# Socioeconomic inequalities in access and use of skilled birth attendants during childbirth in Ghana: a decomposition analysis – BMC Pregnancy and Childbirth

Dec 31, 2021

### Data source

The data used in this study was derived from the 2014 Ghana Demographic and Health Survey, which was collected by the Demographic and Health Surveys (DHS) Program. Information for the analysis was drawn from the DHS Women’s Questionnaire focusing on women who gave birth in the preceding the survey year. A two-stage sampling method was employed. A total of 427 clusters were selected in the first stage. Employing a systematic sampling for the second stage, 30 households were selected from each cluster totalling 12,831 households being selected of which 11,835 households were successfully interviewed. From the 11,835 households that were successfully interviewed, 9656 women aged 15–49 years were eligible for an individual interview at a response rate of 97% [16]. For this study, 1305 women between the ages of 15–49 years who had live births the year preceding the survey year were selected. The choice of a year before the survey date was to avoid the issue of recall bias from respondents.

### Study variable

#### Outcome variable

The outcome variable for this study is whether women who had a live birth the year preceding the interview year had deliveries assisted by skilled birth attendants or not. The outcome variable is a binary outcome where a value of “1” was given if the delivery was assisted by a skilled birth attendant and “0” if the delivery was not. A skilled birth attendant in this study was defined as a trained and licenced health professional i.e., a doctor, nurse/midwife or community health officer who provides basic and emergency healthcare services to women and their new-borns during pregnancy, delivery, and the immediate postpartum period (i.e., the first 48 h after delivery). Information on delivery assisted or attended by a skilled birth attendant was based on the question “Who assisted in the delivery of (NAME OF CHILD)” in the women’s questionnaire?

### Socioeconomic status (SES)/ factors

Socioeconomic factors are non-medical factors that influence health outcomes. They are social and economic experiences and realities that help mould one’s personality, attitude, and lifestyle. These factors can also define regions and neighborhoods and have an important influence on health inequities – the unfair and avoidable differences in health status seen within and between countries. In countries at all levels of income, health and illness follow a social gradient: the lower the socioeconomic position, the worse the health.

Research has shown that social determinants can be more important than health care or lifestyle choices in influencing health [31,32,33].

Economic status is measured by income. Social status is measured by education, and work status is measured by occupation. Each status is considered an indicator, although they are related, they do not overlap [34].

### Household wealth index

The household wealth index is a measurement of the cumulative living standard of a household such as ownership of selected assets (e.g., televisions, radio, bicycles, etc.), sanitation facilities, types of water access, among others using the principal component analysis [31]. The household wealth index has been used in many demographics and health survey (DHS) reports to measure inequalities in household characteristics, access and use of health services, and health outcomes [32, 33]. The household wealth index is considered a more reliable measure than income and consumption because it represents a long-term standard of living of a household which allows for the identification of problems particular to the poor, such as unequal access and use of health services, etc. [34]. The household wealth index is calculated using a household’s ownership of selected items such as televisions and bicycles; materials used for housing construction; and types of water access and sanitation facilities [35]. A technique known as the principal component analysis was developed by Filmer and Pritchett to calculate the wealth index [36]. The household wealth index is usually divided into five wealth quintiles making the difference between the poor and rich very evident [31]. For this study, wealth was grouped into 5 quintiles – poorest (Q1), poorer (Q2), Middle (Q3), richer (Q4) and richest (Q5).

### Education

Education is a process (occurring at home, school, family, and community) and a product (attained through formal and experiential learning), which is considered as one of the most widely used indicators of socioeconomic status [37]. Education is an attribute of a person and an essential factor of a person’s health.

There has been growing global recognition of the interdependency between education and health. The World Health Organization (WHO) posits that a person is unhealthy if he/she is unable to conduct him/herself effectively and achieve some level of ‘social well-being’. The Incheon Declaration states that quality education develops skills, values, and attitudes that enable an individual to lead a healthy and fulfilled life and make informed decisions [38].

There is good evidence that education is strongly linked to health outcomes and determinants of health such as healthy lifestyles and behaviours, health service utilization, etc. Therefore, people with higher educational levels may have better economic conditions which help them afford better and quality healthcare services, develop better information processing and abilities required to make better-informed decisions about their health [39]. One major reason educational level is used as a measure of socioeconomic status for an adult is the reduction in the likelihood of reverse causation as education is complete before health status declines [16]. Inequality in education opportunities is found not only regarding individuals and social classes but also in terms of regions and territorial regions such as urban and rural areas [40].

According to the 2014 GDHS, education is self-reported which collects the highest level of education attained by both women and their husbands/partners [16]. For this study, education will be regrouped into three (3): 1) no education, 2) primary, and 3) secondary +.

### Occupation/employment

Occupation is used interchangeably with employment as a measure of socioeconomic status, embodies both income and education hence, its influence on health. Occupation/Employment reflects the educational attainment required to obtain the job and income levels that vary with different jobs and within ranks of occupations. This is used to measure the effect of socioeconomic status on health due to its role in positioning individuals within the social structure [41].

Employment has social, psychological, and financial benefits to improve one’s health. This implies that having a well-paying job provides an individual with the financial means to access nutritious foods, quality healthcare, safe housing, etc. all of which impacts health directly. The established correlation between employment and health is that having employment leads to income and eventually having the means to seek better healthcare thereby improving health status [42]. For example, with employment one can seek better health care services on time because they can afford it. In this study, occupation was measured using maternal employment status categorized into two groups: employed or unemployed.

### Other socioeconomic variables

Mother’s autonomy is an important predictor variable [43]. A mother’s autonomy was defined in the GDHS as a mother’s ability to decide on their health. This was derived from the question: a person who usually decides on a mother’s health care. The response options were: (a) mother alone, (b) mother and husband/partner, (c) husband/partner alone and (d) other (i.e., any other person besides the aforementioned). However, for this study, the responses were limited to three (3): (1) mother alone, (2) mother and husband/partner, (3) husband/partner alone.

Other predictor variables of interest for this study include mother’s age at birth, mother’s marital status, sex of household head, region of residence, area of residence, mother’s health insurance ownership, mother’s educational level, husband/partner’s educational level, mother’s autonomy, household wealth index, and mother’s employment status. The selection of predictor variables in this study was based on existing literature that reported a significant association with different maternal health care services.

### Statistical analysis

#### Data analysis

The study analysed the data using STATA 14 statistical software. The Adept software version 6 was used to calculate the concentration indices and curves of the socioeconomic inequalities in access and use of skilled birth attendants during childbirth, the concentration of the problem in the selected population and the contribution of the socioeconomic factors to the observed inequality among the population.

### Measuring inequalities

The study estimated and measured inequality in the health outcome (access and use of skilled birth attendants during childbirth) using the concentration indices (CI) and concentration curves (CC). Before inequity can be measured, the following are essential:

• An indicator of the health outcome of interest (dependent variable) i.e., delivery by a skilled health professional.

• a stratifying factor capturing the socioeconomic status against which the distribution is to be assessed (household wealth index), and

• a measure of socioeconomic inequality to quantify the degree of inequity in the indicator variable of interest (dependent variable).

The study used concentration curves and indices to measure socioeconomic status and inequalities that are essential in understanding the risk, burden, and impact of socioeconomic factors in accessing skilled birth attendants in Ghana. A concentration index (CI) is a relative measure (− 1 to + 1) of the extent to which a health outcome/variable is concentrated among the poor or the rich groups. The larger the absolute value of the concentration index (CI), the greater concentration of the inequality. A concentration curve (CC) plots the aggregate per centage share of health in a population against the aggregate per centage share of the population ranked according to their socioeconomic status (wealth) from the lowest to highest [44, 45]. As illustrated in Fig. 1, the concentration curves blue and red named C(p) and C(p*) respectively, are known as the line of inequality and may fall above or below the line of equality illustrated as green 45° line in the diagram. A concentration index is defined as twice the area between the line of equality (the green 45° line) and the concentration curve (either the blue C(p) or the red C(p*)). However, in a case where there is no income-related inequality, the concentration index is zero (0) meaning there will be no concentration curve above or below the line of equality as the concentration curve will be the same as the line of equality depicted in Fig. 1.

In this study, concentration indices (CI) were calculated to measure the magnitude of the inequality in the socioeconomic factors. The concentration index is defined as twice the area between the concentration curve and the line of equality (the 45-degrees line check the definition please). This is estimated as twice the covariance of the health care utilization and a person’s relative rank in terms of socioeconomic status, divided by the outcome mean [46]. This is presented in the formula below.

$$C=frac{2}{mu}mathit{operatorname{cov}}left({h}_i,{r}_iright)$$

(1)

Where C is the concentration index; hi is the health variable index; ri is the fractional rank of the individual i in the distribution of socioeconomic position; μ is the mean of the health variable and cov denotes the covariance.

The value of the CI measures the severity of socioeconomic inequality and vary between − 1 to + 1. A negative value implies that the health outcome is concentrated among those with lower socioeconomic status (i.e., the poor) showing a concentration curve above/to the left of the line of equality. A positive value indicates concentration among the higher socioeconomic status (i.e., the rich) showing a concentration curve below/to the right the line of equality. A CI value of zero implies no inequality. The larger the absolute value of CI, the greater the concentration of inequality [46, 47]. For example, if the health variable is ‘bad’ such as ill-health, a negative value of the concentration index means ill-health is higher among the poor, and vice versa.

### Decomposing the concentration index

Understanding and explaining the extent to which an underlying factor contributes to socioeconomic inequality has become of great interest to researchers and policymakers. The concentration index is commonly used to examine socioeconomic inequality in health [48, 49].

Decomposition estimations have mostly been used when the health outcome is a continuous variable (a numerical value that can be measured) using the Ordinary Least Square (OLS) regression model. However, given a situation where the dependent variable is binary like the use of skilled birth attendance during delivery or not as used in this study, the following need to be considered.

1. 1.

Regress the health outcome against its determinants using an appropriate model. This helps in finding the coefficients of the predictor variables (βk) as seen in eq. (2) below:

$$y=alpha+{textstylesum_k}beta_kx_k+varepsilon$$

(2)

Where y is the concentration index (C), α is the y-intercept, β and χ are the predictor variable of health care demand and ɛ is the error term. Since most health outcomes are binary, a few studies have used different methods – Probit analysis [45] and logit analysis [49, 50]. Given the dichotomy nature of the dependent variable, the normalization process ensures that the CI is quantified in the range of-1 to 1 for any given health outcome [46]. Calculate the concentration indices of the health utilization outcome variable and the determinants using the equation below:

$$C={textstylesum_k}left(beta_kappafrac{{overline{mathcal X}}_kappa}muright)C_k+frac{GC_varepsilon}mu$$

(3)

Where μ is the mean of the outcome variable y in eq. 2 (i.e., mean of the deliveries by SBA) ({overline{mathcal{X}}}_k) is the mean of ({mathcal{X}}_k), Cκ is the concentration index of determinant ({mathcal{X}}_k) xk (defined analogously to C) and GCε is the generalised concentration index for the error term of (ɛ). This equation shows that C is equal to the weighted sum of the concentration indices of the κ regressors, where the weight for ({mathcal{X}}_{kappa }) is the elasticity of y for ({mathcal{X}}_{kappa}left({eta}_{kappa }={beta}_{kappa}frac{x_k}{mu}right)). The residual component as captured by the last term reflects the income-related inequality in health that is not explained by systematic variation in the regressors, which should approach zero for a well-specified model.

### Ethical clearance

The GDHS 2014 sought ethical approval from the GHS Ethical Review Committee, Ghana and ICF Macro International Review Board, Maryland, USA. Further, written informed consent from each participant before enrolment was sought. For this study, ethical approval was received from the University of Cape Town Human Research Ethics Committee (HREC).