The study, published in the journal Environmental Advances, was conducted by a team of toxicologists led by Aryatara Shakya of TRC Environmental, working in collaboration with Charles River Laboratories in Edinburgh, United Kingdom. Rather than relying on bulk content measurements alone — the approach that dominated earlier research, including a widely publicized 2024 study that found metal(loid)s in every tampon tested — the team built an experimental framework designed to translate trace metal content into an estimate of actual absorbed dose. That distinction matters because, as critics of the earlier work pointed out, finding lead at nanogram-per-gram levels in a product tells you nothing about whether the body ever takes it up.

The researchers focused on six elements chosen for their toxicological relevance and their routine appearance in regulatory biocompatibility evaluations: arsenic, cadmium, cobalt, chromium, lead, and mercury. Trace amounts of these metals are ubiquitous in plant-derived textiles like cotton and rayon, introduced through uptake from soil, atmospheric deposition, agrochemicals, and industrial processing — not intentional addition. The team confirmed this baseline, detecting all six metals in dry tampon material at low nanogram-per-gram concentrations, with lead the only element reliably quantifiable, at a mean of about 88.7 ng/g — consistent with the geometric mean of 120 ng/g reported across 30 tampons from 14 brands in the 2024 study.

The first experimental phase simulated actual use. Whole commercial tampons were incubated in 12 milliliters of artificial menstrual fluid for 12 hours at 37 degrees Celsius — deliberately longer than the average wear time of roughly 4.4 to 4.6 hours and even the labeled maximum of 8 hours, to maximize per-tampon metal release. The artificial menses fluid was formulated to be physiologically realistic, containing hemoglobin, albumin with sulfhydryl groups, and ferrous iron at a slightly alkaline pH of 7.4, mimicking the chemistry of real menstrual blood. The results were striking: metal release into the fluid was minimal across all six elements, with most concentrations falling at or below the limits of quantification. Lead was again the only metal quantifiable, peaking at 3.4 ng/mL — a mass-transfer figure directionally consistent with the modest drop in lead measured in the soaked tampon material itself.

The second phase addressed the critical unknown: whether metals that do leach out can penetrate vaginal tissue. The team used the EpiVaginal full-thickness model, a three-dimensional reconstructed human vaginal-ectocervical tissue built from primary human epithelial cells and fibroblasts cultured at an air-liquid interface. The model, in use since 2003, forms stratified, multilayered tissue with tight junctions and desmosomes that closely mimics the non-keratinized, highly vascularized vaginal mucosa — the only commercially available model of its kind. Tissues were exposed for 12 hours to the raw leachate and to two surrogate solutions prepared at 1,000-fold and 10,000-fold the measured leachate concentrations, deliberately creating worst-case bounding scenarios.

After exposure, the researchers measured metal distribution across three compartments: a rinse fluid capturing surface-associated material, the tissue itself, and a receptor fluid representing the bloodstream side of the epithelial barrier. The pattern was unambiguous. For every metal at both surrogate concentrations, the vast majority of recovered mass — between 50 and 95 percent — stayed in the rinse, meaning it remained surface-bound and would simply drain away with menstrual flow. Tissue-associated fractions were modest, and transfer into the receptor fluid was nearly absent: detectable only for arsenic, chromium, and lead, and even then representing just fractions of a percent to a few percent of the applied dose. Transepithelial electrical resistance measurements confirmed the tissue barrier remained intact throughout, with no exposure-related impairment.

The final step was quantitative risk characterization using the margin of safety (MoS) approach codified in ISO 10993-17, the international standard for toxicological risk assessment of medical device constituents — a framework particularly apt here because the U.S. FDA classifies tampons as medical devices. The team calculated estimated daily exposure doses, combining the maximal single-tampon release with an aggressive assumption of six tampons per day, and compared those doses against permitted daily exposure values from the ICH Q3D guidance on elemental impurities. Parenteral PDEs were deliberately chosen as conservative benchmarks, since they assume direct delivery into systemic circulation without any reduction from incomplete absorption.

Under the measured leachate conditions, the results were emphatic. Margin of safety values — where a value of at least 1 indicates acceptable risk — ranged from approximately 97,000 for cobalt to roughly 64.5 million for chromium. Even under the deliberately extreme 10,000-fold surrogate scenario, which has no plausible physiological basis, all margins remained above unity, from about 24 for cadmium to nearly 4,000 for chromium. A sensitivity analysis that assumed six tampons used every single day of the year — an obviously impossible scenario — still left all margins above the acceptance threshold, with cadmium, the most conservative case, at approximately 3.9.

The implications for lead, the metal that generated the most public concern following the 2024 findings, are particularly significant. Because no threshold exists for lead neurotoxicity, the team evaluated absorbed dose rather than product content. They estimated systemic lead exposure of roughly 0.00001 micrograms per day under real leachate conditions — thousands of times below the FDA’s interim reference level of 8.8 micrograms per day for females of childbearing age. The predicted contribution to blood lead would be well below 0.01 micrograms per deciliter, analytically indistinguishable from background. Blood lead levels, the authors conclude, are unlikely to differ measurably between tampon users and non-users. Their findings also align with independent mechanistic modeling published in Toxicological Sciences in 2026, which predicted that less than 1 percent of lead released from a tampon would be taken up by vaginal tissue, and that lead naturally present in menstrual fluid exceeds the amount absorbed from the product itself.

For chromium, the team addressed the speciation question by attributing measured total chromium to Cr(III), the form that predominates in plant-derived materials and is actively generated by the reductant-rich chemistry of menstrual fluid, which converts any Cr(VI) to Cr(III) at physiological pH. Even under a precautionary assumption that all chromium were the more hazardous Cr(VI), the margin of safety would still be roughly 6,030 under actual use conditions.

The authors are careful to situate their findings in a broader risk-communication context. Ultra-trace detection at nanogram levels, they note, reflects the extraordinary sensitivity of modern analytical instrumentation, not evidence of harm. Professional toxicology organizations, including the American College of Medical Toxicology, have already cautioned that detection-only messaging can foster undue alarm — and in some cases dangerous responses such as unnecessary chelation therapy. Tampons, they emphasize, serve an important public health function, supporting education, employment, and daily life for hundreds of millions of people worldwide.

The study does have limitations: it evaluated a single commercial product, did not analytically determine chromium speciation, and, like all in vitro systems, cannot fully replicate living vaginal physiology with its hormonal cycling, mucus turnover, and microbiome activity. But the researchers argue these factors cut toward conservatism — the static design likely overestimates contact time, and the tissue model may be more permeable than native mucosa. The regulatory timing is notable: the FDA has initiated bench studies of metal release from tampons, and the International Organization for Standardization’s technical committee TC 338 is developing global safety standards for menstrual products. The framework established here — leaching under physiological conditions, direct measurement of epithelial permeation, and absorbed-dose-based risk characterization — offers regulators a reproducible template. The arithmetic, the authors note, is ultimately constraining: a tampon containing a metal at even one part per million holds only a few micrograms in total, and only a fraction leaches, and only a fraction of that crosses tissue. Realistic worst cases simply cannot approach toxicological thresholds — and now, for the first time, that conclusion rests on direct experimental evidence rather than assumption.

A new study addresses this blind spot by blending multiple data sources, including low-cost air quality sensors and satellite observations, to build a higher-resolution picture of PM2.5 across Taiwan. The approach is designed to complement traditional networks rather than replace them, using inexpensive instruments to expand spatial coverage where conventional monitors are lacking.

The researchers focus on integrating sensor measurements with satellite-derived aerosol information. Satellites can observe broad areas, but translating their signals into ground-level PM2.5 requires careful modeling to account for atmospheric conditions and calibration differences. Low-cost sensors, meanwhile, can capture local variation but may drift or underperform in complex environments unless corrected.

To overcome these limitations, the team developed a high-resolution PM2.5 model that fuses data streams into a unified framework. The method leverages the satellites’ wide-area perspective while using ground-based low-cost sensors to anchor predictions at finer spatial scales. By doing so, the model aims to reduce uncertainty caused by both monitoring gaps and satellite retrieval biases.

High resolution matters because urban pollution patterns can change dramatically over short distances due to traffic, industry, and meteorology. Capturing that variability can improve exposure assessment, helping researchers and decision-makers identify areas at greater health risk rather than relying on citywide averages.

Importantly, the study highlights a practical path for countries with limited monitoring infrastructure. As low-cost sensing networks become more common, their value increases when paired with satellite data and robust statistical correction. The result is a modeling system that is both data-rich and scalable.

With PM2.5 linked to respiratory and cardiovascular outcomes, better spatial accuracy could support earlier warnings and more targeted interventions. The framework demonstrated in Taiwan may offer a template for other regions facing similar constraints in monitoring coverage.

Subject of Research: Development of a high-resolution PM2.5 model using integrated low-cost sensors and satellite observations.

Article Title: Development of a high-resolution PM2.5 model in Taiwan using integrated low-cost air quality sensors and satellite observations.

Article References: Jung, CR., Chuang, WH., Chen, WT. et al. J Expo Sci Environ Epidemiol (2026). https://doi.org/10.1038/s41370-026-00953-9

Image Credits: AI Generated

DOI: 10.1038/s41370-026-00953-9

Keywords:

The impetus behind this innovative research lies in the complex realities of human chemical exposure. Individuals are simultaneously subjected to a variety of chemicals present in air, water, food, consumer products, and occupational environments. Previous models frequently failed to capture the cumulative burden imposed by these varied sources, potentially underestimating adverse health outcomes linked to chronic low-dose exposure. By employing a Bayesian perspective, the study harnesses probabilistic modeling to synthesize heterogeneous data streams, yielding a high-resolution picture of total chemical intake.

At its core, the Bayesian approach offers a natural and rigorous framework for integrating prior knowledge and empirical data to estimate exposure distributions with inherent uncertainties accounted for. This contrasts sharply with deterministic methods that often simplify complex exposure dynamics into static, average metrics. The probabilistic nature of Bayesian inference permits detailed characterization of variability across populations and temporal fluctuations in exposure, enhancing the fidelity of risk projections.

The study unfolds with a detailed mathematical formulation that models exposures from multiple pathways concurrently. These include inhalation of ambient pollutants, ingestion of contaminated food and water, dermal absorption from consumer products, and accidental occupational contacts. By structuring each source as an input variable with associated likelihoods, the Bayesian model calculates posterior distributions representing aggregated exposures. This allows for nuanced investigation into the interplay between sources that can combine in additive or synergistic fashions.

To validate their framework, the researchers applied it to a comprehensive dataset integrating biomonitoring results, environmental measurements, and consumer behavior surveys. This multi-layered dataset underscored the importance of considering aggregated exposures; for instance, chemicals present at seemingly innocuous levels in individual media collectively reached concentrations in blood and urine biomarkers suggestive of significant health risk. The Bayesian model successfully reconstructed these biomarker levels, thereby affirming its predictive power.

One transformative aspect of this research lies in its capability to pinpoint dominant exposure sources within the composite profile. Regulators and public health officials can leverage such insights to prioritize interventions strategically, focusing on pathways that contribute disproportionately to total chemical load. This targeted approach contrasts with blanket policies and can optimize the allocation of limited resources for maximum health benefit.

Furthermore, the Bayesian framework inherently accommodates incorporation of emerging data, facilitating dynamic updates to exposure assessments as scientific understanding and environmental conditions evolve. This adaptability is crucial for addressing rapidly changing exposure landscapes, exemplified by the proliferation of novel chemicals and shifting lifestyle practices. By embracing continuous refinement, this methodology fosters responsiveness in public health decision-making.

The paper also delves into the challenges of scaling aggregated exposure assessments across diverse demographic groups and geographic regions. Variations in chemical use patterns, socioeconomic status, and environmental quality contribute to heterogeneous exposure profiles. The Bayesian method elegantly manages such complexity, enabling stratified analyses that reveal vulnerable subpopulations otherwise obscured in aggregate statistics.

Importantly, the authors highlight the implications of their approach for cumulative risk assessment frameworks, which have garnered increasing regulatory attention globally. By providing robust quantification of total chemical burdens, the model facilitates integration of mixture toxicity considerations into risk evaluations. This aligns with a growing consensus that addressing single chemicals in isolation fails to capture the real-world context of human exposure.

Technical rigor permeates the study’s methodology, including sophisticated algorithms for Markov Chain Monte Carlo (MCMC) simulations used to approximate posterior distributions. The researchers also conducted thorough sensitivity analyses to evaluate model responsiveness to varying assumptions and data quality. These methodological assurances instill confidence in the model’s applicability for both research and regulatory purposes.

Beyond regulatory science, the insights derived from this Bayesian approach bear relevance for individual risk communication and personalized exposure management. As wearable sensors and personal monitoring devices improve, coupling their data streams with such probabilistic frameworks could empower individuals with actionable exposure profiles, fostering preventative health behaviors.

The research also paves the way for integrating chemical exposure assessments with biomarker-based health effect studies, opening avenues for causality inference in environmental epidemiology. By aligning predicted aggregate exposures with observed health endpoints, scientists can better elucidate dose-response relationships and refine safety thresholds based on real-life complexities.

The study’s broader impact extends to informing industry practices and chemical management policies. Companies can use these modeling tools to proactively assess cumulative exposures arising from product lines or manufacturing processes, enabling safer design and marketing strategies that minimize health risks.

In conclusion, the Bayesian approach to aggregated chemical exposure assessment delineated in this study represents a transformative leap toward capturing the full spectrum of human chemical burdens. By uniting multiple exposure sources within a dynamic, probabilistic framework, it addresses longstanding limitations of conventional assessments and enhances our capacity to safeguard public health in an increasingly complex chemical landscape. As this methodology gains adoption, it promises to catalyze interdisciplinary collaborations, refine risk governance, and ultimately contribute to healthier environments worldwide.

Subject of Research: Human chemical exposure assessment using a Bayesian probabilistic framework to model aggregated exposure from multiple environmental and consumer sources.

Article Title: A Bayesian approach to aggregated chemical exposure assessment.

Article References:
Van Den Neucker, S., Grigoriev, A., Demaegdt, H. et al. A Bayesian approach to aggregated chemical exposure assessment. J Expo Sci Environ Epidemiol (2026). https://doi.org/10.1038/s41370-026-00900-8

Image Credits: AI Generated

DOI: 21 April 2026