Pages within the "support" section:

19. Interpretation of results

It will be clear from the preceding sections that interpreting the results of a quantitative risk-benefit assessment requires great care, full consideration of the associated assumptions and uncertainties, and significant expertise in the various scientific disciplines involved (toxicology, epidemiology, nutrition, intake modelling, etc.).

Due to all the uncertainties involved, the results should not be presented as estimates of “real” impacts on real individuals. Instead, they should be communicated as providing an indication of the potential average annual health impact of the dietary change for the population as a whole. Furthermore, it should be emphasised that the estimate represents the steady-state outcome over long time periods. This is because the directly attributable health loss approach implicitly considers a steady-state scenario and does not consider transition effects.

Among other considerations, when using the directly attributable health loss approach, the assessor will need to evaluate how the overall health impact might be affected by the way the individual effects combine. The calculation simply sums the DALYs or QALY losses for all the effects that onset in a given year. This means that:

1.       The combination of two effects is estimated as the sum of their individual effects. In reality, the combined effect could be larger (i.e. having both diseases has a greater impact on quality of health than the sum of their individual impacts) or smaller (e.g. if both diseases affect the same bodily function, and the additional impact of having both diseases is marginal). If there are many concurrent effects, or if there are a small number of major impacts, simply taking the sum of the DALY weights or QALY losses might exceed the maximum value of 1 (= death).

2.       The DALYs or QALY loss caused by an effect will be overestimated to some degree because Equations (1)-(4) imply an assumption that individuals not affected by the dietary change will remain in full health until their normal life expectancy whereas, in reality, health generally declines in later years due to background diseases. However, this only affects the contribution from years lived with the disease, because the years of life lost to background diseases should already be accounted for in the normal life expectancy.  

The assessor will need to consider how complications (1) and (2) are affecting their assessment, and take this into account together with other uncertainties when drawing conclusions (general approaches for evaluating uncertainties are discussed in section 15.1). The combined effect of (1) and (2) across all the effects may often cause over-estimation of combined health impact, unless some of the effects show substantially more than additive impact. If effects are overestimated, this implies the change in health impact between the two scenarios is likely to be overestimated also (subject to the effect of other uncertainties). However, the overestimation of health impacts is likely to cancel out to some extent when calculating the difference between the two scenarios. The assessor should consider carefully these possibilities and take them into account as part of the qualitative evaluation of unquantified uncertainties (i.e. as part of Table 2, section 15.1).

If the uncertainties affecting the assessment results are too great for the assessor and risk manager to form a sufficiently firm conclusion about the impact of the dietary change, two options are open: seek more data to reduce one or more of the uncertainties and repeat the assessment, or attempt to improve the clarity of the outcome by quantifying more of the key uncertainties probabilistically.