Non-Inferiority Flashcards
Superiority
- What is it
To prove if an intervention is better than another
- Superiority vs Placebo
Superiority
- Hypothesis Testing
Null Hypothesis: Assume at first that treatment has no effect
Alternative Hypothesis: Assume there is a difference between groups
Superiority
- Rejecting Null Hypothesis
Just because we do not reject the null hypothesis does not mean that the null hypothesis is true
Equivalence
- What is it
To prove if an intervention is similar to another
Equivalence Trials
Seeks to rule out differences between two treatments
- To establish bioequivalence
Bioequivalence
If CI is equal to 1 equivalent
If CI does not contain 1 equivalent
If CI contains 1 and is not equal to 1 NO equivalence
Non-inferiority
To prove if an intervention not worse than another drug
When are NIFT preferred
CVOT in diabetes
Antibiotics
New Inhaled COPD Therapy
New Oral anticoagulants in atrial fibrillation
NIFT
- Hypothesis Testing
Null Hypothesis
- There is a difference between the groups
Alternative Hypothesis
- Difference between new and old intervention is less than ___
Non-Inferiority and Superiority
A non-inferiority trials can show superiority
- Must show non-inferiority first and then superiority
Interpreting NIFT
Three Interpretations
1. Both drugs are equal
2. Both drugs are ineffective
3. The trial could not find a differnce
–> However, does not mean they are equivalent
Common errors in NIFT
- Just because the trial fails to find a difference does not make the groups equivalent
Intention to Treat vs Per Protocol Analysis
Use Intention To Treat for Non-Inferority Trials
- ITT favours no difference (underestimates)
- PPA favours a difference (favours against non-inferiority) (overestimates)
How to tell if a trial has Non-Inferiority Goal
The intervention is not that much worse than the control
Margin of Inferiority
50% of the minimal treatment effect of the active control should be preserved
- Want to use the smallest acceptable degree of clinical inferiority (Want to show no important loss of efficacy)
How to calculate margin of inferiority
Use relative risk or odds ratio
(Should be less than 1.3)
- Absolute risk reduction can influence bias and result in underpowered trial
Non-Inferiority and Sample Size
The smaller the inferiority margin the more study subjects needed
Additional Questions to ask for Non-Inferior Trials
- Did the investigators guard against unwarranted conclusions of non-inferiority
- Was the effect of standard treatment preserved
- Did investigators analyze patients to the groups they were assigned to
Additional Analysis for Risk of Bias
- Ensure active control was proven to be effective
- Ensure patients and outcomes are similar to original trials of active control
- Ensure both regiments were applied optimally
- Ensure patients are analyzed to treatment they received AND the group they were assigned to
- Ensure Non-Inferiority margin was specified
- Ensure sample size is large enough
Type I Error
Concluding there is a difference when there is no difference
- False Positive
Alpha is the probability of making a type 1 error
(Is set at 0.05)
Type II Error
Concluding there was no difference when there was a difference
- False Negative
Beta is the probability of making a type 2 error
(Main determinant is sample size)
Issues with Superiority Trials
- Unethical to do placebo
- Regulatory agency wants clinical endpoint trials instead of surrogate endpoint trials
- Benefit of therapies is getting smaller
- May want to prove drug is efficacious
CI is to the left of HR (1)
Superior and Non-Inferior
CI touches HR (1)
Non-Inferior
CI touches Non-Inferiority Boundary and is to the right of HR (1)
Inferior
CI touches Both Non-Inferiority Boundary and is to the right of HR (1)
CI touches HR (1) and Both Non-Inferiority Boundary
- Underpowered
- Not approvable
CI is to the right of HR (1)
Non-inferior
- Though new treatment is worse technically
FRISBE
Follow Up
Randomization / Allocation Concealment
Intention to Treat
Similar Base Line
Blinding
Equal Treatment
Superiority vs Non-Inferiority
- F
Superiority
- Follow Up should be complete as possible
Non-Inferiority
- Follow Up should be complete as possible
- May need to be longer to see a difference
Superiority vs Non-Inferiority
- R
Superiority
- Randomization and Allocation Concealment are both desirable
Non-Inferiority
- Randomization and Allocation Concealment are both desirable
Superiority vs Non-Inferiority
- I
Superiority
- Use ITT to preserve randomness
- Harder to find differences when using ITT
Non-Inferiority
- ITT benefits Non-inferiority
- Wants to use ITT and PPA with consistient results
- Use PPA first to assess hypothesis and then use ITT to support
Superiority vs Non-Inferiority
- S
Superiority
- Verifies balanced randomization
- Wants similar groups
Non-Inferiority
- Verifies balanced randomization
- Wants similar groups
Superiority vs Non-Inferiority
- B
Superiority
- To prevent placebo
- To prevent contamination
Non-Inferiority
- To prevent placebo
- To prevent contamination
Superiority vs Non-Inferiority
- E
Superiority
- Wants no differences in treatment except intervention
- Wants NO contamination
- Wants NO co-intervention
Non-Inferiority
- Wants no differences in treatment except intervention
- Wants NO contamination
- Wants NO co-intervention
- Even more picky about contamination as it makes it harder to find differences
CI touches HR (1) and Non-Inferiority Boundary
Non-Inferiority is NOT shown
