Microplastic risk models may be built on a mathematical illusion, review warns

A critical review published in the journal Microplastics and Nanoplastics argues that human health risk assessments for microplastics rest on an assumption that the available evidence cannot support: that estimates of how much plastic people ingest can be translated into internal tissue doses and lifetime body burdens. The paper, authored by Timothy Yu Yee Ong, Long Chiau Ming, Raja Affendi Raja Ali and Lai Ti Gew, contends that the obstacle is not simply a shortage of data but a structural flaw in how the underlying toxicokinetic systems have been measured, leaving the key parameters of any human microplastic toxicokinetic model mathematically unidentifiable.

The authors conducted a scoping review of 22 eligible studies spanning human biomonitoring, ex vivo organoid experiments, animal work and in vitro mechanistic research. Their central finding is stark: no dataset in the literature co-measures intake and elimination in the same individuals. Tissue detections, meanwhile, aggregate burdens from multiple unattributed routes, including ingestion, inhalation and dermal contact. Without knowing how much plastic enters a person, by which route, and how much leaves or accumulates, the fixed-fraction retention models commonly used in the field are being fitted to structurally under-determined systems.

To demonstrate the problem concretely, the team built a synthetic dependency and sensitivity analysis using a literature-anchored synthetic dataset and probabilistic Monte Carlo simulations. When excretion was defined as a fixed fraction of intake, the simulation produced a seemingly robust, near-perfect relationship between estimated ingestion and faecal excretion, with Pearson correlation coefficients ranging from approximately 0.88 to 0.98 across Monte Carlo conditions. In the headline dependency simulation the correlation reached r = 0.949, and the authors note that no significance test is reported for it because the variables are dependent by construction.

That near-perfect correlation, the review argues, is a purely mathematical artifact of the model’s structural architecture rather than evidence of biological coupling. In other words, a model that assumes excretion is a fixed proportion of intake will inevitably show intake and excretion rising and falling together, no matter what biology does. The authors formalize the issue with a convolution model of multi-route tissue burden, showing that individual clinical biomonitoring correlations, such as those observed in tissue biopsies or bodily fluids, are fundamentally uncoupled from ingestion pathways because of unmeasured multi-route confounding.

The quantitative analyses in the review are restricted to microplastics of 1 micrometre or larger. Nanoplastics, defined as particles smaller than 1 micrometre, are treated qualitatively throughout, because the analytical methods that detect them and the toxicokinetic evidence available for them differ substantially from those for microplastics. This distinction matters for risk assessment, since the two particle classes cannot simply be pooled under a single exposure-dose framework.

From this diagnosis, the authors derive the minimum identifiability conditions that any valid human microplastic toxicokinetic model must satisfy. These are longitudinal, same-subject, route-resolved measurements of intake, elimination and retention. In practical terms, a valid model would require tracking the same individuals over time, measuring how much microplastic they take in through each exposure route, how much they excrete, and how much remains in their tissues. The review translates these conditions into concrete study-design and reporting requirements for future research.

The reframing the authors propose shifts the field’s central problem from “insufficient data” to “structurally uninformative data.” The distinction is consequential. Collecting more of the same cross-sectional biomonitoring measurements would enlarge the literature without resolving the identifiability barrier, because such data cannot disentangle ingestion from inhalation or dermal contributions to tissue burden. What is needed instead are measurement designs specifically constructed to make quantitative body-burden modelling possible.

The review is published as open access and was released on 3 November 2026 under a share-first arrangement, meaning the version available is subject to further edits before the final Version of Record replaces it. As a Critical Review, it does not report new human measurements but evaluates whether the existing human evidence base can support the translation from external exposure to internal dose that risk assessment currently assumes. Its conclusions apply to microplastics of 1 micrometre and larger, with nanoplastics addressed only qualitatively.

For the risk assessment community, the paper’s message is that the near-perfect intake-excretion correlations that have lent apparent credibility to fixed-fraction retention models are consequences of model construction, not evidence that human microplastic kinetics behave that way. Until studies meet the identifiability conditions the authors set out, any quantitative estimate of lifetime microplastic body burden derived from ingestion exposure estimates remains, in their framing, unsupported by the human evidence base.

Subject of Research: Structural non-identifiability of human microplastic toxicokinetic models

Article Title: When exposure isn’t dose: parameter non-identifiability as the central barrier to human microplastic toxicokinetics

Article References: Ong, T. Y. Y., Ming, L. C., Ali, R. A. R., & Gew, L. T. (2026). When exposure isn’t dose: parameter non-identifiability as the central barrier to human microplastic toxicokinetics. Microplastics and Nanoplastics. https://doi.org/10.1186/s43591-026-00245-x

Image Credits: AI Generated

DOI: 10.1186/s43591-026-00245-x

Keywords: microplastics, toxicokinetics, risk assessment, parameter identifiability, biomonitoring, Monte Carlo simulation, exposure routes, body burden, scoping review, nanoplastics, exposure, dose