Net Benefit Approach

A framework for economic evaluation that converts health outcomes into money at a set value per unit of effect, then subtracts cost.

Source: Stinnett & Mullahy 1998

The health effect is multiplied by a stated value per unit, converting it into money, and the cost is subtracted. The result is a single monetary figure for each option, and the option with the highest figure is preferred. An equivalent formulation divides cost by the same value to express it in health units and subtracts it from the health gained, producing net health benefit. Both give identical rankings and differ only in which dimension is converted into the other. Because the result is a monetary or health quantity rather than a ratio, it can be added across subgroups or across parts of a pathway, which ratios cannot.

Source: Stinnett & Mullahy 1998

To overcome the statistical difficulties of ratios. A ratio becomes undefined as the difference in effect approaches zero, cannot be averaged meaningfully across simulations, and produces the same value in situations carrying opposite recommendations. Net benefit is linear in the underlying parameters, so it can be averaged, aggregated, and given conventional confidence intervals. These properties made probabilistic analysis tractable and are why most modern methodological work is expressed in net benefit terms. It also permits conventional statistical inference, so confidence intervals around the difference between options have their usual interpretation rather than requiring specialised methods.

Source: Stinnett & Mullahy 1998

A value per unit of health, which is the threshold, and this must be supplied rather than derived from the analysis. Because the figure changes with the threshold, results are presented across a range rather than at a single point, and the threshold at which the sign changes is the incremental cost-effectiveness ratio. The approach therefore does not remove the need for a threshold; it makes the dependence on it explicit rather than implicit. Plotting net benefit against the threshold is the standard presentation, since the line shows both the recommendation and the point at which it would change.

Source: Briggs, Claxton & Sculpher 2006

Directly, by calculating the figure for each and selecting the highest, without the sequential removal of dominated and extendedly dominated options that ratio-based analysis requires. This is one of its practical advantages, since the ordering procedure is error-prone and unnecessary once options can be ranked on a single quantity. Dominance still holds arithmetically, since a dominated option cannot have the highest net benefit at any threshold. Dominated options can therefore be identified without the sequential procedure, simply by observing that their net benefit is lower at every threshold examined. Extended dominance is likewise handled automatically, since a mixture cannot achieve higher net benefit than the best single option at any threshold.

Source: Drummond et al. 2015

It requires the threshold to be treated as known, and where the threshold is itself uncertain that uncertainty is separate from parameter uncertainty and is handled by presentation rather than simulation. Converting health into money makes the exchange rate explicit, which some find objectionable even though the same assumption underlies any threshold comparison. And the figure is not comparable between analyses using different thresholds, so it travels less well than a ratio does. Reporting the threshold used alongside any net benefit figure is therefore essential, since the number is meaningless without it in a way that a ratio is not.

Source: Stinnett & Mullahy 1998