In the last post Clinical Trials (P-1), we read about the term Clinical trial and few terminologies like population and subject encountered in these types of studies. In this post, let’s further put light on the term: Randomization.
As mentioned in the last post, clinical trials are interventional research studies that explore the safety and effectiveness of a medical treatment or device in humans.
To carry out a clinical trial, subjects are chosen from a population, who actually participate in the study. The subjects are divided into test group and control group. Each group is either treated or not treated with the test treatment (depending on the study), and ultimately the results of the two groups are obtained. The results are compared using a suitable biostatistical tool and the conclusion is made if the test treatment/ intervention is good enough and if it can be used on humans on a large scale or no.
For such comparison, between the groups, the two groups should have similar baseline readings. That is the two groups should be similar at the start. So, the difference at the end of the study can be attributed solely to the intervention (treatment).
However, in some cases the researcher may be biased (regarding subject or outcome) and can deliberately manipulate the baseline readings to change the final outcome. That is the researcher can put his favored patient (friend or relative) into either the test or control group (whichever he feels is beneficial). A researcher may also put comparatively healthier patients into the test group to show that the test material is better, and hence manipulate the outcome (concluding that the group participants heal faster or better).
To achieve an honest outcome, the baseline must be as uniform as possible. This required comparability or balance at start (baseline) can be achieved by randomly (without any bias) allocating subjects to either of the groups; this is known as randomization.
The participants are chosen randomly and have equal chance to be assigned to either of the group. This helps in avoiding bias during allotment of the patients. A randomized trial is therefore a vital means for testing the efficacy of the treatment.
The procedure includes:
- To generate an unpredictable random sequence.
- To implement the sequence in a way that conceals the treatments until patients have been formally assigned to their groups.
– Benefits of randomization
- Eliminates selection bias.
- Balances arms with respect to prognostic variables (known and unknown).
- Forms basis for statistical tests, a basis for an assumption-free statistical test of the equality of treatments.
– Disadvantages of Randomization
It leads to increase in the complexity of the study and hence the overall cost.
• Types of Randomization
- 1. Simple Randomization:
These are easiest and most basic methods. This includes manual methods, tossing coin, drawing lots or throwing a dice.
(tossing a coin: for each subject that enters a trial. Such as Heads = Active, Tails = Placebo. Throwing dice: even number: group A; odd number: group B)
It also includes, generating random allocation sequence using a random-number table or a computer software program (like Research Randomizer) that generates the random sequence. The participants are then allocated to the different groups as per the random sequence generated.
Advantage: Simple, Easy to implement, Completely unpredictable.
Disadvantage: Imbalance in treatment assignment, especially in smaller trials (in 10 tosses there could be 7 tails and 3 heads, leading to imbalance in assigning the treatments). Cannot be checked or reproduced.
2. Block Randomization
Here the researcher divides participants into subgroups called blocks. Each block can be considered as a sequence of different treatment in equal number. Subjects within each block are randomly assigned to either of the treatment group. Hence the participants are randomized within blocks such that an equal number are assigned to each treatment.
Suppose there are 12 patients, who have to be assigned to two treatment groups A and B. The researcher, say designs blocks of block size 4. Supposedly the blocks are AABB, ABAB and BABA. The patients are alloted into the 2 groups based on the sequence of the blocks. Hence considering the first block AABB; the first two patients are allocated to the treatment A then the two to treatment B. Next 4 patients will be allocated to A, B, A and B treatments respectively. And so on. Its necessary for the researcher to be blind to the sequence or the sequence can be predicted, again making space for bias.
Block size depends on the number of treatments, it should be short enough to prevent imbalance, and long enough to prevent guessing allocation in trials. The block size should be at least 2x number of treatments. The block size is not stated in the protocol. The volunteers and investigators are blind to the block size.
Advantage– It divides participants equally into the treatment groups.
Limitation– If blocking is not masked in open-label trials, the sequence becomes somewhat predictable (e.g. 2n= 4): B A B? Must be A.
This could lead to selection bias. The solutions to avoid selection bias are:
(1) Do not reveal blocking mechanism.
(2) Use random block sizes.
- a. Stratified Randomization:
Most of the medical conditions are affected by prognostic factors, such as, sex, age, race, severity of condition, etc. In case of such medical conditions, alloting subjects randomly to the groups may not result in balanced baseline readings. In such cases the trial should be well-balanced across prognostic factors. Hence the subjects are stratified according to different prognotic factor to produce relatively homogenous, non-overlapping and comparable groups with regard to certain characteristics. Thus, produce valid statistical tests.
Eg: if one had to find the average height of school children, they have to be divided into strata of certain class groups, like Class 1 to 3, class 4 to 6 and so on. Here the prognostic factor ‘age’ influences the ‘height’ of the subjects.
The subjects are first divided/stratified based on such prognostic factors and then the participants can be allocated to the different treatment groups using block randomization method. Hence in this, the subjects should have baseline measurements taken before randomization. Increased number of stratification variables or increased number of levels within strata leads to fewer patients per stratum. The block size should be relatively small to maintain balance in small strata.
Large clinical trials don’t use stratification. It is unlikely to get imbalance in subject characteristics in a large randomized trial.
Eg: use block randomization separately to stratify
- Cancer patients into diabetics and non-diabetics. (Then diabetes and non-diabetes can be divided into test or treatment group)
- Age Group: < 40, 41-60, >60; Sex: M, F: Total number of strata = 3 x 2 = 6
Stratification Randomization can hence balance subjects on baseline covariates.
- 2. Unequal Randomization
Randomized trials with equal numbers of patients to experimental and control groups give the most statistically efficient randomization ratio as it maximizes statistical power for a given total sample size.
However, this may not be the most economically efficient or ethically/practically feasible in some cases.
– Economical Reasons:
When two or more treatments under evaluation have a cost difference it may be more economically efficient to randomize fewer patients to the expensive treatment and more to the cheaper one. Some costly drug or surgery cannot be tested on a large number of patient.
– Ethical Reasons:
When one arm of the treatment saves lives and the other such as placebo/medical care can delay the treatment and hence be harmful. Here the subject survival time depends on which treatment they receive. Hence extreme allocation may be used in these trials to allocate fewer patients into the placebo group and provide immediate medical help to the patients.
Example, in case of some serious cardiac disorders, it would be really unethical to give a placebo to the patient.
This is all for this post. Few more terms used in the clinical studies will be discussed in next post, before we start with the different phases involved in the clinical trials.
Other posts by The Biotech Notes;
Please comment, question or give feedback.
– Unpublished paper, Randomization in Clinical Trial Studies. Shen & Lu.
– Basic & Clinical Pharmacology, 12th Ed. Bertram Katzung, Susan Masters, Anthony Trevor.