Blocking

Another common method employed to decrease the background variability is blocking. Like stratification, blocking involves the subdivision of the subject population into homogeneous subgroups. The experimenter defines blocks of subjects and randomizes the subjects within each block to the study treatments, such that the same number of subjects are assigned to each treatment within each block. The blocks are defined so that the intrablock variability is minimal, e.g. to determine whether a drug is carcinogenic, rats of the same litter are randomized to received several doses of the drug or placebo. In this way, the variability due to genetic variation is minimized.

To take advantage of the block design, the treatments are compared within each block and then the information is pooled across blocks. When the 'within-block' or 'intrablock' variability is substantially smaller than the 'between-block' or 'interblock' variability, blocked designs could be very efficient in the use of subject resources. One disadvantage of blocked designs is that they do not allow for missing data. If data from one subject in the block are missing, the entire block may be disqualified.

A variation on the idea of blocking is the crossover design. Here, each block consists of one subject, who receives the study treatments in a random order. Cross-over experiments are frequently used in bioavailability and pharmacokinetic studies. The reason is that the pharmacokinetic parameters that determine the absorption, distribution, and metabolism of the drug in the body and its elimination from the body depend on the biological make-up of the subject and vary, often considerably, from subject to subject. Thus, the intersubject variability is typically much higher than the intrasubject variability. In cross-over studies the treatments are compared within each subject and then summarized across subjects. The cross-over design is different from the blocked design described above, in that each block consists of a single subject, which means that measurements within each block are not independent of each other. Furthermore, it is possible that a residual effect of one drug carries over to impact the effect of another drug administered subsequently. Statistical analytical methods are limited in their ability to adjust and correct for such effects. This is why the use of cross-over designs in clinical research is limited.

In summary, the design of a clinical trial incorporates methods of minimizing noise and the prevention of bias. This is done through the use of appropriate subject allocation procedures, such as randomization and blinding, or through the use of stratification and blocking.

0 0

Post a comment