Discussion: The number of patients one needs to treat (NNT) to expect to prevent one patient having the adverse target event expresses the effect of an intervention in a clinically compelling way. Absolute risk reduction and its inverse, NNT, varies with baseline risk, and one can conceptualize an "NNT curve" which describes the continuous relation between baseline risk and NNT. A recommendation to treat at any level of baseline risk depends on the relation between the NNT and the "threshold" NNT at which gains from therapy balance the human and monetary costs. We explored the determinants of the threshold NNT. We conceived the threshold NNT as a point on the NNT curve at which the value of treatment inputs equals the value of treatment outputs. We identified the various elements comprising the treatment inputs and outputs and derived the following equation for computing the threshold NNT:


Threshold = Cost of treating one target event + Value of one target event
NNT [Total treatment cost+Value of treatment attributable side effects per patient treated]

To solve the equation, actual costs or explicit values (e.g., using willingness-to-pay) are needed for each of the four terms. Alternatively, no absolute costs or values need be explicitly measured, the equation being solved using only relative costs and values. The intersection of the threshold NNT with the NNT curve identifies the critical baseline risk above which treatment can be recommended based on this model.