eduKateSG · Why Science?
Turn a finished clinical study into a report another scientist can inspect
Read FDA’s final ICH E3 guidance and 2013 Q&A through report structure, synopsis, protocol, analysis populations, efficacy, safety, deviations, narratives, tables and appendices.
A clinical study is not finished when the database is locked. Its value depends on whether another reader can understand what was planned, what happened, what changed and how the conclusions follow from the evidence.
ICH E3 provides a common structure for an integrated clinical and statistical study report. FDA’s January 2013 final Q&A clarifies that E3 is guidance, not a rigid template, and encourages adaptations that improve clear communication.
Did you know? The study synopsis should work as a stand-alone document. It compresses the design, population, methods, results and conclusions, but its brevity must not turn uncertainty or important safety information invisible.
For students, a clinical study report is the grown-up version of a science practical: question, method, observations, analysis, limitations and evidence are organised so a new reader can check the reasoning.
Chapter index
Chapters 1–9: purpose, synopsis, ethics, protocol and analysis planning
- 1. A CSR integrates clinical and statistical reporting
- 2. The July 1996 guidance remains final
- 3. The January 2013 Q&A clarifies flexible use
- 4. The title page and synopsis orient the reviewer
- 5. Ethics and administrative structure remain visible
- 6. The introduction explains why the study was done
- 7. Objectives and endpoints should match the protocol
- 8. Study design needs enough detail for interpretation
- 9. Participants and eligibility define the evidence population
Chapters 10–17: participant flow, efficacy, safety, tables, narratives, appendices and reviewability
- 10. Treatments and concomitant care need context
- 11. Statistical methods should be prespecified and reproducible
- 12. Participant flow reveals conduct quality
- 13. Efficacy results need estimates, uncertainty and context
- 14. Safety reporting needs exposure and participant detail
- 15. Tables, figures and listings support different questions
- 16. Appendices preserve essential supporting material
- 17. A complete CSR receipt makes conclusions reviewable
CHAPTER 1 OF 17 · Orient the reader
1. A CSR integrates clinical and statistical reporting
ICH E3 describes a full report of an individual therapeutic, prophylactic or diagnostic study in human participants. The aim is a document complete, free from ambiguity, well organised and easy to review, with methods and results presented together rather than split into disconnected narratives.
For a csr integrates clinical and statistical reporting, the report should help a sceptical reader retrace the study. ICH E3 describes a full report of an individual therapeutic, prophylactic or diagnostic study in human participants. The aim is a document complete, free from ambiguity, well organised and easy to review, with methods and results presented together rather than split into disconnected narratives. Keep this boundary visible: Integration should clarify reasoning, not hide discordant analyses or inconvenient observations.
Try this
Build a report crosswalk for a csr integrates clinical and statistical reporting. Choose one conclusion and trace it to the protocol objective, dataset, analysis, table and participant-level evidence. Add the source file, version and responsible author. Keep this limit: Integration should clarify reasoning, not hide discordant analyses or inconvenient observations.
A strong science note separates observation from interpretation. Choose one conclusion and trace it to the protocol objective, dataset, analysis, table and participant-level 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
Choose one conclusion and trace it to the protocol objective, dataset, analysis, table and participant-level evidence. Keep this boundary visible: Integration should clarify reasoning, not hide discordant analyses or inconvenient observations.
FDA identifies the original ICH E3 guidance as final, dated July 1996, under docket FDA-1995-D-0201. It was designed to facilitate one core report acceptable across ICH regions, with regional additions in appendices when required.
Good reporting of the july 1996 guidance remains final joins compression with auditability. FDA identifies the original ICH E3 guidance as final, dated July 1996, under docket FDA-1995-D-0201. It was designed to facilitate one core report acceptable across ICH regions, with regional additions in appendices when required. Preserve this guardrail: A long-standing final guidance should be used with current technical standards, regulations and later Q&A clarifications.
Try this
Audit the july 1996 guidance remains final for selective visibility. Create a source receipt with original title, final status, date, docket, purpose and later clarifications. Mark what is primary, supportive and exploratory. Respect this boundary: A long-standing final guidance should be used with current technical standards, regulations and later Q&A clarifications.
A strong science note separates observation from interpretation. Create a source receipt with original title, final status, date, docket, purpose and later clarifications. 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 original title, final status, date, docket, purpose and later clarifications. Keep this boundary visible: A long-standing final guidance should be used with current technical standards, regulations and later Q&A clarifications.
FDA’s final January 2013 E3 Q&A states that E3 is guidance rather than a rigid required template. Sections may be adapted, regrouped, added or removed when that improves display and communication for a particular study.
A reviewer tests the january 2013 q&a clarifies flexible use by moving from claim to method to participant-level evidence. FDA’s final January 2013 E3 Q&A states that E3 is guidance rather than a rigid required template. Sections may be adapted, regrouped, added or removed when that improves display and communication for a particular study. The conclusion stops here: Flexibility does not justify omitting relevant methods, results or limitations.
Try this
Explain the january 2013 q&a clarifies flexible use as if the original team is unavailable. Redesign the outline for an unusual study and explain where every relevant E3 topic remains accessible. Ask whether another analyst could reproduce the conclusion. Preserve this caution: Flexibility does not justify omitting relevant methods, results or limitations.
A strong science note separates observation from interpretation. Redesign the outline for an unusual study and explain where every relevant E3 topic remains accessible. 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
Redesign the outline for an unusual study and explain where every relevant E3 topic remains accessible. Keep this boundary visible: Flexibility does not justify omitting relevant methods, results or limitations.
CHAPTER 4 OF 17 · Orient the reader
4. The title page and synopsis orient the reviewer
Identification details, study title, sponsor, product, protocol number, investigators, dates and report version help control the record. The synopsis summarises objectives, design, participants, treatment, endpoints, statistics, results and conclusions so it can be interpreted on its own.
For the title page and synopsis orient the reviewer, the report should help a sceptical reader retrace the study. Identification details, study title, sponsor, product, protocol number, investigators, dates and report version help control the record. The synopsis summarises objectives, design, participants, treatment, endpoints, statistics, results and conclusions so it can be interpreted on its own. Keep this boundary visible: A stand-alone synopsis must not become a promotional abstract detached from the full results.
Try this
Build a report crosswalk for the title page and synopsis orient the reviewer. Compare every numerical claim in the synopsis with its table and analysis in the body. Add the source file, version and responsible author. Keep this limit: A stand-alone synopsis must not become a promotional abstract detached from the full results.
A strong science note separates observation from interpretation. Compare every numerical claim in the synopsis with its table and analysis in the body. 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 every numerical claim in the synopsis with its table and analysis in the body. Keep this boundary visible: A stand-alone synopsis must not become a promotional abstract detached from the full results.
CHAPTER 5 OF 17 · Explain the methods
5. Ethics and administrative structure remain visible
The report should identify investigators, centres, ethics review, informed consent and important administrative arrangements. This information helps reviewers understand accountability, study settings and whether participant protections were established.
Good reporting of ethics and administrative structure remain visible joins compression with auditability. The report should identify investigators, centres, ethics review, informed consent and important administrative arrangements. This information helps reviewers understand accountability, study settings and whether participant protections were established. Preserve this guardrail: Listing approvals does not prove that consent and protocol conduct were adequate at every site.
Try this
Audit ethics and administrative structure remain visible for selective visibility. Build an accountability map linking centres, investigators, committees, dates and responsibilities. Mark what is primary, supportive and exploratory. Respect this boundary: Listing approvals does not prove that consent and protocol conduct were adequate at every site.
A strong science note separates observation from interpretation. Build an accountability map linking centres, investigators, committees, dates and responsibilities. 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 accountability map linking centres, investigators, committees, dates and responsibilities. Keep this boundary visible: Listing approvals does not prove that consent and protocol conduct were adequate at every site.
CHAPTER 6 OF 17 · Explain the methods
6. The introduction explains why the study was done
Scientific background should connect prior evidence to the study objective without turning the report into a marketing narrative. The rationale should explain the disease, treatment, dose or strategy and the uncertainty the study was designed to reduce.
A reviewer tests the introduction explains why the study was done by moving from claim to method to participant-level evidence. Scientific background should connect prior evidence to the study objective without turning the report into a marketing narrative. The rationale should explain the disease, treatment, dose or strategy and the uncertainty the study was designed to reduce. The conclusion stops here: A strong rationale cannot compensate for a design incapable of answering the question.
Try this
Explain the introduction explains why the study was done as if the original team is unavailable. Write the evidence gap in one sentence and identify which planned comparison closes it. Ask whether another analyst could reproduce the conclusion. Preserve this caution: A strong rationale cannot compensate for a design incapable of answering the question.
A strong science note separates observation from interpretation. Write the evidence gap in one sentence and identify which planned comparison closes 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
Write the evidence gap in one sentence and identify which planned comparison closes it. Keep this boundary visible: A strong rationale cannot compensate for a design incapable of answering the question.
CHAPTER 7 OF 17 · Explain the methods
7. Objectives and endpoints should match the protocol
Primary and secondary objectives, endpoints and hypotheses should be stated as prespecified, with any changes dated and explained. Reviewers need to know which analyses were confirmatory, supportive or exploratory.
For objectives and endpoints should match the protocol, the report should help a sceptical reader retrace the study. Primary and secondary objectives, endpoints and hypotheses should be stated as prespecified, with any changes dated and explained. Reviewers need to know which analyses were confirmatory, supportive or exploratory. Keep this boundary visible: Relabelling an unsuccessful secondary or exploratory finding as primary after the study misrepresents the evidence.
Try this
Build a report crosswalk for objectives and endpoints should match the protocol. Create a protocol-to-report crosswalk for objectives, endpoints, timing and analysis status. Add the source file, version and responsible author. Keep this limit: Relabelling an unsuccessful secondary or exploratory finding as primary after the study misrepresents the evidence.
A strong science note separates observation from interpretation. Create a protocol-to-report crosswalk for objectives, endpoints, timing and analysis status. 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 protocol-to-report crosswalk for objectives, endpoints, timing and analysis status. Keep this boundary visible: Relabelling an unsuccessful secondary or exploratory finding as primary after the study misrepresents the evidence.
CHAPTER 8 OF 17 · Explain the methods
8. Study design needs enough detail for interpretation
The report explains design, control, randomisation, blinding, treatment, duration, visits and stopping rules. It should justify design choices and describe changes made during conduct. Diagrams can show periods, sequences and decision points efficiently.
Good reporting of study design needs enough detail for interpretation joins compression with auditability. The report explains design, control, randomisation, blinding, treatment, duration, visits and stopping rules. It should justify design choices and describe changes made during conduct. Diagrams can show periods, sequences and decision points efficiently. Preserve this guardrail: Naming a design does not reveal whether allocation, masking or follow-up worked as intended.
Try this
Audit study design needs enough detail for interpretation for selective visibility. Draw the study timeline and mark every point where bias could enter. Mark what is primary, supportive and exploratory. Respect this boundary: Naming a design does not reveal whether allocation, masking or follow-up worked as intended.
A strong science note separates observation from interpretation. Draw the study timeline and mark every point where bias could enter. 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 study timeline and mark every point where bias could enter. Keep this boundary visible: Naming a design does not reveal whether allocation, masking or follow-up worked as intended.
CHAPTER 9 OF 17 · Show the results
9. Participants and eligibility define the evidence population
Inclusion and exclusion criteria, recruitment, withdrawals and protocol deviations help reviewers judge who entered and who contributed data. Baseline characteristics should be shown for relevant analysis populations and treatment groups.
A reviewer tests participants and eligibility define the evidence population by moving from claim to method to participant-level evidence. Inclusion and exclusion criteria, recruitment, withdrawals and protocol deviations help reviewers judge who entered and who contributed data. Baseline characteristics should be shown for relevant analysis populations and treatment groups. The conclusion stops here: The enrolled sample should not be assumed to represent all future patients.
Try this
Explain participants and eligibility define the evidence population as if the original team is unavailable. Build a flow diagram from screened people to each analysis set and record reasons for loss. Ask whether another analyst could reproduce the conclusion. Preserve this caution: The enrolled sample should not be assumed to represent all future patients.
A strong science note separates observation from interpretation. Build a flow diagram from screened people to each analysis set and record reasons for loss. 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 flow diagram from screened people to each analysis set and record reasons for loss. Keep this boundary visible: The enrolled sample should not be assumed to represent all future patients.
The report describes investigational and control treatments, dose, route, schedule, compliance and relevant concomitant therapy. Rescue treatment and changes can alter interpretation and should be connected to endpoint and estimand decisions.
For treatments and concomitant care need context, the report should help a sceptical reader retrace the study. The report describes investigational and control treatments, dose, route, schedule, compliance and relevant concomitant therapy. Rescue treatment and changes can alter interpretation and should be connected to endpoint and estimand decisions. Keep this boundary visible: A treatment label is incomplete when exposure and accompanying care differ materially between groups.
Try this
Build a report crosswalk for treatments and concomitant care need context. Create an exposure table including assigned, received, duration, adherence, rescue and switching. Add the source file, version and responsible author. Keep this limit: A treatment label is incomplete when exposure and accompanying care differ materially between groups.
A strong science note separates observation from interpretation. Create an exposure table including assigned, received, duration, adherence, rescue and switching. 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 an exposure table including assigned, received, duration, adherence, rescue and switching. Keep this boundary visible: A treatment label is incomplete when exposure and accompanying care differ materially between groups.
CHAPTER 11 OF 17 · Show the results
11. Statistical methods should be prespecified and reproducible
Analysis populations, models, covariates, missing-data rules, multiplicity, interim analyses and sensitivity analyses should be described clearly. Changes from the plan and additional analyses should be labelled with reasons and timing.
Good reporting of statistical methods should be prespecified and reproducible joins compression with auditability. Analysis populations, models, covariates, missing-data rules, multiplicity, interim analyses and sensitivity analyses should be described clearly. Changes from the plan and additional analyses should be labelled with reasons and timing. Preserve this guardrail: A significant result is not self-validating when the model or endpoint was selected after unblinding.
Try this
Audit statistical methods should be prespecified and reproducible for selective visibility. Reconstruct the primary result using only the report, analysis plan and referenced dataset definitions. Mark what is primary, supportive and exploratory. Respect this boundary: A significant result is not self-validating when the model or endpoint was selected after unblinding.
A strong science note separates observation from interpretation. Reconstruct the primary result using only the report, analysis plan and referenced dataset definitions. 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
Reconstruct the primary result using only the report, analysis plan and referenced dataset definitions. Keep this boundary visible: A significant result is not self-validating when the model or endpoint was selected after unblinding.
Disposition tables and diagrams show enrolment, treatment exposure, completion, withdrawal and follow-up. Reasons and timing can expose differential loss or operational problems that affect efficacy and safety interpretation.
A reviewer tests participant flow reveals conduct quality by moving from claim to method to participant-level evidence. Disposition tables and diagrams show enrolment, treatment exposure, completion, withdrawal and follow-up. Reasons and timing can expose differential loss or operational problems that affect efficacy and safety interpretation. The conclusion stops here: Combining unlike withdrawal reasons can hide treatment-related patterns.
Try this
Explain participant flow reveals conduct quality as if the original team is unavailable. Compare withdrawal timing and reasons across groups and connect them to endpoint collection. Ask whether another analyst could reproduce the conclusion. Preserve this caution: Combining unlike withdrawal reasons can hide treatment-related patterns.
A strong science note separates observation from interpretation. Compare withdrawal timing and reasons across groups and connect them to endpoint collection. 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 withdrawal timing and reasons across groups and connect them to endpoint collection. Keep this boundary visible: Combining unlike withdrawal reasons can hide treatment-related patterns.
CHAPTER 13 OF 17 · Preserve the evidence
13. Efficacy results need estimates, uncertainty and context
The report presents primary and secondary analyses with effect estimates, confidence intervals or other uncertainty, relevant tests and sensitivity analyses. Subgroups and exploratory work should be distinguished from prespecified confirmatory evidence.
For efficacy results need estimates, uncertainty and context, the report should help a sceptical reader retrace the study. The report presents primary and secondary analyses with effect estimates, confidence intervals or other uncertainty, relevant tests and sensitivity analyses. Subgroups and exploratory work should be distinguished from prespecified confirmatory evidence. Keep this boundary visible: A small p-value does not by itself show clinical importance or robustness.
Try this
Build a report crosswalk for efficacy results need estimates, uncertainty and context. For each major result, record effect size, uncertainty, prespecification, missingness and clinical meaning. Add the source file, version and responsible author. Keep this limit: A small p-value does not by itself show clinical importance or robustness.
A strong science note separates observation from interpretation. For each major result, record effect size, uncertainty, prespecification, missingness and clinical meaning. 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
For each major result, record effect size, uncertainty, prespecification, missingness and clinical meaning. Keep this boundary visible: A small p-value does not by itself show clinical importance or robustness.
CHAPTER 14 OF 17 · Preserve the evidence
14. Safety reporting needs exposure and participant detail
Adverse events, serious events, deaths, discontinuations, laboratory changes, vital signs and other safety observations should be analysed with treatment exposure and timing in mind. Narratives may be needed for deaths, serious events or other clinically important cases.
Good reporting of safety reporting needs exposure and participant detail joins compression with auditability. Adverse events, serious events, deaths, discontinuations, laboratory changes, vital signs and other safety observations should be analysed with treatment exposure and timing in mind. Narratives may be needed for deaths, serious events or other clinically important cases. Preserve this guardrail: Counting events without exposure, severity, causality or participant context can mislead.
Try this
Audit safety reporting needs exposure and participant detail for selective visibility. Trace one serious event through source narrative, coding, table, analysis and conclusion. Mark what is primary, supportive and exploratory. Respect this boundary: Counting events without exposure, severity, causality or participant context can mislead.
A strong science note separates observation from interpretation. Trace one serious event through source narrative, coding, table, analysis and conclusion. 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
Trace one serious event through source narrative, coding, table, analysis and conclusion. Keep this boundary visible: Counting events without exposure, severity, causality or participant context can mislead.
CHAPTER 15 OF 17 · Preserve the evidence
15. Tables, figures and listings support different questions
Summary tables reveal patterns, figures show trajectories or time relationships, and listings preserve participant-level detail for review. Each should use clear populations, denominators, units and definitions and should link back to the text without unnecessary duplication.
A reviewer tests tables, figures and listings support different questions by moving from claim to method to participant-level evidence. Summary tables reveal patterns, figures show trajectories or time relationships, and listings preserve participant-level detail for review. Each should use clear populations, denominators, units and definitions and should link back to the text without unnecessary duplication. The conclusion stops here: A dense appendix is not transparency when the reader cannot connect it to claims in the report.
Try this
Explain tables, figures and listings support different questions as if the original team is unavailable. Select one claim and identify the best text, table, figure and listing evidence for it. Ask whether another analyst could reproduce the conclusion. Preserve this caution: A dense appendix is not transparency when the reader cannot connect it to claims in the report.
A strong science note separates observation from interpretation. Select one claim and identify the best text, table, figure and listing evidence for 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
Select one claim and identify the best text, table, figure and listing evidence for it. Keep this boundary visible: A dense appendix is not transparency when the reader cannot connect it to claims in the report.
CHAPTER 16 OF 17 · Preserve the evidence
16. Appendices preserve essential supporting material
Appendices may include protocol and amendments, sample case-report forms, investigator information, ethics documents, technical statistical material, data listings and other supporting records. Regional or electronic-submission conventions may determine placement.
For appendices preserve essential supporting material, the report should help a sceptical reader retrace the study. Appendices may include protocol and amendments, sample case-report forms, investigator information, ethics documents, technical statistical material, data listings and other supporting records. Regional or electronic-submission conventions may determine placement. Keep this boundary visible: Moving evidence to an appendix should not make a key method or limitation invisible in the main report.
Try this
Build a report crosswalk for appendices preserve essential supporting material. Create an appendix index with purpose, file name, version, owner and cross-reference from the report body. Add the source file, version and responsible author. Keep this limit: Moving evidence to an appendix should not make a key method or limitation invisible in the main report.
A strong science note separates observation from interpretation. Create an appendix index with purpose, file name, version, owner and cross-reference from the report body. 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 an appendix index with purpose, file name, version, owner and cross-reference from the report body. Keep this boundary visible: Moving evidence to an appendix should not make a key method or limitation invisible in the main report.
CHAPTER 17 OF 17 · Preserve the evidence
17. A complete CSR receipt makes conclusions reviewable
The receipt records report version, protocol, amendments, analysis plan, data cut, populations, methods, deviations, participant flow, efficacy, safety, sensitivity work, narratives, tables, listings, appendices and approvals. It lets a reviewer verify the path from plan to conclusion.
Good reporting of a complete csr receipt makes conclusions reviewable joins compression with auditability. The receipt records report version, protocol, amendments, analysis plan, data cut, populations, methods, deviations, participant flow, efficacy, safety, sensitivity work, narratives, tables, listings, appendices and approvals. It lets a reviewer verify the path from plan to conclusion. Preserve this guardrail: A beautifully organised report cannot transform a biased or underpowered study into reliable evidence.
Try this
Audit a complete csr receipt makes conclusions reviewable for selective visibility. Ask an independent reader to find one unexpected result, its source evidence and the limitation attached to it. Mark what is primary, supportive and exploratory. Respect this boundary: A beautifully organised report cannot transform a biased or underpowered study into reliable evidence.
A strong science note separates observation from interpretation. Ask an independent reader to find one unexpected result, its source evidence and the limitation attached to 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
Ask an independent reader to find one unexpected result, its source evidence and the limitation attached to it. Keep this boundary visible: A beautifully organised report cannot transform a biased or underpowered study into reliable evidence.
Clinical study report map
| Question | Evidence to capture | Boundary to retain |
|---|---|---|
| What was planned? | Protocol, objectives, endpoints and analysis plan | Later changes need dates and reasons |
| What happened? | Sites, participant flow, exposure, deviations and data | Disposition can reveal bias |
| What was found? | Estimates, uncertainty, efficacy, safety and sensitivity | Significance is not clinical importance |
| Can it be reviewed? | Tables, figures, listings, narratives and appendices | Organisation cannot rescue weak design |
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.
A strong clinical study report is generous to the future. It gives reviewers, clinicians and later scientists enough structure and evidence to understand the study without asking them to trust the original team’s memory.
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 E3 Structure and Content of Clinical Study Reports — Final guidance, July 1996; docket FDA-1995-D-0201.
- FDA: Full ICH E3 guidance PDF — Official recommended structure and content for integrated clinical study reports.
- FDA: ICH E3 Questions and Answers (R1) — Final Q&A, January 2013; clarifies flexible rather than rigid use.