The challenge of measuring plastics in humans

Some of the hardest questions in microplastics science are not about interpretation. They are about measurement. Before we can know what a level means, we have to know how confidently we found it.

That framing matters because the public conversation about microplastics has sprinted ahead of the measurement. Studies report plastic particles in blood, brain, placenta, lung, testis, and breast milk. Consumer blood tests ship to homes for $150. Regulators debate thresholds. And somewhere underneath all of it sits a technical question that the headlines rarely pause on: how do you actually find a polymer particle inside a complex biological sample, how confident are you that what you found is plastic, and how confident are you in the number attached to it?

The answer is harder than most non-specialists realize. No single instrument spans the size range that matters. Every method has confounders that can generate signals where there is no plastic, or miss plastic that is there. Sample preparation alone can change results by orders of magnitude. Eighty-four laboratories handed identical reference samples in a recent interlab study disagreed with each other by factors of two to ten. The highest-profile studies of plastic in human tissue are now being directly challenged by the same community that produced them, not because the researchers were careless but because the technique they relied on may not have been up to the matrix.

None of that means the fundamental findings are wrong. Plastic particles really do appear in human samples, reproducibly, across multiple independent techniques. The qualitative picture has held up – and, in itself, is startling. What has not yet held up is the quantitative picture: exactly how much, of which polymer, at what size, in which tissue. That distinction is the point of this piece. It is not a dismissal of the field, and it is not a call for alarm. It is an argument for a particular kind of patience. Measurement infrastructure is being built. Until it is finished, the honest answer to "how much plastic is in me" sits somewhere between a tiny amount and a meaningful amount.

This is a dive into what detection actually looks like today. It walks the instruments, the preparation steps, the failure modes, and the critiques – in roughly the order a sample would pass through them – to give a reader the tools to evaluate any new microplastic study on its own terms.

Why measurement is the hard part

A six-order-of-magnitude target

The first thing to understand is the size range. Microplastics are conventionally defined as particles between 1 μm and 5 mm, and nanoplastics as particles below 1 μm. The range from a visible plastic shard to a particle smaller than a virus spans six orders of magnitude. To put that in perspective, a size range spanning from a marble to a mountain. No one instrument covers it. Stereomicroscopes stop working below about 500 μm. Fourier-transform infrared spectroscopy hits a diffraction wall around 10 μm. Classical Raman microscopy reaches about 1 μm before diffraction defeats it. Electron microscopy goes deeper but loses chemical identification. Only a handful of frontier techniques (e.g., stimulated Raman scattering, optical photothermal infrared, atomic-force-microscope-coupled infrared spectroscopy) can identify a polymer particle below 500 nm, and each of them trades something else (throughput, cost, matrix compatibility, validation) to get there ^[1-2].

That tradeoff has a direct consequence. If a study reports "the concentration of microplastics in human blood," the first question to ask is which size range was measured, and the second question is what was invisible to the instrument. A technique that cannot see below 10 μm is not measuring the nanoplastic fraction that most of the recent biological interest centers on. A technique that can see down to 100 nm is often too slow to generate population data.

Counts, mass, and presence are not the same number

The second thing to understand is that almost every detection method answers one of three distinct questions, and the answers are not interconvertible ^[1-3]. Imaging plus vibrational spectroscopy (i.e., the FTIR and Raman families) answer how many particles, what size, what shape, what polymer. Thermal decomposition into mass spectrometry answers how much polymer mass is in this sample, by class. And the nanoparticle sizing tools borrowed from nanomedicine (i.e., dynamic light scattering, nanoparticle tracking analysis, tunable resistive pulse sensing) answer is something there, and roughly how big is it, usually without telling you what polymer it is.

A study that reports both particle counts and polymer masses from a single method has almost certainly extrapolated. The two numbers cannot be derived from each other without assumptions about particle shape, density, and size distribution, and each assumption multiplies the uncertainty. The strongest work in the field pairs a particle-based method with a mass-based method on matched sample aliquots, and treats any divergence between them as information.

Before the instrument, there is the preparation

The third thing to understand is that what the instrument sees is not the sample. It is what survived preparation. And what happens between the sample tube and the instrument is where most errors in the field are born.

Every microplastic workflow has to do three things: remove the biological matrix so the particles can be seen, concentrate or isolate the polymer fraction so the instrument can find it, and avoid contaminating the sample with plastic from the lab itself. Those three steps each have a signature failure mode. They happen in different orders for different techniques. And almost every published critique of a human-biology microplastic finding ultimately lands on one of them.

The preparation problem

Loss

Aggressive chemical digestions dissolve biological matrices, but they also dissolve the analyte. Potassium hydroxide, the most common digestion reagent, partially hydrolyzes polyethylene terephthalate at elevated temperatures and completely degrades polylactic acid. Hot Fenton's reagent (i.e., iron-catalyzed hydrogen peroxide) can run away thermally and melt polyethylene and polypropylene at the same temperatures that clear lipids and proteins ^[4]. Enzymatic digestions (e.g., proteinase K, lipase, cellulase) are gentler but slow, and their polymer-specific recoveries depend on matrix and incubation conditions that are rarely standardized across labs.

Density separation, the other workhorse of prep, uses salt solutions of known density to float polymer particles away from denser inorganic material. Sodium chloride at 1.2 g/mL is cheap and benign but cannot float polyvinyl chloride or polyethylene terephthalate, which are denser. Zinc bromide at 1.8 g/mL recovers those polymers but is corrosive, expensive, and damages sensitive plastics. The choice of density salt is not neutral: it silently selects which polymers survive the workflow and which are lost.

The consequence is that a sample which started with a mix of seven common polymers can, by the time it reaches the instrument, contain only four of them. A study that reports "no detectable PVC in blood" may be reporting a prep artifact rather than a biological absence.

Masking

Incomplete digestion leaves the organic matrix behind, and that residual matrix has two effects that matter. Under Raman microscopy, organic residues fluoresce at intensities orders of magnitude higher than the polymer Raman signal, effectively blinding the instrument. Under infrared microscopy, they broaden and shift polymer absorption bands, degrading library matching. And under pyrolysis-GC/MS, the technique behind most of the human-biology numbers, residual lipids pyrolyze into the same alkane and alkene marker compounds used to quantify polyethylene.

That last problem is the one that has come into sharp focus in the last year, and it is covered in detail below. The short version: if the thing you are measuring and the thing you are trying to remove both produce the same signal, cleanup is not a matter of being more careful. It is a structural limit of the method.

Addition

Every surface the sample touches is a potential plastic source. Ambient laboratory air contains synthetic fibers from clothing and HVAC filters. Deionized water carries polymer fragments from tubing and storage bottles. Reagent bottles, collection tubes, pipette tips, filter membranes, and, most recently documented, the gloves worn by the analyst all shed material into the sample before the instrument ever sees it.

The scale of the addition problem is not small. A meta-analysis of procedural blanks across the microplastics literature found that blanks intended to be zero contained between 7 and 511 particles, with a mean around 80 ^[5]. 82% of published studies did not report clean-air controls. 66% did not blank-correct their results. A study that reports 100 particles per milliliter of sample may be reporting an 80-particle lab background plus a 20-particle biological signal, with blanks run at a different scale on a different day, on a different instrument, by a different person.

The discipline that separates credible from preliminary work has a recognizable shape. Matched procedural blanks travel with the sample from collection through analysis. Density separation and digestion are run with documented per-polymer recovery in the specific matrix. An isotope-traced spike, such as ¹³C- or deuterium-labeled polymer particles added to the sample at the start, allows the analyst to measure how much of a known input made it through the entire workflow. Limits of detection are defined as blank mean plus three standard deviations, not as vendor specifications on a clean reference material. Results are reported alongside the raw blank values they were corrected against. Studies that omit any of these steps are preliminary regardless of how striking the headline number is.

The instruments

The FTIR family, and the wall at ten microns

Fourier-transform infrared spectroscopy identifies polymers by their characteristic absorption bands in the mid-infrared (i.e., carbon-hydrogen stretches, carbonyl stretches, carbon-oxygen vibrations) matched against a reference library. It is the quantitative workhorse of environmental microplastics work, and it comes in four variants that form a natural progression from slow to fast.

Attenuated total reflection FTIR, or ATR, presses a single particle against a high-index crystal and measures the reflected infrared. It is the cheapest variant, and it works well for particles larger than about 500 μm that can be placed by tweezers. Below that, particles are too small to handle one at a time and ATR becomes impractical.

Micro-FTIR couples the spectrometer to a microscope and brings the lower size limit down to about 10 to 20 μm, measuring one point at a time across a filter ^[1]. Focal plane array FTIR replaces the single detector with an array of 64 by 64 or 128 by 128 elements and images a whole filter area at once, gaining roughly two orders of magnitude in throughput. It is the current standard for quantitative microplastic surveys of water and sediment. Quantum cascade laser infrared imaging (e.g., the Agilent 8700 LDI) replaces the broadband source with a tunable laser and scans only at the wavenumbers informative for polymer classification, reaching 10 to 50 times the throughput of FPA-FTIR on realistic samples by skipping the empty filter regions that make up most of the image ^[6].

All four variants share the same hard limit. Mid-infrared wavelengths span roughly 6 to 20 μm, and a particle cannot absorb light at wavelengths larger than itself. That puts a diffraction wall at about 10 μm below which no FTIR variant can reliably identify a particle, regardless of how fast or sensitive the instrument is. For environmental work, which is dominated by larger fragments, this wall is not a severe constraint. For the biological size range that most of the recent human-tissue interest focuses on, such as particles smaller than a few micrometers, reaching into the nanoscale, it means FTIR is simply not the right tool. Anything FTIR does not see in that size range, it is not seeing because the physics of the instrument will not let it.

Raman, and the climb into the nanoscale

Where FTIR measures absorbed infrared, Raman spectroscopy measures inelastic light scattering at visible or near-infrared wavelengths. The shorter wavelength pushes the diffraction limit down to about 1 μm in conventional microscopy, and Raman's chemical specificity is if anything higher than FTIR's. It should be the obvious tool for small particles. Two things stand in the way.

The first is fluorescence. Many real-world polymer particles carry adsorbed organic matter, pigments, or weathering-induced surface chromophores that fluoresce far more intensely than any Raman scattering signal, drowning the spectrum. The frontier of Raman method development is largely a set of workarounds for this problem. Surface-enhanced Raman spectroscopy uses nanostructured metal substrates to amplify the Raman signal by six to ten orders of magnitude, outrunning fluorescence, but enhancement depends on uncontrolled particle-substrate geometry and quantification suffers accordingly ^[7]. Stimulated Raman scattering uses coherent pulse pairs to drive a specific vibrational transition, producing signals orders of magnitude stronger than spontaneous Raman and bringing single-particle detection into the sub-100 nm range at microseconds per particle. The 2024 Rutgers study that reported hundreds of thousands of nanoplastic particles per milliliter in commercial bottled water used stimulated Raman, and it was that technique that made the measurement possible at all ^[8]. Optical photothermal infrared spectroscopy uses a visible probe beam to detect the tiny thermal expansion caused by infrared absorption, producing FTIR-like spectra at about 500 nm spatial resolution, crossing the mid-infrared diffraction wall without contact, and without the throughput cost of atomic-force-microscope methods ^[2].

The second obstacle is throughput. Mapping a filter at sub-micrometer resolution with conventional Raman takes hours to days. That matters because any population-scale study, like the kind that would establish a reference range for blood microplastic burden, needs to measure many samples quickly. Raman's resolution advantage comes at a throughput cost that has not yet been solved for routine work.

Mass spectrometry, and the technique that carries most of the biology literature

Pyrolysis-GC/MS is the technique behind most published numbers for microplastic content in human blood and tissue. The workflow is straightforward: a few micrograms of sample are flash-heated to around 600 °C in an inert atmosphere, cracking the polymer into low-molecular-weight marker compounds. Those markers are separated by gas chromatography and identified by mass spectrometry. Each polymer has a characteristic set: styrene monomer and dimer from polystyrene, terephthalic acid derivatives from polyethylene terephthalate, propylene oligomers from polypropylene, ethylene oligomers from polyethylene. With a calibration curve against reference polymers, the instrument reports micrograms of each polymer per sample.

The technique is destructive. It reports mass, not particle count, and it tells you nothing about size, shape, or morphology. In exchange, it offers sensitivity that particle-based methods cannot match for the smallest fractions, and on clean reference materials the published limits of detection reach as low as 0.01 μg for polystyrene and a few micrograms for polyethylene ^[9]. It is also the only practical way to quantify the true nanoplastic fraction at population scale with existing infrastructure. Those two facts — that mass spectrometry reaches into the nanoscale where imaging cannot, and that it is what most of the human-biology literature has used — are closely linked.

Related variants extend the reach. Thermal extraction-desorption GC/MS uses a thermogravimetric analyzer to gradually heat milligram-scale samples, relaxing the homogeneity problem that dogs flash pyrolysis on heterogeneous environmental matrices ^[10]. Single-particle ICP-MS, adapted from elemental analysis, can in principle count and size individual polymer particles via their ¹³C signal as they pass through an inductively coupled plasma torch ^[2]. Both are developmental for biological samples but represent the likeliest bridges to routine nanoplastic quantification.

The limitation of mass spectrometry methods for biology is not primarily sensitivity. It is specificity in a lipid-rich matrix, and that is the problem the field has just begun to confront seriously.

The two questions, and why credible work asks both

Any honest measurement of microplastics in a biological sample has to answer two questions in parallel: what is there at the particle level, and how much is there at the mass level. Each answer constrains the other. If a particle-based method reports a thousand particles of polystyrene in a milliliter of blood, a mass-based method run on the same sample should produce a mass consistent with the count multiplied by a plausible particle-size distribution. If it does not, one of the two measurements is wrong, or there are more particles below the imaging method's resolution limit than the workflow accounted for.

This is the two-method principle, and across the credible modern literature it is the closest thing the field has to a reliability test. Pair a particle-based method (an FTIR variant, a Raman variant, or an O-PTIR/AFM-IR technique depending on size class) with a mass-based method (pyrolysis-GC/MS, TED-GC/MS, or single-particle ICP-MS). Run both on matched aliquots of the same sample. Run both against matched procedural blanks. Validate the entire prep chain with an isotope-traced spike recovery of at least 70% for every polymer you intend to report. Define limits of detection from the blank distribution on the actual matrix, not from vendor specifications on clean reference materials. Report blank-corrected results next to the raw blanks they were corrected against.

The principle is necessary but not sufficient. Two methods run on the same contaminated extract produce two contaminated numbers. If the prep chain introduces background, orthogonal detection does not remove it. And if the matrix itself produces signal that the instruments cannot distinguish from polymer — which, as the next section shows, is exactly what happens in blood — two methods can agree on an answer that is not real.

The reckoning

Reproducibility: eighty-four labs, identical samples, and the variance problem

In 2023 and 2024, the Versailles Project on Advanced Materials and Standards ran an 84-laboratory interlaboratory study on identical microplastic reference samples ^[11]. Each participating lab received the same material, used its normal workflow, and reported back. The coefficients of variation across the participating labs ranged from 45% to 129%, meaning that for a given reference sample, the standard deviation across labs was roughly half to one and a half times the mean value. Put differently, the same material, measured by eighty-four qualified labs using published methods, produced results that disagreed with each other by factors of two to ten.

That is not the variance signature of a clinical assay. It is the variance signature of a field still negotiating what the best way to measure. The VAMAS result is consistent with smaller round-robin studies from the Southern California Coastal Water Research Project, the Joint Research Centre of the European Commission, and the NORMAN network, all of which have found that interlab variance on real samples dwarfs within-lab variance. It is also consistent with the long-observed phenomenon that the same person, tested by two consumer microplastic blood tests in the same month, can receive different results.

The correct reading of the VAMAS result is not that the participating labs are doing bad work. It is that the field does not yet have the standardized prep protocols, reference materials, and validation infrastructure that would make their work commensurable. Every lab is doing something reasonable. The reasonable things do not yet converge.

Contamination: the gloves the community did not believe

In 2020, Witzig and colleagues published a paper showing that standard nitrile laboratory gloves, the gloves worn by essentially every microplastics analyst in the world, shed stearate salt particles into wet-contact samples ^[12]. Stearates of calcium, magnesium, sodium, and zinc are standard mold-release agents used to keep the glove from sticking to its form during manufacture. In the specific infrared region where polymer analysts look for polyethylene, stearates produce carboxylate peaks that can be distinguished from polymer by their specific position. But they are distinguishable only if you are looking for them. If you are not, they count as polymer.

The paper should have changed prep protocols across the field. It largely did not. A literature survey of 26 microplastic method reviews published after Witzig 2020 found that 81% still recommended gloves, a drop of only 7% from pre-Witzig recommendations. The finding had been made, published, and forgotten.

In 2026, Clough and colleagues replicated and extended the Witzig result to dry contact, which had not been directly measured ^[13]. Under 30 newtons of sustained dry contact, the kind of pressure a glove exerts while a technician holds a pipette or presses a sample against a filter, they measured approximately 2,000 false-positive particles per square millimeter of glove contact area. The contaminants were the same family of stearate salts, with a diagnostic carboxylate peak in the infrared that is visible in FTIR but Raman-inactive. That asymmetry matters because a Raman-based workflow could be silently contaminated without the analyst ever seeing a distinguishing peak. Clough also measured a twenty-fold reduction in contamination with specialized nitrile cleanroom gloves compared with standard nitrile gloves, released an open-access stearate spectral library so other labs could screen their existing data for the contaminant, and described a conformal-prediction rescue workflow for re-analyzing studies that had not accounted for it.

Pyrolysis-GC/MS in blood: the critique that cuts deeper

The most consequential recent critique of the microplastic-in-human-biology literature is Rauert and colleagues' 2025 paper, which addresses pyrolysis-GC/MS directly in biological matrices ^[14]. The critique matters because it comes from inside the pyrolysis-GC/MS community — from Minderoo Foundation's clean-lab infrastructure, which sets the current methodological bar — and because it does not challenge anyone's technique or cleanliness. It challenges whether the technique is fit for the matrix at all.

The central finding is that lipids, when pyrolyzed, produce the same alkane and alkene marker compounds that define the calibration curve for polyethylene. Fatty acid chains cracking at 600°C generate homologous series of alkanes and alkenes at exactly the retention times and mass-to-charge ratios that polyethylene pyrolysis produces. A blood sample contains lipids. A cleaner blood sample still contains lipids. Cleanup cannot remove the problem because the cleanup target and the analyte signal are chemically identical in the relevant channel. The implication is that published polyethylene concentrations in blood, which depend entirely on those marker compounds, have been measuring some unknowable mixture of polymer and endogenous lipid pyrolyzate.

Polyvinyl chloride has a different problem. PVC pyrolysis proceeds through HCl elimination to polyene intermediates to polyaromatic hydrocarbons, and the polyaromatic products are not specific to PVC. Any aromatic-rich matrix produces them. In biological samples there is no unique PVC pyrolyzate at all. Published PVC concentrations in blood rest on markers that cannot distinguish PVC from matrix.

Rauert's group also re-derived limits of detection using blank-matched matrix standards rather than Milli-Q water, and found that realistic LODs for polyethylene and PVC in blood are 10 to 20 times higher than the nominal LODs reported in the earlier literature. Spike recoveries ranged from 7% to 109% across polymers. Carboxylated polystyrene nanoparticles recovered at only 17% in initial runs, improving to 52% after prep optimization, an improvement, but still not quantitative.

The paper included a pilot measurement of eight Australian participants, fasting and non-fasting, using the optimized workflow and the new LOD convention. Across all eight participants, no polymer was detected above the realistic LOD. The authors' conclusion, in their own words, is that pyrolysis-GC/MS is not suitable for quantifying polyethylene or polyvinyl chloride in biological matrices.

That is a substantial critique. It does not refute the qualitative observation that polymer particles appear in human samples, which rests on multiple independent techniques. It does refute the claim that the community's dominant mass-quantification method is delivering trustworthy quantitative numbers for certain polymer types.

What the reckoning does not mean

None of this means microplastic exposure is not real. The qualitative evidence for microplastic and nanoplastic presence in human tissue comes from more than one technique, and the non-pyrolysis methods have their own reproducibility problems but do not share pyrolysis's lipid confound. The 2024 human olfactory-bulb detections used micro-FTIR with matched negative procedural blanks and found particles in eight of fifteen decedents ^[15]. Stimulated Raman imaging of bottled water has produced particle counts that are independent of mass attribution. Isotope-labeled model-particle studies in animals have directly tracked polymer uptake. These lines of evidence are not individually conclusive, but they converge on the qualitative observation that plastic particles enter human tissue through multiple routes. That observation is stable.

What is not stable is the quantitative overlay. Saying "we find plastics in the human body" is a defensible statement today. Saying "the average person carries X micrograms of polyethylene per milliliter of blood" is not. The gap between those two sentences is the entire point of this piece.

The view from the patient, the clinician, and the consumer

Consumer blood tests in 2026

A small but increasingly visible direct to consumer market now offers microplastic blood tests. Blueprint, Bryan Johnson’s protocol, sells one for about $150, and several smaller providers now market similar services. Some, such as Arrow Lab Solutions, appear to rely on microscopy to count particles in roughly the 1 to 70 μm range. That approach can detect certain visible microplastics, but it has important limits: it captures only a narrow slice of the particle size spectrum and is blind to nanoplastics.

Even at the upper end of the market, where Py-GC/MS is genuinely used, the same structural limits remain. There is still no validated population reference range for normal microplastic content in human blood, because no large cohort has been measured with a single standardized method. A result such as “12 μg/mL of polyethylene” therefore cannot be interpreted as high, low, or average in any meaningful clinical sense. The field is still too early for those answers.

Consumer cholesterol testing in the 1950s had the same shape: the underlying chemistry was real, the link to disease was suspected, and a number could be produced, but normal ranges, fasting protocols, instrument standardization, and the connection between measurement and clinical action took decades to build.

What a clinically useful measurement would require

The technical requirements for a clinical-grade microplastic assay are easy to specify and almost entirely unmet. Both mass and particle information per sample, via paired methods rather than a single technique. Polymer-specific reporting for at least the seven most common polymers (e.g., polyethylene, polypropylene, polystyrene, polyethylene terephthalate, polyvinyl chloride, polyamide, polycarbonate) with per-polymer limits of detection defined against blank-matched matrix standards. Validated, isotope-traced spike recovery for every matrix. End-to-end procedural blanks that travel with every sample from collection through analytical reporting. Reference standards traceable to a national metrology institute, which currently do not exist for polymer particles in biological matrices and are being worked on at the National Institute of Standards and Technology and several European metrology institutes.

And then, even once the analytical layer is solved, the clinical layer is still further away. Population reference distributions of normal burden across age, sex, geography, and exposure group, an epidemiological undertaking on the scale of NHANES. Longitudinal stability data to distinguish a single result from noise. Linkage of measured burden to health outcomes that an intervention could influence. Validated interventions whose effect on burden is itself measurable. None of these exist as of today for microplastics and nanoplastics.

The ARPA-H STOMP program, and what it concedes

The United States Advanced Research Projects Agency for Health launched the Systematic Targeting Of MicroPlastics program (STOMP) as an explicit response to this gap ^[16].

STOMP is structured in two sequential phases. The first, running 24 months, is dedicated to building laboratory measurement systems capable of detecting and characterizing nano-sized microplastic particles in biological tissue — research-grade infrastructure with the validation the current literature lacks.

The program's structure is itself a diagnosis of the field. By explicitly separating research-grade lab measurement from affordable clinical-grade measurement, and by giving the first one two years of dedicated funding before the second begins, STOMP is acknowledging that today's research methods are not yet validated to the point of driving clinical decisions.

For a reader trying to make sense of the current state, STOMP is the clearest single statement of where the field actually is. The problem is real, the measurement is not yet ready for clinical use, and the path to closing the gap is a coordinated multi-year program of work that has only just begun.

The honest summary

The qualitative observation that microplastic and nanoplastic particles reach human tissue is supported across multiple independent techniques and is, at this point, the stable finding of the field. The quantitative overlay — specific concentrations of specific polymers in specific tissues — is not yet stable for the most common polymers in the most important matrices. The toxicological link between measured burden and health outcomes in humans at realistic exposures remains.

The honest reading is that this is a new field. These phases are normal for a measurement problem this difficult at this early stage. They do not justify dismissal, and they do not justify alarm about specific numbers. What they do justify is patience with the process, skepticism about precise concentration claims, and continued investment in the measurement infrastructure that would let the quantitative layer catch up to the qualitative one.

The fact that we find plastic inside us is, in the most direct sense, disturbing. What we do not yet know is exactly how much, in which tissues, of which polymers, at which sizes, compared with what baseline. Both of those sentences are true at the same time. Holding them together, neither dismissing the finding nor overreading its magnitude, is the only honest posture available today.

That is the work. It will take years. In the meantime, the best response to a new microplastic headline is not to ask what the number is. It is to ask how it was measured, whether the prep chain was validated, whether the blanks were reported, and whether a second orthogonal method agrees with the first. A reader who asks those questions will be better equipped to interpret any future study than a reader who takes the headline number at face value. And, if the field moves in the direction STOMP is pointing it, that reader will eventually have numbers worth taking at face value.