nigeria. Subsistence farming, fishing, and trading are the main occupations of people in the state. The state has a higher proportion of urban dwellers compared to rural dwellers. Ondo State has the worst maternal outcomes in the region, according to the 2013 and 2008 NDHS reports. In 2013, up to 15% of pregnant women did not attend antenatal care, and about 44% of them did not utilise a skilled birth attendant for child delivery [6]. According to the Ondo State Ministry of Health, 40% of women of reproductive age are using modern contraception, and 86% of HIV+ pregnant women are on anti-retroviral therapy [14].
Study design and sampling
This cross-sectional evaluation population survey was conducted between May and September 2016. The current survey is referred to as the endline survey in this study, and the 2013 NDHS is referred to as the baseline survey. The sample size of 409 was estimated using the sample size calculator (http://web1.sph.emory.edu/users/cdckms/samplesize%20icc%20deff2.html), at a confidence coefficient of 95%, a confidence interval width of ±5, 800,000 live births over the 5 years (2011–2015), design effect of 2, 15 observations per enumeration areas and a total of 27 enumeration areas. A two-stage cluster random sampling method facilitated the selection of a representative sample of women included in the study. The state was clustered into enumeration units and stratified based on rural and urban areas. Through simple random sampling, the research staff identified Enumeration Areas (EAs) from the list of EAs in the 2006 census, with probability proportional to size. We selected a minimum of 15 households in each EA until they reached a sample of 409. To accommodate new houses built since 2006, which were not included in EAs, they selected every 10th household in each EA. We skipped households without women who gave birth over the period of study, and we only selected one woman per household, irrespective of the number of eligible women living in the household. The overview of the participants’ selected is presented in Fig. 1.
Baseline survey
Since there was no baseline survey available, we used the data from the 2013 NDHS as a baseline to enable us to assess improvement in the use of maternal health care services post-implementation of the “Abiye” programme. The 2013 NDHS data is publicly available and could be obtained from Measure DHS (https://www.dhsprogram.com/Data/). The sampling method used in the DHS is similar to that of the current study. We downloaded the women recode data and extracted relevant variables on maternal health care utilisation and socio-demographic characteristics. A total of 432 women in Ondo State selected using 2-stage cluster random sampling were included in the NDHS 2013 survey. Our comparison is possible given that we used the DHS questionnaire for the current study.
Participants
Participants were women within the reproductive age (15–49 years) that gave birth to at least one child over the 5 years (2011–2015) post-implementation of the “Abiye” programme in Ondo state. Women who did not deliver between 2011 and 2015 were excluded. Women aged 50 years or more were excluded.
Method of data collection
Trained research assistants, together with the first author, administered the questionnaire to the study participants. The questionnaire consisted of pre-validated questions adapted from the Nigeria Demographic and Health Survey [6]. Before administering the questionnaire, research staff conducted a pilot study among 20 women in a different state.
Outcome measures
The two main outcome measures in this study are the use of antenatal care and child delivery care services. Use of antenatal care was measured by asking the following questions: When you got pregnant after your last birth, did you see anyone for prenatal care for this pregnancy? Whom did you see? Where did you receive antenatal care for this pregnancy? Each question had a nominal level response category (yes/no) measure and assessed an important dimension of the use of prenatal care. All participants were asked where they delivered their index pregnancy to assess utilisation of facility-based child delivery,. Responses were categorised as follows: home, traditional birth attendants, faith-based facilities, private health facilities and public health facilities.
Covariates
The covariates included in the study are; age, place of residence, level of education, religion, ethnic groups, income, employment status, marital status, watching of television, and wealth status. Age was measured as a continuous variable but later categorised into three categories (15–24, 25–34, and 35–49 years). To measure the level of education, two questions were posed: 1) “Have you ever attended school?” and, 2) “What was the highest level of school you attended?” The first question had a nominal response category (that is, yes or no), while the second question had an ordinal level of measurement (primary, secondary, and higher education) as the response category. To measure employment and income, participants were asked: “Do you have a job? How much is your income monthly?” A categorical response (yes or no) to the first question would suffice, and an interval measurement summarised the answers to the second question. An income between 1 to 20,000 Naira scored one; two indicated an income above 20,000 Naira. Participants were asked if they watch television, and a ‘yes’ or ‘no’ response category was assigned.
Socioeconomic status
In this study, ten questions on levels of education, income, employment status, ownership of mobile phones, regular watching of television, use of bank accounts and the Internet were used to measure socioeconomic status. Participants’ socioeconomic status was derived by summing up the scores accrued to each participant from the questions assessing the level of education, employment status, income, ownership of mobile phones, regular watching of television, use of bank accounts and the Internet. A total score of 10 represented the highest socioeconomic status. A score between 0 and 4 was regarded as low socioeconomic status, scores between 5 and 7 as a moderate socioeconomic status, while a score between 8 and 10 reflected high socioeconomic status.
Statistical analysis
The data analysed here were drawn from a larger study, which examined maternal outcomes in the context of free maternal health care [18]. Obtained data were coded and captured into the Statistical Package for the Social Sciences (SPSS version 24). Mean and frequency distribution of all variables of interest were computed. Chi-square statistics and Fisher’s exact test were used to examine determinants of maternal health care services utilisation. Also, adjusted and unadjusted logistic regression models were computed to examine the determinants of maternal health care services utilisation. Alpha values less than 0.05 were considered to be statistically significant. To account for the complex sampling strategy, sampling weight was applied, and analysis was performed using the complex sample feature of SPSS.
Ethical considerations
The University of Fort Hare’s Research Ethical Committee (UREC) approved the study protocol (AKP031SAJA01). Furthermore, the Ondo State Health Research Ethics Committee (OSHREC) reviewed and approved the study protocol [NHREC/18/08/2016]. Community leaders and household heads in the study settings granted the researchers permission to conduct the study. Participation was voluntary, and all study participants signed written informed consent. Rights to privacy and anonymity were respected throughout the study. Parent consent and assent were obtained for the inclusion of a few respondents less than 18 years in the DHS data.