LAUSR

Health disparities research in low- and middle-income countries (LMICs) is hampered by the difficulty of measuring economic status in low-resource settings. We previously developed the EconomicClusters k-medoids clustering-based algorithm for defining population-specific economic models based on few Demographic and Health Surveys (DHS) assets. The algorithm previously defined a twenty-group economic model for Cameroon. The aims of this study are to optimize the functionality of our EconomicClusters algorithm and app based on collaborator feedback from early use of this twenty-group economic model, to test the validity of the model as a metric of economic status, and to assess the utility of the model in another LMIC context. We condense the twenty Cameroonian economic groups into fewer, ordinally-ranked, groups using agglomerative hierarchical clustering based on mean cluster child height-for-age Z-score (HAZ), women’s literacy score, and proportion of children who are deceased. We develop an EconomicClusters model for Ghana consisting of five economic groups and rank these groups based on the same three variables. The proportion of variance in women’s literacy score accounted for by the EconomicClusters model was 5–12% less than the proportion of variance accounted for by the DHS Wealth Index model. The proportion of the variance in child HAZ and proportion of children who are deceased accounted for by the EconomicClusters model was similar to (0.4–2.5% less than) the proportion of variance accounted for by the DHS Wealth Index model. The EconomicClusters model requires asking only five questions, as opposed to greater than twenty Wealth Index questions. The EconomicClusters algorithm and app could facilitate health disparities research in any country with DHS data by generating ordinally-ranked, population-specific economic models that perform nearly as well as the Wealth Index in evaluating variability in health and social outcomes based on wealth status but that are more feasible to assess in time-constrained settings.