eduKateSG · Why Science?
Build one global trial that can still explain meaningful regional differences
Read FDA’s July 2018 final ICH E17 guidance through regional variability, intrinsic and extrinsic factors, common protocols, sample allocation, pooling, consistency, analysis and generalizability.
A medicine may be developed through one trial spanning countries, healthcare systems, ethnic groups and treatment practices. A multi-regional clinical trial can produce evidence for several authorities at once, but only when global efficiency and regional meaning are designed together.
FDA’s final July 2018 ICH E17 guidance gives general principles for planning and designing multi-regional clinical trials so their evidence is more acceptable across regions and countries.
Did you know? E17 does not demand a statistically significant treatment effect within every region. It asks for scientifically justified representation and a careful evaluation of whether clinically relevant differences exist and what might explain them.
For students in Singapore, E17 is a global science lesson: human biology travels with genetics, diet, care, culture, measurement and access. Good research respects both shared mechanisms and local conditions.
Chapter index
Chapters 1–9: global questions, regional factors, common protocols and recruitment
- 1. A multi-regional trial uses one protocol across regions
- 2. The July 2018 E17 guidance is final
- 3. The development strategy should define the MRCT role
- 4. Regions should be scientifically meaningful
- 5. Intrinsic factors can change exposure or response
- 6. Extrinsic factors travel through healthcare and culture
- 7. Regional experts should shape the protocol early
- 8. Eligibility should support the target population
- 9. Endpoints need common meaning and measurement
Chapters 10–17: sample allocation, pooling, consistency, interpretation and regulatory use
- 10. Treatment and comparator choices need regional relevance
- 11. Recruitment should be monitored while the trial runs
- 12. Sample allocation needs a scientific basis
- 13. Prespecified pooling can connect similar populations
- 14. One primary analysis should integrate the full trial
- 15. Consistency is a clinical and statistical judgement
- 16. Generalizability depends on people and settings represented
- 17. A complete MRCT receipt preserves global and local meaning
CHAPTER 1 OF 17 · Plan globally
1. A multi-regional trial uses one protocol across regions
An MRCT is conducted under a single protocol in more than one geographical region. It can support a common primary question while enrolling populations across countries or regulatory areas. The common framework can reduce duplication, speed evidence generation and reveal variability that separate local studies might not compare directly.
The global question in a multi-regional trial uses one protocol across regions becomes credible when regional differences are planned rather than discovered with surprise. An MRCT is conducted under a single protocol in more than one geographical region. It can support a common primary question while enrolling populations across countries or regulatory areas. The common framework can reduce duplication, speed evidence generation and reveal variability that separate local studies might not compare directly. Keep this boundary visible: One protocol does not make all participating settings biologically, clinically or operationally identical.
Try this
Build a global-local evidence card for a multi-regional trial uses one protocol across regions. List the elements that must remain common and the local adaptations that require justification and documentation. Add one intrinsic and one extrinsic factor that could alter the conclusion. Keep this limit: One protocol does not make all participating settings biologically, clinically or operationally identical.
A strong science note separates observation from interpretation. List the elements that must remain common and the local adaptations that require justification and documentation. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
List the elements that must remain common and the local adaptations that require justification and documentation. Keep this boundary visible: One protocol does not make all participating settings biologically, clinically or operationally identical.
FDA identifies E17 as final guidance dated July 2018, issued by CDER and CBER, with docket reference 2018-15395. Its purpose is to improve the planning and design of MRCTs and increase their acceptability as primary evidence in global regulatory submissions for medicinal products.
For the july 2018 e17 guidance is final, geography is only the beginning; mechanism, care and measurement explain why results might travel. FDA identifies E17 as final guidance dated July 2018, issued by CDER and CBER, with docket reference 2018-15395. Its purpose is to improve the planning and design of MRCTs and increase their acceptability as primary evidence in global regulatory submissions for medicinal products. Preserve this guardrail: The guidance gives general principles and does not replace disease-specific or regional requirements.
Try this
Run a representation audit for the july 2018 e17 guidance is final. Create a source receipt with title, final status, date, docket reference, issuing centres, purpose and scope. Mark the region or subpopulation with the weakest information. Respect this boundary: The guidance gives general principles and does not replace disease-specific or regional requirements.
A strong science note separates observation from interpretation. Create a source receipt with title, final status, date, docket reference, issuing centres, purpose and scope. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Create a source receipt with title, final status, date, docket reference, issuing centres, purpose and scope. Keep this boundary visible: The guidance gives general principles and does not replace disease-specific or regional requirements.
CHAPTER 3 OF 17 · Plan globally
3. The development strategy should define the MRCT role
Before designing one trial, sponsors should explain how early pharmacology, dose finding, prior regional experience and confirmatory evidence fit together. The MRCT may be the main confirmatory study or one part of a larger programme. Early consultation across participating regions helps align the question, endpoint, comparator and evidentiary expectations.
A good MRCT reading of the development strategy should define the mrct role holds the overall effect and regional context at the same time. Before designing one trial, sponsors should explain how early pharmacology, dose finding, prior regional experience and confirmatory evidence fit together. The MRCT may be the main confirmatory study or one part of a larger programme. Early consultation across participating regions helps align the question, endpoint, comparator and evidentiary expectations. The inference stops here: A large global trial is not efficient when authorities need incompatible evidence that was never reconciled.
Try this
Explain the development strategy should define the mrct role using map, mechanism and uncertainty. Draw the development programme and label which uncertainty each regional or global study resolves. Let a reviewer from another region challenge the transfer. Preserve this caution: A large global trial is not efficient when authorities need incompatible evidence that was never reconciled.
A strong science note separates observation from interpretation. Draw the development programme and label which uncertainty each regional or global study resolves. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Draw the development programme and label which uncertainty each regional or global study resolves. Keep this boundary visible: A large global trial is not efficient when authorities need incompatible evidence that was never reconciled.
A region may be geographic, regulatory or defined by countries with relevant common features. The choice should connect to factors that could influence treatment response, clinical practice or evidence interpretation. Administrative convenience alone may group populations that differ in important ways or split populations that share the same drivers.
The global question in regions should be scientifically meaningful becomes credible when regional differences are planned rather than discovered with surprise. A region may be geographic, regulatory or defined by countries with relevant common features. The choice should connect to factors that could influence treatment response, clinical practice or evidence interpretation. Administrative convenience alone may group populations that differ in important ways or split populations that share the same drivers. Keep this boundary visible: The map boundary used for operations is not automatically the right boundary for scientific analysis.
Try this
Build a global-local evidence card for regions should be scientifically meaningful. Justify each proposed region using disease, population, care, measurement and regulatory factors rather than geography alone. Add one intrinsic and one extrinsic factor that could alter the conclusion. Keep this limit: The map boundary used for operations is not automatically the right boundary for scientific analysis.
A strong science note separates observation from interpretation. Justify each proposed region using disease, population, care, measurement and regulatory factors rather than geography alone. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Justify each proposed region using disease, population, care, measurement and regulatory factors rather than geography alone. Keep this boundary visible: The map boundary used for operations is not automatically the right boundary for scientific analysis.
CHAPTER 5 OF 17 · Represent regions
5. Intrinsic factors can change exposure or response
Intrinsic factors include genetic variation, age, sex, body size, organ function, disease severity and other characteristics within participants. These can affect pharmacokinetics, pharmacodynamics, baseline risk or outcome measurement. Relevant factors should be identified during planning and collected consistently so observed variation can be investigated.
For intrinsic factors can change exposure or response, geography is only the beginning; mechanism, care and measurement explain why results might travel. Intrinsic factors include genetic variation, age, sex, body size, organ function, disease severity and other characteristics within participants. These can affect pharmacokinetics, pharmacodynamics, baseline risk or outcome measurement. Relevant factors should be identified during planning and collected consistently so observed variation can be investigated. Preserve this guardrail: Broad ethnic labels are weak substitutes for the biological or clinical factor actually expected to matter.
Try this
Run a representation audit for intrinsic factors can change exposure or response. Name the hypothesised mechanism for each intrinsic factor and the measurement needed to test it. Mark the region or subpopulation with the weakest information. Respect this boundary: Broad ethnic labels are weak substitutes for the biological or clinical factor actually expected to matter.
A strong science note separates observation from interpretation. Name the hypothesised mechanism for each intrinsic factor and the measurement needed to test it. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Name the hypothesised mechanism for each intrinsic factor and the measurement needed to test it. Keep this boundary visible: Broad ethnic labels are weak substitutes for the biological or clinical factor actually expected to matter.
CHAPTER 6 OF 17 · Represent regions
6. Extrinsic factors travel through healthcare and culture
Extrinsic factors include medical practice, concomitant treatments, diet, environment, adherence, healthcare access, diagnostic criteria and social or cultural conditions. They may affect who enters a study, how treatment is delivered and how outcomes are recorded. Standardisation can reduce avoidable variation, while excessive standardisation may weaken real-world generalizability.
A good MRCT reading of extrinsic factors travel through healthcare and culture holds the overall effect and regional context at the same time. Extrinsic factors include medical practice, concomitant treatments, diet, environment, adherence, healthcare access, diagnostic criteria and social or cultural conditions. They may affect who enters a study, how treatment is delivered and how outcomes are recorded. Standardisation can reduce avoidable variation, while excessive standardisation may weaken real-world generalizability. The inference stops here: Operational uniformity can hide whether the result will transfer to ordinary regional practice.
Try this
Explain extrinsic factors travel through healthcare and culture using map, mechanism and uncertainty. Map the care pathway in each region and flag differences that could modify treatment effect or data quality. Let a reviewer from another region challenge the transfer. Preserve this caution: Operational uniformity can hide whether the result will transfer to ordinary regional practice.
A strong science note separates observation from interpretation. Map the care pathway in each region and flag differences that could modify treatment effect or data quality. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Map the care pathway in each region and flag differences that could modify treatment effect or data quality. Keep this boundary visible: Operational uniformity can hide whether the result will transfer to ordinary regional practice.
CHAPTER 7 OF 17 · Represent regions
7. Regional experts should shape the protocol early
E17 recommends consulting investigators and experts representing participating regions during planning. They can identify differences in disease patterns, care, feasibility, language, outcome meaning and recruitment. Patient and site input can reveal burdens or exclusions that a central design team may miss.
The global question in regional experts should shape the protocol early becomes credible when regional differences are planned rather than discovered with surprise. E17 recommends consulting investigators and experts representing participating regions during planning. They can identify differences in disease patterns, care, feasibility, language, outcome meaning and recruitment. Patient and site input can reveal burdens or exclusions that a central design team may miss. Keep this boundary visible: Consultation adds value only when decisions and unresolved disagreements are recorded.
Try this
Build a global-local evidence card for regional experts should shape the protocol early. Create a regional input log showing evidence raised, protocol response, owner and remaining uncertainty. Add one intrinsic and one extrinsic factor that could alter the conclusion. Keep this limit: Consultation adds value only when decisions and unresolved disagreements are recorded.
A strong science note separates observation from interpretation. Create a regional input log showing evidence raised, protocol response, owner and remaining uncertainty. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Create a regional input log showing evidence raised, protocol response, owner and remaining uncertainty. Keep this boundary visible: Consultation adds value only when decisions and unresolved disagreements are recorded.
CHAPTER 8 OF 17 · Represent regions
8. Eligibility should support the target population
Common eligibility criteria help align participants across regions, yet criteria should not exclude important users merely to create an artificially homogeneous sample. Disease definitions, prior treatment and diagnostic procedures need consistent interpretation. Training and adjudication can reduce variation without erasing clinically relevant diversity.
For eligibility should support the target population, geography is only the beginning; mechanism, care and measurement explain why results might travel. Common eligibility criteria help align participants across regions, yet criteria should not exclude important users merely to create an artificially homogeneous sample. Disease definitions, prior treatment and diagnostic procedures need consistent interpretation. Training and adjudication can reduce variation without erasing clinically relevant diversity. Preserve this guardrail: A globally uniform sample can still be unrepresentative of patients who will use the medicine.
Try this
Run a representation audit for eligibility should support the target population. Compare the target population with enrolled participants in each region and record the largest representation gap. Mark the region or subpopulation with the weakest information. Respect this boundary: A globally uniform sample can still be unrepresentative of patients who will use the medicine.
A strong science note separates observation from interpretation. Compare the target population with enrolled participants in each region and record the largest representation gap. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Compare the target population with enrolled participants in each region and record the largest representation gap. Keep this boundary visible: A globally uniform sample can still be unrepresentative of patients who will use the medicine.
The same endpoint label can be measured differently across languages, healthcare systems or laboratory networks. Instruments may require translation and cultural adaptation; event definitions, timing and adjudication should be consistent. Central laboratories or review committees may reduce variation, but their procedures and remaining local influences should be documented.
A good MRCT reading of endpoints need common meaning and measurement holds the overall effect and regional context at the same time. The same endpoint label can be measured differently across languages, healthcare systems or laboratory networks. Instruments may require translation and cultural adaptation; event definitions, timing and adjudication should be consistent. Central laboratories or review committees may reduce variation, but their procedures and remaining local influences should be documented. The inference stops here: Literal translation does not guarantee conceptual equivalence or equal measurement performance.
Try this
Explain endpoints need common meaning and measurement using map, mechanism and uncertainty. Build an endpoint equivalence table covering concept, language, instrument, timing, training, adjudication and validation. Let a reviewer from another region challenge the transfer. Preserve this caution: Literal translation does not guarantee conceptual equivalence or equal measurement performance.
A strong science note separates observation from interpretation. Build an endpoint equivalence table covering concept, language, instrument, timing, training, adjudication and validation. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Build an endpoint equivalence table covering concept, language, instrument, timing, training, adjudication and validation. Keep this boundary visible: Literal translation does not guarantee conceptual equivalence or equal measurement performance.
CHAPTER 10 OF 17 · Analyse variation
10. Treatment and comparator choices need regional relevance
Dose, regimen and comparator should make sense across the intended regions. Differences in standard care, approved therapies, metabolism or prescribing can complicate one global protocol. The plan should explain how background treatment is controlled and whether the comparison remains ethical and interpretable everywhere.
The global question in treatment and comparator choices need regional relevance becomes credible when regional differences are planned rather than discovered with surprise. Dose, regimen and comparator should make sense across the intended regions. Differences in standard care, approved therapies, metabolism or prescribing can complicate one global protocol. The plan should explain how background treatment is controlled and whether the comparison remains ethical and interpretable everywhere. Keep this boundary visible: A comparator accepted in one region may be unavailable, outdated or inappropriate in another.
Try this
Build a global-local evidence card for treatment and comparator choices need regional relevance. Audit standard care and comparator relevance by region, then document the harmonised rule and any exceptions. Add one intrinsic and one extrinsic factor that could alter the conclusion. Keep this limit: A comparator accepted in one region may be unavailable, outdated or inappropriate in another.
A strong science note separates observation from interpretation. Audit standard care and comparator relevance by region, then document the harmonised rule and any exceptions. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Audit standard care and comparator relevance by region, then document the harmonised rule and any exceptions. Keep this boundary visible: A comparator accepted in one region may be unavailable, outdated or inappropriate in another.
CHAPTER 11 OF 17 · Analyse variation
11. Recruitment should be monitored while the trial runs
Planned representation can drift when some regions recruit faster than others. Central monitoring should track enrolment, baseline characteristics, protocol adherence, outcomes and important intrinsic or extrinsic factors. Corrective action may be needed before one region dominates or key subpopulations become too small for useful evaluation.
For recruitment should be monitored while the trial runs, geography is only the beginning; mechanism, care and measurement explain why results might travel. Planned representation can drift when some regions recruit faster than others. Central monitoring should track enrolment, baseline characteristics, protocol adherence, outcomes and important intrinsic or extrinsic factors. Corrective action may be needed before one region dominates or key subpopulations become too small for useful evaluation. Preserve this guardrail: Rapid total recruitment is not success if the final sample cannot support the intended regional interpretation.
Try this
Run a representation audit for recruitment should be monitored while the trial runs. Design a recruitment dashboard with representation targets, drift signals, escalation thresholds and accountable owners. Mark the region or subpopulation with the weakest information. Respect this boundary: Rapid total recruitment is not success if the final sample cannot support the intended regional interpretation.
A strong science note separates observation from interpretation. Design a recruitment dashboard with representation targets, drift signals, escalation thresholds and accountable owners. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Design a recruitment dashboard with representation targets, drift signals, escalation thresholds and accountable owners. Keep this boundary visible: Rapid total recruitment is not success if the final sample cannot support the intended regional interpretation.
E17 says regional allocation should support consistency evaluation and regulatory decisions while remaining feasible. Disease prevalence, regional size, accrual, relevant factors and logistics all matter. There is no universally optimal formula. Arbitrary fixed minimums or separate local-significance targets can inflate sample size without improving the global question.
A good MRCT reading of sample allocation needs a scientific basis holds the overall effect and regional context at the same time. E17 says regional allocation should support consistency evaluation and regulatory decisions while remaining feasible. Disease prevalence, regional size, accrual, relevant factors and logistics all matter. There is no universally optimal formula. Arbitrary fixed minimums or separate local-significance targets can inflate sample size without improving the global question. The inference stops here: Equal allocation is not automatically fair or informative, and proportional allocation may leave small regions poorly characterised.
Try this
Explain sample allocation needs a scientific basis using map, mechanism and uncertainty. Compare proportional, equal and balanced allocation strategies using feasibility and precision for the planned regional questions. Let a reviewer from another region challenge the transfer. Preserve this caution: Equal allocation is not automatically fair or informative, and proportional allocation may leave small regions poorly characterised.
A strong science note separates observation from interpretation. Compare proportional, equal and balanced allocation strategies using feasibility and precision for the planned regional questions. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Compare proportional, equal and balanced allocation strategies using feasibility and precision for the planned regional questions. Keep this boundary visible: Equal allocation is not automatically fair or informative, and proportional allocation may leave small regions poorly characterised.
CHAPTER 13 OF 17 · Translate responsibly
13. Prespecified pooling can connect similar populations
Regions or subpopulations may be pooled in advance when relevant intrinsic or extrinsic factors make them sufficiently similar. Pooling can improve precision and flexibility, but the rationale, membership and analysis should be prespecified. Cross-regional subpopulations defined by genotype or another causal factor may sometimes be more meaningful than geography.
The global question in prespecified pooling can connect similar populations becomes credible when regional differences are planned rather than discovered with surprise. Regions or subpopulations may be pooled in advance when relevant intrinsic or extrinsic factors make them sufficiently similar. Pooling can improve precision and flexibility, but the rationale, membership and analysis should be prespecified. Cross-regional subpopulations defined by genotype or another causal factor may sometimes be more meaningful than geography. Keep this boundary visible: Post hoc pooling chosen after viewing effects can manufacture apparent consistency.
Try this
Build a global-local evidence card for prespecified pooling can connect similar populations. Write the pooling rule, supporting evidence and a sensitivity analysis for plausible alternative groupings. Add one intrinsic and one extrinsic factor that could alter the conclusion. Keep this limit: Post hoc pooling chosen after viewing effects can manufacture apparent consistency.
A strong science note separates observation from interpretation. Write the pooling rule, supporting evidence and a sensitivity analysis for plausible alternative groupings. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Write the pooling rule, supporting evidence and a sensitivity analysis for plausible alternative groupings. Keep this boundary visible: Post hoc pooling chosen after viewing effects can manufacture apparent consistency.
CHAPTER 14 OF 17 · Translate responsibly
14. One primary analysis should integrate the full trial
E17’s standard is a single primary analysis approach specified in the protocol and agreed before the trial. For most MRCTs it uses data across all included regions and subpopulations to test and estimate the overall treatment effect. Stratification or prespecified pooling should be reflected appropriately, and the plan should be final before unblinding.
For one primary analysis should integrate the full trial, geography is only the beginning; mechanism, care and measurement explain why results might travel. E17’s standard is a single primary analysis approach specified in the protocol and agreed before the trial. For most MRCTs it uses data across all included regions and subpopulations to test and estimate the overall treatment effect. Stratification or prespecified pooling should be reflected appropriately, and the plan should be final before unblinding. Preserve this guardrail: An overall estimate can be precise while masking a clinically important regional difference.
Try this
Run a representation audit for one primary analysis should integrate the full trial. State the target population, estimand, endpoint, regional adjustment and summary measure in one primary-analysis receipt. Mark the region or subpopulation with the weakest information. Respect this boundary: An overall estimate can be precise while masking a clinically important regional difference.
A strong science note separates observation from interpretation. State the target population, estimand, endpoint, regional adjustment and summary measure in one primary-analysis receipt. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
State the target population, estimand, endpoint, regional adjustment and summary measure in one primary-analysis receipt. Keep this boundary visible: An overall estimate can be precise while masking a clinically important regional difference.
CHAPTER 15 OF 17 · Translate responsibly
15. Consistency is a clinical and statistical judgement
The trial should evaluate whether treatment effects show clinically relevant differences across regions. Regional estimates, intervals and forest plots can reveal patterns, but small regional samples create noise. Interpretation should consider biological plausibility, internal patterns across endpoints, external evidence, clinical importance and statistical uncertainty together.
A good MRCT reading of consistency is a clinical and statistical judgement holds the overall effect and regional context at the same time. The trial should evaluate whether treatment effects show clinically relevant differences across regions. Regional estimates, intervals and forest plots can reveal patterns, but small regional samples create noise. Interpretation should consider biological plausibility, internal patterns across endpoints, external evidence, clinical importance and statistical uncertainty together. The inference stops here: A non-significant interaction does not prove identical effects, and an extreme small-region estimate may be chance.
Try this
Explain consistency is a clinical and statistical judgement using map, mechanism and uncertainty. Review a hypothetical forest plot and separate precision, direction, clinical importance and plausible explanation for every region. Let a reviewer from another region challenge the transfer. Preserve this caution: A non-significant interaction does not prove identical effects, and an extreme small-region estimate may be chance.
A strong science note separates observation from interpretation. Review a hypothetical forest plot and separate precision, direction, clinical importance and plausible explanation for every region. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Review a hypothetical forest plot and separate precision, direction, clinical importance and plausible explanation for every region. Keep this boundary visible: A non-significant interaction does not prove identical effects, and an extreme small-region estimate may be chance.
CHAPTER 16 OF 17 · Translate responsibly
16. Generalizability depends on people and settings represented
The overall result may support participating and nonparticipating settings when relevant populations, care pathways, exposure and outcome mechanisms are adequately represented. Extrapolation becomes less certain when a region has different disease patterns, genetics, standard care or access. The report should explain which evidence supports transfer and what remains unknown.
The global question in generalizability depends on people and settings represented becomes credible when regional differences are planned rather than discovered with surprise. The overall result may support participating and nonparticipating settings when relevant populations, care pathways, exposure and outcome mechanisms are adequately represented. Extrapolation becomes less certain when a region has different disease patterns, genetics, standard care or access. The report should explain which evidence supports transfer and what remains unknown. Keep this boundary visible: Global enrolment does not guarantee universal applicability.
Try this
Build a global-local evidence card for generalizability depends on people and settings represented. Build a transfer table for a new country covering population, exposure, care, outcome measurement and missing evidence. Add one intrinsic and one extrinsic factor that could alter the conclusion. Keep this limit: Global enrolment does not guarantee universal applicability.
A strong science note separates observation from interpretation. Build a transfer table for a new country covering population, exposure, care, outcome measurement and missing evidence. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Build a transfer table for a new country covering population, exposure, care, outcome measurement and missing evidence. Keep this boundary visible: Global enrolment does not guarantee universal applicability.
CHAPTER 17 OF 17 · Translate responsibly
17. A complete MRCT receipt preserves global and local meaning
The receipt records development role, participating regions, regional definitions, expert input, intrinsic and extrinsic factors, eligibility, endpoints, comparator, allocation, recruitment, pooling, protocol deviations, primary analysis, regional estimates, consistency assessment, generalizability and regulator discussions. It lets reviewers see how one trial became evidence for several decisions.
For a complete mrct receipt preserves global and local meaning, geography is only the beginning; mechanism, care and measurement explain why results might travel. The receipt records development role, participating regions, regional definitions, expert input, intrinsic and extrinsic factors, eligibility, endpoints, comparator, allocation, recruitment, pooling, protocol deviations, primary analysis, regional estimates, consistency assessment, generalizability and regulator discussions. It lets reviewers see how one trial became evidence for several decisions. Preserve this guardrail: A detailed receipt cannot compensate for a region or population that the design failed to represent.
Try this
Run a representation audit for a complete mrct receipt preserves global and local meaning. Ask reviewers from two regions to read the receipt and identify where their regulatory or clinical conclusions could differ. Mark the region or subpopulation with the weakest information. Respect this boundary: A detailed receipt cannot compensate for a region or population that the design failed to represent.
A strong science note separates observation from interpretation. Ask reviewers from two regions to read the receipt and identify where their regulatory or clinical conclusions could differ. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Ask reviewers from two regions to read the receipt and identify where their regulatory or clinical conclusions could differ. Keep this boundary visible: A detailed receipt cannot compensate for a region or population that the design failed to represent.
E17 global-to-regional evidence map
| Question | Evidence to capture | Boundary to retain |
|---|---|---|
| What is shared globally? | Protocol, primary question, endpoints and analysis | Common does not mean identical context |
| What may vary? | Intrinsic factors, care, culture, exposure and measurement | Geography is a proxy, not a mechanism |
| Is representation adequate? | Allocation, recruitment, pooling and regional precision | Speed should not erase diversity |
| Will the result travel? | Consistency, plausibility, uncertainty and transfer evidence | Global evidence still has boundaries |
Science learning becomes useful when a student can show where a claim came from, why it applies and where it stops. Continue with the eduKate Science Learning Hub, compare choices in the Education Hub, and connect these habits to career pathways built from subject capabilities.
E17 offers an optimistic way to do global science: share one rigorous question, invite regional knowledge into the design and treat variation as information. The result can be faster development with a more honest account of who the evidence represents.
Primary sources and verification date
Source status checked 26 October 2026. FDA guidance describes the agency’s current thinking and does not replace applicable statutes, regulations or product-specific communication with FDA.
- FDA: ICH E17 General Principles for Multi-Regional Clinical Trials — Final guidance, July 2018; issued by CDER and CBER.
- FDA: Full ICH E17 guidance PDF — Official planning, allocation, pooling, analysis and consistency principles.
- FDA: ICH Guidance Documents — Current FDA index connecting E17 with related ICH guidance.