A spatial analysis of social and economic determinants of tuberculosis in Brazil

• We studied standardized municipality tuberculosis case notification rates in Brazil.
• Between 2002 and 2009, rates fell due to decreases amongst older individuals.
• Tuberculosis rates were strongly positively spatially autocorrelated.
• Higher rates were associated with poorer social and economic conditions.
• The association between tuberculosis and poverty was confounded by space.

We investigated the spatial distribution, and social and economic correlates, of tuberculosis in Brazil between 2002 and 2009 using municipality-level age/sex-standardized tuberculosis notification data. Rates were very strongly spatially autocorrelated, being notably high in urban areas on the eastern seaboard and in the west of the country. Non-spatial ecological regression analyses found higher rates associated with urbanicity, population density, poor economic conditions, household crowding, non-white population and worse health and healthcare indicators. These associations remained in spatial conditional autoregressive models, although the effect of poverty appeared partially confounded by urbanicity, race and spatial autocorrelation, and partially mediated by household crowding. Our analysis highlights both the multiple relationships between socioeconomic factors and tuberculosis in Brazil, and the importance of accounting for spatial factors in analysing socioeconomic determinants of tuberculosis.