Premium
An efficient design for a study comparing two drugs, their combination and placebo
Author(s) -
Wei Lynn
Publication year - 2005
Publication title -
statistics in medicine
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.996
H-Index - 183
eISSN - 1097-0258
pISSN - 0277-6715
DOI - 10.1002/sim.2336
Subject(s) - placebo , sample size determination , latin square , mathematics , clinical study design , medicine , statistics , research design , computer science , mathematical optimization , clinical trial , chemistry , rumen , alternative medicine , food science , pathology , fermentation
A novel design for a study comparing two drugs (A and B), their combination (AB) and placebo (P) in a stable disease such as chronic asthma is proposed. The primary objectives of the trial are to compare A (a new drug) with placebo and AB with B (an old drug). Other between‐treatment comparisons are secondary. The new design consists of the first 2 sequences of a single 4 × 4 Latin square: (A, AB, P, B|P, B, A, AB|AB, A, B, P|B, P, AB, A). The washout period between A and AB and that between P and B are eliminated based on the assumptions that there is no carry‐over effect from placebo and that the effect of the combination would be approximately same at steady state either starting the 2 drugs at the same time or adding one later. Equal period effect for Periods 1 and 2, and for Periods 3 and 4 are also assumed for all treatment effects being estimable. Advantages and disadvantages of this design were evaluated and compared with other possible designs. Compared with other possible designs, the new design is more efficient. For example, the new design requires less than ¼ of the sample size a parallel design would need and it requires ⅔ of the sample size a 4‐treatment, 3‐period incomplete block design would need for the same power at the same detectable difference. It has a similar efficiency to a 4‐treatment, 4‐period cross‐over design but with a shorter duration. In this design, all between‐treatment comparisons are based on direct within‐subject information. The new design could also potentially decrease the probability of patient's drop‐outs and use of rescue medication due to lack of efficacy during the placebo periods by reducing the number of placebo washout periods. The benefits reaped from the new design seem to outweigh its risks, which include potential bias for the secondary between‐treatment comparisons if the additional assumption about period effects are not satisfied and the incapability of directly assessing the onset of action of the combination of 2 drugs. Copyright © 2005 John Wiley & Sons, Ltd.