Hi There,
This is my first posting to the list but I've been watching the NNT discussion
with interest. I think there is a basic problem in answering the question
posed in that NNT/NNH are not generic measures. There might be an acceptable
answer to the question "What is a good cost per quality adjusted life year
saved?" but NNT does not serve the same purpose as cost per QALY saved. It is
intended as a tool to present intuitively understandable figures to patients,
decision makers or clinicians. Thus it is not intended that NNTs for different
treatments should be comparable.
As Rod pointed out the NNT needs to be interpreted in the light of the nature
of the treatment, the nature of the outcome and the time period under
consideration. Thus the use of NNT may be presented in terms of "if we were to
treat x patients with treatment y then there would one fewer of outcome z over
the next N years". All the variables - x, y, z and N are needed to interpret
the NNT and most of the discussion so far has concentrated on the "x" and "N".
This is best illustrated by examples. There would probably be huge demand for
a single shot immunisation against AIDS with an NNT in the thousands whereas
it would be a perfectly reasonable decision for a patient to decline to
undergo a major surgical procedure with an NNT of 2 (e.g. "if we were to treat
two 85 year-old patients with an asymptomatic aortic aneurysm the same size as
yours by surgical repair then one of them would be saved from death due to
their aneurysmal disease over the next five years"). Note - both examples are
prophylactic treatments.
There are all sorts of other issues that NNT does not address, such as
discounting, risk aversion etc. and, at least in theory, one could conceive a
"generic NNT" with standardised outcomes (e.g. discounted quality adjusted
life years) weighted treatment severity etc.
But wouldn't that defeat the whole object of the exercise!
Jonathan Michaels
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|