The case against annual profiles for the valuation of disability weights
The quantification of disability in burden of disease studies relies on cou- pling severity weights to time spent in less than full health. In the Austra- lian Burden of Disease studies (Mathers et al. 1999; Vos and Begg 2000) a conscious choice was made to spend the available research time in col- lecting plausible and internally consistent estimates of incidence, preva- lence and duration. For the severity weights we relied mostly on the weights generated by the Dutch Disability Weight Study (Stouthard et al. 1997) and supplemented these with weights from the Global Burden of Disease (GBD) Study (Murray and Lopez 1996) or generated new, inter- polated weights for conditions not covered by the Dutch study. The Dutch weights were attractive to use as they: i) covered the range of conditions contributing most to the burden of disease in Australia; ii) allowed more detailed disease modelling because weights for the most common disabling conditions were available by several levels of severity; iii) had described each of the disease states valued with a EuroQol—with a sixth dimension for cognitive functioning (EQ5D +)—notation which facilitated matching disability weights to epidemiological data for comparable health states.
A small set of weights from the Dutch study appeared to be outliers. These are the conditions of short duration for which annual profiles were presented to the valuation panels. For example, symptomatic acute gonnorrhoea was described as “one year of which during one week symp- tomatic acute gonorrhoea is present” with an EQ5D + description of 111211 (indicating moderate pain/discomfort and no problems in the other five dimensions of mobility, self-care, usual activities, anxiety/depres- sion and cognitive functioning). The Dutch weights were subsequently applied in a national burden of disease calculation for the Netherlands, which estimated the burden from non-fatal conditions by multiplying prevalence with a one-year duration and the corresponding disability weights (Melse and Kramers 1998). This is a different approach from that used in most other burden of disease studies, including the GBD, which calculate the disability burden from the multiplication of incidence, du-