Distributional Cost-Effectiveness

An extension of cost-effectiveness analysis examining how an intervention's costs and health effects are distributed across population subgroups, such as by income.

Source: Cookson et al. 2021

Conventional cost-effectiveness analysis reports the total health gained and the total cost, treating a quality-adjusted life year as equally valuable whoever receives it, so it is silent on who gains and who loses. Distributional analysis disaggregates both by population subgroup, most commonly by socioeconomic position, and reports how the health gains and the health displaced by the associated spending fall across those groups. It then combines the efficiency result with a measure of how unequally health is distributed, so a decision maker can see the trade-off between producing more health overall and producing a more equal distribution of it. The analysis makes explicit a consideration that appraisal frameworks otherwise handle informally or not at all.

Source: Cookson et al. 2021

It requires a measure of how health is distributed across the population before and after the intervention, usually expressed as quality-adjusted life expectancy by socioeconomic group. Changes in that distribution are summarised using an inequality index, and the analyst specifies how much aggregate health society would be prepared to give up to achieve a more equal distribution, expressed as an inequality aversion parameter. The result is an equally distributed equivalent level of health, which is the amount of health distributed perfectly equally that would be regarded as equivalent to the actual unequal distribution. Comparing that figure across options identifies which is preferred once both efficiency and equity are weighed.

Source: Cookson et al. 2021

It requires evidence on how the intervention's effects vary across subgroups, including differences in baseline risk, in uptake, in adherence and in the effect itself, none of which trials routinely report by socioeconomic position. It requires an estimate of how the displaced spending would have been distributed, which depends on assumptions about where budget reductions actually fall. And it requires an inequality aversion parameter, which is a value judgement rather than an empirical quantity and is normally varied across a range rather than fixed.

Source: healtheconomics.wiki

The data requirements exceed what most evaluations have available, so subgroup effects are frequently assumed rather than observed. The distribution of displaced health is poorly evidenced, and assumptions about it can drive the equity conclusion. The inequality aversion parameter has no agreed value, so the analysis reports how the conclusion changes across a range rather than delivering a single answer. And subgroups defined by socioeconomic position are only one dimension of equity, with others including geography, ethnicity and disability handled less readily.

Source: healtheconomics.wiki

It is most useful where an intervention plausibly affects groups differently, which covers most public health programmes, screening, and any intervention whose uptake varies with deprivation. It is also useful where a conventional analysis and an equity consideration point in opposite directions, since it quantifies the trade-off rather than leaving it to unstructured judgement. Where effects are uniform across the population it adds little, and the effort is better spent elsewhere.

Source: Cookson et al. 2021