DisMod-MR also made use of the data available to estimate prevalence for countries and regions for which no raw data were available. Given that the aim of GBD 2010 was to compare burden caused by diseases and injuries between countries, this approach was considered preferable to the alternative, which was to entirely exclude parts of the world where no local data was available from GBD estimations, thereby assuming the prevalence of mental, neurological, and substance use disorders in those countries was zero. The final results provided prevalence estimates for 187 countries, 21 regions, 7 super regions, 20 age groups, men and women, for 1990, 2005, and 2010. Uncertainty around the raw epidemiological data was propagated to the final DisMod-MR model to provide 95% uncertainty intervals around all prevalence estimates [11].
