A practical, ordered workflow for critically appraising a clinical study report structured to ICH E3 — what to read first, what to cross-check, and where reviewers most often find the report and the data diverging.
Updated June 2026 · Educational resource — not a substitute for professional review
Read in this order
A CSR is large, and reading it linearly wastes the most valuable thing you have — fresh judgment about whether the conclusion is earned. Anchor your expectations from the protocol and statistical analysis plan first, then let the report try to meet them.
Step 1 — Synopsis, protocol & SAP
Read the synopsis for the claimed result, then read the protocol's objectives, primary endpoint, and the pre-specified primary analysis in the SAP.
Note the protocol version and date: amendments after first-patient-in are a frequent source of endpoint or population drift.
Step 2 — Design & population integrity
Confirm the report's stated design, randomization, and blinding match the protocol.
Check the analysis-set definitions (ITT, per-protocol, safety) and how each patient was assigned to them.
Step 3 — Patient disposition & deviations
Reconcile the disposition: screened → randomized → treated → completed → analyzed. Do the numbers tie out across the CONSORT-style flow, the disposition table, and the analysis-population counts?
Read the protocol deviations section: are important deviations classified, and could they bias the primary result (e.g. eligibility violations concentrated in one arm)?
Is dropout balanced across arms, and is its handling consistent with the SAP's missing-data strategy?
Step 4 — Efficacy
Verify the primary analysis is the pre-specified one — same population, model, covariates, and handling of intercurrent events. A switch from ITT to per-protocol, or an added covariate, is a red flag unless pre-specified.
Check effect size and precision (confidence intervals), not just the p-value, and confirm multiplicity control across secondary endpoints.
Distinguish pre-specified from post hoc analyses; post hoc findings are hypothesis-generating.
Step 5 — Safety
Confirm the extent-of-exposure denominator before interpreting any rate.
Reconcile the narrative for each death and serious adverse event against the listings; check that AE coding (MedDRA) is consistent and that severity vs. seriousness aren't conflated.
Step 6 — Conclusions vs. evidence
Do the discussion and synopsis conclusions stay within what the primary result and its uncertainty support?
Are limitations stated honestly, and is generalizability claimed only to the studied population?
PUT THIS INTO PRACTICE
TrialScope runs the checks below automatically — findings, severities, and a clear verdict in under 90 seconds, calibrated to your professional role and sharpened by corrections from verified clinical professionals.
Frequently asked questions
What is a clinical study report (CSR)?
A clinical study report is the formal, integrated account of a single clinical trial's design, conduct, and results, submitted to regulators. Its structure follows ICH E3, which defines the standard sections from the synopsis through efficacy, safety, and appendices.
What does ICH E3 cover?
ICH E3 (Structure and Content of Clinical Study Reports) specifies the expected sections of a CSR: title page and synopsis, ethics, investigators and administrative structure, introduction, objectives, investigational plan, study patients (disposition and protocol deviations), efficacy evaluation, safety evaluation, discussion and conclusions, tables and figures, and appendices.
How do you audit a CSR efficiently?
Read the synopsis and protocol first to fix expectations, then verify that the report's design, endpoints, and analysis populations match the protocol and SAP. Trace the patient disposition, confirm the primary analysis was conducted as pre-specified, reconcile the safety narratives against the tables, and check that the conclusions are supported by — and not overstated relative to — the results.