Downloaded 77 times
Uploaded byTerry Shaneyfelt
Blinding in Clinical Trials
Blinding in clinical trials refers to keeping participants and investigators unaware of treatment group assignments after randomization. This helps reduce performance bias and ascertainment bias. Blinding is accomplished through placebos, sham interventions, and coding of group assignments. It is important for subjective outcomes but not always possible, and sometimes must be purposefully broken.
More Related Content
More from Terry Shaneyfelt
Blinding in Clinical Trials
- 1. Blinding in ClinicalTrials Terry Shaneyfelt, MD, MPH Associate Professor of Medicine UAB School of Medicine ebmteacher.com
- 2. • Keeping peopleunaware of treatment group assignment after randomization Blinding
- 3. • Participants – Favorableexpectations & assumptions; deprived – Compliance & retention • Investigators – Attitudes for or against transferred to patients – Differential • use of ancillary interventions, • decisions to withdraw patients, • adjustments to dosing, • assessment of outcomes Knowing the group assignment…
- 4. • Can helpreduce – Performance bias – Ascertainment or detection bias Why Blind?
- 5. • Placebo • Shamintervention • Coding of group assignment How is postrandomization blinding accomplished?
- 6. • Not asimportant for objective outcomes • Cant always be accomplished! • Sometimes is purposely broken 3 caveats…
- 7. How might thislook in a study report? Could patients and researchers figure out if a patient was on spironolactone? Probably… • Hyperkalemia • Breast pain
Editor's Notes
- #2 These slides will discuss what blinding is and why its important
- #3 Blinding is keeping various people unware as to which group the study patient is enrolled into. As shown in the comic there are several types of blinding and different ways to define the types of blinding: Single Blind: only the patient is blinded Double Blind: patient and researchers are blinded Triple Blind: everyone including the statisticians are blinded
- #4 If the group assignment is known then several unfavorable things can happen. Participants (or patients) can affect their responses to the intervention. They expect a new treatment to be better than the older therapy (many times it isn’t) so they have favorable expectations on outcomes. If they know they received standard therapy they might feel deprived. This knowledge can also affect compliance with study procedures and how long they stay in a study. Investigators also have assumptions that can affect how they treat participants. They usually believe the new intervention is better and hope to prove so. They might order more tests or give different therapies to patients in certain groups. They might w/d patients from the study earlier in one group. They might adjust meds differently or assess outcomes differently in one group or another.
- #5 Blinding is useful to prevent or at least reduce the things I discussed on the previous slide. Performance bias refers to systematic differences between groups in the care that is provided, or in exposure to factors other than the interventions of interest. Detection bias refers to systematic differences between groups in how outcomes are determined.
- #6 For drug trials placebos are usually used to accomplish blinding. Placebos look, smell, taste, feel identical to active drugs but have no active component. For surgical interventions or physical activities (like chiropractic manipulation) you could do a sham intervention. For example, for a surgical study an incision could be made but the active part of the surgery is not done. The patient would have a surgical wound and wouldn’t know if they had the real operation or not. These interventions push the ethical limits of study design but have been done and have led to important findings. All patient assignments should be coded (for example, A for one intervention and B for another) and the code should be hidden until after the study is over and analyzed.
- #7 Blinding isn’t as important if outcomes are objective like death. Knowing what group a patient is in shouldn’t affect determination of death. That’s pretty straightforward. Now assigning a cause of death is more subjective and could be affected by knowledge of group assignment. Subjective outcomes, for example pain scores, are more prone to be affected by knowing the group assignment. Blinding cant always be accomplished. It is hard to blind a surgical study (not impossible, just hard). So what do you do in this case? Many patients and physicians can also figure out what arm of the study a patient is in. If a drug has a physiological effect (like lower cholesterol or blood pressure) its not hard to figure out if the patient is on it. The best you can do in these situations is to make sure those who assess outcomes are blinded. Sometimes we purposely unblind a study for safety or treatment based reasons. For example, we might find during an interim analysis that one group is gaining amazing benefits and it wouldn’t be ethical to withhold this treatment from others. The study would be unblended and the effective treatment offered to all. Alternatively, harms might be occurring at a greater rate in one group and the study must be unblended to stop the harmful therapy
- #8 The study was double blinded which usually means the patients and the researchers were blinded BUT you never really know who was blinded unless they tell you. Also the adjudication committee was blinded. So I feel pretty good about the blinding. Could patients and researchers figures out who was likely on spironolactone. Probably because spironolactone elevates potassium in many patients and many develop breast pain and gynecomastia. If blinding can be broken then you need to ask yourself how important that becomes. In this study its less important because the outcome is objective: death from any cause.