Menu
Menu
Menu
Menu
N/A
Although many countries are making progress towards achieving the global sustainable development goals, sub-Saharan Africa (SSA) lags behind. SSA bears a relatively higher burden of maternal morbidity and mortality than other regions despite existing cost-effective interventions. This paper assesses antenatal care (ANC) service utilisation among women in the Southern African Development Community (SADC) countries, one of the four SSA regions. Specifically, it assesses socioeconomic inequality in the number of ANC visits, use of no ANC service, between one and three ANC visits and at least four ANC visits, previously recommended by the World Health Organization (WHO). Data come from the most recent Demographic and Health Surveys in twelve SADC countries. Wagstaff’s normalised concentration index (CI) was used to assess socioeconomic inequalities. Factors explaining these inequalities were assessed using a standard method and similar variables contained in the DHS data. A positive CI means that the variable of interest is concentrated among wealthier women, while a negative CI signified the opposite. The paper found that wealthier women in the SADC countries are generally more likely to have more ANC visits than their poorer counterparts. Apart from Zambia, the CIs were positive for inequalities in at least 4 ANC visits and negative for between 1 and 3 ANC visits. Women from poorer backgrounds significantly report no ANC visits than wealthier women. Apart from the portion that was not explainable due to limitations in the variables included in the model, critical social determinants of health, including wealth, education and the number of children, explain socioeconomic inequalities in ANC coverage in SADC. A vital policy consideration is not to leave any woman behind. Therefore, addressing access barriers and critical social determinants of ANC inequalities, such as women’s education and economic well-being, can potentially redress inequalities in ANC coverage in the SADC region.
Data come from the latest Demographic and Health Surveys (DHS) for SADC countries with available data (twelve of the sixteen SADC countries) as of October 2021. The Union of Comoros was not included in the analysis because the latest data are for 2012, and it only became a full member of the SADC countries in August 2018 (SADC, 2021). The DHS use standardised questions to collect information mainly from women of reproductive age (i.e. aged between 15 and 49 years) (Rutstein & Rojas, 2006). The DHS datasets are cross-sectional and nationally representative, with information on women’s sociodemographic and socioeconomic characteristics and maternal health service utilisation (DHS Program, 2021). Table 2 contains a summary of the DHS datasets for available countries. Sample size per SADC country. Notes: * Sample size = number of women aged 15–49 years. Three mutually exclusive variables were created to assess socioeconomic inequality in each of the variables critically: 1) No ANC visits (i.e. when a woman with a live birth in the specified period did not have any ANC visit; 0 ANC) 2) At least one but less than four ANC visits (i.e. having between one and three visits; 1–3 ANC), and 3) At least four ANC visits (i.e. a woman with at least four ANC visits; ≥ 4 ANC or 4+ ANC). A fourth encompassing category (ANC intensity) uses the total number of ANC visits that a pregnant woman had received. The DHS does not directly report a household’s expenditure or income but contains information on household assets or a wealth index developed based on a method by Rutstein and Johnson (2004). This paper uses the wealth index as a proxy for socioeconomic status (SES). This index was constructed from household asset data, including access to sanitation facilities, type of flooring material and source of drinking water. A comparative analysis of ANC utilisation in the twelve SADC countries was done to give a descriptive assessment of inequalities in the use of antenatal care. This analysis uses equity stratifiers such as type of residence, highest education level, respondents’ occupation and wealth quintiles. Socioeconomic inequality in the distribution of ANC utilisation was assessed using concentration indices (Wagstaff et al., 1991). Two key variables used to estimate the concentration index are ANC utilisation as a health variable of interest (i.e. 0 ANC, 1–3 ANC, 4+ ANC or ANC intensity) and SES using the wealth index. The standard concentration index is estimated as twice the covariance between the ANC utilisation variable (Hi) and the relative rank of women using the SES measure (Ri), divided by the mean of the ANC utilisation variable (μH) (Wagstaff et al., 1991). This standard concentration index was used to assess socioeconomic inequalities in the number of ANC visits (i.e. ANC intensity). However, because the other key mutually exclusive variables are dichotomous (i.e. 0 ANC, 1–3 ANC, 4+ ANC), the standard concentration index will not range from −1 to +1 (Wagstaff, 2005). The standard concentration index in Equation (1) was normalised using the approach proposed by Wagstaff (2005). Generally, a negative valued concentration index (including the normalised index) corresponds to a higher distribution of ANC service utilisation among women from poorer socioeconomic backgrounds. A positive-valued index signifies a higher utilisation distribution among wealthier women (Kakwani et al., 1997). Also, for interpretation, a positive-valued concentration index can be interpreted as “pro-rich” while a negative index value as “pro-poor.” The concentration index for ANC intensity was decomposed to identify factors that explain observed socioeconomic inequalities in ANC coverage in SADC countries (Wagstaff et al., 2003). Let us define the relationship between ANC intensity (Hi) and a set of explanatory variables or factors (zji) as: where α and β are ordinary least squares parameter estimates and ε is the error term. Wagstaff et al. (2003) use the relationship in Equation (2) to decompose the concentration index in Equation (1) (CH) into two major components: where Cj is the j-th contributing factor’s concentration index, and βjz‾jμH is the elasticity of ANC intensity to marginal changes in the j-th explanatory variable or factor. The generalised concentration index of the error term is denoted by GCε. The explained component (i.e. (βjz‾jμH)Cj) is factor j’s contribution to socioeconomic inequality in ANC intensity. Explanatory variables or factors used in this paper include the woman’s age, education, employment, urban or rural location, region of residence, socioeconomic quintiles, and the total number of children for each woman. These variables featured prominently in previous studies (Obse & Ataguba, 2021; Rosário et al., 2019; Nagdeva, 2009; Shibre et al., 2020; Yaya et al., 2016; McTavish et al., 2010). A woman’s total number of children was included in the model to capture multigravida and a woman’s previous ANC utilisation experiences that may affect current service utilisation. Interpreting the contributions for each factor ((βjz‾jμH)Cj)) is straightforward. With a positive concentration index, for example, a positive contribution of a factor means that the factor contributes to the concentration of inequalities in ANC utilisation among wealthier women. The unexplained component, (GCεμH), is also called the residual and accounts, among other things, for unexplained factors. The value of the unexplained component should be close to zero for a well-specified model that includes all relevant variables. The values of each component, including their associated standard errors, were computed in Stata using a user-developed computer routine (Bilger et al., 2017). Specifically, bootstrap methods are used to obtain standard errors in Equation (3) with 500 replications (Efron, 1987; Efron & Tibshirani, 1986), accounting for the sampling structure of each DHS. Stata 15 was used to perform all analyses in the paper (StataCorp, 2017).
