Crossover studies are a better format for comparing equivalent treatments than parallel‐group studies.
- 157 Downloads
Background: In controlled clinical trials the new treatment may be a slight modification of the standard or be equivalent to it with the addition of a new component. In this situation there is a positive correlation between the response to the new and the standard treatment. Objective: The influence of the correlation between the treatment responses on the statistical sensitivity of testing was studied. Randomized trials of equivalent treatments in psychiatry and hypertension research were studied for their design in relation to their level of correlation. Results: With equivalent treatments and, thus, a positive correlation a paired analysis provides better power than an unpaired one. Conclusions: Crossover studies are a better format than parallel‐group studies for comparing equivalent treatments. The scientific community is technically largely unaware of the mechanisms by which correlation levels influence or lack to influence the statistical power of controlled clinical trials.
Unable to display preview. Download preview PDF.
- 1.Cleophas TJM. Clinical trials: specific problems associated with the use of a placebo-control group. J Mol Med 1995;73:421–24.Google Scholar
- 2.Senn SJ. Problems with two stage analysis of crossover trials. Br J Clin Pharmacol 1991;32:133.Google Scholar
- 3.Senn SJ. Crossover trials in clinical research. Chichester, England: John Wiley & Sons Ltd, 1993.Google Scholar
- 4.Nies AS, Spielberg SP. Individualization of drug therapy. In: Hardman JL, Limbird LE, eds. Goodman and Gilman's Pharmacological Basis of Therapeutics. New York: McGraw-Hill, 1996: 43–63.Google Scholar
- 5.Fleiss JL. A critique of recent research on the two-treatment crossover design. Control Clin Trials 1989;10:237–43.Google Scholar
- 6.Freeman PR. The performance of the two-stage analysis of two-period, two-treatment crossover trials. Stat Med 1989;8: 1421–32.Google Scholar
- 7.Cleophas TJM. Relevance of correlation between treatment responses. Eur J Clin Pharmacol 1996;50:1–6.Google Scholar
- 8.Grieve AP. Bayesian analysis of two-treatment crossover studies. Stat Meth Med Res 1994;3:407–29.Google Scholar
- 9.Cleophas TJM, Fennis JFM, van't Laar A. Alpha-and beta-blockade and beta-stimulation in Raynaud's syndrome. Angiology 1985;36:219–26.Google Scholar
- 10.Cleophas TJM. Crossover trials are only useful when there is a positive correlation between the response to different treatment modalities. Br J Clin Pharmacol 1996;41: 235–9.Google Scholar
- 11.Jackson PR, Yeo WW, Cleophas TJM. Letters to the editor. Br J Clin Pharmacol 1996;42:399–404.Google Scholar
- 12.Willan AR, Pater JL. Carryover and the two-period crossover clinical trial. Biometrics 1986;42:593–99.Google Scholar
- 13.Scheffe H. Mixed models. In: Scheffe H, ed. The analysis of variance. New York: John Wiley & Sons, 1959, 261–91.Google Scholar
- 14.Gram LF. Fluoxetine. N Engl J Med 1994; 331:1354–61.Google Scholar
- 15.Cardiology Dialogue. Edited by Rapid Literature Service. Cologne, Germany: Limbach GMBH, 1993.Google Scholar
- 16.Cleophas TJM, Tavenier P. Fundamental issues of choosing the right type of trial. Am J Ther 1994;1:327–32.Google Scholar
- 17.Lee ML, Lusher JM. The problem of therapeutic equivalence with paired qualitative data: an example from a clinical trial using haemophiliacs with an inhibitor to factor VIII. Stat Med 1991;10:433–41.Google Scholar
- 18.Blackwelder WC. Proving the null hypothesis in clinical trials. Control Clin Trials 1982;3:345–53.Google Scholar
- 19.Cleophas TJM, van Lier HJ. Crossover studies with a binary response: a powerful method in spite of crossover effect. J Clin Pharmacol 1996;36:198–204.Google Scholar