Control in the Middle (CIM) for Three-Period Crossover Studies

Abstract

Three-period crossover studies can be efficient and convenient methods of conducting Phase II clinical trials. Non-randomly placing control in the middle (CIM) has not been practiced but may be extremely useful in studies testing herbal products for which placebos are not available, or for distinguishing between behavioral and biological effects. Furthermore, this design can serve as a valuable addition to classical studies of either (a) two competing treatments or (b) treatment versus placebo versus an open label “nothing” as the control. Therefore, we propose rigorous designs that will help practitioners efficiently answer research questions where (1) two active treatments need to be compared against each other with treatment vs. placebo comparisons being of secondary importance; (2) a single active treatment needs to be tested where no placebo is available; or (3) the placebo effect is of interest in a treatment vs. placebo trial. For studies where no placebo is available, deception will be required, with participants told that in one randomly selected period (#1 or #3) they will receive the active treatment, and that they will receive a new experimental inert placebo in the other period. Assuming this design is approved by an ethics committee, it can be very useful in biomedical research.

Key words

blinding - crossover study - deception - phase II clinical trial - placebo - washout

References

• 1 Senn S J. Cross-over trials in clinical research. Chichester; Wiley 2002
• 2 Shuster J J. Design of experiments. Ambrosius WT Methods in molecular biology: elementary biostatistics, chapter 12. Totawa; Humana Press 2007
• 3 Tudor G, Koch G G. Review of nonparametric methods for the analysis of crossover studies.  Stat Methods Med Res. 1994;  3 345-381
• 4 Shuster J J. Diagnostics for assumptions in moderate to large simple clinical trials: do they really help?.  Stat Med. 2005;  24 2431-2438
• 5 Shuster J J. Student t-tests for potentially abnormal data.  Stat Med. 2009;  28 2170-2184
• 6 Chow S, Shao J. Statistical methods for two-sequence three-period crossover designs with incomplete data.  Stat Med. 1997;  16 1031-1039
• 7 Kuratsune H, Umigai N, Takeno R, Kajimoto Y, Nakano T. Effect of crocetin from Gardenia jasminoides Ellis on sleep: a pilot study.  Phytomedicine. 2010;  17 840-843
• 8 Mohammed Abdul M I, Jiang X, Williams K M, Day R O, Roufogalis B D, Liauw W S, Xu H, McLachlan A J. Pharmacodynamic interaction of warfarin with cranberry but not with garlic in healthy subjects.  Br J Pharmacol. 2008;  154 1691-1700
• 9 Kaptchuk T J, Stason W B, Davis R B, Legedza A T R, Schnyer R N, Kerr C E, Stone D A, Nam B H, Kirsch I, Goldman R H. Sham device vs. inert pill: randomised controlled trial of two placebo treatments.  Br Med J. 2006;  332 391-397
• 10 Mayberg H S, Silva J A, Brannan S K, Tekell J L, Mahurin R K, McGinnis S, Jerabek P A. The functional neuroanatomy of the placebo effect.  Am J Psychiatry. 2002;  159 728-737