Pages within the "support" section:
3.2 Alternative approaches for calculating DALY/QALYs
The most accurate way to calculate QALYs would be to obtain a record of the actual history of health changes for each individual in a population, assign the relevant QALY weight for their health state in each year of their life, and calculate the sum of years weighted by their associated QALY weights. However, risk-benefit assessment requires the estimation of QALYs for future health states for different dietary scenarios (reference and alternative).
The most obvious approach for this is to simulate future diet and the resulting health states for each individual under the two scenarios. This has the very important advantage of allowing the assessor to take account of how different diseases or health effects combine, including how effects associated with the dietary change under assessment combine with “background” diseases. How different health effects combine has an important impact on the calculation: e.g. if a person dies from one disease at 50 then another disease that would commence at 60 will have no effect.
This simulation approach has been used in DALY calculations, e.g. in the RIVM Chronic Disease Model for the Netherlands (van Kreijl et al., 2006; Hoogenveen et al. 2009). However, this type of modelling is extremely complex and requires data on many parameters including the normal incidences, durations and severities of major background diseases as well as the health effects of the dietary change under consideration, and the interactions between them. It also requires modelling the demographic development of the population over the period of dietary change (births, deaths, immigration, emigration).
van Kreijl et al. (2006) also present a much simpler, but less realistic, approach which estimates the “annual directly attributable health loss”. This was used in a risk-benefit assessment for folate by Hoekstra et al. (2008). This approach considers only a single year, and only health effects that have their onset during that year. The potential impact of each health effect starting during the year is considered in isolation, ignoring interactions with other effects starting in the same or subsequent years, and ignoring interactions between these effects and background diseases. Because of this, the output is interpreted as measure of potential impact per year , rather than as an estimate of actual health outcome. Furthermore, the potential impacts of alternative disease outcomes (recover, die early as a result of the disease, or survive with the disease until the normal life expectancy) are combined as a weighted average, weighted by their respective probabilities. The result should therefore be interpreted as estimating the average of the potential impacts for a (large) number of similar individuals , rather than a specific outcome for a single individual.
A comparison of these two approaches showed that health impacts estimated by the Chronic Disease Model were somewhat lower than those estimated by the directly attributable health impacts approach (van Kreijl et al., 2006). This occurs because the Chronic Disease Model takes account of combination effects (comorbidity, substitution and delayed effects) that prevent the full potential impacts of effects being realised.
Because of the complexity and resource and data requirements of simulating health effects over whole lifetimes, the Qalibra framework and software is based instead on the simpler approach of calculating annual directly attributable health loss. A practical view is to consider this as a basic option for Tiers 3 and 4 of the Brafo tiered approach, with simulation of health over whole lifetimes being reserved as a higher tier option for use only with risk-benefit questions that cannot be satisfactorily resolved by the simpler, directly attributable health impacts method. The assessor would then start with the directly attributable health impacts approach, interpreting the results very carefully to take account of its limitations. In particular, the assessor will need to consider how the overall health impact might be affected by the way the individual effects combine (see interpretation of results, below).