In answering questions about whether or not a treatment should be offered to your patient, whether or not the treatment is worth the cost, etc., relative risks, relative risk reductions and the like are not terribly helpful in making those decisions -- that is, they don't give you useful, CLINICALLY relevant information -- save that of the ABSOLUTE RISK REDUCTION.
This is a relatively useful number because it reflects the treatment effect for the population of individuals in the study. By simply taking the inverse of the absolute risk reduction
1/ARR = NNT
we can arrive at the Number Needed to Treat (NNT). This is to say that you would need to treat that many patients to prevent or cause one outcome. For example, if the risk of outcome (O) for treatment (y) was 50%, and the risk of (O) for the control (x) was 75%, then the absolute risk reduction (ARR) is x-y, or 25%. By taking the inverse of this number, we arrive at the NNT, in this case 1/0.25, or 4. You would only have to treat four patients to achieve the outcome (O). Intuitively, the larger the ARR, the fewer the patients who will need to be treated to achieve the outcome. Conversely, the smaller the ARR, the greater the number of patients who will need to be treated.
Each physician will have his or her own threshold for treating using the above method -- for instance, if the outcome is significantly severe enough in terms of morbidity/mortality, cost, and the like, you may want to INCREASE your threshold (an NNT of 100, say, instead of 4). It is not quite as black and white as I'm making it sound here, because you also need to consider the cost/benefit/harm of treatment as well. This really constitutes Decision Analysis, a topic that requires the knowledge of critical appraisal, but will be discussed elsewhere. However, this process can help in making a decision when you are in complete control of the variables!
Sackett DL, Haynes RB, Guyatt GH, Tugwell P. CLINICAL EPIDEMIOLOGY: A Basic Science For Clinical Medicine. 2nd ed. Little, Brown & Co., 1991.
Laupacis A, Sackett DL, Roberts RS. An Assessment of Clinically Useful Measures of the Consequences of Treatment. NEJM, 1988;318:1728-.
Rajkumar SV, Sampathkumar P, Gustafson AB. Number Needed to Treat Is a Simple Measure of Treatment Efficacy for Clinicians. JGIM, 1996;11:357-359.