Schematic of methods used to estimate populations at risk of morbidity. Age-specific prevalence estimates were generated using geostatistical modelling (SSA) or on an empirical basis (all other regions). Geographical variation was approximated using modelled logit-normal distributions, and numbers exceeding burden thresholds estimated using negative binomial distributions. These results were fed into the 2010 GBD framework to estimate years lived with disability (YLD) and Disease Adjusted Life Years (DALYs). Steps contained within shaded grey areas are completed within a Bayesian framework.