Improving health information systems in Guatemala using weighted correlation network analysis

Guatemala has the fifth worst child stunting prevalence - low-height-for-age - in the world, at 49%. Child stunting is associated with negative short and long term health effects and the contributing factors are complex, interrelated, and potentially non-linear. Current health information systems (HIS) in Guatemala are disaggregated, overly complex, and have limited scalability. This paper demonstrates the use of weighted correlation network analysis to visualize and explore data in a way that provides useful information for future HIS. The methods generate a holistic causal factor model for stunting that explores how cofactors relate to stunting and each other. The demonstration here is based on a Guatemala regional data set obtained from the USAID Open Data Website. First, the overall correlation network structure is observed and compared to generalized structural models proposed by the WHO and USAID. Next, quantile comparisons are performed using the outcome variable z-score height-for-age, and distinct child age groups. The comparisons demonstrate how these networks can be used as an extension of widely used methods while also providing advantages that are important for exploratory analysis. This work is an important first step in evaluation of a novel analysis method for health information systems currently being developed in Guatemala.