This is lesson 2.1.3, Blinded Designs. In the previous lesson, we discussed the importance of randomization in clinical trials. In this lesson, we'll discuss blinded designs and the importance of the trial design in order to ensure that there's a minimization of bias in the outcomes. And the results are interpretable and usable for technology assessment. Blinding is an important element of trial design. It's the masking the identity of the assigned interventions. The main goal is to avoid potential bias caused by conscious or subconscious factors. Now, there's three types of blinding that can be applied in a trial design. The first is single blind. That's when the patient is blinded to the intervention. So they don't know whether they received the intervention or the non-intervention. However, both the physician treating the patient or administering the intervention as well as the researchers do know which patients received which. The second type of blinding is double blind. Here, both the patient and the assessing investigator are blinded. So, this would mean that the patient and the physician providing the treatment are blinded. And they don't know, both the patient and the doctor don't know whether the patient's receiving the treatment or intervention or if they're receiving the non-treatment protocol. The third type is triple blind. Here the committee monitoring the response variables or the statistician is also blinded. So this means that the patient doesn't know whether they receive the intervention or the control. The treating physician or investigator doesn't know and the statistician doesn't know. Which means that any type of analysis or statistical evaluation of the results are dependent only on the results and not on what anyone thinks might be going on. It's important that patients should be blinded. Patients, if they know if they are receiving a new or experimental intervention, may report more or less side effects. And this may be intentional or subconsciously. Patients not on the newer experimental treatment may also be more or less likely to drop out of the study, depending on what the treatment is. Then the study would be biased, because one group would no longer be the same as the other group. Also, a patient may have a preconceived notion about the benefits of the therapy, or not receiving the therapy, which can bias the results. And finally, there are instances of patients trying to get well or please physicians because they think they're receiving the treatment. They think they should be getting better results and that's not necessarily the case. It's also important to consider blinding the investigators when possible. The treating physicians and investigators are often the same people. The possibility of an unconscious bias in assessing the outcome is difficult to rule out. If the physicians or investigators expect an intervention to have an effect, they may subconsciously impose that supposition on the patients they're treating. Decisions about compensatory treatment are often made by someone who knows the treatment assignment. So there also may be treatment that's more likely given to patients on a protocol arm that's perceived to be less effective. In other words, because some physicians know which intervention is being given to which patients, they may tend to compensate and treat different patients in different groups differently. Which then results in non-random assignment to groups, because the groups are no longer equal in the treatments they're receiving except the intervention. Both of these are sources of bias that can affect the final results and how the intervention is evaluated. So, how to blind a trial is by identifying a placebo treatment or intervention that can be used to blind the patient and/or the investigator and statistician. So for example, if it's a pill, it's easy to think of a placebo treatment which would be a pill of the same size, color and shape that's simply a sugar pill that's given at the same intervals as the intervention. It's a little more difficult to identify a placebo operation or device, but it is possible. In this case, you could blind the patient, but not the treating physician. But a sham operation, which still involves anesthesia for the patient and an incision, where the incision would occur for the same group that's receiving the treatment, would occur for the patients. And so when patients wake up from surgery, every patient, whether they're in the intervention group or not, would believe that they received the operation. Note, there are ethical considerations to placebo treatments or blinded studies, and this goes back to the discussion we had in previous lessons about trial design. It's not always possible to blind a trial, just like it's not always possible to create randomly assigned groups. If blinding a trial is not possible, it's important to standardize the treatment groups as much as possible and use objective reliable outcomes. So, we may not want to use a survey following the non-blinded study where patients discuss, or answer questions about how they felt that the treatment worked unless we're sure that that's a valid, reliable survey instrument. We can also consider duplicate assessment. In other words, we would have multiple investigators assessing the outcomes in order to prevent, or at least minimize, the amount of any intuition or influence that one investigator may have about what they believe should be happening. It's also important to acknowledge limitations of any randomized control trial, but it's especially important when blinding is not possible.
