[MUSIC] So now let's talk about some of the nuts and bolts of masking. Just by way of review, the reason that we mask treatment assignments is to reduce the bias in data collection and outcome assessment. It's also synonymous with blinding. What are the factors we should consider when we're deciding whether to mask the drugs in a clinical trial or not? Well, the first consideration is ethical. Is it ethical to mask the treatments? Is it ethical to expose people to the risk of no treatment if we were using a placebo or a sham surgery? There's also the potential for an increased risk if investigators and data collectors don't know what treatment participant has been assigned to in a double mask trial. And we also need to make sure that we have a viable plan to unmask the treatment assignments at the end of the trial or during the trial if need be. The first consideration is whether it's possible to mask. So drugs come in a number of different formulations and the idea of masking is to create two basically identical looking, feeling, smelling drugs for the participants to take. So is that going to be possible? Are there distinct characteristics that would prevent us from masking? For example, it's difficult to mask many of the nutrients that have been tested in clinical trials, like fish oil tends to have a bad taste and causes people to burp. I've done a trial of zinc versus placebo and that leaves an aftertaste. So there's a number of features that you have to deal with in masking and you have to think about the complexity of the treatment. Are dose adjustments going to be part of the protocol? Masking makes that more difficult. As well as the route of administration of treatment. An oral drug is easier to mask than an inhaled drug. A drug that is delivered by IV is very easy to mask. Is it feasible to mask? Do you have the resources to do it because it can be very costly. And is that cost in time and money worth it? You should also consider the participant burden. One of the reasons people don't like to enroll in clinical trials or a barrier to enrollment in clinical trials is masking. As you can imagine if you're telling someone, we're going to select your treatment by chance and then you and your doctor are not going to know your treatment, that is a big hurdle to get over. And you also have to make sure that the masking regimen is something people would adhere to. Because it may be more difficult to actually adhere to a treatment regimen in a clinical trial that involves masking than it would be for the normal patient. You have to think about the type of controls. Are you going to have a placebo control? And if you are going to have a placebo control, consider the importance of a placebo effect. Are you doing a comparative effectiveness trial where all of the agents are active agents. So that may decrease the need to mask in the sense that all of the participants will know they're receiving an active drug. Other issues that arise have to do with pharmacology. Does masking in any way interfere with the absorption of the drug or how it's metabolized or regulatory considerations? So I mentioned before that drugs can come in many formulations and I have listed some here on this slide to give you an idea of how complicated masking can be just because treatments come in a lot of different shapes and sizes. And this slide is just a picture of many different pills and injections and IVs and dermal patches and inhalers. So some considerations for masking are, how are you going to mask, where are you going to get the product and who was going to do the masking? How is it going to be packaged and labeled, and how will it be distributed both to sites and to participants? Decisions made about one aspect of masking can influence other aspects of masking. So, the first thing is how will you do it? As I mentioned before, you have to attempt to match all the physical characteristics of a drug, so it must taste the same smell the same, be the same weight. You have to remember that investigators and sometimes participants are going to be curious about what the treatment assignment is. So they may try to figure out which treatment is which and they can use any of those physical characteristics to help them figure it out. It's also important that the masking process in itself doesn't alter the activity of a drug. Common choices are over encapsulation. So one of the problems is that makes the pills larger. On the right I have an illustration of over encapsulation where you see how a small pill might be put in a gel cap and have some backfill in it. And then the control pill will just be the gel cap with the backfill in it. That's a fairly simple process. You can go to a more complex process of actually manufacturing placebos that look exactly like the drug that you're testing in your trial. That can be quite expensive and you have to be cognizant of all the regulatory issues and testing that will need to be done if you are manufacturing something for people to consume. Other strategies that we use are double dummy designs, which I'll show you an example of later on. Double dummy designs are used often in comparative effectiveness trials where you have- More than one active agent as an intervention and it may be difficult to make these two different drugs look exactly the same. The double dummy design increases complexity and also increases participant burden. You can use sham devices that are devices that are not active. For example, we did a trial of CPAP which is the machine that people with sleep apnea used to help them breathe evenly all through the night. And we use those machines to see if they would help asthmatics. And we had a sham machine that did not deliver the constant airflow, and you can have a sham controls or procedures as well. So one of the first things with masking is that you have to be able to label the drugs so that you know the difference between the treatment arms. And that needs to be communicated to both the site staff and the participant. So you could use a code or number for each treatment and make it very simple and just use A and B. A for placebo, B for active treatment. But there's a real problem with such a simple system, because once you unmask one person, either because they came to the end of the trial or they had some sort of adverse events, you've effectively unmasked the entire trial. A little bit more complicated system that protects against unmasking is to use multiple codes for each treatment, commonly referred to as a bin system. So in that case we might have three or four different codes for placebo, and three or four different codes for the active treatment. So then if we unmasked someone assigned to bin D, we've only unmasked people in bin D. It doesn't tell us anything about the contents of the other bins. The most robust way to label drugs to ensure that they are masked is to use a unique treatment ID for each drug container. So, a participant will be assigned to one or multiple drug kits that would have all of their supplies for the trial. Some of the considerations are that this is a more expensive way to package drug for a trial that you have to be thinking about resupply issues. So, if a participant is assigned to a particular drug and they're going to need refills during the trial, you have to have a system that ensures that you provide participants with the right drug throughout the follow up. Unique IDs are a strong protection against inadvertent unmasking. If you unmask one individual, it only tells you the treatment assignment of that particular individual. However, one problem I have noticed in my own career as we have moved from using bin systems to unique IDs, that it's easier to mix up the treatment assignment at the site to somehow pick up the wrong bottle and give that to the participant. So even though a unique ID system offers a lot of protection against inadvertent unmasking, it may increase the risk of other types of mistakes. So, let's think about some of the pros and cons of doing an unmasked trial. In an unmasked trial, an investigator may be able to provide better care to a participant and it is a better reflection of the real world. How the treatments will be used in clinical practice, allowing for dose adjustments and other concomitant treatments. It's really much simpler to conduct and it really decreases the complexity of providing study interventions to participants. However, it does not offer any protection against observer bias which can influence all of the things that we have talked about in the conduct of the trial. I mentioned earlier the idea of masking the outcome assessors. So, not masking the participant or the investigators, just the person who is assessing the outcome. This is easier to do in things where you might be using some kind of image to grade the outcome. We've used that in ophthalmology trials where we take pictures of the interior of the eye and those are sent to a grading center. However, you can have masked graders at each site evaluating patients. For example, in those same ophthalmology trials we typically mask the visual function examiner. So the person that is measuring visual acuity does not know the treatment assignment of the participant, and the participant is instructed not to discuss that with the technician. Single masking has similar advantages as we discussed for an unmasked trial and similar risk. Because you still have the risk of observer bias and you've only eliminated it on outcome assessment, all of the other factors involved in the conduct of the trial are still at risk of being influenced by observer bias. If we mask at the participant and clinic level, what we traditionally called double masking, we have the protections against bias in the conduct of the trial. And the disadvantage is that we have discussed, negative effects on recruitment, increased cost and complexity, and we have to ensure that we have appropriate procedures for unmasking. There's also masking of the interim analysis. So that is when you look at results as the trial progresses to ensure that you're still doing an ethical experiment, should the people who are looking at those results be masked. So, masking in data monitoring reports, masking statisticians or the people reviewing those reports. So, masking of analyst or reviewers has the purpose of keeping those people objective. So, they are not influenced in their decisions by the treatment assignment. The problem with that is that Masking Interim Analyses may reduce the competency of the data review because it's more difficult to balance the benefits against the risk of a treatment. This is especially true in situations where you have surprising results so that it would be harder for a reviewer of a mast report to think about. Well that increase in risk could be the placebo group. The tendency is to assume that any side effects are going to be from active agents and any good effects are going to be from active agents. And it's hard to construct the different scenarios that could lead to a specific set of results. And masking the interim analyses and the reports just adds another layer of complexity, and you have to have a means to unmask the reviewers, if there are concerns about the benefit risk ratio. So if you mask a trial, you also need to have a plan to unmask participants so they know what treatment assignment they received during the trial. And that strategy for unmasking depends somewhat on the clothes out design of the trial. If you have an anniversary close out where everybody is followed for a specific length of time, for example, one year, but you enroll in the trial over a couple of years. So if you have an anniversary close out, that means that each participant, regardless of when they are enrolled, are followed for a specific length of time, after enrollment, let's say one year. So as participants complete the trial, you need to let them know what treatment they were on. But you also at the same time need to maintain the masking of the clinical staff who were still conducting the experiment. That's commonly done with unmasking envelopes or messages to the participants. And participants are instructed not to discuss their treatment assignment with the clinical staff. And often you also need a letter to the participants physician so they know what their patient had been treated with. In a common close out design, we may enroll people over 2 or 3 year period, but everyone gets followed to a certain calendar date. So a person who is enrolled earlier in the trial may be followed quite a bit longer than someone who is enrolled later in the trial. However, the masking is easier to manage because you don't have to be concerned about unmasking the clinical staff. When you unmask a participant, you also need to provide them with instructions on continuing the treatment or what treatment they should go on after being in this trial. You need to collect data on what the participant and the clinical site think we're the treatment assignments to see how well the masking worked. And you also need to document the whole process to ensure that you have provided everyone with the necessary information. You also have to plan for unplanned unmasking, when you need to know the treatment assignment right away. So that may be an emergency situation where a child has swallowed a lot of the pills or a participant has inadvertently taken too much of the study drug. So you need to know right away whether that was placebo or an active agent. Often we think that you need to unmask participant when they have an adverse event or they stopped the treatment because it hasn't been effective. But you really usually don't need to unmask in those situations, you just need to stop the treatment, and the adverse event will likely subside. Sometimes it's a little bit more complicated if the treatment assigned in the study will influence the next steps and treatment. So the key is to make it easy to unmask the treatment assignment, but not too easy. So it could be something that's on the study drug container, now, that would be very easy, if you had a label that you could rip off. You can have someone at the clinic who is a semi independent person responsible for unmasking and have unmasking envelopes, or you could have a 24 hour call line or website. Again, this would go back to really what the active agents are and what are the risk profiles of those agents to determine what steps would need to be taken to unmask a participant. And as I said before, you need to document everything and ensure that everyone who needs to know about the treatment assignment is provided with that information. [MUSIC]
