How to reduce plagiarism similarity in a medical manuscript, ethically

A high similarity score is one of the fastest routes to desk rejection. Most journals run every submission through similarity-checking software before an editor reads it, and a concentrated match, a whole paragraph from one source, can end the process regardless of the science. The good news is that most high scores in medical manuscripts come from fixable writing habits, not deliberate copying.

First, understand what the software does

Tools such as iThenticate (used by journals through Crossref Similarity Check), Turnitin and DrillBit compare your text with a database of published articles, web pages, theses and previously submitted work. They report the percentage of your text that matches existing sources and list each source.

Similarity is not the same as plagiarism. A match may be a correctly quoted and cited sentence, a standard phrase ("this study was approved by the institutional ethics committee"), or your reference list. Editors look at what matched and where, not just the total.

How to read a similarity report

1. Look at the source list, not the headline number. A score of 18% spread across fifty sources at under 1% each is usually common phrasing. A score of 12% where 9% comes from one paper is a serious problem.
2. Check which sections match. Overlap in methods descriptions is treated more leniently than overlap in the introduction or discussion, where authors are expected to write in their own words.
3. Identify your own earlier work. Matches to your thesis or previous papers need disclosure or rewriting, depending on the journal.
4. Exclude what does not count (bibliography, properly quoted text, very small matches) to see the overlap that genuinely needs work.

Ethical techniques that actually reduce similarity

Write from understanding, not from the source

The most reliable method: read the source, close it, and write what it means in your own words, for your argument. Then go back and check that you have represented it accurately and cited it. Patchwriting (keeping the original sentence structure and swapping a few words) still produces matches and is still a form of plagiarism.

Synthesise instead of summarising one paper at a time

Introductions often match because each sentence restates one paper. Group studies by finding instead: "Three cohort studies in South Asian populations reported higher rates of X [refs], whereas a European registry found no difference [ref]." Synthesis is better science and naturally original.

Quote what must stay exact

Definitions, diagnostic criteria and guideline recommendations sometimes cannot be paraphrased without changing their meaning. Put them in quotation marks with a citation, and keep quotations short; most medical journals expect very little direct quotation.

Cite every idea that is not yours

A well-paraphrased idea without a citation is still plagiarism. Conversely, adding the missing citation does not fix copied wording; you need both your own words and the reference.

Rewrite thesis-derived text

If the manuscript comes from your thesis and the thesis is in a university repository or Shodhganga, rewrite rather than copy, and disclose the thesis in the cover letter. See converting a thesis into a journal article.

Handle methods text carefully

For standard procedures, describe them briefly and cite the original method or your previous paper ("as described previously [ref]") instead of repeating long passages.

Techniques to avoid

• Automated paraphrasing ("spinner") tools. They distort technical meaning ("myocardial infarction" becoming "heart muscle death event") and produce text editors recognise instantly.
• Character tricks. Replacing letters with look-alike characters from other alphabets, inserting hidden text or converting text to images are detected by current software and treated as misconduct.
• Translating and back-translating. This is still plagiarism of ideas and structure, and the resulting language is usually poor.
• Deleting citations to lower matches. Removing references to sources you used makes the problem worse, not better.

What score is acceptable?

Journals do not use a single universal threshold, and policies differ. Many editors begin to look closely once overall similarity rises above roughly 15–20%, or when any single source contributes more than a few percent, but a low score does not protect a manuscript with a copied paragraph. For theses at UGC-regulated institutions in India, the UGC Regulations, 2018 define similarity up to 10% (excluding quoted, cited and standard content) as minor, with graded consequences above that. Our plagiarism check page summarises the levels.

A pre-submission routine

• Run a similarity check on the final draft, not an early one
• Review every source above 1% individually
• Rewrite matched introduction and discussion text from scratch
• Quote and cite exact definitions; cite previous methods
• Disclose thesis or preprint origins in the cover letter
• Re-run the check and keep the report in case the editor asks

Sources and further reading

• COPE guidance and flowcharts on plagiarism and text recycling
• ICMJE Recommendations on overlapping publications
• UGC (Promotion of Academic Integrity and Prevention of Plagiarism in Higher Educational Institutions) Regulations, 2018