Modern Randomization Methodologies and Technologies
Prior to the adoption of newer, more advanced systems we use in clinical trials today, bias was a common theme. As noted in the article: “FDA and Clinical Drug Trials: A Short History” by Suzanne White Junod, Ph.D., this was often observed when physicians would treat sicker patients with known controls, while seemingly stronger patients received experimental products. Furthermore, when not blinded, knowledge of a patient’s treatment regimen tended to affect the level of care a health care worker would provide and also influenced their observations. With the introduction of modern methodologies, such as a randomization and blinding, sponsors have been able to gain much more control over their clinical trial studies. In this article, we’ll review some of the major milestones in technology and strategy that have helped pave the way for a more adequate, well-managed study.
Technology Brings New Benefits
By the early 1990s, biopharma companies and CROs began using technology more frequently and as a result, new methods of randomization were used to maintain the blind to avoid bias, minimize wasted medication and greatly improve the clinical supply chain. Today, these systems are generally referred to as Interactive Response Technology (IRT) systems – which utilize innovative technologies and user interfaces via PC, tablet, phone and similar devices.
An important advantage of IRT is a more efficient supply chain. Using a six-month clinical trial as an example, suppose each patient visits the site once per month of the study. Instead of sending six months of medication per patient to each site, only one or two visits worth per patient could be supplied initially. When a patient is randomized via the IRT, the investigative staff is told the pack number to give the patient for that visit. For subsequent visits, the system knows which visit packs at the site will be valid for the patient and again, tells the staff which pack to dispense. Thus, if a patient discontinues early, hardly any drug is wasted as the remaining packs at the site can still be used by other patients on the same regimen.
Another advantage of IRT randomization is the ability to maintain balance over the duration of a study to ensure accuracy. There are a variety of methods for doing so and these are often referred to as “minimization of imbalance” designs, or simply as “minimization.” The design examples mentioned so far have been very basic, but clinical trials generally involve stratification factors, such as gender, smoking status, age and other considerations. IRT is indispensable in ensuring balance across the study and across various stratification factors by intelligently managing the randomization scheme according to the design and needs of the study.
Minimization of Imbalance Methods
One well-established method for minimization of imbalance is called the “biased coin.” For example, when the first patient enrolls in a study, there is an equal chance of getting drug A or B, much like tossing a coin and coming up with heads or tails. If the first patient is assigned drug A and the second patient gets B, we have balance. But if both patients are assigned drug A, we would have an imbalance with two A’s and zero B’s. For the next randomization, the algorithm would weigh one side of the coin to give a higher probability that it will result in drug B, thereby minimizing the imbalance.
Of course, clinical trials rarely have such simple designs and usually have multiple treatment arms, varying treatment balance ratios, multiple stratification factors and other variables that must be considered. Even for highly complex designs, minimization methods can help to maintain balance. This is generally achieved by using formulas to calculate an imbalance score as each successive patient is enrolled and then weighting the outcome in favor of the treatment assignment that will result in the lowest imbalance score. There are many methods for calculating such a score and we will illustrate two common concepts. Consider the example in Table 1, where 15 patients have already been randomized and now the 16th arrives at the site – a male, non-smoker in the “low” age group (sub-group MNL for Male/ Non-smoker/Low age group).
Table 1. Fixed List at Patient Kit Level
The first method that could be used is to sum the number that would now be in the MNL group for treatment A and the same sum for treatment B. In this example, assigning the patient to A would result in a score of 6+6+4=16. Assignment to B would score 4+5+5=14. Since the score for assignment to treatment B is lower, we would weigh the coin in favor of that treatment group. Another method is to sum the differences within each stratification factor. In this example, if the patient is randomized to A, the differences total as follows: ABS(6- 2)+ABS(2-6)+ABS(4-4)=8. The differences for B would be: ABS(4-5)+ABS(4-5)+ABS(4-4)=3. Again, assignment to treatment B results in the lower imbalance score and we would weigh the coin in favor of that treatment group.
The imbalance calculations presented so far have been rather straightforward. Suppose, though, that maintaining balance in gender is far more important than smoking status or age group. Then we could apply a weighting or multiplier to that factor in our calculation to help ensure that we continue to have approximately equal numbers of males and females in each treatment group.
Simulation
As randomization methods grow more complex, how can we be certain that they will maintain balance and lead to the desired ratios in treatment groups? Biostatisticians typically run a number of simulations through the study design and then analyze the behavior of the randomization schema. These simulations are based on the expected patient characteristics while also introducing some randomness to account for real-world unknowns. Using our previous example from Table 1, a simulation run may begin randomly creating patients with a 50/50 probability that the next patient will be male or female, 30/70 probability for smoker/ non-smoker and 40/60 for low-age/high-age group. Further, some “what if” scenarios can be modeled to see what would happen if more or less randomness is introduced into the algorithms. Simulations give the statistical team insights into the likely outcomes and also give clinical teams strong evidence that the algorithms will lead to the desired results in terms of balance and treatment ratios.
Regardless of the randomization method used, it is vital to have a provider who has the tools to monitor how the study is performing once it goes live. While simulation results provide a solid reassurance that the randomization algorithm is set up according to the needs of the study, even the best simulation techniques cannot account for every real-world factor.
Adaptive Trial Designs
Sometimes studies utilize an adaptive trial design that allows for modification of specific components during the trial, which also require careful consideration and planning in relation to the randomization methods to avoid bias. Adaptive trial designs are utilized to ensure certain changes to the study will be made based on data analyzed during the trial with a focus on getting results more quickly and at times, involving fewer patients. Any and all potential changes are clearly identified in advance in the protocol and statistical plan. Further, the design of such potential changes is generally based on results from previous trials, research on similar trials, feedback from regulators on safety issues and other considerations.
One of the more common uses of an adaptive trial has been dose finding. For example, a study may start with several arms, perhaps at doses of 10 mg, 20 mg, 30 mg, 40 mg, 50 mg and a placebo. Initially, the IRT may randomly assign patients to one of these dosing arms with equal probability. Then over time, as safety and efficacy factors are examined, either in an automated fashion or by an unblinded team, the randomization schema is altered by either changing the ratio in favor of the doses with better outcomes or by dropping an arm altogether. In this example, if the 50 mg dose was causing a high level of adverse events, it might simply be dropped and the IRT would no longer randomize patients to that dose. On the other hand, if the 10 mg dose was producing no health benefit, it too could be dropped. By the end of the trial, the majority of patients will have been randomized to one of the optimal dose levels.
Staying with this simple example of an adaptive design, if the medication was in tablet form, would it make sense to manufacture five separate tablets for the active medication? Doing so could be very expensive and if we rule out use of the 10 mg and 50 mg early in the trial, all that medication would be wasted. An IRT vendor with experience in these study designs can collaborate with the sponsor and the drug supply vendor to make recommendations that could easily be built into the system to optimize the entire supply chain for this, and for future studies. In this case, perhaps the team would suggest that manufacturing tablets in 10 mg, 20 mg and 30 mg doses would suffice. Then each patient could be given two vials with the clear instructions to take one tablet from each vial when it is time to take their medication. Table 2 shows a simple design for this scenario.
Table 2. Permuted Blocks
The table also shows how the IRT system could manage alternatives if stock levels of a particular dose were running low. Of course, several other configurations are also possible and again, an experienced team can present the optimal set up and can readily implement it in the IRT.
Other Considerations
The unexpected is bound to happen once a study is launched into the real world, but a thoughtful design scheme will help to ensure randomization methods aren’t compromised. For example, on occasion, a site may encounter a stock-out issue, meaning that the treatment to which a patient has been randomized is currently not in stock at the site. When this occurs, it’s a good protocol to have the patient return to the site, but this is inconvenient for both parties. A better solution may be to allow “forcing,” which means that the system automatically puts this patient onto a treatment that is currently available. Then the system would assign a higher probability that the next patient in the study would be assigned to the treatment group that was just skipped due to the stock-out. While forcing is not technically random, it is permitted by regulators as long as it is not occurring frequently. An experienced vendor can provide the necessary guidance to help ensure the quality and integrity of the study design and its execution throughout the trial’s lifecycle.
On the topic of human expertise, there is a current trend toward rapid development of IRT systems and this is bringing many benefits to clinical research. Chief benefits include establishing better, more consistent standards as well as validated and reusable libraries and modules that reduce the time and burden in system testing and that lead to higher quality and reliability. However, there are many other factors involved in the design, deployment and conduct of a study and having a team of experts who understand the critical success factors, from manufacturing and packaging through accountability, returns and destruction, is essential.
Summing It Up
Randomization is a key element in the success of a trial. An effective randomization plan built into an IRT system forms the basis of an adequate and well-controlled trial that is free of bias, maintains the study blind and ensures that the desired balance and treatment allocation ratios are preserved throughout the trial. Modern systems and simulation techniques are in place that have also greatly reduced the timelines to implement an IRT system while also providing highquality, reusable algorithms with a greatly reduced validation effort. Finally, there is no replacement for a qualified and experienced team who can provide guidance on all the critical factors involved in running a successful trial.