Hello, my name is Florencia Noll. I'm a gynecologist-oncologist from Argentina, and today we are going to discuss controlled clinical trials. This is our agenda, so let's start with some basics. There are two types of research studies. We have the observational studies, where the goal is to observe and collect data without influencing the participant's environment or disease scores, and interventional or experimental studies, where the researcher deliberately influences the course of the event and investigates the effects of that intervention on a carefully selected population of patients. In the observational studies, we have different designs, cross-sectional case control, retrospective cohorts, and prospective cohorts. The interventional or experimental studies are a specific type of cohort in which the investigator assigns the exposure to one of the groups. It is always prospective, and it allows to infer causality. So what are we learning about randomized controlled trials? Well, because these trials are the gold standard in evidence-based medicine and are the highest possible level of evidence in our field. So, why randomized? Because we randomly assign patients to one group or the other. Controlled because we are going to perform interventions to one of the groups, so we are going to have a control group and a treatment group. And clinical trial because this experimental study aims to establish the superiority, equivalence, or non-inferiority of our intervention. So, at the very beginning, we should specify the kind of hypothesis we are going to be testing. So, we have superiority studies aiming to demonstrate that treatment A is superior to treatment B. We have equivalent studies that aim to assess the therapeutic equivalence between drug A and drug B. After choosing an interval of clinical equivalence, the therapeutic equivalence between two drugs, for example, can be affirmed if the observed effect of A versus B fall in the same equivalent margin. And then the non-inferiority trials, where we aim to show that the effect of drug A is not inferior to drug B. These kinds of trials are different from the equivalence trials because we are not trying to show that drugs are equivalent, but we need to demonstrate that the new treatment is not unacceptably inferior to the active control. The structure of our trial depends on the question. One strategy is the PICO question, where we include always the population or disease we are going to be testing, the intervention or variable of interest we are going to perform, the comparison groups, the outcome, and the time. So, how good is the primary research question? Well, the question to answer that question is, at the end of the day, when the clinical research is completed and the data analyzed, will the answer, whatever it is, to the primary research question advance scientific knowledge or our clinical practice? Well, once we have our question, these are the ingredients of the clinical trial. First, we're going to start with the pre-randomization stage. Patient recruitment source is one of the most important steps before we start the recruitment, because we need to choose which centers are we going to be including or inviting to our trial. So, we need adequate volume of patients in the expected time. Also, we need resources and enthusiastic researchers, obviously. The sample selection is not randomized, but we should state clear the inclusion criteria and exclusion criteria. What type of intervention we are going to perform to our treatment group, and what is the intervention we are going to offer to our control group? So, finally, we have all the individuals with a disease or condition, then those who are interested in participating in our trial, those who met the inclusion and exclusion criteria, and those who consent to participate. Once we complete this sample, we can start with the randomization. The randomization is the process of assigning participants to treatment and control groups, assuming that each participant has an equal chance of being assigned to any group. So, both branches should be comparable and interchangeable. So, how to randomize? We have different strategies. The first one is the simple randomization. This is based on a single sequence of random assignment. The basic method is flipping a coin, and it works for large trials, but when we have small sample size clinical trial, we have to choose a different and better method of randomization. An alternative is the block randomization. It is used to ensure the balance in sample size across groups over the time. We are creating blocks of patients, small and balanced, and then we are going to divide them into the treatment and control group, the whole block. Each block could be four patients or six patients. Then, we have the stratified randomization. This balances not only the number, but also the co-variates. This is really important, and this randomization is achieved by generating a separate block of each combination of co-variates. The limitation of this strategy is that it's complicated to implement if we need to control several co-variates, and it only works if we know all our baseline characteristics of the patients we included. Finally, we have the last one, the co-variate adaptive randomization. This is a specific way of randomization, and it works when we are going to be including patients in a continuous way. We don't have the whole sample size that we are going to be including over the time, so each new participant is sequentially assigned to a particular treatment group by taking into account the specific co-variates and the previous assignment of participants. The limitation of this kind of randomization is that treatment assignments sometimes become highly predictable, so this is going against our basic concept of randomization. For this strategy, we have a software online, a free user-friendly computer program, so we can use it to randomize our sample. So, several factors need to be considered. This is the flowchart for selecting the appropriate randomization technique. First, the sample size. With a large sample size, simple randomization is okay. With a small sample size, we need to ask, do we have co-variates that need to be controlled? If the answer is no, then we can go with the block randomization, and if we need to control our co-variates, then we need to change and to use a different way of randomization, and we can choose if we are going to include patients. All at the same time, they stratify randomization, and if we are going to include patients in a continuous way, we are going to use the co-variate adaptive randomization. The allocation concealment is the process by which investigators and participants enrolled in a clinical study are kept unaware of the upcoming group assignment until the moment of the assignment. It's not blinding. They don't know in which group they are going to be until the start of the assignment. Then, we have the post-randomization. The post-randomization stage starts by defining the outcomes. So, our outcome should be patient-centered, has quality of life or survival, measurable and evaluable in the same way, and it should occur frequently enough for the study to be adequately powered. We have different kinds of outcomes. The primary outcome is the most relevant to answer our question, our trial question. The secondary outcome, there are additional outcomes monitored to help interpret the primary outcome. The surrogate outcomes, these are biomarkers intended to substitute our clinical outcome, are continuous variables, so they occur earlier in the clinical trial, so it reduces cost and study duration and size. Then, we have the composite outcomes that are made of multiple variables, and the advantage is that we increase the power of the study when each event is rare. It's important that the study outcomes should be stated a priori before the researcher looks at the results. The sample size calculation should be carried out to detect the clinical relevant effect of intervention on our primary outcome, and the choice of the most suitable outcome should be based on the research question. Then, blinding. This is the process by which information that has the potential to influence the study results is with help from one or more participants involved in the research study. Why blinding? Because the blinding eliminates the possible computers due to co-intervention and to prevent bias in adjudicating the event to one group or the other. Classically, we have single blind where the participants are blinded, double blind where the blinding is to the participants and the researcher, and the triple blind where all are blind, participants, researchers, and data analysts, but we should explicitly report blinding status. We need to say who is going to be blind and who is not going to be blind and what information is consumed and how blinding is performed. About the follow-up, once the study is started and patients are enrolled, everything can happen. Patients can end up in their treatment arms, then patients can cross over to another treatment arm, drop out of the treatment, or lost on follow-up. And finally, the analysis plan. This is how we are going to interpret our results. We have two main complementary strategies we can use to analyze our data, the intention to treat analysis and the peer protocol analysis. The intention to treat analysis is to assess the average effect of assigning a drug to a group of patients, for example. When a participant violates the protocol, for example, early in the study, they are considered as belonging to the treatment arm they were originally randomized to, so that it maintains the effect of randomization and avoids the selection bias resulting from treatment allocation based on patient prognosis or respective drug response. The disadvantage of this strategy is that it is difficult to demonstrate the adverse effect of an intervention and it favors the equivalence of the results. The peer protocol subpopulation consists of a subgroup of patients of the intention to treat who completed the trial without significant deviation from the protocol requirements. It captured the true exposure of the treatment versus placebo. Other parameters to analyze our data are the risk difference. This is the difference in risk of a condition such as a disease between an exposed group and unexposed group. Then we have the number needed to treat. Sorry, this is the number of patients who need to treat to prevent one additional bad outcome is the inverse of the risk difference and the relative risk. This is a measure of the risk of a certain event happening in one group compared to the risk of the same event happening in another group. If we have a relative risk of one, there is no difference between two groups in terms of the risk of cancer, for example. If we have under one or less than one, the group being exposed to certain substance or factor decreased the risk of developing the disease. With a relative risk of more than one, the group being exposed to certain substance or factor increased the risk of developing some disease. But what happens when we are readers, which are the critical reading tools? Well, first, we need to note what outcome specific information should be included and published in a clinical trial report. Everything should be transparent. So, every clinical trial should include the consort statement. This flowchart is a harmonized evidence and consensus-based standard for reporting outcomes in every clinical trial. Also, we have the checklist, the consort checklist, where we can check if every point of the checklist is being covered in the clinical trial. From the title, abstract introduction methods, to the way of organization, the way of blinding, and also if the authors register their protocol. So, the registration of a clinical trial is a must. We should register our trials in clinicaltrial.gov, also in the World Health Organization webpage. The importance is that by registry, we fulfill the ethical obligations to participants and research community. We provide information to potential participants and referring clinicians. We reduce the publication bias, and we promote more efficient allocation of the research funds. So, it's really important to register our trials always. So, summarizing what are the advantages or reasons for conducting a clinical study. First, we know, as we mentioned at the beginning, this is the gold standard for producing primary evidence. The randomization reduces the risk of bias, and we can measure the impact of one intervention in multiple events simultaneously. The disadvantage of this kind of trials is that we need resources, we need a large sample of patients, and a long follow-up. So, it's easy to write the study, we'll use a randomized double blind control parallel arm design, and intended to treat analysis, but it's not easy to implement and maintain the integrity of randomization, blinding, multiple study arms, data collection, and transfer data to regulatory or other groups. So, it's important to identify the question you're trying to answer, the design and the outcome of your clinical trial, to recognize and avoid the possible bias, and always look for mentors and for teamwork, because this kind of trials requires time and effort. Thank you so much.

Florencia Noll

gynecologist-oncologist

controlled clinical trials

observational studies

interventional studies