Targeting Health Inequalities


Health inequalities are differences in health status that are experienced by certain population groups. Most health inequalities are due to poor socioeconomic conditions. Many programmes try to alleviate heath inequalities by targeting services to the most deprived households. Poor households can be categorized according to cut off points of the household expenditures distribution, but collecting expenditure data is too complex for routing targeting of households. For this reason, proxy means tests are used to identify poor households according to presence of households' characteristics that are a proxy of household's wealth. Several statistical methods can be used to assign weights to household variables, the sum of which produces a household score. Households having a household score under a certain cut off point can be considered at risk of deprivation for purpose of targeting with health programmes.

Another potential cause for health inequalities is the variation in Social Capital. Individuals gain resources in terms of information and support they get through social interactions, and by participation into formal and information associations. These resources are categorized under SC to differentiate them from the physical capital (i.e. tools) and the human capital (i.e. education). However, it is difficult to measure SC through a metric score, while multitudes of categorical variables, such as political activism and membership to social networks are used instead. As with proxy means testing, several analytical methods can help to measure SC to identify its role in causing health inequalities.

I can help to analyse data from household surveys to assigned deprivation weights to critical variables, the sum of which will amount to a household scores. These variables can include the educational level and the type of employment of the head of the household, house's building materials, possession of durable goods, access to services and many other variables. I used several analytical techniques to reduce tens of categorical variables into a 3 or 4 metric dimensions that summarize most of the total variation. This allows to eliminate redundant information and transform the most important variables into provide weighted scores. This analysis can be done on the data sets of past surveys and the goodness of the household score in correctly identifying deprived households can be validated against household expenditures cut off points, and against morbidity and malnutrition among households' members.

Examples include:

I analysed national household surveys carried out by the World Bank in Ethiopia and South Africa to assign weighted scores to individual households' variables to identify deprived households. The predictive values of several cut off points of the proxy means test score were tested against households cut off points for expenditures and presence of malnutrition among households' members.

I applied Non Linear Principal Component and Cluster Analysis to the data from a household survey carried out in Kampala, Uganda, to transform household variables into dimension of Social Capital. The first four dimensions explained more than 70% of the variance and were related to social networking, trust in neighbours and institutions, and social norms. Cluster analysis was applied on these dimensions to assign households into different groups characterized by different scores of SC.