Flashcards in Clinical Trials Deck (27):
Whats the ABC's of clinical trial design?
A = Assignment
B = Blinding
C = Comparison
s = Sequence
What are the two types of assignment?
- First come first served
- Randomized (inc; balanced, stratified)
What does the first come first served method involve?
- Range of possibilites that will induce bias
- Main source of bias is the lack of blinding I.e everyone knows whos getting treatment vs placebo etc
What is the randomized assignment method?
Subjects are randomized into control and subject groups
- Can be done evenly (balanced)
- Or stratified based on age, gender etc
What is blinding used for?
To reduce bias
What are the types of blinding?
- Single Blind (subjects dont know)
- Double blind
-Double dummy (two physically different treatments i.e inhaler vs tablet)
- Triple blind (randomized sequence is lost or misinterpreted) - nobody ever knows what treatment was given
What are the types of comparison?
- Standard treatment
What is is active comparison?
- Dose controlled
- Concentration controlled
- Biomarker controlled i.e cholesterol level
Comparing the treatments based on their differences. To learn about the relationship between intensity of treatment and outcome
What is standard treatment comparison?
- All subjects recieve the treatment and then some get an add-on (testing drug)
(non-inferiority) refers to trial that shows that the new treatment is no worse than the standard.
What is sequence?
The sequence of treatments can influenced what is learned from a trial and the kind of bias that can araise
What are the types of sequence?
What is parallel sequence?
Different treatments given to different groups at the same time
Whats the upside and downside to parallel sequence?
Answers the does the drug work question but cant answer learning questions such as shape of the dose response curve
What is crossover sequence?
Two or more treatments in each subject
Allows formation of dose-reponse curve and may be more statistically valuable
Whats wrong with the crossover design?
There may be carryover affect i.e during the time of off medication
What is the titration design?
Fixed sequential doses (Forced)
Change in dose till therapeutic response level is achieved (flexible)
What is analysis?
Does it work? (confirming Qs) i.e does it lower BP
How well does it work? (learning Qs) - scales i.e what does is needed to produce this response..
What are some methods for determining does it work?
Simple comparison i.e drug vs placebo
- t-test (two groups)
- Anova (several)
- log rank test (survival)
What factors are considered in the size of the response?
How much Q's - i.e how much drug to produce this response
What are the methods behind how well does it work questions?
PK and PD methods
Why are PK, PD and PKPD models for how well does it work questions, better than hypothesis testing?
They account for:
Execution differences i.e
- Adherence (stop taking meds)
- Drop out
Co-variates (inter-individual differences)
- Renal function, size, gender
What is a power analysis?
A power analysis is used to determine the likelihood that a clinical trial will detect a drug effect if the drug effect really exists
Why would we use a power analysis?
We would use a power analysis because of human variability that means the testing does not show the true effect of the drug because its effect is lost to "noise"
It generates a % likelihood that the null hypothesis will be rejected (good) based on criteria
What questions are asked in a power analysis?
- What is the size of the effect
- What is the variability in the response
- What is the desired p value
- What is the desired power i.e 0.8 (good)
What are the two analysis perspectives for analyzing clinical trials?
- Intention to treat
- As treated
What does the intention to treat perspective encompass?
- Only takes into account treatment assessment
- Doesnt take into account if the subject took the treatment or not "use effectiveness"
- Good for pharmacoeconomics