The concept of a randomized controlled trial
Randomized controlled trial (RCT) is an experiment in which subjects are randomized into groups, commonly referred to as intervention and control groups, in order for them to undergo or have not undergone experimental preventive or therapeutic interventions.
The results are evaluated by comparing the incidence of the disease, death, recovery, or other relevant outcome in the study and control groups. A randomized controlled trial is usually regarded as the most scientifically rigorous hypothesis testing method available in epidemiology. Some authors refer to this method as a randomized control trial.
The confidentiality of the results of randomization is combined with the disguised provision of placebo controls that are indistinguishable from the intervention being investigated, thereby achieving blindness of study subjects and personnel (double-blind trial).
A distinctive feature of modern RCT is strict adherence to ethical standards. Before starting any RCT, its protocol should be approved by a national or regional ethical committee (often both). In the future, the ethics committee is necessarily informed of the serious side effects that have emerged during the RCT and all changes in the protocol. Prior to inclusion in the study, its potential participant should sign, on a voluntary basis, informed consent, where the objective of the study, possible complications or inconveniences, the benefits associated with the participation of the patient in the study, and alternative therapies should be stated in an accessible form. The patient should be informed that the decision on his participation or non-participation in this clinical study at any stage of implementation will not affect his further tactics, and he can at any time stop participating in the RCT. Only after receiving informed consent the patient can participate in the study. In a number of cases, such as the conduct of RCTs in children, informed consent is signed by the parents or guardians.
Types of control
Four types of control are used:
- control of the initial state;
- Placebo-controlled;
- active control;
- control over archive statistics.
Baseline control is used in one form or another for all clinical trials. Clinical measurements before the start of treatment (baseline measurements) are performed in each subject before treatment for the purpose of obtaining the initial data, which will then be compared with the results after the end of treatment. The baseline can be assessed either with a non-drug-free treatment period, or, preferably, taking into account the placebo treatment period before randomization.
The placebo control method, known as negative control, consists in assigning a placebo to the subject, an inactive substance that can not be distinguished from the experimental drug by any indication (appearance, taste, smell).
In active (positive) control, treatment with the use of a drug that is effective relative to the test indication (active drug control) is included. As with the placebo, the active drug control does not differ from the study drug.
Control over archival statistics, or historical control, allows you to compare the experimental treatment with existing data on the outcomes of a particular disease. For many years of observation, the method of control over archival statistics is used when there is no other effective method of treating a known pathology or a rare disease.
Research models
The following typical models of clinical trials are distinguished:
- research in one group (single group study);
- Studies in parallel groups (parallel group study);
- Studies in the “crossover group study” groups.
In the study in one group, all subjects receive the same experimental treatment. Instead of comparing the results within the same group with the results of the control group, this research model is aimed at comparing the results of treatment for each subject with its baseline state before treatment or, if possible, with the results of control over archival statistics. Thus, the subjects are not randomized to treatment groups, and there is no need to “mask” the drug. A single-group model can be used in phase I of clinical trials in which healthy volunteers participate, but are usually not used in phase III trials. Nevertheless, such models are useful for carrying out some experimental courses of treatment.
When conducting studies in parallel groups, subjects in two or more groups receive different courses of treatment or different doses of drugs. To achieve statistical reliability, subjects are randomly assigned to groups. Models of research in parallel groups are considered optimal for determining the effects of treatment and formulating conclusions based on the results. In most cases, clinical trials are conducted in parallel groups. Supervisory authorities most often prefer this model of drug research, which is mainly used in Phase III trials.
In certain situations of clinical testing in parallel groups, two types of simple studies can be used:
- Factorial models;
- inhomogeneous models.
Models are implemented in parallel groups, and the data obtained are justified by specific methods of comparison. The factorial model is a model of studies in which several parallel groups participate. Studies on this model are useful when it is necessary to study the effect of a combination of different drugs and develop treatment regimens.
A non-uniform (interruptible) “withdrawal (discontinuation) design” model is a variant of studies in parallel groups where all subjects first receive the treatment being studied; then to continue the study treatment, patients with appropriate reactions are randomized into groups using blinded study with double control or placebo. This model is usually used to assess the effectiveness of experimental treatment by stopping the drug immediately after the appearance of the reaction and recording the recurrence or remission.
В отличие от исследований в параллельных группах, перекрестные модели позволяют оценить эффекты как изучаемых ЛС, так и сравнительных курсов лечения на одних и тех же испытуемых. Испытуемых рандомизируют в группы, в которых проводят одинаковое курсовое лечение, но с различной последовательностью. Как правило, между курсами необходим ликвидационный период, чтобы пациенты вернулись к исходным показателям, а также чтобы исключить нежелательное влияние остаточных явлений предшествующего лечения на эффекты последующего.
Stages of RCTs
Goals and objectives of the study
Planning of RCT begins with the formulation of the purpose and objectives of the study, which differ in their novelty and practical significance. The aim of the research should differ both in scientific novelty and in practical significance. For example, just such a study was SYST-EUR, the results of which largely rehabilitated calcium antagonists in the treatment of arterial hypertension, when they were attacked by a wave of criticism based on the results of an incorrectly meta-analysis.
Therefore, as a rule, large multicenter RCTs are planned by well-known specialists in their field. Each new RCT is original in design and the population of patients participating in it (determined by inclusion and exclusion criteria), therefore existing protocols can not be fully reproduced, although they can serve as a base for adaptation. As an example of a solution to a similar problem, two very different ones can be cited, both in design and in the results of the ALLHAT and ASCOT research protocols. The precise wording of the main goal is crucial. For example, an ELITE I study comparing the efficacy of the captopril ACE inhibitor and angiotensin II receptor antagonist losartan in patients with chronic heart failure can be cited as an example. There were no significant differences between the two regimens of therapy, but the aim of the study was to “prove the benefits of losartan” and it was not achieved. However, a new class of drugs, which turned out to be no worse than the reference ACE inhibitors, became an important event for cardiologists. If the aim of the study was to prove the equivalent efficacy of losartan and captopril in chronic heart failure, no further controversy simply arose.
Problems solved in RCTs should not be numerous, as this can lead to false-positive results and make interpretation of the data obtained difficult for practice.
Choice of patients
The second most important section of the RCT (after formulating the goal and objectives of the study) is to determine the criteria for inclusion and exclusion from the study. The inclusion criteria are predetermined by the purpose of the study, the broader ones facilitate the recruitment of patients and allow extrapolating the results to a large population of patients. However, in this case, there is a danger of forming heterogeneous groups of patients, both in terms of initial clinical and demographic indicators, and the effectiveness of the intervention being tested. Typically, RCTs include patients with moderate severity of the disease, although patients with an initially higher risk of complications can, at shorter times, assess the effect of intervention on hard endpoints. A group of patients with a low risk and mild course of the disease requires long follow-up and is potentially dangerous in obtaining a statistically significant intervention effect, whereas in fact it can be effective.
The exclusion criteria should minimize the possibility of errors during the study (for example, excluding patients with terminal stage of the disease, reduced liver and kidney function). If this contingent of patients is not the subject of a RCT, it usually does not include minors, women during pregnancy and lactation or who do not use contraception, oncological and psychiatric patients. An important task of exclusion criteria is to ensure the same treatment and management of patients in the study and control groups, as well as the exclusion of the use of drugs similar to the subject in the study.
In general, the larger the RCT and the faster it is planned to finish, the inclusion criteria should be broader and the less the exclusion criteria.
Randomization
The reliability of these RCTs depends directly on the comparability of the groups being compared. Categorically it is impossible to compare groups, one of which included patients with the analyzed intervention, and the other – those who refused to participate in RCTs and received “traditional” therapy. Also, the results of the new treatment method can not be compared in different clinics if they did not evaluate it according to one general protocol (differences in technical equipment, personnel qualifications and accepted treatment standards). The “historical control” method has similar drawbacks.
Randomization is a key issue in conducting RCTs. It should ensure random distribution of patients, regardless of the physician’s desire or any other factors, and comparability of the compared groups according to the clinical and demographic characteristics of patients, the severity of the underlying disease under study, the concomitant pathology and the therapy.
With a small number of patients in groups, even a correctly conducted randomization can not ensure their homogeneity. In this case, a preliminary stratification is used (stratum – layer, layer), in which the distribution of intervention options occurs in more homogeneous groups of patients, initially formed by one or more important characteristics. Differences between patients for these key characteristics should be minimal. This approach practically guarantees the receipt of reliable results and the absence of a systematic error. The absence of a systematic error in a RCT is called validity.
The most common reason for the incompatibility of the compared groups after randomization may be the inclusion in the analysis of not all patients participating in the study.
There is a large number of methods of randomization: the method of opaque sealed and sequentially numbered envelopes, the computer method (the procedure for randomization based on the random number generation method is performed by a specialist who is not directly involved in the treatment), the method of numbered identical containers prepared by a pharmaceutical firm (code and the true contents of containers are known neither to patients nor to physicians participating in the study), centralized randomization by telephone (IVRS) firma firma. The most objective is the last one, and the least objective is the envelope method.
At present, there are several reasons why randomisation is not used in scientific research:
- ignorance by doctors-researchers of the basics of evidence-based medicine;
- blind adherence to traditional practice and fear of obtaining results that are different from the generally accepted and opinion of authorities;
- Fear of using a less effective method of intervention in the patient, but it is for the sake of revealing its real effectiveness that RCTs are conducted;
- existing negative public opinion that identifies RCTs with trials in humans.
Analysis of interpretation and publication of results
The main task of statistical analysis of RCT is to establish the difference and the degree of its reliability by the outcomes (end points) between the group with the intervention being analyzed and the control group. At present, there are many packages of static programs for analyzing the results (BMDP, SOLO, Statistica and others). To obtain objective information about the effectiveness of the intervention in the analysis, it is necessary to include all the initially randomized patients (intention-to-treat analysis), and not just those treated strictly in accordance with the protocol of the study (on protocol analysis). This is one of the main ways to minimize possible errors, when the intention-to-treat analysis is based on the assumption that all patients received treatment prescribed during randomization.
The results of the study are presented using the following indicators: the number of patients who need to be treated in order to prevent one undesirable outcome of the disease, an absolute / relative risk of an undesirable outcome and their decrease, indicating statistical reliability.
Often the article with the results of the RCT precedes the publication of its most significant results and conclusions in the materials of major international conferences and congresses. The structure of the article is of a usual nature. A mandatory requirement is the widest possible representation of the design and statistical aspects of RCTs. Of particular interest is the discussion of the results, since they can provide additional information on the controversial points in its conduct.
Ethical and methodological standards have been developed for conducting the RCT. It should be remembered that placebo-controlled studies are justified when there is no alternative method of treatment, the effectiveness of which is beyond doubt.
Conclusion
In this paper, I reviewed the concept of a randomized controlled trial, the types of control and model of the study, the features, the steps in the RCT, and the methodology for conducting it.
Today, doctors, researchers and research organizers need to continuously improve not only professionalism, but also the level of knowledge in the field of evidence-based medicine, jurisprudence and GCP. At the same time, the conduct of RCTs is not only a science, but also a special (and rather profitable) medical business, which puts even higher ethical demands on all its participants.
