CliniGA CRO

CliniGA CRO Blog

CSR and Manuscript Numbers: Distinguishing Errors from Definition Differences

Explain numeric differences between the CSR and manuscript through population, period, denominator and analysis definitions.

Representative research photograph: CSR and Manuscript Numbers: Distinguishing Errors from Definition Differences
Photo: U.S. Navy CLDJ by Petty Officer 2nd Class JEROME FJELD · License: Public domain · Cropped and resized.
Published
Updated

A clinical study report and manuscript may describe the same trial while using different scopes or displays. Rather than forcing every number to match, explain whether a difference arises from definitions, calculations or an actual error.

The publication decision

Make population, period, unit, denominator and rounding rules visible during comparison. If the manuscript addresses a narrower question, tell readers clearly. Copying a number from the CSR without its context does not establish a valid comparison. Relevant experts should own the explanation and evidence used to resolve material discrepancies between the two outputs.

Worked example

For example, the CSR includes the full safety population while a manuscript displays participants completing a specified follow-up period. Event proportions differ. The team verifies population and time-window differences against source tables and improves methods and footnotes. If the cause remains unknown, it does not simply change one number to create apparent agreement; the issue remains an open scientific question.

Coordination checks

  • Compare population and denominator definitions for each central number.
  • Separate rounding differences from record or analysis differences.
  • Check that explanations appear in methods or the relevant table footnote.

Acceptance and boundaries

Acceptance means important numeric differences are understandable through sources and rationale. The manuscript need not repeat every CSR detail, but its selected scope should be balanced. If a new calculation is used, do not present it as an unchanged result from the older report. A reconciliation record supports traceability rather than replacing statistical approval.

Frequently asked questions

Must every number be identical?
Expect agreement for the same definition and scope; explain the reason when uses differ.
Do rounding differences matter?
Their significance depends on context. First establish that they do not conceal a real scope or calculation difference.

Sources

  1. EMA — ICH E3 structure and content of clinical study reports
  2. ICMJE — Preparing a manuscript for submission to a medical journal