Incremental Cost-Utility Ratio

The difference in cost between two interventions divided by the difference in quality-adjusted life years they produce.

Source: Weinstein & Stason 1977

It is a particular case of the incremental cost-effectiveness ratio in which the health effect is measured in quality-adjusted life years rather than in a natural clinical unit. Because that measure combines survival with the quality of the years lived, the resulting ratio can be compared across disease areas and across services, which a ratio expressed in cases detected or events avoided cannot. The two terms are frequently used interchangeably, and the distinction matters only when the outcome unit is in question. Where a framework specifies quality-adjusted life years as the reference outcome, the incremental cost-utility ratio is what submissions are expected to report, and analyses using other outcome measures are treated as supporting evidence.

Source: Weinstein & Stason 1977

Because a system allocating a fixed budget must compare interventions across conditions, and only a common outcome measure permits that. Cost per quality-adjusted life year allows a hip replacement, a cancer treatment and a mental health service to be assessed against the same threshold. The price is that health must be reduced to a single index, which requires the quality weights to be credible and the descriptive system to be sensitive to the changes at issue. It also allows a single threshold to be applied consistently across the whole health service, which is the practical requirement that drove adoption of the measure.

Source: Drummond et al. 2015

Survival and quality of life must both be estimated for each option over the horizon, which normally requires a model rather than trial data alone, since trials rarely follow patients for a lifetime. Health state utility values are needed for every state in the model, obtained from a preference-based instrument or from published valuation studies. Costs must be estimated on a consistent basis for both options. Each of these carries uncertainty that propagates into the ratio. Where survival is extrapolated beyond the observed data, the assumed shape of the extrapolation frequently contributes more to the ratio than any parameter estimated from the trial itself.

Source: Brazier, Ratcliffe, Salomon & Tsuchiya 2017

The quality-adjusted life year treats a unit of health as equally valuable whoever receives it, so the measure is silent on severity, disadvantage and the distribution of gains. Descriptive systems are insensitive to changes that matter in some conditions, so interventions in those areas appear less effective than they are. Different instruments produce different values for the same state, so the ratio depends partly on the measure chosen. And older patients gain fewer quality-adjusted life years from an identical intervention. These limitations are why frameworks applying the ratio attach additional considerations to it rather than deciding on the figure alone, and why the ratio is treated as one input to a judgement.

Source: Neumann, Sanders et al. 2016

The comparator, perspective, horizon and discount rate, as for any incremental ratio, together with the instrument used to obtain utility values and the population from which those values were elicited. Where the utilities came from a source other than the trial population, that should be stated. Uncertainty should be presented probabilistically rather than as a confidence interval, and the sensitivity of the ratio to the utility values should be reported, since they are frequently the input to which it is most sensitive. The source of the survival extrapolation and the alternative shapes examined should also be reported, since that assumption is frequently the second most influential input after the utility values.

Source: Drummond et al. 2015