5 - Clinical trials Flashcards
Difference b/w clinical trials & controlled lab investigations
- Ethics → experiment involving human subjects ?!
- Bias → “ on intelligent subjects requires new measures of control
What is the hierarchy of scientific evidence?
case reports –> case series –> ecologic studies
–> cross-sectional studies –> case-control studies –> cohort studies –> RANDOMISED CONTROLLED TRIALS
(from generating hypotheses to establishing causality)
How is a hypothesis formed?
- Primary Q
Eg: “In population X, is drug A at daily dose Y more efficacious in reducing K, over a period of time T, than drug B at daily dose Z?” - Secondary Q
Study-related Qs in the whole gp or its subgroups
Which 2 groups are compared in a study?
Treatment gp = allocated the agent under study
Comparison gp = not allocated the agent under study
(may receive no treatment, an inactive treatment aka PLACEBO or another active treatm (e.g standard treatm))
Types of study designs
- Simple randomised design
- Stratified randomised design
- Crossover design
- Factorial design
What is the basis of stratified randomised studies?
If prognostic factors are known & pts can be grouped into prognostic categories, comparability among treatm gps is better achieved w/ stratification
Within each gp, pts are randomly assigned to treatments
What is the basis of crossover studies?
Pts serve as their own controls
Eg: subjects may undergo therapy for 6w & then “cross over” to control therapy for another 6w
What is a -ve of crossover study design?
potential for “carryover” effects (i.e treatm given during 1st period may carry over into 2nd period)
What are the types & what is the basis of factorial studies?
Motive = ask 2 or + Qs in same clinical trial
- 2x2 –> simplest factorial design: 2 treatments studied for their relationship to response & each is given at 2 levels (eg high dosage & low dosage or drug A & B)
- 2x2x2 etc
The 5 elements of research protocol
- Sample size
- Funding
- Approval from ethics committee
- Eligibility criteria
- Signed informed consent
What are the main parameters of sample size estimation?
- Level of significance/ α → how big a risk can be taken that the 2 treatm are incorrectly shown as statistically different?
p Β error & 1-β (power of a trial)→ how “ “ “ as not statistically different?
Randomisation is the preferred method of assigning pts to receive treatments being compared. True or false?
True
What are the 3 main randomisation methods?
Method 1: toss a coin
Method 2: use a table of random numbers
Method 3: stratified block randomisation
What are the +ves of randomization?
- Produces study gps comparable w/ respect to known & unknown risk factors
- Removes investigator bias in allocation of participants & guarantees results have valid significance
What is the purpose of a placebo?
match experience of comparison gp w/ that of treatm gp
What is blinding/masking (bias control) ?
= keeping identity of treatm masked for:
- Subject → single-blind
- Subject + investigator → double-blind
- Triple-blind → extension of double-blind → committee monitoring response is not told identity of gps, simply given data for groups A & B
What is the purpose of blinding/masking?
bias reduction
What may adversely affect follow-ups?
pts dropout, or cant be located/contacted → lost to follow-up
What are the reasons for non-compliance & the consequences?
Adverse reactions, waning interest, desire for other therapies
Consequences: smaller diff b/w treatm & comparison gps than truly exists → dilutes real impact of a treatm
What can be done to prevent non-compliance?
- Simple & easy-to-follow regimen
- Present realistic pic of req tasks during consent process
- Frequent contact w/ participants
- Run-in period before enrollment & randomization
What is the run-in period?
- To ascertain which potential participants can comply to study regimen
- Only compliant ppl are enrolled in trial
What is an interim analysis?
Compares intervention gps at any time before formal completion of trial
What is an ITT (intention-to-treat) analysis?
A specific strategy for generating results of a randomized trial
All subjects are compared in treatm gps to which they were originally randomized, regardless of any treatm they later received
(more informative & gives results closer to those that would’ve been seen if the treatm were given to the pop as a whole)
Main advantages of ITT analysis
Preserves strengths of randomization, i.e minimize bias
Captures what happens in real-life
Gives a pragmatic estimate of effect of a treatm
