A Clinical Study Report (CSR) is the full scientific account of a single clinical trial: what you set out to test, how you ran the study, who took part, what the efficacy and safety data showed, and what it all means. It is the primary evidence document for a trial, the thing every summary and overview above it traces back down into. In a marketing application it sits in Module 5 of the CTD, one CSR per study, and its structure is set by ICH E3, "Structure and Content of Clinical Study Reports."

E3 gives you a 16-section skeleton and a description of what belongs in each part. It does not, despite how it is often treated, give you a rigid fill-in-the-blanks template. Writing a good CSR is less about reproducing that skeleton and more about assembling a large set of source documents — protocol, statistical analysis plan, tables and listings, safety data — into one internally consistent narrative where every number agrees with the number it came from. This guide walks the workflow first, then goes section by section through E3.

What a CSR is, and what E3 actually governs

E3 governs the structure and content of the report for one controlled clinical study. It applies most cleanly to the pivotal efficacy-and-safety trials that carry a submission, though the same skeleton is scaled down for smaller or earlier-phase studies. It tells you which sections a reader expects, in roughly what order, and what each should contain. It does not tell you how to run the analysis (that is E9 and the SAP) or how to conduct the trial (that is E6, Good Clinical Practice).

The single most important thing to understand about E3 is that it is guidance, not a mandatory format. In 2012 ICH published an E3 Questions and Answers document specifically to make this clear: the section numbering and headings are a recommended way to organize content, and you may combine, reorder, or relocate material to suit the study as long as the expected content is present and a reviewer can find it. A useful, practical companion resource is CORE Reference (Clarity and Openness in Reporting: E3-based), a jointly developed EMWA and AMWA resource that builds on E3 and two decades of accumulated expectation to serve as a working authoring guide. E3 tells you the shape; CORE Reference tells you what the shape has come to mean in practice.

The writer's workflow, before you write a word

The mistake new CSR writers make is opening a blank template and starting at the title page. The report is a synthesis of documents that already exist, and most of the work is upstream of the prose.

Assemble the sources. A CSR is built from a stack of inputs: the final protocol and every amendment, the statistical analysis plan (SAP), the tables, figures, and listings (the TFLs, produced by the statistics group from the locked database), the patient-level safety listings, the investigator's brochure for the reference safety information, and the trial's regulatory and ethics documentation. You cannot write Section 11 without the efficacy tables, and you cannot write Section 12 without the safety tables and listings. Final CSR drafting is generally based on locked data and final or approved TFLs, though drafting of the template and data-independent sections often begins earlier.

Confirm versions. This sounds clerical and is not. You need the final protocol, the final SAP, and the TFLs generated from the locked database. Drafting against a draft SAP or an interim data cut is how a CSR ends up describing analyses that were never run, or reporting counts that change under you. Record the version and date of every source and check they are the ones the report is meant to reflect.

Map E3 sections to sources. Before drafting, build a map of which source feeds which section. The protocol and SAP feed the investigational plan and objectives. The disposition and demographic tables feed the study patients and baseline sections. The efficacy TFLs feed Section 11. The safety TFLs and listings feed Section 12, and the individual patient listings feed the narratives. Knowing where each section's facts live means you are transcribing and interpreting, not inventing.

Draft, then QC, then review. The first draft renders the sources into E3's structure. Then comes quality control: a systematic check that every number in the text matches the source table it came from, that the synopsis agrees with the body, and that the same result is stated the same way everywhere it appears. Then the review cycles — clinical, biostatistics, safety or pharmacovigilance, and regulatory each read and comment, often more than once. A CSR is not written once; it is reconciled repeatedly against its sources and its reviewers until the whole document is consistent.

The E3 structure, section by section

E3 lays out 16 numbered sections. Here they are, grouped the way the work actually clusters.

Front matter (Sections 1–6)

The opening sections orient the reader and establish the study's provenance. They are lower-effort than the analytical sections but they are where consistency errors are cheapest to catch and most embarrassing to miss.

  • Section 1, Title Page — study title, drug, indication, sponsor, protocol number, phase, key dates, and the sponsor's responsible medical officer.
  • Section 2, Synopsis — a self-contained summary of the whole report, usually held to a few pages, covering objectives, methods, results, and conclusions. Every figure in it must match the body exactly, which is why writers often draft the synopsis last, once the numbers underneath have stopped moving.
  • Section 3, Table of Contents for the individual report.
  • Section 4, List of Abbreviations and Definitions of Terms.
  • Section 5, Ethics — the independent ethics committee or IRB review, the statement that the study was conducted in accordance with ethical principles including the Declaration of Helsinki and GCP, and how informed consent was handled.
  • Section 6, Investigators and Study Administrative Structure — who ran the study and how it was organized, with the detailed investigator lists usually carried in the appendices.

Introduction and objectives (Sections 7–8)

  • Section 7, Introduction — a short scientific and regulatory context for the study: the disease, the drug, the development rationale, and why this trial was run. A page or so, no literature review.
  • Section 8, Study Objectives — a plain statement of the study's objectives, primary and secondary. These have to align precisely with the endpoints described later and with the estimands in the SAP; the objective, the endpoint, and the analysis are three views of the same thing and they cannot disagree.

Section 9, Investigational Plan

This is the first of the heavy sections and it describes how the study was designed and conducted. E3 organizes it into subsections covering the overall study design and plan; a discussion of the design and choice of control; selection of the study population, including inclusion and exclusion criteria and how patients were withdrawn; the treatments, meaning the investigational product, doses, regimens, randomization, and blinding; the efficacy and safety variables and the schedule of assessments; data quality assurance; the statistical methods, including the planned analysis and how the sample size was determined; and — importantly — any changes in the conduct of the study or the planned analyses.

Two things make Section 9 hard. First, if the study changed mid-flight through amendments, the changes have to be described here honestly, not quietly smoothed over. The subsection on changes to the conduct or planned analyses is where reviewers look to understand whether the trial that ran is the trial that was designed. Second, the statistical methods subsection has to render the SAP faithfully, including the estimand framework under E9(R1): not just "the ITT population was analyzed" but the full description of the treatment effect, including how intercurrent events like discontinuation or switching were handled. Describe the estimator without the estimand and a statistical reviewer will notice.

Section 10, Study Patients

Section 10 accounts for everyone who entered the study. It covers the disposition of patients — how many were screened, randomized, treated, completed, and discontinued, and the reasons for discontinuation, usually anchored to a CONSORT-style flow — and protocol deviations. It is short but exacting: the numbers here set the denominators for everything that follows, and the analysis populations in Section 11 are carved out of this disposition. If the disposition table and the efficacy population counts don't reconcile, the whole report is suspect.

Section 11, Efficacy Evaluation

Section 11 is where the trial's central question gets answered. E3 structures it around the data sets analyzed (which patients contributed to which analysis, and the definition of the ITT, per-protocol, and other populations); demographic and baseline characteristics; measurements of treatment compliance; and then the efficacy results themselves — the analysis of the primary and secondary endpoints, the statistical and analytical issues (handling of dropouts and missing data, multiplicity, subgroups, interim analyses), and tabulations of the response.

The hard part of Section 11 is not producing numbers; the statisticians produce those. It is presenting them as a coherent argument that matches the pre-specified analysis. The primary endpoint result has to tie back to the objective in Section 8 and the estimand in Section 9, the handling of missing data has to match what the SAP said would happen, and any analysis that was not pre-specified has to be flagged as post hoc. Writers get into trouble when the text quietly upgrades an exploratory finding into something that reads like a confirmed effect. The efficacy section reports what the pre-specified analysis found, in the frame the SAP set.

Section 12, Safety Evaluation

Section 12 is usually the longest section and often the hardest to get fully consistent. E3 organizes it around the extent of exposure (how much drug, for how long, in how many patients); adverse events, summarized by frequency, severity, and relationship to treatment; deaths, other serious adverse events, and other significant adverse events; clinical laboratory evaluations; vital signs and other safety observations; and overall safety conclusions.

Inside the deaths and serious-events subsection sits the piece writers most associate with CSR work: the patient narratives. E3 asks for narratives of each death, each other serious adverse event, and certain other significant adverse events — brief, factual, chronological accounts of what happened to individual patients — though E3 notes that serious events clearly unrelated to the investigational product may be omitted or described only briefly. Writing them is a discipline of its own: reconciling the clinical database, the safety listings, and sometimes the source case report forms into a consistent story for each patient, and keeping the narrative's dates, doses, and outcomes in agreement with the summary tables above them. We cover the craft in detail in how to write a patient safety narrative. A single trial can require dozens or hundreds of them, and they are a common bottleneck in the CSR timeline.

The broader difficulty in Section 12 is reconciliation: the same adverse event count appears in the summary tables, the narratives, the synopsis, and often the discussion, and every appearance has to agree. When the safety database is updated late, that consistency is easy to break and tedious to restore.

Section 13, Discussion and Overall Conclusions

Section 13 is where the report steps back and interprets, weighing the efficacy findings against the safety profile and stating what the study concluded in the context of the drug's development. This is the one section that is genuinely interpretive rather than descriptive, and it has to stay disciplined: the conclusions can only rest on what Sections 11 and 12 actually showed, and the benefit-risk read should not outrun the data above it. A discussion that claims more than the results support is the fastest way to draw a reviewer's skepticism onto the whole report.

Tables, references, and appendices (Sections 14–16)

  • Section 14, Tables, Figures and Graphs referred to but not included in the body text — the in-text supporting displays collected together.
  • Section 15, Reference List — the literature cited in the report.
  • Section 16, Appendices — the large supporting documentation: the protocol and amendments, the sample case report form, the statistical documentation and SAP, the full investigator and IEC information, the patient data listings, and the individual patient data. Section 16 is where the bulk of the CSR's supporting content lives, though how it is packaged for submission varies: datasets and large patient listings are commonly provided as separate eCTD components rather than embedded in one document. The report body interprets the findings; the appendices hold the primary record that the body is accountable to.

Where CSR writing actually gets hard

Across the whole document, the recurring difficulty is not any single section — it is consistency between all of them. The same fact surfaces in the synopsis, the body, a table, and a narrative, every surfacing has to agree, and a late data change or a reviewer's comment can ripple through all of them. Most of the professional skill in CSR writing is holding that web of cross-references true while the sources shift underneath you. That reconciliation problem is one that CSR automation is designed to help with: holding the E3 structure and the source data as constraints, rendering the draft against them with the numbers traced back to their sources, and leaving the writer the interpretation E3 can't encode. For the wider map of where the CSR sits among the other documents you'll write, see the regulatory documents field guide.

FAQ

How long does a CSR take to write?

It varies widely with the size and complexity of the trial, but a full pivotal CSR is generally measured in weeks to a few months from TFL delivery to a final approved document, not days. The drafting is a fraction of it; the review cycles and the reconciliation against source data usually take longer than the first draft. Small early-phase studies move faster; large multi-arm trials with hundreds of safety narratives move slower.

How long is a CSR?

The report body is often in the range of a hundred pages or so, but the complete CSR with all Section 16 appendices — protocol, SAP, patient listings, individual data — commonly amounts to hundreds or thousands of pages of content. Most of that bulk is the appendices, not the narrative sections. How it is submitted varies, though: the datasets and large listings are frequently packaged as separate eCTD components rather than assembled into a single embedded document.

What is the hardest section to write?

Most writers point to Section 12, the safety evaluation, and specifically the patient narratives inside it, both for volume and for the reconciliation they demand. Section 11 is intellectually demanding in a different way, because it has to present the efficacy results exactly as the pre-specified analysis framed them. Section 9's statistical methods can be deceptively hard too, since it has to render the estimand and the SAP faithfully.

Do I have to follow E3's section numbering exactly?

No. The 2012 E3 Q&A makes clear that E3 is guidance on content, not a mandatory format. You can combine, reorder, or relocate material to suit the study, as long as the expected content is present and a reviewer can find it. What you should not do is drop expected content or let the structure hide it.