ANIMAL experiments may offer evidence of the value of a substance for human nutrition, but only well-conducted clinical trials can result in a verdict. Unless substances are tested on human subjects, judgment should be withheld. Though the design of experiments varies, certain general principles pertain to all experimentation; in addition, special considerations apply in the field of nutrition. The outline which follows presents some of these principles and considerations and should be useful to those conducting clinical trials.
A statement of specific objectives should be prepared as the first step. In order to estimate the necessary sample size, estimates of expected results must be obtained, or deduced, either from the literature or from a pilot experiment. All possible sources of bias, such as age, sex, race, etc. should be listed in advance and considered in planning the study. Decisions must also be made concerning the following: 1) population for which the results will be relevant; 2) controls; 3)criteria for judging results; 4) sample size and duration of study; 5) form of reporting results. Whenever possible the people who give treatments and make measurements should be unable to distinguish between the study and the control subjects.
The type of population for which the conclusion is intended will determine the choice of subjects. For example the value of a foodstuff for all infants should be studied in subjects from a heterogeneous population. Alternatively, a dietary supplement for treating deficiency must be assayed in deficient subjects. Such a supplement would probably produce no signficant changes in normal subjects but even if it did the results would not necessarily apply to treatment of deficiency.
- Copyright © 1957 by the American Academy of Pediatrics