The Millennium Development Goals (MDGs) have put maternal health in the mainstream, but there is a need to go beyond the MDGs to address equity within countries. We argue that MDG focus on maternal health is necessary but not sufficient. This paper uses Demographic and Health Survey (DHS) data from Kenya, Ethiopia and Ghana to examine a set of maternal health indicators stratified along five different dimensions. The study highlights the interactive and multiple forms of disadvantage and demonstrates that equity monitoring for the MDGs is possible, even given current data limitations.

We analyse DHS data from Ghana, Kenya and Ethiopia on four indicators: skilled birth attendant, contraceptive prevalence rate, AIDS knowledge and access to a health facility. We define six social strata along five different dimensions: poverty status, education, region, ethnicity and the more traditional wealth quintile. Data are stratified singly (e.g. by region) and then stratified simultaneously (e.g. by region and by education) in order to examine the compounded effect of dual forms of vulnerability.

Almost all disparities were found to be significant, although the stratifier with the strongest effect on health outcomes varied by indicator and by country. In some cases, urban-dwelling is a more significant advantage than wealth and in others, educational status trumps poverty status. The nuances of this analysis are important for policymaking processes aimed at reaching the MDGs and incorporating maternal health in national development plans.

The article highlights the following key points about inequities and maternal health: 1) measuring and monitoring inequity in access to maternal health is possible even in low resource settings—using current data 2) statistically significant health gaps exist not just between rich and poor, but across other population groups as well, and multiple forms of disadvantage confer greater risk and 3) policies must be aligned with reducing health gaps in access to key maternal health services.

The authors declare that there are no conflicts of interest. The lead author had full access to the data used in this paper and the final responsibility for the decision to submit for publication.