CSR Writing Template for Sponsors: ICH E3 Structure, Best Practices, and What to Prepare Before Day One

Every NDA, BLA, and MAA rests on clinical study reports. Regulators at the FDA, EMA, and PMDA do not simply accept a sponsor's conclusions; they work through the CSR to verify them independently, checking that stated endpoints match the protocol, that statistical methods in the report align with the pre-specified SAP, and that adverse event narratives support the safety conclusions in the regulatory summary. A Phase III CSR can run to several hundred pages, and for large pivotal trials with extensive safety data and patient listings, the full document package including appendices can exceed 1,000 pages ^[1]. Getting the structure wrong, or starting the writing process without the right materials assembled, pushes submission dates back and extends the period before regulatory approval, with schedule and cost implications for the sponsoring organization ^[2].

This guide covers the CSR writing template sponsors should use, grounded in ICH E3 ^[3] and supplemented by the TransCelerate Clinical Template Suite ^[4] and CORE Reference ^[5]. It is written for sponsor regulatory and clinical operations teams preparing to commission or internally produce a submission-ready CSR.

Why CSR Structure Matters More Than Most Sponsors Realize

The International Council for Harmonisation finalized ICH E3, "Structure and Content of Clinical Study Reports," in November 1995, and it remains the operative global standard today ^[3]. ICH E3 is widely used across ICH-aligned regulatory submissions globally; its stated purpose is to allow a single core clinical study report to satisfy all regulatory authorities in the ICH regions without reformatting for each jurisdiction ^[3].

What is often misunderstood is ICH E3's design philosophy. The ICH E3 Q&As (R1) explicitly state that the guideline is "not a set of rigid requirements or a template" and that modifications that lead to better display and communication of information are encouraged ^[6]. Sponsors are not locked into a fixed sequence; they are required to cover all relevant topics and to present them with clarity. In practice, that flexibility is narrower than it sounds, because regulatory reviewers have come to expect the standard section numbering. Departing from it requires a documented rationale.

The ICH E3 Section Map: A Quick Reference for Sponsors

Before working through each section, it helps to have the full official numbering in view. The table below maps every ICH E3 section to its name and the primary sponsor inputs that section requires ^[3].

Common ICH E3 numbering mistakes to avoid:

• Introduction is Section 7, not Section 1.
• Efficacy Evaluation is Section 11, not Section 5 or Section 8.
• Patient narratives belong at Section 12.3.2 (body text) and Section 14.3.3 (post-text). Section 16.2.3 is the listing of patients excluded from the efficacy analysis.
• Sections 16.2.4 to 16.2.8 are patient data listings (demographics, compliance, efficacy response, adverse events, lab measurements). Case report forms for individual patients are Section 16.3.
• Section 15 is the Reference List. Appendices are Section 16.

The ICH E3 Template: A Section-by-Section Breakdown

Front Matter: Sections 1 to 6

Section 1 (Title Page) identifies the investigational product, protocol number, sponsor's name, and report date. The protocol number must match the eCTD backbone exactly; a mismatch creates a cross-reference error during filing review ^[7].

Section 2 (Synopsis) is typically two to three pages and functions as a standalone summary of objectives, design, population, endpoints, and principal results. The ICH E3 Q&As (R1) note that for complex or large pivotal studies, a longer synopsis is acceptable; the 10-page example in ICH M4E should not be exceeded considerably ^[6]. In regulatory submissions with multiple clinical study reports, the synopsis is frequently the first section a reviewer reads. Cross-references to other CSR sections should be avoided within it; the synopsis is intended to stand independently ^[6].

A common sponsor error is leaving the synopsis as the final writing task. It should be drafted in parallel with the body sections, then finalized after all results are confirmed, not written from scratch at the end of an already compressed timeline.

Section 3 (Table of Contents) must map every section, appendix, table, and figure with accurate page numbers. In multi-volume CSRs submitted in eCTD format, the table of contents also needs to function within the electronic navigation structure required by FDA ^[7].

Sections 4 to 6 (Abbreviations, Ethics, Investigators) are frequently treated as administrative formalities, but each carries its own risk. Section 5 (Ethics) must confirm IEC or IRB approval was obtained before study initiation and that informed consent procedures were consistent with applicable GCP requirements and the ICH E6 version and regional requirements in effect for the study ^[8]. Section 6 (Investigators and Study Administrative Structure) must list all investigators along with brief CVs or equivalent summaries of training and experience relevant to the clinical study ^[3].

Section 7: Introduction

This section explains the clinical and scientific rationale for the study. It covers the pharmacological background of the investigational product, the unmet medical need addressed, and the regulatory context of this particular study within the broader development program. The introduction must be factual rather than promotional and must reflect the state of knowledge at the time the study was conducted, not at the time of writing.

Section 8: Study Objectives

Primary and secondary objectives are stated precisely as they appear in the approved protocol. Any wording divergence between Section 8 of the CSR and the protocol is a signal to reviewers that something changed mid-study, which triggers scrutiny of the protocol deviation log and amendment history.

Section 9: Investigational Plan

This is one of the most structurally detailed sections in the CSR. ICH E3 divides it into eight subsections ^[3]:

9.1 Overall study design and plan description (parallel group, crossover, adaptive, etc.)
9.2 Discussion of study design, including choice of control group
9.3 Selection of study population (inclusion criteria, exclusion criteria, removal from therapy)
9.4 Treatments (doses administered, IP identity, randomization, blinding, prior and concomitant therapy, compliance)
9.5 Efficacy and safety variables (the assessment schedule, primary efficacy variable, PK measurements where applicable)
9.6 Data quality assurance
9.7 Statistical methods planned in the protocol and sample size determination
9.8 Changes in the conduct of the study or planned analyses

The SAP must be finalized and locked before database lock to preserve the pre-specified nature of all primary and secondary analyses ^[9]. Section 9.8 is where the CSR documents any deviations between the protocol's planned analyses and what was actually carried out, with justifications. Any analysis conducted after database lock that was not pre-specified must be clearly identified as post-hoc in the CSR ^[9].

Section 10: Study Patients

This section documents the number of patients enrolled, randomized, treated, and completing the trial, plus those who withdrew and the reasons. The disposition table (typically numbered 14.1.x in the post-text tables) is among the first things a reviewer checks to assess dropout patterns and how they might affect the primary analysis ^[1]. Consistency between the disposition numbers in the body text and the corresponding table is mandatory; discrepancies between body text and tables are a common source of reviewer queries; cross-checking every data-dependent statement against the corresponding table or listing before submission is standard practice ^[12].

Section 11: Efficacy Evaluation

The efficacy section reports primary and secondary endpoint results in the sequence specified in the SAP. Primary endpoints come first. The narrative must mirror the logic of the statistical analysis: primary results, then secondary outcomes, then subgroup and exploratory analyses, each with clear labeling of pre-specified versus post-hoc status ^[10].

For each endpoint, the section should include the estimated treatment effect, confidence intervals, the analysis population used (ITT, PP, mITT), and any sensitivity analyses conducted. P-values are reported where the SAP specifies hypothesis testing; not every endpoint context requires formal significance testing. Subgroup analyses must be clearly labeled as pre-specified or exploratory; unqualified subgroup claims are among the most frequent sources of regulatory information requests.

Section 12: Safety Evaluation

The safety section is typically the most scrutinized component of the CSR. ICH E3 structures it across six subsections ^[3]:

12.1 Extent of Exposure: Number of patients, treatment duration by group, total patient-years of exposure.

12.2 Adverse Events: Brief summary; tabular display by MedDRA system organ class and preferred term; analysis by causality, severity, and action taken; individual patient listings.

12.3 Deaths, Other Serious Adverse Events, and Other Significant Adverse Events: This subsection requires particular attention to its two reporting formats:

• Section 12.3.1 carries the listings of deaths, other serious adverse events, and other significant adverse events, organized by patient.
• Section 12.3.2 carries the individual patient narratives for deaths, serious adverse events, and certain other significant adverse events ^[3]. These narratives typically run one to four pages each, depending on event complexity and the volume of data available for the patient ^[11]. In a large Phase III safety analysis, the number of required narratives can reach into the hundreds ^[11]. These same narratives are mirrored in the post-text tables section under 14.3.3 ^[3].

12.4 Clinical Laboratory Evaluation: Individual lab measurements by patient (which feed into 16.2.8) and each abnormal value (which feeds into 14.3.4).

12.5 Vital Signs, Physical Findings, and Other Observations Related to Safety.

12.6 Safety Conclusions.

Section 13: Discussion and Overall Conclusions

The discussion integrates efficacy and safety findings into a benefit-risk assessment in the context of the existing evidence base for the therapeutic area. This section is narrative-heavy and requires the medical writer to understand the indication deeply enough to position the trial results against prior literature. Conclusions must follow directly from the data reported in the CSR body; conclusions not grounded in what Sections 11 and 12 actually report are a significant source of regulatory review questions.

Section 14: Post-Text Tables, Figures, and Graphs

Under ICH E3, Section 14 contains the post-text data displays, numbered systematically ^[3]:

• 14.1.x: Demographic data (population summary, baseline characteristics, disposition)
• 14.2.x: Efficacy data (primary and secondary endpoint tables, figures, Kaplan-Meier curves where applicable)
• 14.3.x: Safety data (14.3.1 adverse event displays; 14.3.2 listings of deaths and SAEs; 14.3.3 narratives of deaths, serious adverse events, and certain other significant adverse events; 14.3.4 abnormal laboratory value listings)

These output shells are generated by the biostatistics team and must be QC-checked by the medical writer against the source TLF package before the CSR is finalized ^[12]. A complete cross-check of every number in the body text against the corresponding table or listing is standard practice for submission-ready CSRs ^[12].

Section 15: Reference List

Section 15 carries the complete bibliography for all sources cited in the CSR body. References should be formatted consistently and must be accessible to regulatory reviewers. Sources cited only in the appendices are typically listed separately within those appendices.

Section 16: Appendices

The appendices carry supporting materials that are required to be available but are not always submitted in full to every authority. Standard subsections under ICH E3 Section 16 include ^[3]:

16.1 Study Information:

• 16.1.1 Protocol and protocol amendments
• 16.1.2 Sample case report forms (unique pages only)
• 16.1.3 List of IECs or IRBs and representative consent form documentation
• 16.1.4 List and description of investigators with brief CVs
• 16.1.5 Signatures of principal or coordinating investigator(s) or sponsor's responsible medical officer
• 16.1.6 Listing of patients receiving investigational product from specific batches, where more than one batch was used
• 16.1.9 Documentation of statistical methods
• 16.1.11 Publications based on the study

16.2 Patient Data Listings (individual-level data by subject):

• 16.2.1 Discontinued patients
• 16.2.2 Protocol deviations
• 16.2.3 Patients excluded from the efficacy analysis (this is a patient exclusion list, not narratives; narratives appear at Sections 12.3.2 and 14.3.3)
• 16.2.4 Demographic data by patient
• 16.2.5 Compliance and drug concentration data by patient (where applicable)
• 16.2.6 Individual efficacy response data by patient
• 16.2.7 Adverse event listings by patient
• 16.2.8 Listing of individual laboratory measurements by patient (cross-referenced from Section 12.4.1 of the CSR body) ^[3]

16.3 Case Report Forms for patients who died during the study, discontinued due to adverse events, or for whom case report forms are otherwise required by the relevant regulatory authority ^[3]

16.4 Individual Patient Data Listings (US Archival Listings): a US FDA-specific requirement for individual patient data in standardized archival formats submitted as part of NDA and BLA applications ^[3]. This section is not required for submissions to the EMA or PMDA; sponsors preparing US NDA or BLA applications should confirm 16.4 requirements with their regulatory team before finalizing the document package.

Different regulatory authorities have different requirements for which appendices must be physically included in the submission versus merely held available for inspection. Sponsors preparing simultaneous FDA and EMA submissions should confirm these requirements separately for each dossier before the submission plan is finalized. Submission expectations for individual appendix subsections can also vary by study phase, therapeutic area, and whether the study is pivotal or supportive.

Industry Templates: TransCelerate and CORE Reference

Two non-regulatory resources have become standard reference points for CSR authoring teams.

TransCelerate BioPharma's Clinical Template Suite includes templates for the Common Protocol Template (CPT), Statistical Analysis Plan (SAP), and Clinical Study Report (CSR). Released in November 2018, the CSR template received its final update in the 2024 CTS Release ^[4]. TransCelerate engaged with regulatory authorities and aligned the templates with ICH guidelines throughout the project's lifecycle, ultimately aligning with ICH M11 guidance in 2026 ^[4]. The eCSR template is a structured MS Word document with metadata and content-reuse capabilities: shared definitions entered in the CPT can carry forward into the SAP and CSR automatically, reducing the risk of inconsistent terminology across documents. The 2024 release was confirmed as the final supported version of the CTS eTemplates ^[4].

CORE Reference (Clarity and Openness in Reporting: E3-based), released in 2016 by the European Medical Writers Association (EMWA) and the American Medical Writers Association (AMWA), is an open-access authoring guide registered with the EQUATOR Network ^[5]. A peer-reviewed critical review published in 2019 found that opportunities remained to refine the TransCelerate CSR Template's structure and instructional text, enhance content clarity, and improve transparency for the broad readership of CSRs ^[5]. CORE Reference Version 2 incorporated the estimand framework introduced by ICH E9(R1) as a defined term ^[5].

Medical writers working on CSRs typically use both. The TransCelerate template provides structural scaffolding; CORE Reference provides detailed authoring guidance on what each section must contain and how to express it accurately.

What Sponsors Must Have Ready Before Writing Begins

One of the most consistent operational causes of CSR delays is the absence of required inputs when medical writers start drafting ^[2]. Sponsors who commission a CSR before having the following materials finalized extend both the writing timeline and the submission date.

The Pre-Writing Input Checklist

The CSR Shell Approach

One established practice that compresses post-database-lock timelines is drafting a CSR shell before TLF availability. The shell populates all sections that do not depend on data: the introduction (Section 7), objectives (Section 8), the protocol-based portions of the investigational plan (Section 9), abbreviations, ethics, and investigator information (Sections 4 to 6). The sponsor team reviews and approves these sections before data arrives. Once TLFs are available and key messages have been aligned across the writing, statistics, and medical teams, the data-dependent sections (11, 12, and 13) can be drafted against an already-reviewed skeleton, which saves material review time after the data is in.

Regulatory and Documentation Considerations

The CSR lives within Module 5 of the Common Technical Document (CTD). FDA mandates eCTD format for all commercial NDAs, ANDAs, BLAs, and INDs ^[7]. As of September 16, 2024, FDA began accepting eCTD v4.0 for new NDA, BLA, ANDA, IND, and Master File submissions; version 3.2.2 remains supported in parallel for existing applications ^[7]. The transition to mandatory v4.0-only submissions will be announced by FDA with advance notice; sponsors managing large programs should assess v4.0 readiness now.

Cross-document consistency is a requirement that sponsors routinely underestimate. Every population definition, efficacy endpoint label, and adverse event term used in the CSR must match the corresponding definition in the protocol and SAP. Inconsistent terminology across documents is one of the most common sources of regulatory information requests. The TransCelerate eTemplates address this specifically by enabling content reuse from the CPT through the SAP and into the CSR ^[4].

ICH E6(R3), the Good Clinical Practice guideline, was endorsed at Step 2 in May 2023 and adopted at Step 4 as final on January 6, 2025 ^[8]. It reinforces the sponsor's responsibility for oversight and documentation quality throughout the trial lifecycle. The CSR is the terminal record of how that oversight was exercised, and regulators assess it with that accountability in mind.

AI and Automation in CSR Generation

Automated document generation has become an active area of investment across the industry. The AI-in-clinical-trials market was valued at approximately $2 billion in 2024 according to commercial market analyses ^[13], though independently audited figures for this specific segment are limited. Industry and vendor reports have cited CSR time reductions in the range of 30% or more with AI-assisted drafting ^[14]; these figures come from commercial sources, have not been tested in controlled studies, and should be treated as directional only. A 2025 survey cited by Medidata found broad executive interest in AI adoption across trial operations, though specific CSR-generation data was not disaggregated ^[14].

For CSR writing specifically, the most established use cases involve: population of standard sections from structured inputs (protocol text, SAP specifications, TLF metadata), draft generation of patient narratives from safety database records, and consistency checking between the CSR narrative and source tables across large documents. These tasks are well-suited to AI-assisted approaches because they are heavily structured, rule-governed, and data-dependent.

What AI does not replace is the interpretive work of Section 13 (Discussion and Overall Conclusions), which requires clinical judgment about how trial results sit within the existing evidence base. It also does not replace the medical writer's responsibility to verify every data-dependent claim in the narrative against its source TLF before submission. AI-generated CSR sections require review by qualified medical writers before regulatory submission. Under ICH E6(R3) ^[8], sponsors retain full accountability for the quality of clinical trial documentation; no current regulatory framework transfers that accountability to a software tool. In practice, sponsors and CROs assign a named medical writer or responsible medical officer to each CSR, and that person is accountable for the document's accuracy regardless of the drafting tools used.

Kitsa's KScribe assists regulatory and medical writing teams with structured document generation across the clinical trial document lifecycle, including CSRs, with particular focus on cross-document consistency across the protocol, SAP, and CSR. Where sponsors are managing multiple CSRs in parallel across a development program, consistency in population definitions and endpoint terminology is one of the areas where structured AI assistance reduces accumulated error risk. More detail is available at kitsa.ai/regulatory-document-generation.

Key Takeaways

• ICH E3 defines 16 numbered sections. The body runs from Section 7 (Introduction) through Section 13 (Discussion); post-text tables are Section 14; the reference list is Section 15; appendices are Section 16 (16.1 study information, 16.2 patient data listings, 16.3 case report forms, 16.4 US archival patient data listings (NDA/BLA)).
• Patient narratives appear in the body at Section 12.3.2 and in the post-text tables at Section 14.3.3. Section 16.2.3 is the listing of patients excluded from the efficacy analysis, not narratives.
• ICH E3 is guidance, not a rigid template. The Q&As (R1) explicitly permit structural adaptations that improve clarity; standard section numbering is practically expected by reviewers even so.
• The synopsis (Section 2) is often the first thing a reviewer reads and should be drafted in parallel with the body, not written last.
• Sponsors who start writing before the SAP is finalized, the database is locked, and TLFs are validated introduce errors that require re-verification under time pressure.
• The TransCelerate CSR eTemplate (2024, final release) and CORE Reference (Version 2) are the two principal non-regulatory authoring resources; they are complementary rather than interchangeable.
• ICH E6(R3) was adopted at Step 4 on January 6, 2025. Sponsors should track jurisdiction-specific implementation timelines, since ICH Step 4 adoption does not automatically equate to immediate enforceability in every region; align documentation practices with the requirements in each applicable jurisdiction.
• FDA has accepted eCTD v4.0 for new submissions since September 2024. Mandatory v4.0-only submissions will follow with advance notice from FDA.

Frequently Asked Questions

References

Regulatory structure claims in this article are sourced to ICH, FDA, and TransCelerate primary documents. Commercial and industry sources are included for operational context only and are noted accordingly.

1. [1]Clinion. "Clinical Study Report (CSR): Structure, ICH E3 Format and Submission." Clinion Insights, December 2025. https://www.clinion.com/insight/clinical-study-reports-csr-complete-guide/
2. [2]James Lind Institute / Biotech Ink Spots. "Documents Needed by Medical Writers to Write a Clinical Study Report." See also Bernstein L. "48 Things Medical Writers Need for CSRs." Biotech Ink Spots, 2016. https://jliedu.ch/documents-needed-by-medical-writers-to-write-a-clinical-study-report/
3. [3]International Council for Harmonisation (ICH). "E3: Structure and Content of Clinical Study Reports." Step 4 Guideline, November 1995. https://database.ich.org/sites/default/files/E3_Guideline.pdf
4. [4]TransCelerate BioPharma Inc. "Clinical Content and Reuse Initiative." (accessed June 2026) https://www.transceleratebiopharmainc.com/initiatives/clinical-content-reuse/
5. [5]Bernstein AB et al. "Critical review of the TransCelerate Template for clinical study reports (CSRs) and publication of Version 2 of the CORE Reference Terminology Table." Research Integrity and Peer Review 4:12, August 2019. https://researchintegrityjournal.biomedcentral.com/articles/10.1186/s41073-019-0075-5
6. [6]International Council for Harmonisation (ICH). "E3 Questions and Answers (R1)." EMA/CHMP/ICH/435606/2012, July 2012. https://database.ich.org/sites/default/files/E3_Q%26As_R1_Q%26As.pdf
7. [7]U.S. Food and Drug Administration. "Electronic Common Technical Document (eCTD)." CDER/CBER. Updated September 2024. https://www.fda.gov/drugs/electronic-regulatory-submission-and-review/electronic-common-technical-document-ectd
8. [8]International Council for Harmonisation (ICH). "E6(R3): Guideline for Good Clinical Practice." Final version, adopted January 6, 2025. https://database.ich.org/sites/default/files/ICH_E6%28R3%29_Step4_FinalGuideline_2025_0106.pdf
9. [9]Stevens G, Dolley S, Mogg R, Connor JT. "A template for the authoring of statistical analysis plans." Contemporary Clinical Trials Communications 34:101100, June 2023. https://pmc.ncbi.nlm.nih.gov/articles/PMC10300078/
10. [10]Billeveast. "ICH E3: How to Prepare a Regulator-Ready Clinical Report." January 2026. Industry commentary; verify against ICH E3 primary source. https://billeveast.com/ich-e3-preparing-a-regulator-ready-clinical-report/
11. [11]Sakhare SR et al. "Narrative Writing: Effective Ways and Best Practices." Perspectives in Clinical Research 8(2):68-73, 2017. https://pmc.ncbi.nlm.nih.gov/articles/PMC5384400/
12. [12]Precision for Medicine. "Inside a CRO: Growing Role of Medical Writers for an Effective CSR." https://www.precisionformedicine.com/blog/growing-the-role-of-medical-writer-for-an-effective-csr/
13. [13]Research and Markets / GlobeNewswire. "Artificial Intelligence (AI) in Clinical Trials Market Research Report 2025." November 2025. Commercial market analysis; treat figures as directional. https://www.globenewswire.com/news-release/2025/11/24/3193588/28124/en/Artificial-Intelligence-AI-in-Clinical-Trials-Market-Research-Report-2025.html
14. [14]IntuitionLabs. "Clinical Study Report Automation: AI Opportunities and Risks." 2025. Industry/vendor synthesis; reduction figures are not from controlled studies. https://intuitionlabs.ai/articles/clinical-study-report-automation-ai-risks