Non inferiority

Describe when a NI study is appropriate

Describe when a NI study is appropriate

•Equivalence Trial: Designed to show if a new therapy is similar to a control therapy

Non inferiority trial: Designed to show a new therapy is not worse than the control therapy by more than an acceptable amount

•Use when unethical to compare to a palcebo

What is Hh and Ha in Non inferiority

H0: New therapy- Control therapy >= Margin

HA:New therapy- control therapy < Margin

Margin is what would be unacceptably worse

What is a type 1 error in non inferiority

Found new agent to be non inferior when it actually is inferior

If Upper bound of 95% confidence interval crosses margin….

then it is unacceptably worse

Is ITT or Per protovol less conservative in NI

ITT is less conservative in NI, and we look at per protocol first

Can NI be interpreted to mean “the same” “equivalent” “ no difference”

NO! >:(

six supplemental CONSORT questions used to evaluate an NI trial

1)Non inferiority margin

2)Control

3)Sample size

4)Confidence Interval

5)Analysis of the results

6)Conclusion

Non inferiority margin considerations

What is the non inferiority margin ?

Was it established a priori?

Is it justified as being clinicallt relevant and sensible?

Control

What is the control?

Has the control been proven to be efficacious vs placebo?

Confidence interval

Is a two sided confidence interval presented around the difference between the intervention and control and presented in relationship to the non inferiority margin

Analysis of the results

Is it presented for both per protocol and ITT?

Conclusion

Does it reference the non inferiority margin?

FACTS eval for NI studies

1)Formualte null hypothesis

2)Assess the NI margin

3)Compare the ITT and PP analysis

4)Translate CIs graphically

5)Summarize the clinical relevance of results

Translate CIs graphicallt

1)NI margin

2)Type of CIs assocaited with primary endpoint

3)Numerical results of primary endpoint

Proper phrasing for non inferiority

“Met the criteria for non inferiority because the upper bound of the 95% confidence interval did not cross the non inferiority margin of ___”