Differential Discounting

The practice of applying different discount rates to costs and health outcomes within an economic evaluation, rather than discounting both at the same rate.

Source: Parsonage & Neuburger 1992

The argument advanced for it is that money and health are not the same kind of thing, so there is no reason the rate appropriate for one should apply to the other. Discounting costs reflects the return available from investing resources elsewhere, which is an observable market quantity. Discounting health reflects how society weighs health accruing to people in the future against health now, which is a social value judgement with no market counterpart. On that reasoning, applying a single rate to both assumes an equivalence that has not been established.

Source: Parsonage & Neuburger 1992

Applying a lower rate to health than to costs raises the present value of future health gains relative to present costs, which systematically favours interventions whose benefits arrive late. Prevention, vaccination and public health programmes gain most, since their costs fall now and their effects accrue over decades. The size of the shift grows with the horizon, so for lifetime analyses the choice between equal and differential rates can change a conclusion by more than any clinical parameter.

Source: Drummond et al. 2015

That it produces internally inconsistent results. If health is discounted at a lower rate than money, then postponing a health-producing programme indefinitely appears increasingly attractive, because its benefits shrink more slowly than its costs. This paradox arises whenever the two rates differ and has no satisfactory resolution within the framework. A further objection is that if health can be converted to money at a threshold, discounting the two differently implies the threshold itself changes over time, which most frameworks do not intend.

Source: Drummond et al. 2015

Practice varies and has changed over time. Several frameworks that once specified a lower rate for health moved to equal rates for both, on consistency grounds. Others retain differential rates or permit them in sensitivity analysis. Where a framework specifies equal rates as the reference case, analyses examining long-horizon prevention frequently report differential rates alongside as a scenario, which makes the sensitivity visible without departing from the required base case.

Source: healtheconomics.wiki

The rates applied to costs and to health should be stated separately and explicitly, since a single stated rate leaves a reader to assume both. Results at equal rates should be presented alongside any differential analysis so the effect of the choice is visible. Where the conclusion changes between the two, that should be identified as a finding rather than left in a sensitivity table, because it means the recommendation rests on a contested methodological judgement rather than on the evidence.

Source: Drummond et al. 2015