Search

TO THE EDITOR:— It is to be hoped human drug trials, present and future, will be viewed with a more critical eye than has been the case in the past. I am writing regarding "A Controlled Double-Blind Study of Trimethobenzamide, Prochlorperazine and Pacebo" (196:796, 1966), by Dr. Philip Bardfeld.The author found "no significant differences between saline and the trimethobenzamide," but made a strong statistical point about "the difference between the prochlorperazine and the placebo." I feel that any impression given by the author that prochlorperazine was almost significantly better than placebo should be corrected. The author's calculation of chi-square is in error, leading as it does to a P value between 0.07 and 0.08. This error arises from a failure to make a continuity correction which is necessary in 2 × 2 chi-square tables. Rather than the author's 3.25, the correct chi-square value is 2.37 which corresponds to