Net Monetary Benefit

Generally, cost-effectiveness analysis expresses the outcome in the cost-effectiveness ratio. This ratio relates the difference in costs of two alternative healthcare interventions to their difference in health effects. Two alternatives may, for example, be two types of pharmacotherapies— that is, a new drug being compared with the old standard treatment. The difference in health effects may be expressed in life years gained, quality-adjusted life years (QALYs) gained, disability-adjusted life years (DALYs) averted, and [Page 812]so on. In a formula, the cost-effectiveness ratio (R) may be written down as

with ΔC the difference in costs and ΔE the difference in health effects.

Obviously, R represents a ratio, limiting its usefulness for understanding the relative sizes of the differences in costs and effects. For example, the ratio does not provide any information on the budget impact. Furthermore, the ratio does not differentiate between the SE and NW quadrants of the cost-effectiveness plane. For example, both the combinations of (ΔC = −100; ΔE = 10) and (ΔC = 100; ΔE = −10) result in the same cost-effectiveness ratio of −10, with the former combination being very acceptable and the latter very unpleasant. Finally, it is well established that statistical analysis on ratios involves some specific problems. Ergo, some drawbacks exist while working with cost-effectiveness ratios.

As one of several options to overcome these drawbacks, the concept of net monetary benefit (NMB) has been developed. To arrive at the NMB, the above equation on the cost-effectiveness ratio is simply rewritten, after having inserted an explicit threshold for cost-effectiveness (often denoted as λ). This threshold explicitly gives the maximum amount of money that society/decision makers want to pay for gaining one unit of health effect, for example, a QALY. So we are interested in the situation that

If we rewrite this as

or

we have formally derived the requirement that the NMB (= λΔE − δC) should be positive. So, for example, for a new drug, we would require that the monetarized difference in health effects (λΔE) exceeds the difference in costs (ΔC). It is immediately clear from inserting just the simple example of (ΔC = −100; ΔE = 10) and (ΔC = 100; ΔE = −10) that the NMB does differentiate between SE and NW quadrants.

Furthermore, if calculated on the exact patient populations within specific countries, the NMB provides exact information on the costs (savings) to be net paid (achieved). As such, it gives policy makers information on the socioeconomic impact at the national macrolevel. Finally, calculus of the NMB enhances possibilities for formal statistical tests and analysis of uncertainty around point estimates of cost-effectiveness.

NMB is a versatile tool to evaluate uncertainty in health-economic analyses. For example, derivation of uncertainty intervals around cost-effectiveness ratios is often based on analysis of NMB. Also, NMB can be used for sample size estimation for cost-effectiveness in clinical trials. Furthermore, subgroup analysis using regression techniques on NMB is very straightforward.

NMB allows also for uncertainty evaluation of cost-effectiveness point estimates from health-economic models using cost-effectiveness acceptability curves. In this case, the probabilistic sensitivity analysis is used to generate cost and effect pairs (e.g., 10,000 simulations) for all alternatives included in the model (see Table 1 for a numerical example). In the next step, the threshold for cost-effectiveness (λ) is varied, and the NMB for each cost and effect pair is calculated (Table 2). The number of simulations with the highest NMB represents the probability of cost-effectiveness for each alternative at the specific threshold (bold in Table 2). Finally, the probability is plotted versus the threshold (λ) and the cost-effectiveness acceptability curve (CEAC) is created (Figure 1). Also, value of the information theory is building on the NMB concept.

...