No absolutes

Anthony Kerrigan states that relative risk reduction does not take into account primary and secondary end points (which are expressed by absolute risk reduction) and that presenting absolute risk reductions would give practitioners more realistic estimates of the benefits of specific interventions. Although this viewpoint has been frequently expressed in letters to the editor and editorials, its premise is flawed.

Patients enrolled in clinical trials are frequently at lower risk of important adverse outcomes than the patients commonly encountered in actual practice, many of whom have risk-increasing comorbidities that tend to exclude them from such studies. Therefore, the absolute risk reduction reported in a pertinent clinical trial cannot readily be applied to such patients. The obvious solution is to calculate a new absolute risk reduction (and a new number needed to treat) based on the relative risk reduction reported in the clinical trial, as applied to the patient's estimated baseline, pretreatment risk.1 Fortunately, the relative risk reductions derived from cardiovascular trials tend to be relatively impervious to the baseline risk of the patient. Therefore, as implied by my summary2 of the Anglo-Scandinavian Cardiac Outcomes Trial (ASCOT),3 it is entirely appropriate for practitioners to apply the relative, not the absolute, risk reduction from such clinical trials to the patients they see.

Daniel Hackam Division of Clinical Pharmacology Department of Medicine University of Toronto Toronto, Ont.