Table of Contents
What Is RCT?
RCT stands for Randomised Control Trial. The results of RCT have gone to the extent of changing the practices in the clinical setting. One of the notable examples to be quoted is the decline in the sales of flecainide and encainide (antiarrhythmics) after the RCT in NEJM on these drugs reported an increase in mortality (opposed to the previous non-randomised studies).
A second notable example is the decline in the prescription of postmenopausal hormone replacement therapy to women for prevention of myocardial infarction. The RCT of HRT in JAMA showed that HRT is associated with increase in myocardial infarction risk.
RCT – advantages and disadvantages
Advantages of RCT
RCT is among the top of the hierarchical pyramid of evidences (though below the meta-analysis and systematic review). The hierarchical pyramid of evidences represents the hierarchical pattern of importance of study and how close the result of the study represents the true effect. The study types which are present in the bottom of the pyramid represent the more bias susceptible studies (for example observational studies.)
Disadvantages of RCT
RCT are very expensive and take a long time to be performed. Secondly, the RCT takes time to be published and some of the findings described in the study might not be relevant at the time of publication.
The variables which were evaluated in the RCT are limited (commonly referred to as the primary and secondary outcomes) and might not represent the complete clinical picture of the patient.
RCT are performed in the most controlled environment and the validity of the same results in the external real life scenario remain a significant debatable question. This is commonly referred to as the external validity of the RCT.
There remains a significant problem of conflict of interest while conducting an RCT. Nowadays, most of the research journals and regulatory bodies like FDA have asked the researcher/principal investigator and the other persons involved in the study to fully disclose the competing conflict of interest, still it remains debatable whether it is fully authenticated.
The concept of withholding the medication to the control group or giving the new medication/intervention to the treatment group (with its possible dangerous risk) remains an ethical question, which is commonly discussed. The RCT requires a proper ethics committee approval of the local governing body to be performed.
Difference Between Pragmatic and Explanatory Trials
Pragmatic trials are the trials which are carried out in order to evaluate the effectiveness of an intervention in the day-to-day real life, whereas the explanatory trials are done under routine stringent condition (the common way in which RCT are performed.)
As already discussed in the disadvantages, there occurs a difficulty in generalisation of the findings of RCT. This is the reason why the pragmatic trials are gaining momentum in the current medical practice.
This is referred to as the efficacy vs effectiveness studies: the experimental trials represent the efficacy and the pragmatic trials represent the effectiveness of the intervention of interest.
There are generally two groups in randomised control trial. One is the treatment or the intervention group and another is a control group. The treatment/intervention group receives the drug or intervention of interest which is intended to be studied in the RCT, whereas the control group does not receive any drug but in the place of drug they receive something known as placebo.
Placebo are the medications which do not have any effect but they match to that of the intervention drug in terms of texture, colour, smell and even in the route of administration. It is also recommended to be given in a similar treatment regimen, in order to avoid the confounding effect. The treatment or intervention group might be a single group or it can be many groups.
The different groups might be tested with number of different medications or different dosing and treatment regimens of the same medications. In addition to using placebo as the control, in some of the studies, they also use the positive control.
The positive control is the usage of a drug which is already established as a treatment and efficacious against the disease of interest. So in this scenario, the research hypothesis is to show that the experimental drug is better than that of the positive control.
The negative controls are not used in human studies, as it obviously would be producing the deleterious effect (it would be against the ethics). It is worth knowing that negative control is used in preclinical experimental studies in animals.
Blinding commonly refers to the concealment of the allocation with regards to the group of individuals who are involved in RCT.
Blinding may involve manifold of individuals. It might either be limited to the persons who are participating in the study (patients or normal volunteers), or the person who is conducting the study, or the person who is evaluating the outcomes of the study (statistician or the person who is not directly involved with the conduct of the study). The blinding employed can also be in a combination of two or all the three groups mentioned above.
The old terms of single blind, double blind or triple blind is not currently recommended to be used following the CONSORT guidelines for performing RCT. The term open or unblinded is used when no blinding is performed in the study.
As the number of groups who are blinded increases in a study, the risk of bias decreases and the authentication of the study also increases.
Different Methods of Randomisation
The aim is to provide the adequate sample size for the subgroup analysis, which would in turn increase and maximum statistical power to completely eliminate the selection bias, to keep it to the bar minimum and to nullify the effect of covariance by means of allocating equally to both groups, so that the covariate effect will get nullified.
Simple randomisation refers to allocating the participants to both groups randomly. This is the preferred method but there is a risk of unevenness between both groups. This chance of unevenness in number between the groups is overcome by means of employing restricted randomisation, which in turn can be divided into two types namely block randomisation and adaptive biased coin randomisation. The permuted block randomisation is relatively more common and similar to stratified randomisation.
Classification of RCT
RCT can be classified based on the study design, based on the hypothesis or based on the pragmatism of the study.
Classification based on study design
The categories are parallel group, crossover, cluster and factorial.
As the name suggests, in parallel design, when we allocate participants randomly, on the basis of randomisation the participant might be allocated to one group and all the members within the group receive one of the studied medication. In the crossover design, the participants would be changed to the opposite group (after the washout period of the first drug). So every participant would receive both the interventions but in a different time and at a random sequence.
In some of the cases, it would be that there exists inherent clustering within the studied participants. This would be utilised in the case of cluster study design, where all the persons within the cluster are assigned to a particular intervention. Over here, the clusters are selected randomly rather than the participant.
In factorial design, the combination of interventions is studied and the groups within the study receives different combinations of interventions which in turn is evaluated.
Difference between the superiority noninferiority and equivalence trial
In most of RCT, the hypothesis is to prove that the intervention which is being studied is better than the control or the positive control which is used in the study.
In addition to the superior difference, the researcher can also show that the intervention is equivalent to that of the positive control (this is referred to as the equivalence trial) or that there is a margin within which the intervention studied can be compared with that of the positive control. This margin is referred to as the non-inferiority margins within which the outcome of interest should be present.
Experimental studies are also known as interventional studies and it is intended to compare the effect of treatment with that of the control in humans. In accordance to the drug discovery, for a new drug to be approved, it has to pass through the pre-clinical stages, followed by four phases of study.
Phase 1 is conducted in healthy human volunteers and the primary aim is to determine whether the drug is safe and also to establish the maximum safe dose along with pharmacokinetic and pharmacodynamics measurements.
Phase 2 is conducted in a relatively small number of patients and is intended to determine the effect of the treatment with the drug.
Phase 3 represents a more elaborate study, in which around 1,000 patients are involved and the new drug is compared with the standard drug.
Phase 4 represents the post marketing surveillance studies, where the drug is checked for any side effects after being approved.
Other types of experimental studies include self-controlled studies where the patients are evaluated as their own controls over a period of time, the crossover study (already described), observational studies (in turn can be divided into prospective or retrospective study).
Prospective represents observation of data from the time of starting of the study, whereas the retrospective represents the determination of the data from the records which have already been taken from the patient.
Ecological studies typically represent the study on the factors which modify the risk on a defined population. The population might be defined based on the geographic location or on a temporal basis.
The results of ecological studies have typically helped in finding many of the risk factors. Some of the notable examples are the cholera study, the link between UV and cancer, the link between diet and Alzheimer and the link between diet and cancer.
Though it has its own advantages, there exists a component of “ecological fallacy” (the findings might be present in the group but they do not represent the individual person’s data.)
Concealment of allocation
It is highly probable that if the investigator would know the groups to which the patients are allocated, he would be favouring the outcome of interest. This is avoided by means of “allocation concealment” at all the times of the conduct of the study. This could be done by means of a central monitoring committee, using sealed and opaque envelopes, central randomisation processes etc.
Intention to treat and per protocol analysis
The essential aim of the randomisation is to eliminate the basement confounding factors which are known or unknown in both the groups but it is very much possible that some of the participants of the study in both groups might not complete the entire study and could have deviated from the protocol.
There are two options in front of us in this scenario: these participants could be eliminated from the analysis and the remaining participants might be taken to evaluate the result. This is known as per protocol analysis but the significant disadvantage with this is that we are essentially eliminating the benefit of the randomisation, which would have taken care about the baseline confounding variables. Secondly, even in the real life, it is very much possible that the patient might not complete the entire treatment. So eliminating might not be the complete solution.
An alternative method is the intention to treat analysis, where after randomisation all the participants in the treatment and the control group, whether they completed the study or not, are taken for analysis and the results are evaluated. This represents a more realistic real life scenario but has its own disadvantage that even the persons, who have not completely taken the drug are included in the analysis leading to some misleading result in some scenarios.
It’s a common scenario in RCT to report the results of both the intention to treat and per protocol analysis, so that the reader can decide themselves which to infer.