Reading the evidence

The Retracted Citrulline Paper Still Being Quoted

The most-cited study in this corner of the supplement aisle carries a retraction notice. It is still quoted across marketing pages, because a citation copied from another marketing page is never checked against the record. Checking takes about thirty seconds.

By MemoVance Pro Team · Last updated October 29, 2026 · 8 min read · Reading the evidence · Reviewed by Dr. Judith Leone-Friedman

The paper everybody cites

Search for citrulline and blood pressure and one paper dominates the results: a 2018 systematic review and meta-analysis in the Journal of Human Hypertension reporting that citrulline supplementation reduced blood pressure.

It is an attractive citation for anybody selling citrulline. It is a meta-analysis rather than a single trial, which sits high in the usual hierarchy of evidence. It is in a specialist journal. It is recent enough to look current. And it reports a positive result on a hard outcome rather than a soft one.

The PubMed record for it carries a flag: the publication type field reads retracted publication, and a retraction notice was published in the same journal in 2021.

What a retraction is, and what it is not

A retraction is a journal formally withdrawing a paper from the scientific record. The paper usually remains available so the literature citing it can still be understood, but it is marked, and the mark means it should no longer be relied on.

Retractions happen for a range of reasons: errors in the analysis, problems with the underlying data, duplicate publication, or misconduct. A retraction notice normally states which, and the honest position for anybody outside the process is to report the fact of the retraction rather than to characterise the reason.

What a retraction is not is a matter of debate about the conclusion. It is a formal status on the record, visible on the record, and it is either there or it is not.

Why it matters more here than elsewhere

In most fields a retracted paper stops being cited fairly quickly, because the people citing it are researchers who work with the record daily. Supplement marketing is not that field. A citation that entered the marketing bloodstream in 2019 propagates by copying, and nothing in that chain ever revisits the source.

Why retracted results keep circulating

• Citations are copied, not read. A page cites the paper because another page cited it, and the chain often runs back years.
• A headline outlives its paper. News coverage from the year of publication stays online, unmarked, and ranks well.
• Secondary databases lag. Reference managers and aggregators do not always carry the retraction flag, so a citation exported in 2019 still looks clean.
• Nobody in the chain has an incentive to check. The result is convenient for everyone quoting it.
• The flag is easy to miss if you arrive at the abstract through a search snippet rather than through the record itself.

None of that requires anybody to be acting in bad faith, which is what makes it worth writing about. A chain of honest people copying each other reproduces a withdrawn result indefinitely.

The replacement, and what it found

A second meta-analysis on the same question was published in 2018, in Current Hypertension Reports, and it carries no retraction flag. It pooled the clinical trials of citrulline supplementation and blood pressure and found no significant effect: the brachial systolic change was a fraction of a millimetre of mercury with a confidence interval crossing zero, and the central pressure result was similar.

That is the paper this website cites, on the home page, on the guides, on the disclaimer and on the pages about the product it sells. It is the least flattering result available on an ingredient that occupies two of seven rows on this bottle.

Citing it is not a display of virtue. It is what happens when the rule is to cite the un-retracted record on a question rather than the convenient one, and the rule produces this outcome whether it suits the page or not.

The two papers, side by side

Two meta-analyses on one question, published the same year, reaching opposite conclusions. Only one of them is still standing.

Two meta-analyses, one question, one year, opposite conclusions. That happens in science and it is usually a reason to read both carefully. Here one of them has been withdrawn, which settles it.

Checking a citation in thirty seconds

Five fields, about thirty-five seconds, and they settle most disputes about a citation in this category.

The first row is the one this article is about and it is the quickest of the five. Open the PubMed record, look at the top of the page, and a retraction, a correction or an expression of concern is displayed there in a coloured banner.

The other four rows are worth the same half-minute for a different reason, covered next.

Three other things a record tells you

The dose. This is the single most useful field for anybody comparing a citation with a label, and it is nearly always in the abstract. The citrulline dose-ranging study tested 2, 5, 10 and 15 grams. The arginine dose-response meta-analysis found its effect at four grams a day and above. Knowing those two numbers makes most citrulline and arginine marketing readable at a glance.

The population. Trained athletes, healthy young adults, older adults, or a patient group with a specific condition. A result in one of those does not automatically describe any of the others, and the abstract always says which.

What actually moved. A study measures several things and reports all of them. The primary outcome is the one it was designed to answer, and the interesting question is whether that one moved or whether a secondary measure did. A meta-analysis of citrulline malate and muscle strength is a useful example: it pools trials on a specific outcome in a specific population, and quoting it for anything else would be a misuse of it.

What this desk does with all of it

1. Every record is confirmed live. Each citation on this website was pulled through the public interface in the session that wrote the page, and the bibliographic details were read off that response rather than typed from memory.
2. The dose is recorded beside it. That is what lets a printed label amount be set against the amount that produced a result, which is the most useful thing this website does.
3. Flagged records are excluded. The retracted meta-analysis is deliberately absent from the reference list, and the un-retracted paper on the same question is used in its place.
4. Unflattering results are cited alongside flattering ones. The flat pooled citrulline result appears on several pages, including commercial ones.
5. Sources that block automated requests are recorded as such rather than dropped, and no figure is taken from a page that could not be read.

The full method is on the about page, and the reference list at the foot of every page on this website is the subset that page actually cites.

The habit worth taking away

A citation in this category is worth exactly as much as the last person who checked it, and the check is not hard.

Open the record. Look at the banner. Read the dose. That is three of the five rows in the table above and it takes under fifteen seconds, and it will separate a real reference from a decorative one faster than any amount of reading about mechanisms.

NCCIH's guidance on using dietary supplements sensibly makes a broader version of the same point, and it is worth reading once by anybody who buys anything in this aisle.

For this product specifically, the category guide carries the quantities each family of ingredients was tested at, which is the other half of the same habit.

References