| Popis: |
To the Editor. —We wish to call attention to several deficiencies in the recent Review of over-the-counter cold medications by Drs Smith and Feldman. 1 The literature on cold medicines is rife with methodological problems, including the use of multiple evaluations with P values at levels appropriate for a single comparison, inattention to the question of clinical relevance, and massaging data to find significance. These problems are illustrated in the three articles cited by Smith and Feldman as supporting the value of chlorpheniramine in the treatment of the common cold. In the 1979 study published in JAMA by Howard et al, 2 the total symptom score of placebo and treatment groups were compared at 24 different times, yet statistical significance was defined at the P equal to .10 level. Although the total number of comparisons is unclear to us, at least 65 comparisons of different outcome measures were made in this |
