In psychiatric research, especially drug trials, investigators frequently use a variety of rating scales. In reporting the results of a study, they may compare scale scores between an experimental group and a control group. Although the use of such rating scales in drug trials may be necessary for approval by the Food and Drug Administration and may generate data that can be easily analyzed, these scales have their limitations. For example, clinicians and patients cannot easily appreciate the practical implications of a small difference in scores on a particular scale.
Dichotomous outcome measures, on the other hand, are often more clinically useful (
+4). Examples of such measures include readmission to the hospital, achievement of full remission, a rating scale score of "improved" or "much improved," or a decrease in score of at least 50 percent. These are all measures that most clinicians and patients would consider to be clinically significant and that can be readily understood. In addition, the use of such dichotomous outcome measures allows for the calculation of a number of useful measures of clinical importance, including NNT.
In illustrating some of these measures and their calculation, we draw on the results of a hypothetical randomized controlled trial comparing an antidepressant (the experimental treatment) with a placebo (the control condition) for the treatment of major depression; the outcome measure was remission. In this hypothetical trial, 60 percent of the patients who received the antidepressant responded and 40 percent did not respond; 40 percent of the patients who received the placebo responded and 60 percent did not.