Main

For decades, air pollution epidemiology has primarily focused on six criteria pollutants: particulate matter, ground-level ozone, nitrogen dioxide, sulfur dioxide, lead and carbon monoxide. These studies have been instrumental in establishing the relationship between air pollution and numerous adverse health outcomes, including respiratory diseases, cardiovascular effects and increased mortality1,2. However, this approach has limitations that restrict understanding of how the airborne exposome affects human health.

The exposome concept, first introduced by Wild3 in 2005, offers a more comprehensive framework for understanding human environmental exposures. The exposome encompasses the totality of environmental exposures from conception onwards, including environmental factors (such as air pollution, diet and lifestyle) and biological responses (such as inflammation, oxidative stress and metabolic changes)4. Unlike the traditional focus on single pollutants, the exposome approach recognizes the complex, dynamic nature of environmental exposures across the lifespan and their cumulative impact on health. In the context of air pollution, the airborne exposome includes not only regulated criteria air pollutants but also the myriad chemical, biological and physical agents that individuals encounter throughout their daily lives5,6,7,8,9.

Beyond criteria pollutants

Current regulatory approaches to air pollution focus on a small subset of pollutants that were identified decades ago as indicators of air quality. These criteria pollutants, though important, represent only a small portion of the airborne substances that humans are exposed to daily. Particulate matter comprising fine particles with a diameter of ≤2.5 µm (PM2.5) or ≤10 µm (PM10), for example, is regulated based on size rather than composition, despite growing evidence that their chemical and non-chemical compositions actually influence health effects10,11 (Fig. 1).

Fig. 1: The airborne exposome.
figure 1

Criteria air pollutants represent only a fraction of the airborne exposome, which is generated from various sources. DEHP, di-2-ethylhexyl phthalate; MNPs, micro- and nanoplastics; PAHs, polycyclic aromatic hydrocarbons; PFAs, per- and polyfluoroalkyl substances; PM, particulate matter; SVOCs, semi-volatile organic compounds; VOCs, volatile organic compounds.

Full size image

Particulate matter is a highly heterogeneous mixture that can include sulfates, nitrates, ammonium, elemental carbon, crustal material, trace elements, volatile and semi-volatile organic compounds, and biological components, with its specific composition depending on the collection time and location12. As such, the specific composition varies widely based on sources, meteorological conditions and atmospheric chemistry. Studies have shown that particulate matter from different sources (such as traffic, industrial emissions, wildfires and indoor) can have dramatically different health effects, even at the same mass concentration13,14.

Additionally, thousands of volatile and semi-volatile organic compounds exist in the atmosphere, many with known toxicity but no regulatory standards or routine monitoring. These include airborne pollutants such as polycyclic aromatic hydrocarbons, phthalates and short-chain per- and polyfluoroalkyl substances as legacy and emerging contaminants15,16,17. The focus on criteria pollutants has created a blind spot in the understanding of air pollution’s health effects (Fig. 2).

Fig. 2: Hidden components of the airborne exposome.
figure 2

This conceptual iceberg schematic illustrates how conventional air pollution research captures only the surface-level issues, such as criteria air pollutants, while concealing numerous interconnected challenges. The variables pictured below the surface of the water represent all of the hidden challenges as equally critical limitations that require attention and further research. The iceberg metaphor emphasizes that the majority of important research challenges remain largely unaddressed in current air pollution studies. A more holistic exposomics approach is needed to uncover these hidden layers and advance environmental health science.

Full size image

Individual exposome variability and diversity

Another crucial limitation of traditional approaches is the reliance on ambient monitoring data as proxies for personal exposure. In reality, individuals experience unique exposome profiles shaped by numerous factors.

People spend approximately 90% of their time indoors, where the pollution profile differs from outdoor air18. Indoor sources, such as cooking, cleaning products, furniture and building materials, contribute additional pollutants, whereas building characteristics alter the infiltration of outdoor pollutants19. The relationship between indoor and outdoor air pollution is complex and varies by pollutant, building type, ventilation practices and occupant behaviours20. Outdoor workers also face unique exposure challenges, including direct contact with traffic emissions, industrial sources and variable meteorological conditions. These occupational groups represent critically vulnerable populations requiring specialized exposure assessment approaches that account for both occupational and ambient sources21.

Air pollution concentrations can vary dramatically over small spatial scales, particularly for traffic-related pollutants22. Traditional monitoring networks often miss these local variations, leading to exposure misclassification. Individual movement patterns throughout the day result in exposure to different microenvironments with varying pollution profiles. Work location, commuting patterns and activity choices all shape personal exposure23.

Individual breathing rates, respiratory tract anatomy and physical activity levels affect the actual dose of pollutants delivered to target tissues24. Additionally, biological responses to identical pollutant exposures can vary dramatically due to individual factors, including (but not limited to) sex, age, genetics, body mass index, visceral fat distribution, metabolic capacity, hormonal status, pre-existing conditions, body temperature, respiratory tract and cellular microenvironment pH balance, and microbiome composition. These physiological and biological factors collectively create substantial inter-individual variability in the biological impact of similar ambient concentrations, contributing to different susceptibilities and health outcomes, even under identical exposure conditions.

These factors create heterogeneity in personal exposomes that is poorly captured by traditional ambient monitoring approaches. Two individuals living in the same city, or even the same building, can have markedly different personal exposome profiles due to differences in activities, behaviours and each individual’s microenvironment due to the lack of ventilation in modern buildings25,26.

Limitations of current research approaches

The current research paradigm in air pollution epidemiology faces several limitations that restrict the understanding of how complex pollution mixtures affect human health. These limitations span three key domains: statistical challenges in analysing multi-pollutant data; inadequacies in exposure assessment methodologies; and knowledge gaps in understanding biological mechanisms. Collectively, these limitations hinder the ability to characterize the true health impact of the airborne exposome and develop targeted interventions. Each domain presents unique challenges that require innovative solutions to advance the field beyond its current constraints.

Statistical challenges in multi-pollutant models

Emerging analytical approaches, such as non-targeted analysis using high-resolution mass spectrometry, generate very high-dimensional datasets containing thousands of airborne component features10, whereas conventional epidemiological studies typically analyse only a handful of pollutants. Indeed, traditional epidemiological studies have attempted to address pollution complexity through multi-pollutant statistical models, but these approaches face limitations. Many air pollutants originate from common sources or form through related atmospheric processes, creating high correlations that complicate the statistical separation of effects27. This collinearity problem makes it difficult to disentangle the independent effects of individual pollutants and can lead to unstable effect estimates. In addition, these highly correlated pollutants from the same sources can have different health effects.

Pollutants can also interact synergistically or antagonistically, with effects that cannot be predicted from single-pollutant studies28. For example, ozone and nitrogen dioxide can interact to form secondary pollutants with different toxicity profiles from either parent compound alone. Sulfur dioxide can increase the acidity of particles, potentially enhancing their toxicity29. Thousands of organic compounds present together can create added complexity. Current statistical approaches struggle to characterize these complex interactions, particularly when more than ten pollutants are involved.

In the presence of dozens or hundreds of potential pollutants, statistical methods for variable selection can produce unstable or inconsistent results, depending on model specification30. Different variable selection approaches may yield entirely different sets of pollutants as results, depending on the research question, complicating interpretation and policy implications. Reliance on ambient monitoring data as exposure proxies introduces various forms of measurement error that complicate the interpretation of health effect estimates31. Non-differential exposure misclassification—wherein the error is unrelated to disease status—typically biases health effect estimates towards the null, potentially masking true associations between pollution mixtures and health outcomes32. However, differential misclassification—wherein the exposure error varies by disease status—can bias effect estimates in either direction, potentially creating spurious associations or exaggerating true effects33. In complex mixtures research, the situation is further complicated because measurement errors may vary across pollutants within the mixture, leading to unpredictable patterns of bias in estimated interaction effects34. Sophisticated statistical approaches such as measurement error models, Bayesian hierarchical modelling and advanced machine learning models can help to address these challenges but require careful specification of error structures and sensitivity analyses to assess robustness35.

These statistical challenges limit the utility of traditional epidemiological approaches for understanding the health effects of complex air pollution mixtures. More sophisticated methods are needed that can accommodate the multicollinearity, interactions and measurement error inherent in these data.

Exposure assessment limitations

Current exposure assessment methods also present limitations in understanding the health effects of complex air pollution mixtures. Regulatory monitoring networks provide limited spatial coverage, missing important local variations in pollutant concentrations36. Most urban areas have only a handful of monitoring stations, which cannot capture the fine-scale spatial heterogeneity of many pollutants, particularly those associated with traffic or local industrial sources.

Most monitoring stations measure only criteria pollutants, missing thousands of potentially harmful airborne substances37. For example, volatile and semi-volatile organic compounds such as pesticides, polycyclic aromatic hydrocarbons, dioxins and furans, and many emerging contaminants are rarely measured in routine monitoring networks despite their potential health relevance. This creates a systematic bias towards studying only those pollutants that are routinely measured.

Biological components of air pollution present additional measurement challenges, including seasonal and diurnal variability in pollen concentrations, distinguishing between viable and non-viable microorganisms, and the need for specialized collection and analysis methods that can differentiate between different microbial species and their metabolic products. Fungal spores, bacteria and viral particles each require different sampling and analytical approaches, and their health effects may depend on viability, concentration and co-exposure with chemical pollutants38,39,40.

Despite people spending most time indoors, indoor air pollution remains poorly characterized in most epidemiological studies41,42. Indoor environments contain unique pollutant sources (such as cooking, cleaning products, furniture and building materials) that contribute to personal exposure but are rarely incorporated into epidemiological models. The relationship between indoor and outdoor pollution is complex and varies by pollutant type, building characteristics and occupant behaviours43.

Many monitoring approaches provide daily or longer averaging times, missing short-term concentration spikes that may have health relevance44. For many pollutants, peak exposures may be more biologically relevant than average concentrations, but these peaks are obscured in most monitoring data. Additionally, the cumulative effects of multiple exposures over time are poorly captured by current approaches.

Personal monitoring approaches, although better representing individual-level exposures compared with geospatial data, also face limitations, including device accuracy variations, subject compliance issues, battery life constraints and cost barriers for large-scale deployment17,45. For example, passive samplers such as silicone wristbands provide time-integrated estimates but miss peak exposures, whereas active monitors offer higher resolution but are bulkier and may influence behaviour.

These limitations in exposure assessment introduce uncertainty and probably underestimate the true health burden of air pollution. Addressing these limitations requires innovative approaches to exposure assessment that can capture the true complexity of human exposure to air pollution mixtures46,47.

Knowledge gaps in biological mechanisms

Understanding the biological mechanisms through which complex pollution mixtures affect health remains a major challenge. Traditional toxicological studies typically examine individual pollutants or low-dimensional mixtures, failing to capture the complexity of real-world exposures27. This approach cannot identify potential synergistic or antagonistic effects that may occur when multiple pollutants interact within biological systems.

Current approaches often lack integration between exposure assessment and biological mechanisms. Epidemiological studies may identify statistical associations between pollution metrics and health outcomes without elucidating the underlying causal pathways48. This limits our ability to develop targeted interventions or identify particularly harmful components of pollution mixtures.

The temporal dynamics of exposure and response are poorly understood for many pollutants and health endpoints. Different components of air pollution mixtures may have varying lag times between exposure and effect, complicating efforts to identify causal relationships49. Some pollutants may cause immediate effects whereas others contribute to long-term disease development, and these temporal patterns may differ by health endpoint.

Susceptibility factors that modify individual responses to pollution exposures remain inadequately characterized. Genetic variations, pre-existing conditions, age and psychosocial factors all influence how individuals respond to their airborne exposome profiles50. Understanding these susceptibility factors is crucial for identifying vulnerable populations and developing targeted interventions.

The biological mechanisms underlying the health effects of emerging pollutants and complex mixtures remain largely unexplored. As analytical techniques improve, researchers continue to identify previously unrecognized components of air pollution that may contribute to health effects51. Understanding the biological impact of these emerging pollutants requires new research approaches that can link exposure to biological effects52.

Inhalation toxicology studies require specialized facilities with controlled atmosphere chambers, precise aerosol generation systems and extensive safety protocols. These studies face challenges in replicating real-world exposure scenarios, including mixture effects, variable concentrations and co-exposures to multiple pollutants51. Air–liquid interface cultures and lung-on-chip models offer promising alternatives but face challenges in replicating the full complexity of human respiratory systems, immune responses and systemic effects. Translating findings from controlled laboratory exposures to real-world health risks remains a substantial challenge, particularly for chronic, low-level exposures to complex mixtures28,53.

These knowledge gaps in biological mechanisms limit the ability to develop effective interventions and regulations for complex air pollution mixtures. Addressing these gaps requires interdisciplinary approaches that integrate exposure science, toxicology, epidemiology and data science54.

Three complementary research directions can collectively offer a more comprehensive framework for the human airborne exposome: mixture-based approaches that move beyond single-pollutant analyses; advanced exposure assessment technologies that better capture individual-level exposures; and systems biology approaches that elucidate complex biological responses. These are not mutually exclusive, but rather represent integrated pathways towards a more holistic understanding of the airborne exposome (Box 1).

Mixture-based approaches

Rather than continuing to focus solely on criteria pollutants, future research should adopt approaches that better reflect the complex nature of the airborne exposome. Source-based analysis represents a promising alternative to traditional pollutant-focused approaches. By characterizing pollution mixtures by their sources (such as traffic, industry or residential heating), researchers can provide more interpretable and policy-relevant information than individual pollutant approaches55. Source identification and apportionment methods can identify the contribution of different emission sources to observed pollution mixtures and link these source-specific exposures to health effects.

Chemical fingerprinting such as isotopic tracing offers another innovative approach to studying complex pollution mixtures. Advanced analytical techniques can identify characteristic patterns in pollution mixtures that may be more strongly associated with health effects than individual components56. These fingerprints can serve as markers for specific pollution sources or atmospheric processes, providing a more holistic characterization of exposure than traditional single-pollutant approaches.

Toxicity-weighted assessment approaches weight pollutants by their toxicity rather than treating all components equally. This approach better reflects the biological impact of pollution mixtures by accounting for the varying potency of different components57. Toxicity-weighted indices can integrate information across multiple pollutants to provide a more biologically relevant measure of the exposome. This approach has been successfully applied in studies of hazardous air pollutants and could be extended to broader pollution mixtures.

Untargeted analysis and broader screening approaches can identify previously unrecognized components of pollution mixtures and reveal overlooked contributors to health effects58. Mass spectrometry-based non-targeted analysis and other advanced analytical techniques can detect thousands of chemical constituents in air samples without a priori selection, revealing the true complexity of pollution mixtures53,59,60. These approaches have identified numerous previously unrecognized components of air pollution that may contribute to health effects.

These mixture-based approaches offer a more comprehensive characterization of the airborne exposome than traditional single-pollutant methods. By embracing the inherent complexity of pollution mixtures, these approaches can provide new insights into the relationship between air pollution and health.

Advanced exposure assessment

Improving exposure assessment can aid the understanding of air pollution’s health effects. Personal monitoring technologies enable direct measurement of personal-level exposure across each individual. Wearable and portable sensors that measure multiple pollutants simultaneously can capture exposure variations throughout the day as individuals move between different environments61. These technologies provide a more accurate representation of personal exposure than traditional stationary monitors and can reveal exposure patterns that would be missed by ambient monitoring alone62.

Comprehensive approaches that integrate indoor and outdoor pollution sources provide more realistic exposure estimates. Indoor air-quality monitoring, coupled with information on building characteristics and occupant behaviours, can help to quantify the contribution of indoor sources to total exposure63. Models that account for the penetration of outdoor pollutants into indoor environments and the generation of pollutants from indoor sources can provide a more complete picture of total exposure43.

Advances in computational modelling allow the estimation of pollutant concentrations at fine spatial and temporal scales that better reflect personal exposure. Land-use regression models, dispersion models and hybrid approaches can predict pollutant concentrations at the individual address level, capturing spatial heterogeneity that would be missed by sparse monitoring networks64. High-performance computing enables the application of these models across large populations and geographic areas, facilitating more accurate exposure assessment in epidemiological studies65.

Biological monitoring can provide integrated measures of actual pollutant doses received by individuals, accounting for factors such as breathing rate and absorption. Exposure biomarkers, such as metabolites in urine or blood, can reflect the internal dose of pollutants and provide a more direct measure of biological impact than environmental monitoring58. These biomarkers can capture the combined effect of multiple exposure pathways and help to identify particularly harmful components of complex mixtures66. However, biomarker validation faces substantial challenges, including establishing population-specific reference ranges, inter-individual variability in metabolism that can span orders of magnitude, and distinguishing between recent versus cumulative exposures, with many biomarkers lacking validation across diverse populations and exposure scenarios.

Recent developments in longitudinal personal exposure monitoring have demonstrated the feasibility of characterizing individual-level exposome profiles over time and connecting these profiles to biological responses52,67,68. These approaches hold promise for elucidating the complex relationships between exposure patterns and health outcomes.

Systems biology and multi-omics approaches

Multiple chemical classes interact simultaneously across different exposure windows. Systems biology and multi-omics approaches offer the analytical depth needed to unravel how complex air pollution mixtures perturb biological networks and reveal individual differences in susceptibility10.

Multi-omics integration combines genomics, epigenomics, transcriptomics, proteomics, metabolomics, adductomics and exposomics to illuminate biological pathways affected by complex pollution mixtures6. These approaches can identify molecular signatures associated with specific exposure patterns and provide insight into the mechanisms underlying observed health effects69. By examining changes across multiple biological levels simultaneously, multi-omics approaches can reveal complex response patterns that would be missed by more targeted analyses70.

Toxicity pathway analysis identifies common biological pathways affected by different pollution components, revealing mechanisms underlying health effects. Advanced in vitro or ex vivo systems, including three-dimensional cell cultures, organ-on-a-chip models and human organoids, can capture complex cellular interactions and provide more physiologically relevant information than traditional cell culture models71. These systems can be used to test the effects of complex mixtures and identify particularly harmful components or combinations72.

Computational models that integrate exposure data with biological response information can help to predict the health effects of complex pollution mixtures. Adverse outcome pathway frameworks link molecular initiating events to adverse outcomes at the individual and population level, providing a mechanistic understanding of how pollution exposures lead to health effects73. These models can incorporate information on multiple pollutants and their interactions, facilitating a more comprehensive assessment of pollution health risks74.

Emerging approaches in exposome research have demonstrated the ability to identify molecular signatures associated with specific exposure patterns and link these signatures to health outcomes52,75,76. These approaches hold promise for elucidating the complex biological responses to the airborne exposome and identifying early markers of adverse effects.

Implications for public health and policy

The current air-quality regulatory frameworks worldwide, whether targeting criteria pollutants, limit values or environmental quality standards, have achieved substantial public health benefits in many regions. Since the implementation of air quality regulations globally, dramatic decreases in ambient concentrations of regulated pollutants have prevented an estimated 1.6 million premature deaths annually worldwide and substantially reduced the burden of respiratory and cardiovascular diseases77,78. Economic analyses indicate that every US dollar invested in air quality improvements yields US $30–90 in health and productivity benefits across developed economies79,80. Conservative estimates suggest that new regulatory approaches, such as the source-based model, could prevent an additional 200,000–400,000 premature deaths globally per year, representing up to US $1 trillion in additional economic benefits when scaled across major economies77,79. These achievements represent one of the most successful public health interventions in modern history. The framework presented here builds on this foundation rather than replacing it, seeking to address remaining knowledge gaps and enhance protection through more comprehensive approaches to air quality management.

Regulatory frameworks that consider the combined impact of multiple pollutants could provide more comprehensive health protection than current single-pollutant standards27. Source-based regulation might focus on controlling pollution sources rather than individual pollutants, thereby capturing co-benefits across multiple harmful substances. This approach aligns more closely with the reality of pollution emissions and exposure patterns, potentially providing more efficient and effective health protection81,82. For example, regulations targeting traffic emissions and industrial processes would simultaneously reduce multiple pollutants associated with these sources28.

Risk assessment methodologies need to evolve to address the complexity of air pollution mixtures. More sophisticated risk assessment methods that can account for the combined impact of multiple pollutants would provide a more accurate picture of public health risks and guide more effective interventions83. For instance, new approach methodologies, including in vitro toxicity screening, computational modelling and system-level adverse outcome pathway frameworks, offer promising alternatives to traditional risk assessment by enabling rapid evaluation of complex mixtures and reducing reliance on animal testing.

Economic considerations represent a critical dimension in transitioning to more comprehensive regulatory approaches. Although source-based regulation may ultimately prove more efficient, initial implementation costs could be substantial, requiring new monitoring infrastructure, enforcement mechanisms and compliance frameworks84. Cost–benefit analyses should account not only for direct healthcare savings but also for productivity improvements, ecosystem services and reduced climate impacts—benefits often omitted from traditional regulatory assessments85. Phased implementation approaches could distribute costs over time while prioritizing interventions with the highest benefit-to-cost ratios. Additionally, economic incentive mechanisms such as pollution taxes or cap-and-trade systems could complement direct regulation, thus potentially reducing overall compliance costs while achieving similar health benefits86. These economic considerations should be integrated into the policy development process alongside the scientific evidence on complex pollution mixtures.

Understanding complex mixtures may reveal specific components that disproportionately affect vulnerable groups, enabling targeted interventions. Children, older individuals, individuals with pre-existing conditions, and socioeconomically disadvantaged populations often experience greater health impacts from hazardous airborne exposures50. Identifying the specific mixture components that drive these disparities could inform more targeted and effective interventions for the vulnerable groups87.

Recommendation and conclusion

These recommendations span near-term technological developments to long-term regulatory transformations, creating a roadmap for implementation.

Technology validation and standardization represent potential bottlenecks requiring community consensus and coordinated effort. Technology roadmaps with standardized protocols, inter-laboratory comparison studies and explicit criteria for method validation should be established before widespread adoption. The community would benefit from resisting the temptation to continuously develop new technologies without adequately validating existing approaches, as this undermines the accumulation of comparable evidence needed to support regulatory decisions. Standardized, cost-effective personal monitoring technologies should be an urgent priority, requiring interdisciplinary collaborations to develop systems that simultaneously track multiple pollutant classes across each individual. Concurrently, statistical frameworks for analysing complex mixture data need advancement, adapting methods from fields such as metabolomics and air pollution research88. These foundational elements can be substantially developed in the near term with targeted funding that balances innovation with rigorous validation.

Within the longer term, longitudinal cohort studies that integrate comprehensive exposure assessment with multi-omics profiling should be established across diverse populations. These studies need to intentionally include vulnerable groups and different geographical contexts to capture variability in exposure profiles and susceptibility factors70. Concurrent method development should focus on source apportionment techniques combined with mixture-based health risk assessment tools that can identify the most harmful pollution sources to guide targeted interventions13.

A long-term vision involves implementing source-based regulatory frameworks informed by exposome research. Achieving this shift requires coordinated implementation across multiple domains with specific funding mechanisms and partnerships. For example, coordinated national research programmes over 5–10 years should prioritize technology development, validation studies and pilot implementations. Industry–academia partnerships can accelerate sensor development and data analytics capabilities, whereas international collaborations through international frameworks such as the World Health Organization can harmonize global approaches and share costs across nations. Success requires collaboration between academic research centres, technology companies, regulatory agencies, healthcare systems and community organizations. Public–private partnerships can leverage commercial innovation while ensuring that public health priorities guide development. Environmental agencies should lead regulatory framework development and validation studies; health ministries should coordinate health research and clinical integration; research councils should support fundamental technology development; public health agencies should oversee population health surveillance systems; and environmental health institutes should focus on mechanistic research and population studies. Implementation could follow a phased approach (Box 2). Table 1 provides the examples of how this framework could revolutionize multiple health outcome domains.

Table 1 Leveraging the airborne exposome to elucidate complex disease mechanisms
Full size table

Physicians can integrate exposome data into clinical practice through electronic health record systems that incorporate environmental exposure histories, geographic risk assessments and personalized exposure counselling89. This enables targeted prevention strategies, early detection of exposure-related health effects and evidence-based recommendations for exposure reduction in high-risk patients. Medical and nursing education needs to evolve by including environmental health training, enabling clinicians to interpret exposome data and provide patient-specific guidance on exposure reduction strategies90.

Implementation will require substantial investments, but these are justified by the potential public health benefits—more targeted interventions, enhanced protection for vulnerable populations, and more efficient regulatory approaches. As air pollution patterns evolve with climate change and urbanization, scientific approaches need to evolve. The exposome paradigm offers a pathway to a more comprehensive understanding of how the air we breathe affects our health, enabling more effective protection for current and future generations91,92. The time has come to move beyond ambient particulate matter to address the complex airborne components of the human exposome.