Study setting
The study was carried out in the Upper East Region of Ghana. Located in the northernmost part of Ghana, the region has a population of about 1,046,545 with an annual growth rate of 1.2% [21]. The region is predominantly rural, with 79% of its residents located in rural areas and as high as 72.3% of households are headed by males, total fertility rate is 3.43 and under-five mortality stands at 128 per 1000 live births [21]. The region is made up of about seven ethnic groups all of which are patriarchal in nature. Social research has documented customs of gender stratification that constrain female autonomy with respect to decision making, household governance and health seeking behavior [22]. Subsistence agriculture constitutes the mainstay of the economy of this region contributing as high as 83.7% of households’ economic activity [21, 22]. The area is characterized by two main seasons, a short rainy season from May to September and a long dry season from October to April. With such pervasive dependence on subsistence rain-fed agriculture, the region ranks among Ghana’s most impoverished regions [19, 22, 23], with low literacy and a per capita income level that is estimated to be about a quarter of the level estimated for Ghana as a whole [24].
Study design
This paper makes use of data from a cross-sectional baseline survey of the Ghana Essential Health Intervention Project (GEHIP) conducted in 2011. The purpose of this survey was to provide the project team with detailed information on maternal and child health, fertility, family planning, universal health coverage and other basic health indicators for eventual evaluation of the project.
GEHIP is a five year health systems strengthening and research program being conducted in the Upper East region of Ghana. It seeks to improve Ghana’s comprehensive primary healthcare initiative, known as the Community-based Health Planning and Services Program (CHPS). CHPS provides a wide range of essential preventive and curative services to some of Ghana’s most rural and impoverished locations but has however been faced with serious service gaps and operational flaws preventing it from achieving its full potential [20]. The GEHIP strategy includes introducing a series of training and technical assistance programs aimed at strengthening the capacity of the health systems. Also, additional maternal and child health interventions are being added and increased support provided to system structures to enhance overall effectiveness [20]. GEHIP is being implemented by the Upper East Regional Health Directorate of the Ghana Health Service (GHS) with technical assistance from the Mailman School of Public Health, Columbia University and the school of Public Health of the University of Ghana. Navrongo Health Research Center (NHRC) has the responsibility to provide data for the evaluation of GEHIP.
With the exception of the Kassena-Nankana East and West Districts which are research sites of the Navrongo Health Research Centre (NHRC), the other seven districts in the region as of 2011 were involved in the survey. The districts include: Bawku East Municipality, Bawku West, Talensi-Nabdam, Garu-Tempane, Bongo, Builsa and Bolgatanga Municipality. The sampling unit of the survey was the household, which is defined as ‘a person or group of persons living together in the same house or compound, sharing the same housekeeping arrangements and being catered for as one unit’ [25]. The sample frame was obtained from the Ghana Statistical Service (GSS). A two-stage sampling procedure was involved, the GSS sampled 66 enumeration areas (EAs) and Probability sampling proportional to population size was used, all women age 15–49 years in sampled households were eligible to be interviewed using a structured questionnaire. A total of about 5400 women were interviewed, however, this paper is based on 3,975 of the women representing those who have ever given birth.
Socio-economic and demographic variables
The variables used in this study included mother’s age, mothers’ highest level of education, rural/urban residence, mother’s marital status, occupation, religion, autonomy, mother’s involvement in polygamous marriage, contraceptive use, national health insurance registration status and household socio economic status.
Data analysis
STATA 11.2 was used for all the analysis. For the purpose of this paper, the unit of analysis is respondents who had experienced child birth and the measurable outcome was the experience of under-five death. Mothers’ age was recoded into three categories (15–19, 20–34 and 35–49). Mother’s highest level of education was also categories into three: those whose highest level of education was primary or junior high school were merged into one category while senior high school and tertiary were also merged into another. Respondents’ marital status was recoded: all who had reported to be single, widowed, divorced or separated were merged and labeled as ‘not married’. Place of residence was also recorded as either urban or rural. Occupation was recoded into five categories: all respondents who were involved in trading, craftsmanship, dressmaking etc. were merged and classified as “self-employed”. The other categories under occupation are: farming, government employed, student and others.
In determining the variable autonomy, a question was posed to respondents as to who often makes decisions about major household purchases and this was recorded, respondents who answered that they personally made those decisions themselves and those who answered that they do that jointly with their husbands were merged together and considered to be autonomous while respondents who said that it was the sole prerogative of their husbands and those who answered that someone else other than they or their husbands, were merged and considered ‘not autonomous’.
We used principal component analysis (PCA) a multivariate approach to compute wealth index by using household assets as a proxy to estimate household socioeconomic status. In all 13 household items (assets) were involved in the analysis and wealth index was classified into 3 socio-economic tertiles namely: relatively poor, middle and rich.
Bivariate analysis using chi square test was used to test for association between the independent outcome variables and ever experiencing under-five death. All variables that showed significant association (p < 0.05) in the bivariate analysis were then included in the multivariate analysis (logistic regression model). We calculated odds ratios with each covariate. Variables used in the logistic regression were first tested for multi co-linearity using the variance inflation factor (VIF) and this was found not to be a problem with an average VIF of 2.54 (VIF more than 20 indicates multi co-linearity). In calculating the odds ratio for each category of the independent variables, the first group was always taken as the reference category.
Ethical consideration
Before the baseline was conducted for GEHIP, ethical clearance was sought from ethics committees of the Ghana Health Service and the Navrongo Health Research Centre Institutional Review Board (NHRC IRB). Inform Consent was obtained from each participant. Also data sets used were anonymous of participant identity.