1. Scott Berry, PhD Berry Consultants scott@berryconsultants.com
REMAP-CAP: A Precision Medicine Embedded Platform Trial for Community Acquired Pneumonia
Scott Berry, PhD
Berry Consultants

2. Are we Prepared for the Next Pandemic?

3. PREPARE is funded by the European Commission under grant number 602525
PI: Herman Goosens
University of Antwerp
PREPARE is funded by the European Commission under grant number

4. Solution: A standing protocol for pandemic X …
We don’t know what is coming!
Pandemic X will come through the ICU’s
So, create a standing platform trial in community acquired pneumonia, enrolling sites, learning about CAP, prepared for pandemic X

5. Solution: A Standing Platform Trial In CAP
Randomized Embedded Multifactorial Adaptive Platform

6. Master Protocol Structure
A set of domains: Antibiotic, Immunomodulation with extended macrolide, Immunomodulation with hydrocortisone, … ?
Each regimen is exactly one intervention from each domain
Each patient comes in as a member of a strata
Domains, interventions, and strata will evolve
Science and new questions… pandemic
Empirical decisions during the trial

7. Multifactorial intervention assignments
Regimen = set of domain-specific interventions
Effect of an intervention is conditional upon
Stratum
Interventions within other domains
Regimen
Domain A
Domain B
Domain C #1 A1 B1 C1 #2 C2 #3 B2 #4 #5 A2
….. #n An Bn Cn
Pre-specified architecture determined by
Choice of domains, strata, etc.
Choice of potential interactions
Choices inform a Bayesian inference model
Pre-trial simulations evaluate performance
Each external adaptation (ex. new domain)
Modify elements in Bayesian model
Re-simulate before ‘live’ deployment
Pre-trial design and construction

8. Multifactorial intervention assignments
Patients
Presumed severe CAP
Different strata (ex. shock or not)
Multifactorial intervention assignments
Regimen = set of domain-specific interventions
Effect of an intervention is conditional upon
Stratum
Interventions within other domains
Regimen
Domain A
Domain B
Domain C #1 A1 B1 C1 #2 C2 #3 B2 #4 #5 A2
….. #n An Bn Cn
Pre-specified architecture determined by
Choice of domains, strata, etc.
Choice of potential interactions
Choices inform a Bayesian inference model
Pre-trial simulations evaluate performance
Each external adaptation (ex. new domain)
Modify elements in Bayesian model
Re-simulate before ‘live’ deployment
Pre-trial design and construction
Collected at sites
Managed at regional data centers
Merged at central statistical center
Data collection
Re-estimate Bayesian inference model with new data to update probabilities
Update and adapt
External adaptations
Statistical trigger
Steering Committee can
Add strata, domains & interventions
DSMB can
Request new external data be incorporated in priors
Overrule statistical triggers
Result declared when, within stratum, an intervention is
Superior >99% likely to be best
Equivalent >90% likely to be D<3%
Inferior <1% likely to be best

9. Multifactorial intervention assignments
Patients
Presumed severe CAP
Different strata (ex. shock or not)
Embedding
Patient identification and enrollment
Tied to clinical ‘point-of-care’
Randomized interventions
Issued as ‘order set’ regimen
EHR embedding
Screen and flag patient
Consent documentation
Generate regimen order set
Flag downstream states
Data collection
Multifactorial intervention assignments
Regimen = set of domain-specific interventions
Effect of an intervention is conditional upon
Stratum
Interventions within other domains
Regimen
Domain A
Domain B
Domain C #1 A1 B1 C1 #2 C2 #3 B2 #4 #5 A2
….. #n An Bn Cn
Pre-specified architecture determined by
Choice of domains, strata, etc.
Choice of potential interactions
Choices inform a Bayesian inference model
Pre-trial simulations evaluate performance
Each external adaptation (ex. new domain)
Modify elements in Bayesian model
Re-simulate before ‘live’ deployment
Pre-trial design and construction
Collected at sites
Managed at regional data centers
Merged at central statistical center
Data collection
Re-estimate Bayesian inference model with new data to update probabilities
Update and adapt
External adaptations
Statistical trigger
Steering Committee can
Add strata, domains & interventions
DSMB can
Request new external data be incorporated in priors
Overrule statistical triggers
Result declared when, within stratum, an intervention is
Superior >99% likely to be best
Equivalent >90% likely to be D<3%
Inferior <1% likely to be best

10. Multifactorial intervention assignments
Patients
Presumed severe CAP
Different strata (ex. shock or not)
Embedding
Patient identification and enrollment
Tied to clinical ‘point-of-care’
Randomized interventions
Issued as ‘order set’ regimen
EHR embedding
Screen and flag patient
Consent documentation
Generate regimen order set
Flag downstream states
Data collection
Multifactorial intervention assignments
Regimen = set of domain-specific interventions
Effect of an intervention is conditional upon
Stratum
Interventions within other domains
Regimen
Domain A
Domain B
Domain C #1 A1 B1 C1 #2 C2 #3 B2 #4 #5 A2
….. #n An Bn Cn
Response-adaptive randomization
Launch with initial weights
Update based on new probabilities
Pre-specified architecture determined by
Choice of domains, strata, etc.
Choice of potential interactions
Choices inform a Bayesian inference model
Pre-trial simulations evaluate performance
Each external adaptation (ex. new domain)
Modify elements in Bayesian model
Re-simulate before ‘live’ deployment
Pre-trial design and construction
Collected at sites
Managed at regional data centers
Merged at central statistical center
Data collection
Re-estimate Bayesian inference model with new data to update probabilities
Update and adapt
External adaptations
Statistical trigger
Steering Committee can
Add strata, domains & interventions
DSMB can
Request new external data be incorporated in priors
Overrule statistical triggers
Result declared when, within stratum, an intervention is
Superior >99% likely to be best
Equivalent >90% likely to be D<3%
Inferior <1% likely to be best

11. Multifactorial intervention assignments
Patients
Presumed severe CAP
Different strata (ex. shock or not)
Embedding
Patient identification and enrollment
Tied to clinical ‘point-of-care’
Randomized interventions
Issued as ‘order set’ regimen
EHR embedding
Screen and flag patient
Consent documentation
Generate regimen order set
Flag downstream states
Data collection
Multifactorial intervention assignments
Regimen = set of domain-specific interventions
Effect of an intervention is conditional upon
Stratum
Interventions within other domains
Regimen
Domain A
Domain B
Domain C #1 A1 B1 C1 #2 C2 #3 B2 #4 #5 A2
….. #n An Bn Cn
Response-adaptive randomization
Launch with initial weights
Update based on new probabilities
Pre-specified architecture determined by
Choice of domains, strata, etc.
Choice of potential interactions
Choices inform a Bayesian inference model
Pre-trial simulations evaluate performance
Each external adaptation (ex. new domain)
Modify elements in Bayesian model
Re-simulate before ‘live’ deployment
Pre-trial design and construction
Collected at sites
Managed at regional data centers
Merged at central statistical center
Data collection
Re-estimate Bayesian inference model with new data to update probabilities
Update and adapt
External adaptations
Statistical trigger
Steering Committee can
Add strata, domains & interventions
DSMB can
Request new external data be incorporated in priors
Overrule statistical triggers
Result declared when, within stratum, an intervention is
Superior >99% likely to be best
Equivalent >90% likely to be D<3%
Inferior <1% likely to be best

12. Multifactorial intervention assignments
Patients
Presumed severe CAP
Different strata (ex. shock or not)
Embedding
Patient identification and enrollment
Tied to clinical ‘point-of-care’
Randomized interventions
Issued as ‘order set’ regimen
EHR embedding
Screen and flag patient
Consent documentation
Generate regimen order set
Flag downstream states
Data collection
Multifactorial intervention assignments
Regimen = set of domain-specific interventions
Effect of an intervention is conditional upon
Stratum
Interventions within other domains
Regimen
Domain A
Domain B
Domain C #1 A1 B1 C1 #2 C2 #3 B2 #4 #5 A2
….. #n An Bn Cn
Response-adaptive randomization
Launch with initial weights
Update based on new probabilities
Pre-specified architecture determined by
Choice of domains, strata, etc.
Choice of potential interactions
Choices inform a Bayesian inference model
Pre-trial simulations evaluate performance
Each external adaptation (ex. new domain)
Modify elements in Bayesian model
Re-simulate before ‘live’ deployment
Pre-trial design and construction
Collected at sites
Managed at regional data centers
Merged at central statistical center
Data collection
Re-estimate Bayesian inference model with new data to update probabilities
Update and adapt
External adaptations
Statistical trigger
Steering Committee can
Add strata, domains & interventions
DSMB can
Request new external data be incorporated in priors
Overrule statistical triggers
Result declared when, within stratum, an intervention is
Superior >99% likely to be best
Equivalent >90% likely to be D<3%
Inferior <1% likely to be best

13. Multifactorial intervention assignments
Patients
Presumed severe CAP
Different strata (ex. shock or not)
Embedding
Patient identification and enrollment
Tied to clinical ‘point-of-care’
Randomized interventions
Issued as ‘order set’ regimen
EHR embedding
Screen and flag patient
Consent documentation
Generate regimen order set
Flag downstream states
Data collection
Multifactorial intervention assignments
Regimen = set of domain-specific interventions
Effect of an intervention is conditional upon
Stratum
Interventions within other domains
Regimen
Domain A
Domain B
Domain C #1 A1 B1 C1 #2 C2 #3 B2 #4 #5 A2
….. #n An Bn Cn
Response-adaptive randomization
Launch with initial weights
Update based on new probabilities
Pre-specified architecture determined by
Choice of domains, strata, etc.
Choice of potential interactions
Choices inform a Bayesian inference model
Pre-trial simulations evaluate performance
Each external adaptation (ex. new domain)
Modify elements in Bayesian model
Re-simulate before ‘live’ deployment
Pre-trial design and construction
Collected at sites
Managed at regional data centers
Merged at central statistical center
Data collection
Re-estimate Bayesian inference model with new data to update probabilities
Update and adapt
External adaptations
Statistical trigger
Steering Committee can
Add strata, domains & interventions
DSMB can
Request new external data be incorporated in priors
Overrule statistical triggers
Result declared when, within stratum, an intervention is
Superior >99% likely to be best
Equivalent >90% likely to be D<20% OR
Inferior <1% likely to be best

14. Multifactorial intervention assignments
Patients
Presumed severe CAP
Different strata (ex. shock or not)
Embedding
Patient identification and enrollment
Tied to clinical ‘point-of-care’
Randomized interventions
Issued as ‘order set’ regimen
EHR embedding
Screen and flag patient
Consent documentation
Generate regimen order set
Flag downstream states
Data collection
Multifactorial intervention assignments
Regimen = set of domain-specific interventions
Effect of an intervention is conditional upon
Stratum
Interventions within other domains
Regimen
Domain A
Domain B
Domain C #1 A1 B1 C1 #2 C2 #3 B2 #4 #5 A2
….. #n An Bn Cn
Response-adaptive randomization
Launch with initial weights
Update based on new probabilities
Pre-specified architecture determined by
Choice of domains, strata, etc.
Choice of potential interactions
Choices inform a Bayesian inference model
Pre-trial simulations evaluate performance
Each external adaptation (ex. new domain)
Modify elements in Bayesian model
Re-simulate before ‘live’ deployment
Pre-trial design and construction
Collected at sites
Managed at regional data centers
Merged at central statistical center
Data collection
Re-estimate Bayesian inference model with new data to update probabilities
Update and adapt
Statistical trigger
Result declared when, within stratum, an intervention is
Superior >99% likely to be best
Equivalent >90% likely to be D<20% OR
Inferior <1% likely to be best

15. Multifactorial intervention assignments
Patients
Presumed severe CAP
Different strata (ex. shock or not)
Embedding
Patient identification and enrollment
Tied to clinical ‘point-of-care’
Randomized interventions
Issued as ‘order set’ regimen
EHR embedding
Screen and flag patient
Consent documentation
Generate regimen order set
Flag downstream states
Data collection
Multifactorial intervention assignments
Regimen = set of domain-specific interventions
Effect of an intervention is conditional upon
Stratum
Interventions within other domains
Regimen
Domain A
Domain B
Domain C #1 A1 B1 C1 #2 C2 #3 B2 #4 #5 A2
….. #n An Bn Cn
Response-adaptive randomization
Launch with initial weights
Update based on new probabilities
Pre-specified architecture determined by
Choice of domains, strata, etc.
Choice of potential interactions
Choices inform a Bayesian inference model
Pre-trial simulations evaluate performance
Each external adaptation (ex. new domain)
Modify elements in Bayesian model
Re-simulate before ‘live’ deployment
Pre-trial design and construction
Collected at sites
Managed at regional data centers
Merged at central statistical center
Data collection
Re-estimate Bayesian inference model with new data to update probabilities
Update and adapt
External adaptations
Statistical trigger
Steering Committee can
Add strata, domains & interventions
DSMB can
Request new external data be incorporated in priors
Overrule statistical triggers
Result declared when, within stratum, an intervention is
Superior >99% likely to be best
Equivalent >90% likely to be D<20% OR
Inferior <1% likely to be best

16. Launch

17. Bayesian Logistic Regression Model
Models main effects of interventions
Models 2-way interactions between interventions
Models interactions with stratum
Separate priors on these interactions
Additively model site (nested within region), age, severe hypoxia, stratum, and time

18. Randomization
Proportional to the posterior probability that the regimen is optimal way to treat the patient from that strata

19. Detailed and Exhaustive Simulations
Improved power; by strata
Improved treatment of patients within the system
Optimized “priors” for interaction factors
Optimized decision criteria
Equivalency criteria by domain

20. REMAP-CAP

21. Summary
On behalf of a very large team: Berry; PREPARE; multiple countries; …
Platform trial to learn about and implement the best way to treat patients with CAP
Stratified by shock/no-shock or potential growing strata (severity if hypoxia)
A standing trial to be utilized to learn of the next pandemic, immediately