Measuring wealth for health research in rural and urban settings in three African countries: Findings and recommendations from a multi-country study

Wealth indices enable socioeconomic status measurement: however, their reliability, validity and data collection and computation efficiency can be improved. This analysis aimed to generate four wealth indices in five geographical settings (rural and urban Gambia, rural and urban South Africa and urban Zimbabwe), to compare and validate these indices and formulate recommendations for wealth measurement in Africa. Population-based cross-sectional data from 5,296 adults aged ≥40 years were analysed. Principal component analyses generated indices based on 18 variables quantifying house ownership, wall and roofing material and asset ownership. Recommendations were based on reflective workshops (totalling 6 hours) held with experienced in-country fieldwork coordinators. In all sites, the 'multi-country' wealth index was associated with greater asset ownership in 44.4% (8/18) of the variables intended to measure wealth i.e. refrigerator, television, working car/truck, tap in the house, flush toilet, decoder/satellite dish, computer and tiled floors, albeit with rural-urban and country differences. Reflective workshops established that the accuracy of wealth measurement for multi-country and multi-site studies, using asset-based indices, needs an unambiguous, differential and clearly defined asset list. Furthermore, consulting local teams to select these assets can reduce data collection burden while increasing index validity, including 'contemporary' asset assessment, e.g. Wi-Fi internet connection.