BMS48 – Clinical Trials
CONTENT:
- Randomization
- Blinding
- Block randomization (Random permuted blocks)
- Stratfication
- What do you have to consider at the designing stage of a clinical trial?
- Drug development process
- Cross-over design
- Clinical trials with interim analysis (RCTs)
- Appropriate care
- Superiority trial
- Equivalent trial
- Non-inferiority trial
- Per Protocol vs Intention-to-treat
- Gatekeeping procedure with hierarchical structure
- Null hypothesis in superiority, equivalent and non-inferiority
- Equipoise (uncertainty principle):
- Reasons for conducting a diagnostic trial:
- Factors that can make intervention studies complex:
- Intra-cluster correlation (ICC)
- Effective cluster size
- Cluster RCT
- Contamination
- Individual RCT vs cluster RCT
- Stepped wedge RCT with cluster randomization
, Lecture 1 & 2: Clinical trials (in drug development)
Randomization
Main purpose of randomization is to create two groups with similar prognostic factors, thus the
clinician cannot predict in advance which treatment subjects will receive (this limits bias). This is done
using an allocation sequence.
Blinding
Blinding is facilitated through randomization. Through blinding bias and the Hawthorne effect
(participants behaviour changes as a result of knowing that he/her is being observed) are prevented.
Lastly, placebo only works when participants are blinded. Using the allocation concealment, it is
ensured that the implementation of a random allocation sequence is done correctly (blinded).
Block randomization (Random permuted blocks)
Used to ensure a balance in sample size, while maintaining the randomizing aspect. The sequence is
built up by using blocks (e.g., blocks of 4; AABB, ABAB, ABBA, BBAA, BABA, BAAB). But you could also
use blocks with different sizes, for example 6 and 4.
Stratfication
Allocation sequence which is separate done for different strata of subjects. Whenever you know
smoking has an effect on the outcome, but is not your independent variable, you stratify groups
based on smoking. As a result, smoking is stratified over both groups, meaning it prevents an
imbalance between the two groups.
What do you have to consider at the designing stage of a clinical trial?
- Prognostic factors; could be used to stratify on.
- Logistics of the trial; How many measurement site/hospitals?
- Analysis; Even though you might randomize, you can’t ignore an observed prognostic covariate
during the analysis. Use literature to identify them, or measure if control and treatment differ in
characteristics.
Drug development process
Preclinical: Testing drugs in non-human subjects to gather information
Phase 0: Very small population for pharmacodynamics and pharmacokinetics (e.g., half time)
Phase 1: Testing of drug on healthy volunteers for dose ranging (20-100 participants)
Phase 2: Testing of drug on patients to assess efficacy and safety (100-300 participants)
Phase 3: Testing of drug on patients to assess efficacy, effectiveness and safety (1000-2000 part.)
Phase 4: Post-marketing surveillance, observe drug use in public (very, very large sample size)
In early clinical development:
- investigate pharmacokinetics, pharmacodynamics, & dose.
- these studies are more exploratory with diverse designs (e.g., cross-over)
In late-stage clinical development:
- Finding therapeutic dose and compare with placebo’s/existing therapies
- these studies are more confirmatory and mostly RCT’s with interim analyses
Cross-over design
Cross over design has wash-out periods between treatment A & B, in order to prevent carry-over
effect (effect of first treatment still active when the second treatment is given). You want to limit all
effects, except for the effect you want to investigate à treatment effect. Cross-over designs have
potential smaller sample sizes, moreover, participants are used as his/her own control, reducing
CONTENT:
- Randomization
- Blinding
- Block randomization (Random permuted blocks)
- Stratfication
- What do you have to consider at the designing stage of a clinical trial?
- Drug development process
- Cross-over design
- Clinical trials with interim analysis (RCTs)
- Appropriate care
- Superiority trial
- Equivalent trial
- Non-inferiority trial
- Per Protocol vs Intention-to-treat
- Gatekeeping procedure with hierarchical structure
- Null hypothesis in superiority, equivalent and non-inferiority
- Equipoise (uncertainty principle):
- Reasons for conducting a diagnostic trial:
- Factors that can make intervention studies complex:
- Intra-cluster correlation (ICC)
- Effective cluster size
- Cluster RCT
- Contamination
- Individual RCT vs cluster RCT
- Stepped wedge RCT with cluster randomization
, Lecture 1 & 2: Clinical trials (in drug development)
Randomization
Main purpose of randomization is to create two groups with similar prognostic factors, thus the
clinician cannot predict in advance which treatment subjects will receive (this limits bias). This is done
using an allocation sequence.
Blinding
Blinding is facilitated through randomization. Through blinding bias and the Hawthorne effect
(participants behaviour changes as a result of knowing that he/her is being observed) are prevented.
Lastly, placebo only works when participants are blinded. Using the allocation concealment, it is
ensured that the implementation of a random allocation sequence is done correctly (blinded).
Block randomization (Random permuted blocks)
Used to ensure a balance in sample size, while maintaining the randomizing aspect. The sequence is
built up by using blocks (e.g., blocks of 4; AABB, ABAB, ABBA, BBAA, BABA, BAAB). But you could also
use blocks with different sizes, for example 6 and 4.
Stratfication
Allocation sequence which is separate done for different strata of subjects. Whenever you know
smoking has an effect on the outcome, but is not your independent variable, you stratify groups
based on smoking. As a result, smoking is stratified over both groups, meaning it prevents an
imbalance between the two groups.
What do you have to consider at the designing stage of a clinical trial?
- Prognostic factors; could be used to stratify on.
- Logistics of the trial; How many measurement site/hospitals?
- Analysis; Even though you might randomize, you can’t ignore an observed prognostic covariate
during the analysis. Use literature to identify them, or measure if control and treatment differ in
characteristics.
Drug development process
Preclinical: Testing drugs in non-human subjects to gather information
Phase 0: Very small population for pharmacodynamics and pharmacokinetics (e.g., half time)
Phase 1: Testing of drug on healthy volunteers for dose ranging (20-100 participants)
Phase 2: Testing of drug on patients to assess efficacy and safety (100-300 participants)
Phase 3: Testing of drug on patients to assess efficacy, effectiveness and safety (1000-2000 part.)
Phase 4: Post-marketing surveillance, observe drug use in public (very, very large sample size)
In early clinical development:
- investigate pharmacokinetics, pharmacodynamics, & dose.
- these studies are more exploratory with diverse designs (e.g., cross-over)
In late-stage clinical development:
- Finding therapeutic dose and compare with placebo’s/existing therapies
- these studies are more confirmatory and mostly RCT’s with interim analyses
Cross-over design
Cross over design has wash-out periods between treatment A & B, in order to prevent carry-over
effect (effect of first treatment still active when the second treatment is given). You want to limit all
effects, except for the effect you want to investigate à treatment effect. Cross-over designs have
potential smaller sample sizes, moreover, participants are used as his/her own control, reducing
