Independence of treatment effects
I thank Tom Marshall for clarifying some of the points raised above.
However, his response to my comment about independence of treatment
effects did not refer to exactly what I had in mind, so apologies if I did
not explain myself very clearly the first time around.
It is indeed true that, if the effects of multiple treatments in a
single patient are not independent, the more expensive treatments will be
even less cost effective, but that was not the point I was trying to make.
My point was this. Can we be sure that the effects of treatment are
independent of the patient's baseline risk? In other words, if a treatment
given to patients with a baseline risk of 15% reduces that risk to, say,
7.5% (50% relative risk reduction), does it necessarily follow that the
same treatment given to patients with a baseline risk of 5% will reduce
the risk to 2.5% (also a 50% relative risk reduction)?
Marshall's analysis is based on this assumption, but if it were not
true, then intensive treatment of patients at high baseline risk might
become a more cost-effective strategy than less intensive treatment of
patients at low baseline risk. Perhaps a sensitivity analysis could shed
some light on this?
Competing interests: No competing interests