When every participant receives one and only one treatment in a random manner, this kind of complete randomized design is called a parallel group design. The most important element of this design is randomization which means participants are randomly placed into a group to lower the risk of statistical bias or other kinds of erroneous results. Generally, researchers use either one of the two types of parallel group designs in a comparative clinical trial, that is, parallel group (or group comparison) designs or matched-pair parallel designs (Figure 1). Based on certain characteristics, participants are first paired in matched-pair parallel design. Within one pair, each member will be assigned to different study subgroups randomly, which makes it possible to compare similar participants undergoing diverse procedures between each other. A treatment group vs. a control group is the simplest two-group comparison parallel-group design. In each treatment group there are usually the nearly same amount of participants that's not always the case. A parallel group design is very different from a crossover design in which one group of participants receives treatment A first followed by B then while the other group does the opposite. However, there is also a significant issue in a parallel trial. Researchers must take into consideration the intra subject variability, the variability within the same participants in responses to treatment.
Figure 1. Parallel group design and matched-pair parallel design.
There are usually five options for parallel group designs (shown in Table 1).
Table 1. Options for parallel group designs.
For a better clinical trial plan, it's better for researchers to have a deep understanding of the advantages and disadvantages of candidate statistical designs before determining an appropriate design at the beginning. The key point is that “Is your chosen design suitable to answer your scientific or medical questions?” As in the comparison of a crossover design and a parallel group design, the former one is a kind of trial design in which participants in different groups receive all treatments in a different order. Which means that participants in group X will receive treatment A then treatment B, while those in group Y receive treatment B then treatment A (Figure 2). Participants in the latter one receive completely separate treatments in parallel, that is, group X receives treatment A while group Y receives treatment B. If the number of participants is the same, a crossover design is with a higher statistical power than a parallel group design. Because participants can act as their own controls while parallel group design requires a separate comparison group, tending to be more expensive. The key strength of parallel group design is the absence of carry over effect. Each participant only receives one treatment and there is of course no carry over effect from a previous treatment. In addition, it is globally accepted as the ‘gold standard’ for phase 3 trials. Generally, if you are intended to study the residual effect that might be carried over from one treatment to another or a bioequivalence trial with single dose or multiple dose, a crossover design is a better choice. However, if you are trying to conduct an effectiveness and safety study of a candidate medicine, a parallel group design is more appropriate.
Figure 2. Crossover design.
At CD BioSciences, we can help you with not only the parallel group design but also the appropriate choice of statistical strategy. If you have any questions, please feel free to contact us.
References:
1. Johnson, T. (2001) ‘Proof of efficacy trials: crossover versus parallel-group studies’, Epilepsy Research, 45(1), 49-51.
2. Campbell, M. J. (2010) ‘Fundamentals of clinical research: bridging medicine, statistics and operations by antonella bacchieri, giovanni della cioppa’, International Statistical Review, 75(3), 411-412.