Descriptive statistics on the socio-economic characteristics of the two samples and primary results of the two-steps cluster analyses can be found in the appendices (supplementary files). Below, we present the key findings of the cluster analyses as well as the results of the regression analyses.
The two-step cluster analysis of family planning service use in Ghana automatically produced 3 distinct clusters. In the two-step cluster analysis of family planning service use in Nigeria, the number of clusters (3 clusters) was fixed in advance to be able to produce meaningful clusters. The clusters are presented in Table 1. The clusters were inspected and labeled as high, medium, and poor access to family planning services based on the services used by women in each cluster (see Methods section). The cluster with high access to family planning services captures 19.1% and 21.4% of women in Nigeria’s and Ghana’s sample respectively. The other extreme is the third cluster that consists of women whose access can be described as poor; 71.5% of women in Nigeria’s sample belong to this cluster and 64.2% of women in Ghana’s sample.
We did not predefine the number of clusters for maternal health services. For both countries, the two-step cluster analyses of maternal health services use resulted in five clusters, which we inspected and labeled as higher, high, medium, low and poor access to maternal health services (see Table 1). The higher-access cluster captures 29.6% of women in Nigeria’s sample and 26.3% of the women in Ghana’s sample. Relative to the other four clusters, a larger proportion of members of this cluster report that they accessed government hospitals for antenatal care and used institutional maternal care more. The high-access cluster consists of 21.6% of women in Nigeria’s sample and 23.0 % of women in Ghana’s sample. For both countries, this cluster has a lower proportion of women who accessed government health centers for antenatal care or got assistance from physicians during childbirth. Members of the medium-access cluster in both countries used private facilities for antenatal care as well as for childbirth. This cluster of women makes up 25.9% of Nigeria’s sample and 18.2% of the Ghana sample. Members of the low-access cluster in both countries mostly report that they accessed government health posts/dispensaries for antenatal care but did not have skilled assistance during childbirth. In Nigeria’s sample, 4.4% of women fall into this cluster and in the Ghana’s sample, this share is 7.1%. Lastly, 18.5% and 25.4% of women from the Nigeria’s and Ghana’s sample respectively are members of the poor-access cluster. Members of this cluster mostly did not receive institutionalized maternal care. For both countries, the poor-access cluster has a high proportion of members who had home childbirth and used traditional birth attendants during childbirth.
The dependent variables in the four multinomial logistic regressions were the four cluster membership variables generated in the cluster analyses. Tables 2 and 3 present the odds ratios for the four regressions. Information about the independent variables used and the full results of the regression analyses can be found in Appendix B of the supplementary file.
For family planning services, the results in Table 2 show that in both countries, women with no education, compared to women with secondary or higher education, have higher odds to belong to the poor-access family planning cluster (in Nigeria OR=2.544, 95% CI:1.907- 3.395, p≤ 0.01 and in Ghana OR=1.527, 95% CI: 1.173- 1.988, p≤ 0.01). Increased odds of having poor-access to family planning services are found for women in Ghana who do not belong to white-collar workers but not among women who live in rural areas, and also not among women in any of the wealth quintiles. Higher odds of poor-access to family planning services are also found for women in Nigeria who belong to the service-occupational category (OR=1.283, 95% CI: 1.002- 1.642, p≤0.05), compared with white-collar workers. The odds of poor-access are as much as three times higher among the poorest quintile (95% CI: 1.825- 6.396, p≤ 0.01) than the richest quintile; and among those who have no insurance (OR=1.374, 95% CI: 1.011- 1.867, p≤0.05) compared to those with insurance.
Table 3 shows the regression results on access to maternal health services in Nigeria and Ghana. In Nigeria’s sample, women with primary or no education have higher odds to have poor-access (OR= 1.387, 95% CI: 1.140- 1.687, p≤ 0.01) or low-access (OR= 1.786, 95% CI: 1.247- 2.557, p≤ 0.01) to maternal health services. In Nigeria’s sample, women who are not working have higher odds to belong to the cluster of poor-access maternal health services only (OR= 1.579, 95% CI 1.081- 2.307, p≤ 0.01). Compared to women in the white-collar occupational group, women in other occupational categories in Nigeria also have higher odds to belong to the poor-access cluster. Women in other occupational categories in Nigeria also have higher odds to belong to the poor-access cluster. Women in all household wealth quintiles have higher odds to have high- or poor-access to maternal health services; women without insurance have higher odds to have high or poor-access to maternal health services. Results for Ghana show that women with primary (OR= 1.38, 95% CI: 1.036- 1.838, p≤0.05) or no education (OR= 1.542, 95% CI: 1.115- 2.132, p≤0.01) have higher odds of poor-access to maternal health services. Only women in the agriculture occupational group have higher odds of high-access to maternal health services compared to women in the white-collar sector. Women without health insurance have higher odds of access to maternal health care services.