Menu
Menu
Menu
Menu
Introduction In the pursuit of achieving the Sustainable Development Goal targets of universal health coverage and reducing maternal mortality, many countries in sub-Saharan Africa have implemented health insurance policies over the last two decades. Given that there is a paucity of empirical literature at the sub-regional level, we examined the prevalence and factors associated with health insurance coverage among women in in sub-Saharan Africa. Materials and methods We analysed cross-sectional data of 307,611 reproductive-aged women from the most recent demographic and health surveys of 24 sub-Saharan African countries. Bivariable and multivariable analyses were performed using chi-square test of independence and multilevel logistic regression respectively. Results are presented as adjusted Odds Ratios (aOR) for the multilevel logistic regression analysis. Statistical significance was set at p<0.05. Results The overall coverage of health insurance was 8.5%, with cross-country variations. The lowest coverage was recorded in Chad (0.9%) and the highest in Ghana (62.4%). Individuallevel factors significantly associated with health insurance coverage included age, place of residence, level of formal education, frequency of reading newspaper/magazine and watching television. Wealth status and place of residence were the contextual factors significantly associated with health insurance coverage. Women with no formal education were 78% less likely to be covered by health insurance (aOR = 0.22, 95% CI = 0.21-0.24), comparedwith those who had higher education. Urban women, however, had higher odds of being covered by health insurance, compared with those in the rural areas [aOR = 1.20, 95%CI = 1.15-1.25]. Conclusion We found an overall relatively low prevalence of health insurance coverage among women of reproductive age in sub-Saharan Africa. As sub-Saharan African countries work toward achieving the Sustainable Development Goal targets of universal health coverage and lowering maternal mortality to less than 70 deaths per 100,000 live births, it is important that countries with low coverage of health insurance among women of reproductive age integrate measures such as free maternal healthcare into their respective development plans. Interventions aimed at expanding health insurance coverage should be directed at younger women of reproductive age, rural women, and women who do not read newspapers/magazines or watch television.
Data for this study were obtained from Demographic and Health Surveys (DHS) conducted between January 1, 2010, and December 31, 2019, in 24 sub-Saharan African countries (see Table 1). Specifically the womens’ files of the DHS were used. The choice of the 24 countries was influenced by the availability of the variables of interest in their datasets. DHS is a nationwide survey undertaken across LMICs every five years [25]. The survey is representative of each of these countries and targets core maternal and child health indicators such as health insurance coverage. In selecting the sample for each survey in the various countries, a multi-stage sampling approach was employed. The first step of this sampling approach involved the selection of clusters (i.e., enumeration areas [EAs]), followed by systematic household sampling within the selected EAs. In this study, the sample size consisted of women aged 15–49 who had complete cases on all the variables of interest (N = 307,611). We followed the ‘Strengthening the Reporting of Observational Studies in Epidemiology’ (STROBE) statement in conducting this study and writing the manuscript (see S1) [26]. The data underlying the results presented in the study are available from https://dhsprogram.com/data/available-datasets.cfm. *DR = Democratic Republic The outcome variable in this study was health insurance coverage. This was derived from the question ‘Are you covered by any health insurance?’ It was coded 1 = “Yes” and 0 = “No” [19, 27]. Both individual and contextual (household and community level factors) level factors were considered as explanatory variables in this study. The individual-level factors were age, marital status, educational level, employment, parity, and exposure to the mass media (radio, newspaper and television). The contextual variables were sex of household head, household wealth quintile, place of residence and sub-region (see Table 2). In the DHS, wealth was computed using data on household ownership of selected assets such as bicycle, materials used for house construction, television, type of water access and sanitation facilities were used. Wealth quintile was then created from these assets through Principal Component Analysis (PCA) by placing households on a continuous measure of relative wealth after which households were grouped into five wealth quintiles namely poorest, poorer, middle, richer and richest [25]. The sub-region variable was derived by aggregating countries based on their geographical location on the African continent (thus Western, Southern, Eastern, and Central). Our choice of the explanatory variables were influenced by their inclusion in the DHS datasets and previous literature which found these variables to be associated with health insurance coverage [5, 19–24, 27]. The data were analysed with Stata version 14.2 for macOS (Stata Corporation, College Station, TX, USA). Three steps were followed to analyse the data. The first step was the use of descriptive statistics to describe the sample and cross-tabulation of all the explanatory variables against health insurance coverage. The second step was a bivariable analysis to select potential variables for the regression analysis. Variables that were statistically significant at the bivariable analysis stage at p<0.05 were moved to the final step, where two levels of multilevel logistic regression models were built to assess the individual and contextual factors associated with health insurance coverage. Clusters were considered as random effects to account for the unexplained variability at the contextual level [28]. We fitted four models (see Table 3). The first model was the empty model (Model I), which showed the variation in health insurance coverage attributed to the distribution of the primary sampling units (PSUs) in the absence of the explanatory variables. Model II had only the individual level factors and health insurance coverage. The purpose of Model II was to look at how the individual level factors are associated with health insurance coverage in the absence of the contextual factors. Model III was developed to look at the association between the contextual level factors and health insurance coverage, in the absence of the individual level factors. The final model (Model IV) was the complete model that had the individual and contextual level factors and health insurance. The purpose was to look at the association between both the individual and contextual level factors and health insurance coverage. For all the models, adjusted Odds Ratios (aOR) and their associated 95% confidence intervals were presented. These models were fitted by the Stata MLwinN software version 3.05 [29]. Exponentiated coefficients; 95% confidence intervals in brackets * p < 0.05 ** p < 0.01 *** p < 0.001 SE = Standard Error; ICC = Intra-Class Correlation; LR Test = Likelihood ratio Test; MOR: Median Odds Ratio Model I is the null model, a baseline model without any determinant variable Model II = individual-level variables Model III = contextual variables Model IV is the final model adjusted for individual and contextual variables Using the variance inflation factor (VIF), the multicollinearity test showed that there was no evidence of collinearity among the explanatory variables (Mean VIF = 1.54, Maximum VIF = 2.09 and Minimum VIF = 1.09). The choice of reference categories were informed by the categories with lower likelihood of using NHIS. For example, in the case of age, those aged 15–19 were chosen as reference category. Where this could not be determined per literature, the category with the greatest number of observations was taken as a reference category. According to Hatt and Waters [30], pooling data can reveal broader results that are ‘‘often obscured by the noise of individual data sets.” To calculate the pooled values an additional adjustment is needed to account for the variability in the number of individuals sampled in each country. This is accomplished using the weighting factor 1/(A*nc/nt), where A is the number of countries asked a particular question, nc is the number of respondents for the country c, and nt is the total number of respondents over all countries asked the question [31]. The DHS receive ethical clearance from the Ethics Review Committee of ORC Macro Inc. and the Ethics Review Committees of partner organizations in the various countries where the surveys are conducted.
N/A
