Page 1
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 1
Optimizing Clinical Trials:Concept to Conclusion™
Randomization: Too Important to Gamble with
A Presentation for the Delaware Chapter of the ASAOct 18, 2012
Dennis Sweitzer, Ph.D., Principal Biostatistician
Medidata Randomization Center of Excellence
Page 2
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 2
Outline
Randomized Controlled Trials• Basics• BalanceRandomization methods • Complete Randomization• Strict Minimization• Permuted Block• Dynamic Allocation (Covariate-adaptive, not Response-Adaptive)Randomization Metrics• Balance• Predictability• Loss of Power /Loss of Efficiency• Secondary Imbalance: drop-outsSimulations comparing methods• Confounding site & treatment effects (small sites)• Overall performance• Discontinuing patients• Weighting stratification factorsMeta-Balance
Page 3
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 3
Why randomize anyway?Some basic principles
Page 4
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 4
Why Gold Standard?
Randomized Controlled Trial
• Trial: Prospective & Specific• Controlled:
• Comparison with Control group • (placebo or active)
• Controlled procedures ⇒ Only Test Treatment Varies
• Randomization: Minimizes biases• Allocation bias • Selection bias • Permits blinding
Page 5
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 5
Eliminating Bias
¿ The Fact of bias ?• (conscious, unconscious, or instinctive)
¿ The Question of bias ?• Always 2nd guessing• Critics will think of unanticipated things
¡ Solution !• Treat it as a game• 1 statistician vs N clinicians• Statistician generates a random sequence• Clinicians sequential guess at each assignment• Statistician wins if clinician guesses are no better than chance
(NB: 75% wrong is just as bad as 75% right)
Page 6
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 6
Randomization Metrics
What do we want in a randomization sequence or system?Randomness Unpredictable
⟶ Reduce Allocation Bias (All studies)
⟶ Reduce Selection Bias (All studies)
⟶ Reduce placebo effects (Blinded studies)
Balance “Loss of Efficiency”⟶ Maximizes statistical power
⟶ Minimize Confounding
⟶ Enhance Credibility (Face Validity)
Page 7
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 7
Balancing
Page 8
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 8
Balanced Study
Control Test
Equal allocation between treatment arms• Maximizes
Statistical Power
Page 9
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 9
ImbalancedStatistical power limited by
smallest arm
• 36 subject simulation with Complete Randomization
⟶average loss ≈ 1 subject
10% lose ≥2 subject• Can add 2 to compensate
• BUT only large imbalances have much effect
on statistical power
Resulting in light weight results….
Severe Imbalances are rare in large studies Pr{worse than 60:40 split} for:
• n=25 <42% n=100 <4.4% n=400 0.006%⟶ ⟶ ⟶
Page 10
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 10
(NB: Planned Imbalance)
1:1 randomization maximizes power per patientBut there are other considerations
• Utility:• Need 100 patients on drug to monitor safety• Study only requires 60 (30/arm)• 2:1 randomization ⟶ 100 Test & 50 Placebo
• Motivation:• Better enrollment if 75% chance of Test drug (3:1)
• Ethics:• 85 Placebo + 255 Test vs. 125 Placebo + 125 Test
Page 11
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 11
Imbalance
• Overall balance• Only an issue for small studies
• Subgroup Balance• Fixed size studies can have variable sized subgroups
⟶ Increased risk of underpowered subgroups
Page 12
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 12
Effective Loss of Sample Size
Effective Loss = Reduction of Power as Reduction in Sample Size
Simulations of: • 36 and 18 subjects,
• males as strata at 33% of population, • randomized 1:1
• (complete randomization)
Pla
Test
Females
Con
Test
Males
Page 13
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 13
Bad Imbalance!
Treatment Imbalances within factors ⟶ spurious findings…..
Pla
Test
Females
Pla
Test
Males
Leads to conversations like:
Higher estrogen levels in patients
on Test Treatment ??
ANCOVA showed no
differences in estrogen
levels due to treatmentHmm…
?
Credibility…..
Page 14
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 14
Randomization Methods
(See Animated Powerpoint Slides…)
Page 15
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 15
Randomization
4 methods• Complete Randomization (classic approach)• Strict Minimization • Permuted Block (frequently used)• Dynamic Allocation (gaining in popularity)
Page 16
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 16
Complete Randomization
Every assignment• Same probability for each assignment• Ignore Treatment Imbalances• No restrictions on treatment assignmentsAdvantages:• Simple• Robust against selection & accidental bias• Maximum UnpredictabilityDisadvantage• High likelihood of imbalances (smaller samples)
.
Page 17
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 17
Minimization
Strict Minimization randomizes to the imbalanced arm
Page 18
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 18
Minimization
Strict Minimization rebalances the Arms• BUT at a cost in
predictability• Random only when treatments are currently
balanced
Page 19
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 19
Permuted Block
Blocks of Patients (1, 2, or 3 per treatment)
Here: 2:2 Allocation
• Balanced
• Some Predictability
T P P ?
T P P T
T P P *
(Unless Incomplete Blocks:
More strata
More incomplete⟶ )
Page 20
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 20
Dynamic Allocation
Biases Randomization to the imbalanced arm
• Unpredictable• Almost Balanced
Page 21
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 21
Dynamic Allocation
Complete Randomization• Optimizes Unpredictability
• Ignores Balance
Strict Minimization• Optimizes Balance
• Ignores Predictability
Dynamic Allocation 2nd Best Probability Parameter
Controls Balance vs. Predictability
Tradeoff
Page 22
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 22
Dynamic Allocation Flexibility
2nd Best Probability= 0
⟶ Strict Minimization
Page 23
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 23
Dynamic Allocation Flexibility
2nd Best Probability= 0.5
⟶ Complete Randomization
(for 2 treatment arms)
Page 24
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 24
Stratification Factors
Page 25
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 25
Stratification Factors
Over all Ages:
Pla
Test
Pla
Test
Pla Test
Over both sexes
PlaTest
PlaTest
PlaTest
PlaTest
Pla Test
Males Females
18-35 yo
35-65 yo
>65 yo
Pla Test
PlaTest
PlaTest
Factors Main Effects≣Strata 1≣ st Order Interactions
Randomizing a 25 yo Male:To PLA
Worsens Male ⟶balance
To Test Worsens 18-⟶
35yo balance
Balance
w/in
6
Strata
?
Marginal
Balance
Marginal
Balance
Pla Test
Overall Balance
Page 26
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 26
Permuted Block Stratified Randomization
• Only balances within strata
• Most strata will have incomplete blocks
• Imbalances accumulate at margins
Males Females
18-35 yo
35-65 yo
>65 yo
T P P T
P T * *
P * * *
P * * *
T T P *
T P T *
Pla Test
Over both sexes
Pla
Test
PlaTest
Over all Ages:
Pla
Test
Pla
Test
Pla Test
Page 27
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 27
Minimization & Marginal Balance
* Only balances on margins
* Useful if too many strata, e.g.:
* Appropriate for a main effects analysis (ie, no interactions)
* *
Balance
w/in
6 S
trata
?
Marginal Balance
Marginal Balance
Over all Ages:
Pla
Test
Pla
Test
Pla Test
Over both sexes
PlaTest
PlaTest
PlaTest
PlaTest
Pla Test
Males Females
18-35 yo
35-65 yo
>65 yo
Pla Test
PlaTest
PlaTest
Pla Test
Overall Balance
Page 28
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 28
Stratification & Dynamic Allocation
Over all Ages:
Pla
Test
Pla
Test
Pla Test
Over both sexes
Pla
Test
PlaTest
Pla
Test
PlaTest
Pla Test
Males Females
18-35 yo
35-65 yo
>65 yo
Pla Test
PlaTest
PlaTest
DA: uses weighted combination
of
• Overall balance
• Marginal balances
• Strata balance
⇒ Flexible
Balance
w/in
6 S
trata
?
Marginal Balance
Marginal Balance
Pla Test
Overall Balance
Page 29
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 29
Site as a Special Subgroup(Max 2 lines, 35 characters)
Page 30
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 30
Imbalance
• Overall balance• Only an issue for small studies
• Subgroup Balance• Fixed size studies can have variable sized subgroups
⟶ Increased risk of underpowered subgroups
• Site as special case of subgroup• Small sites ⟶ Increased risk of "monotherapy” at site
⟶ Confounding site & treatment effects ⟶ Effectively non-informative/”lost” patients
• Actual vs Assumed distribution of site size
Page 31
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 31
Enrollment per Center (Densities)
Data Sample• 13 Studies• 7.7 mo Average Enrollment period • 3953 Obs.Pts • 460 Listed Sites • 372 Active.Sites
Size Categories: {0, 1, 2, 3, 4-7, 8-11, 12-15, 16-19, 20-29, 30-39, 40-49, 50-59, 60-79, 80-99, 100-149, 150-199, ≥200 }
Page 32
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 32
Enrollment per Site (#Sites)
Data Sample• 13 Studies• 7.7 mo Average Enrollment period • 3953 Obs.Pts • 460 Listed Sites • 372 Active.Sites# Sites per Size Category {0, 1, 2, 3, 4-7, 8-11, 12-15,
16-19, 20-29, 30-39, 40-49, 50-59, 60-79, 80-99, 100-149, 150-199, ≥200 }
Page 33
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 33
Site Enrollment Simulation
Simulation based on Observations• 4 mo Enrollment Period• Enrollment ~ Poisson distribution
μ = Obs. Pts/mo (active sites) or
μ ≈ 0.5 / Enrollment period (non-active sites)
• Randomize using CR, PB(2:2), or DA(0.15).• Confounded Pts ≣ Patients at centers with only one
treatment ⇒ treatment & center effects are confounded
Page 34
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 34
Results
• Affected studies had many sites with low enrollment• Studies with fewer sites (and more pts at each) were rarely affected• Dynamic Allocation reduced confounding slightly more effectively than
permuted block
mean ±SD (80%
C.I.)
Page 35
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 35
Randomization Metrics
Page 36
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 36
Randomization Metrics
How do we measure “badness” of a randomization sequence or system?• Predictability
• Goal: an observer can guess no better than chance
⟶ Score based on Blackwell-Hodges guessing rule• Easily calculated
• ImbalanceImbalance ⟶ reduced statistical power
⟶ “Loss of Efficiency”• Measure as effective loss in number of subjects
Page 37
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 37
Blackwell-Hodges
Use Blackwell-Hodges guessing rule• Directly corresponds to game interpretation• Investigator always guesses the most probable treatment
assignment, based on past assignments• “ bias factor F”
F ≣ abs(# Correct – Expected # Correct by chance alone)• Measures potential for selection bias• Modifications:
• Limits on knowledge of investigator (eg, can only know prior treatment allocation on own site)
• Score as percentage
e.g., Score ≣ abs(% Correct – 50%)
Page 38
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 38
Blackwell-Hodges Scoring (1)
For treatment sequence “TCCC”
Initial guess ⟶ Expectation = ½ “T” ⟶ Imbalance =+1 ⟶ Guess C ⟶ Correct “TC” ⟶ Imbalance=0 ⟶ Guess either
⟶ Expectation=½ “TCC” ⟶ Imbalance=-1 ⟶ Guess T ⟶ Wrong“TCCC” ⟶ # Correct= ½ + 1+ ½ +0 =2
Score = #Correct - 2 = 2-2 = 0
Page 39
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 39
Blackwell-Hodges Scoring (2)
For treatment sequence “TCCC”“TCCC” # Correct= ½ + 1+ ½ +0 =2⟶
Complete Randomization Pr{“TCCC”} = 1/16⇒
Dynamic Allocation (p=0.15) Pr{“TCCC”}= 0.5 *0.85 * 0.5 * 0.15 = 0.031875⇒
Permuted Block (length≤4) PR{“TCCC”} = 0⇒Strict Minimization Pr{“TCCC”}=0⇒
Page 40
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 40
Blackwell-Hodges Scoring (3)
Sequence “TCCT”# Correct= ½ + 1 + ½ + 1 = 3
Score = 3 – 2 = 1
• Complete Randomization Pr{TCCT}= 1/16⇒• Strict Minimization Pr{TCCT} = ½*1*½*1 = ¼ ⇒• Permuted Block Pr{TCCT} = 1/6⇒
(NB: 6 permutations of TTCC)• Dynamic Allocation (2nd best prob.=0.15)
Pr{TCCT} = 0.5 * 0.85* 0.5 * 0.85 = 0.180625⇒
Page 41
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 41
Warning!
Blackwell-Hodges • Assesses potential selection bias
― Given known imbalance!¿¿ But which imbalance(s)??(Overall imbalance? Within strata? Within Factors?)
• Henceforth: only use imbalance within strata• Proxy for center • Assume observer only knows
imbalance within “his center”• Simple & unambiguous
M Requires some caution in interpretation
Local Predictability
ONLY
Page 42
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 42
Loss of Efficiency
• Loss can be expressed as equivalent # Patients
• In a 100 patient study:Loss of Efficiency= 5
⇒ A perfectly designed study would require only 95
Inference in Covariate-Adaptive allocation
Elsa Valdes Marquez & Nick Fieller
EFSPI Adaptive Randomisation Meeting
Brussels, 7 December 2006
http://www.efspi.org/PDF/activities/international/adaptive-rando-docs/2ValdesMarquez.pdf
Page 43
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 43
RCT vs DOE
Designed Experiment (DOE): ⟶ Select z and covariate values to minimize Ln
RCT Select only ⟶ z (No control of covariates)
X ≣ design matrix:
⟶n rows, 1 per pt ⟶K columns, 1 per covariate
z ≣ Treatment assignments
Page 44
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 44
Loss of Efficiency (Máquez & Fieller)
Dynamic Allocation
Sequentially assign Z to minimize
Page 45
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 45
Loss of Efficiency (Máquez & Fieller)
Page 46
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 46
Randomization Performance Simulations
Page 47
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 47
Simulation Set up
3 methods:• Complete Randomization• Permuted Block• Dynamic Allocation
Each simulated patient randomized w/ each method
4 Measures:• Loss of Efficiency • B-H Score (“Within Strata”)• Overall Imbalance• Relative Loss of Efficiency vs CR• % Loss of Efficiency (of #pts)
6 Strata (Factors: Sex, Age)• 33% or 50% Males• 1:1:1, 1:1:2, 1:2:3
(Young : Middle : Old)
• 48 subjects Total• With random 25% Dropout
Page 48
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 48
Note on Figures
Plot B-H score
vs
Loss of Efficiency
Median
+
80% C.I. ⇒
10% lower& 10% higher
Page 49
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 49
Simulation Results(1)
⟵Averages of Metrics
But for managing risk, need Worst Case
80% ⟶ Confidence Intervals
Predictability %Imbalance Efficiency LossDA(0.00) 22% 0.6% 0.87DA(0.15) 16% 1.6% 1.45DA(0.25) 13% 2.8% 1.99DA(0.33) 8% 4.3% 2.64DA(0.50) 4% 11.3% 4.99CR 4% 11.4% 5.03PB(8:8) 7% 7.1% 3.00PB(4:4) 13% 4.9% 1.52PB(3:3) 16% 4.2% 1.13PB(2:2) 19% 3.5% 0.79PB(1:1) 23% 2.6% 0.47
Both DA & PB are stratified.Simulation: 48 subjects, 2 stratification factors, 6 strata, uneven sizes
(DA) Dynamic Allocation (PB) Permuted Block (CR) Completely RandomDA( 2nd Best Probability ), PB( Allocation Ratio )
Simulated subjects were randomized by all 3 methods
Page 50
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 50
Randomizations Plotted by Metrics
(Essentially Strict Minimization)
CR
PB(4:4)
DA(0.33)
PB(1:1), DA(0)
PB(2:2), DA(0.15)
PB(8:8)
DA(0.5)
CR
DA(0.5) CR ≣PB CR⟶
Page 51
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 51
Correlation of Metrics
Page 52
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 53
Simulated Comparison
• 1,000 simulations per case
* 48 subjects each
* 6 Strata, 2 factor, Variety of proportions
DA(0.25)PB(3:3)
Page 53
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 54
Simulated Comparison
DA(0.25)PB(3:3)
Page 54
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 55
Relative Loss of Efficiency
•
DA(0.25)PB(3:3)
Page 55
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 56
Special TopicsLocal
Predictability
ONLY
Page 56
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 57
Dynamic Allocation Weighting
DA(0) balanced only within strata Approximates PB(1:1)DA(0) equal weighting Approximates PB(1:1)DA(0) balanced on margins Intermediate propertiesDA(0) balanced only overall Approximates CR (large N)
NB: Predictability is limited to imbalance within a stratum
Local Predictability
ONLY
Page 57
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 58
Dynamic Allocation Weighting
Weighting: (Strata, Margins, Overall)
DA(0) Equal Weighting (1,1,1) Strata Balance Dominates Approximates PB(1:1)
DA(0) Margin & Strata (1:9:0) Separates from PB(1:1)
DA(0) Unequal Weighting (1,6,20)
DA(0) Margin Balance (0,1,0)
DA(0) Overall Balance (0,0,1) Approx. CR
Local Predictability
ONLY
Page 58
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 59
DA Algorithm
Distance function ≣ Weighted Sum of Imbalances
• Relative Imbalance:
• Factor as Union of Strata ⇒
⇒ Strata Imbalances dominate Distance function
Page 59
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 60
Weighting
Over all Ages:
Pla
Test
Pla
Test
Pla Test
Over both sexes
Pla
Test
Pla
Test
Pla
Test
Pla
Test
Pla Test
Males Females
18-35 yo
35-65 yo
>65 yo
Pla Test
Pla
Test
Pla
Test
Pla Test
Over all Ages:
Pla
Test
Pla
Test
Pla Test
Over both sexes
Pla
Test
Pla
Test
Pla
Test
Pla
Test
Pla Test
Males Females
18-35 yo
35-65 yo
>65 yo
Pla Test
Pla
Test
Pla
Test
Pla Test
• Stratified Randomization weights on strata, not margins or overall
• Imbalances within strata tend to dominate in DA
• Minimization weights on margins, not strata.
• DA can weight exclusively on margins
Over all Ages:
Pla
Test
Pla
Test
Pla Test
Over both sexes
Pla
Test
Pla
Test
Pla
Test
Pla
Test
Pla Test
Males Females
18-35 yo
35-65 yo
>65 yo
Pla Test
Pla
Test
Pla
Test
Pla Test
• If a Strata is balanced, the next assignment attempts to balance the margins.
• Since small groups are more likely to have imbalances which reduce efficiency, balancing strata 1st is appropriate
Page 60
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 61
Hierarchical Balancing
Over all Ages:
PlaTest
Pla
Test
Pla Test
Over both sexes
PlaTest
PlaTest
PlaTest
PlaTest
PlaTest
Males Females
18-35 yo
35-65 yo
>65 yo
PlaTest
PlaTest
PlaTest
Pla Test
• While Imbalances within strata tends to dominate in DA,if a Strata is balanced, the next assignment attempts to balance the margins
• Since small group imbalances tend to dominate, balancing tends to be sequential
⟵ This example:(1) Balance within strata (2) If balanced within the strata, balance by age group
(since age groups tend to be smaller than sex groups)(3) If balanced within age group, balance within sex group(4) If balanced within sex group, balance overall
However: cumulative imbalances may change this order
Page 61
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 62
Replacement Randomization
Page 62
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 63
Dynamically Adapting to Dropouts
Patients discontinue ⟶ Imbalances ⟶ Reduced efficiency
“Tight” randomizations(PB with small blocks, DA with small 2nd best Prob.)
⟶ Lose more efficiency
“Loose” randomizations(CR, PB with large blocks, DA with large 2nd best Prob.)
⟶ Lose less efficiency ⟶ Little or no change
No DC
25% DC
Page 63
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 64
Dynamically Adapting to Dropouts
Dynamic Allocation: Can allocate new patients to restore balance
No DC
25% DC
DA Adj.
Page 64
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 65
Dynamically Adapting to Dropouts
“Tight” randomizations(PB with small blocks, DA with small 2nd best Prob.)
⟶ Lose more efficiency ⟶ Benefit most
“Loose” randomizations(CR, PB with large blocks, DA with large 2nd best Prob.)
⟶ Lose less efficiency ⟶ Little or no benefit
No DC
25% DC
DA Adj.
Page 65
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 66
Applications
• High drop-out ⇒ PB, DA ⟶ CR• Drop-out before becoming evaluable• Constrained resources (small sample size, limited drug supply, ….)
• Crossover studies: Requires completers• Evaluable Complete Sequence of Treatments
• Provisional Randomization / Randomize to ship• Screening visit triggers:
• Randomize at screening• If randomized treatment not on-site, ship blinded supplies
• Randomization visit:• If patient eligible ⇒ dispense assigned treatment• If not eligible ⇒store for next eligible patient
• Minimizes on-site drug supply
Page 66
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 67
Randomization Optimization Factors
• Equipose ⇒ (less random is acceptable)• Small Study ⇒ Efficiency important
⟶ Lower 2nd Best Probability• Large Study ⇒ Are there small subgroups?
All subgroups large ⟶ CR is acceptable• Small subgroups ⇒ Need more efficiency
⟶ Smaller 2nd best Prob
Page 67
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 68
Balancing Considerations
Unpredictable ⟵ ⟵⟵ ⟶ ⟶ ⟶ Balanced
• Smaller Studies• Studies with small
subgroups• Early phase studies• Interim Analyses• Equipoise • Strong Blinding• Objective Endpoints• Many Strata / Many centers• Limited blinded supplies
• Large Studies• Studies with large subgroups• Late phase studies• Strong Treatment preferences• Weak Blinding• Subjective Endpoints
Page 68
Optimizing Clinical Trials: Concept to Conclusion™ © 2012 Medidata Solutions, Inc. 69
Bibliography
Elsa Valdes Márquez & Nick Fieller. Inference in Covariate-Adaptive allocation. EFSPI Adaptive Randomisation Meeting, Brussels, 7 December 2006