Objectives To consider the data shortcomings and methodological decisions involved in current burden of disease studies and the potential for these to be overcome and/or standardised.

Method Most burden of disease estimates require considerable assumptions or methodological decisions about factors concerning exposure, the appropriate relative risk to match with the exposure, and/or the size of the exposed population. These assumptions usually arise from a lack of data and could be largely overcome by the provision of better data. It is reasonable to expect that for some areas these data will improve with time, but for other areas the required data will probably never be available.

Other assumptions or methodological approaches vary depending primarily on theoretical considerations that are arguable and unlikely to ever be definitively solved by better data availability. Modelling may sometimes be of use but may not always be appropriate or practical and is still likely to involve some assumptions.

Results For example, some countries have reasonable estimates of asbestos exposure and some have good data on at least one asbestos-related outcome (mesothelioma incidence/mortality). How can this information be validly used for burden estimates where such data are poor?

Conclusions It is helpful to consider the extent to which burden estimates vary depending on the assumptions and methodologies involved when assessing the validity of estimates and their usefulness. Consideration of the potential for future improvements in data and better understanding of theoretical aspects should be an important input into the planning of future burden of disease work.