← Back to Hazard Characterisation

Concept: MNP Hazard Scoring System

A tiered three-domain framework for micro- and nanoplastic risk characterisation — standalone testing tool

NOTE: Calculation models are under active development and may not be accurate.

Overview

This Hazard Scoring System is being tested as a possible replacement for the current semi-quantitative banding approach, implementing a structured, multi-dimensional framework that evaluates three distinct hazard domains. Each domain is rated on a 1 (lowest hazard) to 5 (highest hazard) scale, then combined via life-stage-weighting to produce a composite score.

The system is designed to incentivise high-quality data provision: data quality modifiers adjust domain scores upward when only coarse or missing data are available, while high-quality measurement data incur no penalty. Missing sub-parameters default protectively to 5 (worst-case assumed until proven otherwise).

This tool evaluates one particle type at a time. For complex environmental samples (e.g. house dust) containing multiple particle types, each type should be scored separately. See the Mixture Builder section below for discussion on combining scores. For now, an expert user must manually apply the scores across each domain by drop-down selections. In the future, this approach can be automated based on good quality data stored in the hazard characterisation database.

Interactive Scoring Tool

Select the appropriate sub-parameter levels for your particle type below. Missing (unselected) fields are defaulted to a protective score of 5. Then choose a data quality level and life stage, and click Calculate to see the composite hazard score.

A Physical & Kinetic

B Intrinsic Chemical

C Acquired (Extrinsic) Hazard

Scoring Results

Data Quality Modifier
—
Composite Score (1–5)
—
Scaled Score (0–100)
—
Hazard Tier
—

Domain A — Physical & Kinetic Hazard

Evaluates the physical dimensions of the particle that influence its behaviour in biological systems, including transport, deposition, translocation, and cellular interaction.

Sub-Parameter Score 1 Score 2 Score 3 Score 4 Score 5
Particle Size Well-characterised continuous PSD; validated across 1 nm–100 µm range; nano-fraction quantified Continuous PSD available but validated over narrower range (e.g. 1–100 µm only) Binned PSD (e.g. LDIR bins) with ≥5 bins; nano-fraction not resolved Single discrete size estimate; no distribution information No size information; particle size unknown
Morphology Well-characterised by multiple microscopy methods; aspect ratio and shape distribution quantified Morphology confirmed by microscopy; shape categories known qualitatively Morphology inferred from source or process; no direct imaging Vague description only (“fragments/fibres” reported but not quantified) Completely uncharacterised; morphology unknown
Surface Area Measured BET surface area (m²/g) available Surface area estimated from geometry + dimensions Surface area inferred from polymer type or literature proxy Only qualitative surface description (e.g. “high surface area”) No surface area information at all
Agglomeration State Zeta potential measured; Two-State classification applied; fluid-specific correction Zeta potential measured in reference medium but no fluid correction Zeta potential estimated from literature for similar particles Qualitative stability description only No agglomeration information; unknown stability

Domain B — Intrinsic Chemical Hazard

Assesses the inherent toxicity of the polymer material itself, its additive content, and the potential for chemical migration or leaching in biological environments.

Sub-Parameter Score 1 Score 2 Score 3 Score 4 Score 5
Polymer Hazard Inert, food-grade polymers (PE, PET, silicone) with no known chronic toxicity Low-hazard polymers (PP, PA, PMMA) with minimal additive burden Moderate-hazard polymers (PS, ABS) with known additive leaching potential High-hazard polymers (PVC, PC) with toxic monomer residues or plasticisers Extremely hazardous (PU, epoxy, biocide-embedded) or unknown composition
Additive Load No additives detected or all additives confirmed non-toxic at expected exposure levels Minimal additives; all within regulatory limits; low Log Kow Some additives present (plasticisers, stabilisers); some may be of concern Multiple additives detected; known hazardous plasticisers (DEHP, BPA, PBDEs) present Highly toxic additive burden; restricted substances; endocrine disruptors present at >1% w/w
Chemical Migration Potential No measurable migration under simulated physiological conditions (gastric, lung, plasma) Low migration rate; most additives below detection in leachate Moderate migration; some additives leach in relevant fluids High migration; several additives detected in leachate at >1 µg/cm² Very high migration or no migration data available (assumed worst-case)
Environmental Contaminant Adsorption No detectable adsorbed contaminants (PAHs, PCBs, PFAS, metals) on particle surface Low-level adsorbed contaminants detected; below toxicological thresholds Moderate contaminant load; some contaminants above threshold of concern High contaminant load; multiple priority pollutants detected Extreme contaminant load (Trojan Horse risk) or no data available

Domain C — Acquired (Extrinsic) Hazard

Captures hazards acquired during the particle's environmental lifetime — weathering, fouling, and biological colonisation — that modify its toxicity relative to the pristine material.

Sub-Parameter Score 1 Score 2 Score 3 Score 4 Score 5
Weathering & Ageing Pristine / unweathered; reference material with documented storage Minimal weathering (e.g. laboratory-grade, short-term environmental exposure) Moderately weathered; surface oxidation detectable by FTIR or XPS Heavily weathered; embrittled, cracked, or fragmented; UV-degraded Extreme weathering state or unknown (assumed worst-case)
Biofilm & Colonisation No biofilm detected by microscopy or molecular methods Incidental biofilm; low cell density; non-pathogenic species only Moderate biofilm; mixed microbial community; potential for pathogenic colonisation Dense biofilm; pathogens detected (e.g. Pseudomonas, Legionella); potential for mobile genetic elements Heavy biofilm with confirmed pathogens and MGEs / ARGs or no data available
Environmental Conditioning Particle never exposed to environmental matrices; lab-created pristine Short-term environmental exposure under controlled conditions Environmental sample with moderate residence time; matrix origin documented Environmental sample from polluted matrix; hazardous co-contaminants likely Unknown environmental history; no matrix documentation

Data Quality Modifiers

To incentivise high-resolution data provision and transparent handling of uncertainty, domain scores are adjusted by the following data quality modifiers based on the lowest quality sub-parameter in each domain:

Data Quality Level Description Adjustment
Gold Continuous PSD with validated particle number, surface area, and mass metrics; experimentally determined translocation fractions and zeta potential with fluid correction +0.00
Silver Continuous PSD or well-resolved bins (≥5) without surface area data; zeta potential in reference medium only; morphology by microscopy +0.50
Bronze Binned PSD (<5 bins) or single discrete size; no zeta; morphology inferred +1.25
Limited Mass-only concentration; no particle sizing; no physicochemical characterisation +1.75
Missing Sub-parameter not measured or reported; defaults to protective score of 5 N/A (5)

Protective defaults principle: Any sub-parameter for which no data are available defaults to a score of 5 (worst-case assumption). This creates a strong incentive for data generation: “if you don't measure it, we assume the worst.” Data quality adjustments are applied per domain based on the lowest-quality sub-parameter in that domain. The adjustments represent a transparent penalty for uncertainty rather than a subjective correction factor.

Life-Stage Vulnerability Weighting

Rather than shifting domain weights per life stage (which can produce counterintuitive results depending on the particle profile), the composite hazard score is calculated as the unweighted average of the three domain scores (after data quality adjustments), then multiplied by a life-stage vulnerability multiplier:

Life Stage Multiplier Rationale
Foetus (in utero) ×1.20 Highest vulnerability; placental translocation, organogenesis, immature detoxification
Neonate (0–6 mo) ×1.15 Extremely vulnerable; immature blood-brain barrier, developing immune system
Infant (6–12 mo) ×1.10 Elevated vulnerability; oral exploration behaviour, continued neurological development
Toddler (1–3 yr) ×1.05 Moderately elevated; increased environmental contact, approaching metabolic maturity
Adult (reference) ×1.00 Reference baseline; fully developed detoxification and barrier systems

This multiplier approach guarantees that younger = higher composite score for any particle profile, reflecting the intuitive principle that a foetus or neonate is systematically more vulnerable to any given hazard than an adult. The multipliers are expert-elicited proposals and can be refined as empirical data on life-stage sensitivity become available.

Strengths & Weaknesses

Strengths

  • Tiered, transparent structure: The three-domain architecture separates conceptually distinct hazard dimensions, making the scoring rationale clear and auditable.
  • Data quality incentives: The progressive quality modifiers create a measurable penalty for uncertainty, providing a concrete driver for high-resolution characterisation (continuous PSD, surface area, zeta potential).
  • Protective defaults: “If you don't measure it, we assume the worst” is a scientifically justifiable precautionary approach for vulnerable early-life stages, and it actively encourages data generation.
  • Life-stage sensitivity: A vulnerability multiplier ensures that younger life stages (e.g., foetus, neonate) always receive a higher composite hazard score than older ones for the same particle profile, reflecting their heightened biological susceptibility.
  • Modular & extensible: New sub-parameters can be added to any domain without restructuring the whole framework. The scoring algorithm is straightforward (unweighted average with additive DQ modifiers and a life-stage vulnerability multiplier) and amenable to sensitivity testing.
  • No false precision: The 1–5 integer scale avoids the appearance of exactitude that continuous scoring would imply, given the underlying data limitations.

Weaknesses

  • Expert judgement still required: Mapping real-world data onto the scoring tables requires interpretation. Two experts scoring the same particle may arrive at different scores, particularly for borderline cases.
  • No formal inter-rater reliability testing: The scoring tables have not been validated against a panel of independent raters. Reproducibility is assumed but unproven.
  • Life-stage multipliers are expert-elicited, not empirically derived: The vulnerability multipliers (1.20, 1.15, 1.10, 1.05, 1.00) are reasoned proposals, not the output of a quantitative meta-analysis or formal multi-criteria decision analysis. They can and should be refined as empirical data on life-stage sensitivity become available.
  • Biofilm/pathogen domain is speculative: While microbiological hazard is conceptually important (Noventa et al. 2021), the empirical basis for scoring biofilm on environmental MNPs is thin. Most studies report presence/absence only, not colonisation density or pathogen identification.
  • Composite score can mask domain-specific extremes: A particle scoring 5 in Domain B but 1 in Domains A and C would receive a moderate composite score, potentially understating the chemical risk. Domain-specific reporting (always shown alongside the composite) mitigates this but does not eliminate it.
  • Data quality modifiers are additive rather than multiplicative: An additive modifier (+0.75, +1.75) treats uncertainty as a constant increment regardless of the underlying domain score. A multiplicative modifier (e.g. ×1.2 for binned data) might better reflect that the relative uncertainty scales with the hazard level, but adds complexity and a second layer of expert judgement.

Ideas for Later Expansion

Multi-Metric Reporting

The current scoring algorithm operates on a single composite hazard score (1–5) with a scaled 0–100 index and a Low/Medium/High tier. A natural extension is multi-metric reporting, where the hazard is expressed across four complementary dimensions:

  1. Particle number hazard — the current metric; hazard as a function of exposure to discrete particle entities.
  2. Surface area hazard — hazard normalised to total particle surface area (mm²), which correlates with cellular interaction and catalytic surface activity.
  3. Chemical load hazard — the Trojan Horse dimension; hazard expressed as additive + adsorbed contaminant dose per particle (ng/particle).
  4. Mass hazard — conventional mass-based metric (ng/kg-bw/day) for backward compatibility with chemical risk assessment frameworks.

Each dimension would be derived from the same underlying data (particle count, dimensions, density, chemical loading per particle) via post-hoc conversion functions. The user could toggle between metrics in the results display, seeing how the hazard tier shifts across dimensions. This would address the recurring critique that “no single metric is sufficient” for MNP risk characterisation (Koelmans et al. 2026).

Mixture Builder

Environmental exposure is never to a single particle type — it is always a mixture of diverse polymer types, sizes, and chemical loads. A mixture builder would allow the user to:

  • Select or define an exposure matrix (e.g. “European housedust”, “Bottled water”) from the Hazard Characterisation Database, which specifies the fractional abundance of each particle type.
  • Choose an aggregation rule:
    • Abundance-weighted average: The composite hazard score is the weighted mean of each particle type's score, weighted by its fractional abundance in the matrix. This is appropriate when the hazard is additive (e.g. chemical load).
    • Maximum (toxic driver): The composite hazard score is the maximum score among all particle types in the mixture. This is appropriate when a single highly toxic particle type drives the overall risk (the “weakest link” principle for non-additive endpoints).

The mixture builder would transform the scoring tool from a single-particle evaluator into a realistic exposure-scenario hazard profiler, bridging the gap between the controlled laboratory particle and the complex environmental reality.

Integration with the Exposure Models

The ultimate goal is to feed the hazard scoring output directly into the Monte Carlo exposure models, replacing the current manual polymer + chemical additive sliders with a data-driven hazard score derived from the database. The integration pathway:

  1. The exposure model selects an exposure matrix from the database (or the user defines a custom one).
  2. For each Monte Carlo iteration, the model samples a particle type from the matrix composition (weighted by fractional abundance).
  3. The particle type's stored physicochemical data (size, zeta, chemical load) are used to compute a domain-specific hazard score on-the-fly, using the same algorithm implemented in this standalone tool.
  4. The exposure dose (particles/day) is multiplied by the hazard score to produce a hazard-weighted dose, which replaces the current qualitative “Low/Medium/High” banding with a continuously variable, particle-specific risk metric.

This integration would take the framework from “exposure-led with a qualitative hazard adjustment” to a fully integrated, data-driven risk assessment pipeline — while maintaining transparency at every step.