Identifying vulnerable urban neighbourhoods and their environmental, density, and housing characteristics in Accra, Ghana using census and remote sensing data

Abstract

BACKGROUND AND AIM: Identifying vulnerable urban communities, commonly known as slums, can facilitate targeted policies to reduce urban economic and social inequities in cities, but these data are rarely available. We aimed to identify vulnerable urban neighbourhoods and their environmental and housing characteristics in Accra, Ghana using available training data on the city center (Accra Metropolitan Area - AMA) applied to the Greater Accra Metropolitan Area (GAMA). METHODS: We accessed the following enumeration area (EA)-level data for Greater Accra: slum classification available for a subset of 2,418 EAs in the AMA from the Accra Metropolitan Assembly and UN-Habitat 2011 report; housing conditions from the most recent Ghana Census (2010); and environmental quality attributes from remote sensing data provided by the United States Geological Survey and National Aeronautics and Space Administration. We fitted a Bayesian logistic regression model to evaluate associations between housing, density, and environmental attributes with vulnerable area classification of EAs in the AMA. We then applied the model to predict the probability of each urban EA in GAMA as being vulnerable. RESULTS:We estimated that approximately one-fifth of EAs in the GAMA had a vulnerable area probability greater than 80%, corresponding to a population of 752,367 likely living in suboptimal conditions. The variables associated with a higher probability of an EA being vulnerable included greater use of public toilet facilities [OR: 3.51 (95% credible interval (CI): 1.55,7.53)], higher population density [OR: 5.72 (95% CI: 3.85,8.65)], lower use of improved wall materials [OR: 0.11 (95% CI: 0.03,0.43)], lower elevation [OR: 0.45 (95% CI: 0.35, 0.58)], lower use of indoor piping as a drinking water source [OR: 0.50 (95% CI: 0.25,0.99)], and lower vegetation abundance [OR: 0.25 (95% CI: 0.16,0.39)]. CONCLUSIONS:Our approach can be used in future studies to identify geographic clusters of vulnerable areas where interventions are warranted to improve housing and environmental conditions. KEYWORDS: Built environment, Socio-economic factors, Epidemiology