Premium
Crossover Trials Should Not Be Used to Test One Treatment against Another Treatment with a Totally Different Chemical Class/Mode of Action
Author(s) -
Cleophas Ton J.
Publication year - 2000
Publication title -
the journal of clinical pharmacology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.92
H-Index - 116
eISSN - 1552-4604
pISSN - 0091-2700
DOI - 10.1177/009127000004001223
Subject(s) - crossover , crossover study , correlation , correlation coefficient , statistics , mathematics , medicine , computer science , placebo , alternative medicine , geometry , pathology , artificial intelligence
Comparisons of treatments with totally different chemical class/mode of action are at risk of a negative correlation between treatment responses. The crossover design is flawed when correlation between treatment responses is negative. The objective of this study is to assess whether this flaw is prevalent in the literature. Fourteen randomized controlled crossover studies were assessed for correlation levels in relation to their type of treatment comparison. Correlation coefficient ( r ) was calculated using the t‐ statistic : p‐ values were turned into t‐ values for the degrees of freedom, and r was calculated using the formula for the pooled standard error of the mean ( SEM ): (pooled SEM ) 2 = SEM 1 2 + SEM 2 2 − 2 r SEM 1 •SEM 2 . Crossover studies comparing treatments with totally different chemical class/mode of action were frequently negative, and this was obviously due to their correlation between treatment responses' being negative. Crossover studies comparing similar treatments frequently had a positive correlation, and this added extra sensitivity in the treatment comparison; 27% of the crossover hypertension studies compared completely different treatments, and they should not have been performed in the form of a crossover study. The author's conclusion was that crossover trials should not be used to test one treatment against another treatment with a totally different chemical class/mode of action. Crossover trials are particularly sensitive to compare treatments from one chemical class with one mode of action. It is hoped that this work affects the design of future crossover trials .