The purpose of this paper is to challenge the uncritical use of increm
ental cost-outcomes ratios as decision variables. A scenario is presen
ted which describes conditions under which increasing costs per unit o
f outcome prevail. Marginal costs increase as the proportion of patien
ts treated under a given therapy increases. If the health system's obj
ective is to maximise health benefits then patients will be switched u
ntil the marginal benefits per dollar expended are equal between the 2
therapies. In an example where the costs of the new therapy are great
er, for a given proportion of patients treated, patients are switched
from the existing to the new therapy until an equilibrium is achieved
in the allocation of therapies among the treating population. At this
point, the overall costs of treatment are at a minimum. This outcome c
ould only be predicted if the underlying cost-outcomes functions are k
nown and the consequent patterns of therapy substitution and cost impa
cts assessed. The paper concludes by raising concerns as to the role o
f incremental cost-outcomes ratios as decision variables where increas
ing costs may be expected to prevail and there is failure to consider
the implications of these increasing costs in formulary decision makin
g. If increasing costs are present then conventional cost-outcomes and
incremental cost-outcomes ratios are of limited utility as decision v
ariables in the choice of therapy options.