It is tempting to despair that scientific conditions can ever be substituted for clinical circumstances. Randomized trials were first developed for agricultural research in an attempt to account for the natural variability in plots of ground and species. The results of those trials have been used to produce consistent increases in crop yields.1
While random allocation of treatments and isolation of variables appear natural in agriculture, it can be an invasive procedure in clinical medicine. The randomized clinical trial requires clinicians to set aside their own judgments, asks patients to accept a treatment that may be substandard, introduces arbitrary definitions of disease, encourages recruitment of patients, and requires aggressive follow-up and unusual efforts to ensure compliance. But the clinical milieu is complex and refractory; seldom does the randomized clinical trial succeed in eliminating or bypassing "extraneous" clinical variables.
More than two years after the publication of the β-blocker heart