Until now, however, decision-making bodies of different countries using ICERs
in reimbursement decisions have not formed precisepointestimates of the thresh-
olds (above) below which health technologies would be strictly defined as (not)
cost-effective. This is partly because the cost-effectiveness of a technology is a
necessary but certainly not the sole basis for decision-making. The National
Institute for Health and Care Excellence (NICE) in the UK uses a threshold range
varying between₤20,000 and₤30,000, while the weight of other factors (such as
equity considerations) on the decision to recommend a technology for reimburse-
ment is greater if the ICER is closer to the top of the range.^19 Generally speaking,
the treatments that are considered cost-ineffective have a lesser chance of being
reimbursed than treatments with a much-favorable cost-effectiveness ratio.^20
2.4 Uncertainty
Economic evaluations are usually based on different types of data, and many
calculations are involved in the estimates of the costs and the benefits of health
care interventions. The data included in economic evaluations often requires
modeling, mainly of health- and cost-related data. As defined by Russell ( 1999 ),
a model creates the framework for cost-effectiveness analysis, allowing decision-
makers to explore the implications of using the intervention in different ways and
under different conditions. A model answers a question of what happens to the
estimate of relative cost-effectiveness (ICER) if, for instance, costs increase by
15 % or if benefits are not achieved in 1000 patients (as defined in the foundational
analysis) but only in 100 patients. To serve its purpose, a model must produce
accurate predictions and allow for substantial variation in the factors that influence
costs and effects.^21 In practice, each economic evaluation (and welfare analysis of
any type) involves a certain degree of uncertainty primarily because the estimates
of costs and health benefits produced by new health interventions cannot be exactly
predicted in advance. The uncertainty may stem from various sources, for instance,
from treatment options that are not always clear (at least not initially), from the
failures in treatment adherence and patient behavior, changes in costs, and so
on. Uncertainty can also stem from the model specification or the measurement
issues surrounding the available data. The uncertainty surrounding the main param-
eters used in economic evaluations is usually addressed by means of sensitivity
analysis. Sensitivity analysis refers to techniques that analyze how sensitive the
result of economic evaluations (e.g., ICER) is to the changes in the important
(^19) Further details on the threshold used in the UK, see, for instance, Rawlins et al. ( 2010 ),
pp. 346–349.
(^20) As shown, for instance, by Devlin and Parkin ( 2004 ), pp. 437–452; Dakin et al. ( 2006 ),
pp. 352–367.
(^21) Russell ( 1999 ), pp. 3235–3244.
116 A. Bobinac and M. Vehovec