The exposome concept integrates a large number of external exposures across various contexts (eg, ecosystems, lifestyle, social, physical-chemical) to investigate their cumulative impact on internal biological processes that potentially manifest as adverse health outcomes.¹ This framework can readily incorporate various environmental exposures (eg, air pollution, water contamination). Climate change-related exposures and their subsequent hazards are also a pressing environmental concern. Broadly, we define these exposures to be direct environmental exposures which will intensify with climate change. Climate change-related exposures include direct impacts on weather such as extreme heat, tropical cyclones, and wildfires, among others. Climate change also encompasses exposures such as as air pollution, power outages,² and contaminated or limited food³ and water^4,5 supplies. Climate change can have additional downstream consequences that can largely impact health largely through changes in social factors and infrastructure. In theory, the exposome also includes climate change-related events; however, there has been limited exposome research that incorporates these exposures, despite their demonstrated impacts on health.⁶ Individually, climate change-related events (eg, heat waves) can directly threaten an individual’s health by taxing psychosocial, cardiovascular, respiratory, and other biological processes,^7,8 or by coalescing with other dimensions of the exposome. For example, a heat wave can interact with low household income or poor quality housing to worsen health.^6,9

Climate change endangers population health by increasing the frequency, geographic range, and intensity of natural hazard such as tropical cyclones,¹⁰ wildfires,¹¹ extreme heat,¹² floods,¹³ and drought,¹⁴ among others, that precipitate downstream events such as power outages¹⁵ or displacement.¹⁶ Instead of being isolated events,⁶ climate-related exposures may occur simultaneously. Climate change will further increase the likelihood of co-occurring events.^17,18 Extreme heat, drought, and wildfire, for example, may occur at the same time and location. These exposures can also be causally inter-related and be part of feedback loops. For instance, extreme heat and drought can facilitate dry environments that increase risk of wildfires. These interdependencies complicate investigating the role of multiple climate change-related exposures on biological responses. However, complex relationships commonly occur among all environmental exposures; this issue, therefore, is not limited to the role of climate change-related exposures in the exposome. The exposome concept can aid researchers in understanding the health effects of simultaneous climate-related exposures. It provides a framework to better summarize all exposures affecting health, which we extend to include climate-related exposures. Here, we discuss considerations, methods, and gaps and future directions for applying the exposome framework to characterize a mixture of climate change-related exposures in health analyses. Our discussion may align more with environmental epidemiology’s focus on hypothesis-driven research than with exposome research’s emphasis on discovery, both approaches are needed to understand the impacts of climate mixtures on adverse health outcomes and biological processes.

Recently, the field of analyzing mixtures exposures has advanced rapidly, and numerous methods have been adapted from other fields or developed to evaluate multiple exposures in health studies. Techniques include least absolute shrinkage and selection operator (LASSO) and Bayesian kernel machine (BKMR) regressions.^19-21 Extending the field to include climate change-related exposures, however, requires additional considerations, due to the unique characteristics of mixtures of climate change-related exposures. Each mixtures method has been designed to address a specific research question(s) such as characterizing exposure patterns, identifying the most toxic mixture components, and estimating the overall effect of exposure to the entire mixture.^19,22,23 Environmental health studies have traditionally leveraged these techniques to better understand the health impacts of complex mixtures in the context of various chemical and non-chemical stressors including phthalates,^24,25 pesticides,²⁶ air pollution,²¹ and sociodemographic factors.^27,28 Before using mixtures methods to consider exposure patterns, independent effects, and cumulative impacts, one must first consider the ways in which climate change-related disasters differ from more traditional exposures. The concept of a climate mixture is still in its nascent stages, but for our purposes, we define a climate mixture as at least three co-occurring climate change-related exposures.

One important consideration is that not all climate change-related exposures occur in the same place or time. Traditionally, environmental epidemiology conceptualizes mixtures to involve exposures that individuals or communities continuously experience. For example, although the composition of an air pollution mixture varies seasonally and regionally, individuals are always exposed to a collection of pollutants simultaneously. On the other hand, individual climate change-related exposures may occur transiently and lack distinctive spatial and temporal patterns (Figure 1). Though climate change is facilitating unexpected extreme weather events such as a hurricane near California,²⁹ the tropical cyclone season in the United States spans June through November and mostly affects the coastal eastern United States.³⁰ Wildfires in the western United States, a growing problem as climate change drives drier, warmer environmental conditions,^31,32 predominantly take place from July through October. Electrical power outages in the United States occur throughout the year but vary by time of day, season, and region.³³ Certain climate mixture combinations such as wildfire disasters co-occurring with snowstorms would be unlikely to exist owing to the current spatial and temporal patterns of individual climate change-related exposures. Therefore, researchers interested in questions about climate mixtures should keep in mind positivity violations³⁴ where certain exposures have a zero probability of occurring or co-occurring in specific locations. For studies that require a comparison group, researchers should also carefully consider which populations would be appropriate as control groups.

Because exposomics strives to capture the comprehensive and cumulative effects of physical, chemical, biological, and psychosocial influences on biological systems across the human lifespan, identifying the long-term effects of climate mixtures on health is key.^35,36 For questions examining long-term health impacts, researchers should recognize that climate change-related exposures and chemicals exhibit different distribution patterns across time. Research on climate change-related exposures has predominantly focused on acute outcomes (eg, within days, weeks, or months of exposure), as defining long-term exposure to climate mixtures poses challenges. Longer contexts with long-term exposures and their respective chronic effects on population health involve additional complexity, including the definition of long-term exposures and selection of appropriate health outcomes.¹⁹ Characterizing long-term exposures to chemical mixtures has typically been done by averaging concentrations over an extended period, such as annual averages over several years or maximal values across months. However, doing so for climate change-related exposures is not always appropriate. Some climate change-related exposures, such as tropical cyclones and wildfires, have distinct temporal patterns, and researchers must define long-term exposures for events with a sharp, irregular, and discontinuous pattern across time. For example, a prior study conceptualized wildfire fine particle (PM_2.5) exposure across 15 years using 5 definitions considering varying temporal units (weeks Versus continuous days Versus year), wildfire PM_2.5 thresholds (>5 μg/m³ versus >0 μg/m³), and aggregation methods (cumulative weeks Versus averages across years).³⁷ Given the varying ways to define long-term wildfire PM_2.5, researchers must think through which exposure definition is more biologically plausible and relevant for which health outcomes. The lack of standardized approaches for measuring long-term exposure to wildfire PM_2.5 applies to other climate-related exposures as well. When investigating the long-term health impacts of climate-related mixtures, researchers must grapple with defining exposure for those with irregular temporal patterns and decide on the most relevant health outcome(s).

Estimating the health impacts of climate mixtures is crucial for incorporating climate change-related exposures to exposome research. We briefly present research questions about multiple co-occurring climate change-related exposures, discussing the importance and relevance of these questions, established methods that can be used to answer them, and potential methodological limitations. This discussion will focus on hypothesis-driven techniques to methodologically address climate mixtures, an important step towards expanding the implementation of exposome research to include climate change-related exposures. Throughout the paper, we will use the following example: the climate mixture comprises a tropical cyclone event, power outages, and flooding, and the health outcome is respiratory hospitalizations. Respiratory hospitalizations are extracted from administrative health data that are often used in health studies and can cover large geographic regions (eg, states, entire US), ensuring adequate variability in macro-level climate change-related exposures. Studies have found that tropical cyclones are associated with respiratory hospitalizations.³⁸ One potential pathway, for example, is that tropical cyclones result in power outages, and the loss of electricity may prompt individuals who are reliant on electricity-dependent medical equipment to seek healthcare services.^38-40 Flooding can expose individuals to contaminated water and microbes (eg, bacteria, mold), which can increase risk for acute respiratory infections.⁴¹ Our examples center on short-term exposure and acute respiratory hospitalizations. Certain methods cannot be used for all study designs (eg, BKMR is not appropriate for case-control studies or time-series with aggregated outcome data), but as our goal centers on climate mixtures and mixture methods, we will focus on illustrating the ways in which mixtures methods can be applied to such exposures.

We use this example to illustrate specific concepts and considerations for analyses of climate mixtures. We note that the methods we present below are not an exhaustive list of methods to analyze climate mixtures exposure in health models; rather, we present some examples of existing and widely used methods in environmental epidemiology that can be used to answer specific research questions about exposure to climate mixtures (Figure 2). In addition, many of the methods we present could be used to answer different research questions about climate mixtures. However, to discuss multiple methods, we only include each method in one research question category. Our climate mixture example is not fully generalizable to other climate combinations, but our discussion of concepts and considerations can be applied to future research.

The methods we described above are commonly used in environmental epidemiology applications. Nonetheless, we recognize that other methods may exist, such as sequence and life course analyses, or need to be developed to address some of the aforementioned limitations. Using these methods to incorporate climate change-related exposures in exposome research should be considered in future research, both in simulations and real-life data studies.

Studies can apply mixtures methods to examine multiple co-occurring climate change-related exposures, but a few methodological gaps remain that limit the types of research questions and objectives we can investigate. As previously discussed, certain climate mixture combinations violate the positivity assumption across time and/or space, as not all climate change-related exposures can occur at the same time or place. Thus, creating a standardized exposure index across larger geographic regions (eg, the United States) to understand cumulative exposure is challenging. For instance, tropical cyclones in the United States almost always occur in geographically distinct areas, while extreme heat and snowstorms happen in separate seasons. Researchers have challenges when comparing cumulative exposure to a global set of climate mixture combinations across time and/or space. Instead, they should consider localizing their questions and analyses. For example, researchers may want to focus on creating an exposure index for areas in the Southeast with exposures in our previous examples (hurricanes, floods, and power outages) or for locations across Southern California with wildfire disaster, drought, and power outages.

Positivity violations limit the types of research questions that can be asked, while current practices to define and measure long-term health effects of multiple simultaneous exposures pose challenges. Given that climate change has and will exacerbate related exposures, it is essential to consider long-term health studies. Methodologically, there are not yet standard or widely used metrics to assess long-term climate change-related exposures. Such metrics could capture different exposure domains such as duration, intensity, and frequency,²⁷ so researchers should consider which metric is most appropriate for their research question and health outcome of interest. Earlier, we provided an example of how long-term wildfire smoke measures can drastically vary across years depending on how the metric is originally operationalized, but these quandaries and decisions also apply to other climate change-related exposures. Long-term exposures to the events in our previous example (tropical cyclones, flooding, power outages) can vary (Table 1). The ways in which we define long-term exposures to climate events and mixtures may capture different pathways, resulting in different estimated health effects. For example, specific parameterizations might be more appropriate for one exposure than another. Peak wildfire PM_2.5 in a year could be most strongly related to asthma exacerbation while the number of total weeks exposed to any wildfire PM_2.5 might be most predictive of asthma onset.

Another challenge to assessing long-term climate change-related exposures is the dearth of historic and present-day data. Some climate change-related exposures, such as those related to temperature (eg, heat waves, extreme cold), have measures as early as 1895⁵³ and today, there are robust measures with high spatiotemporal resolution, such as the hourly measures at a 1 km² resolution level.⁵⁴ For power outages, the Department of Energy records power outage events that affected more than 50,000 customers or resulted in a 300-megawatt loss,⁵⁵ but there is a lack of available data on smaller scale outages, hindering investigations of power outage-related health impacts. There have been growing efforts to assess small-scale outages,^33,56 and data on all types of power outages will be useful for future health research. Recent studies have leveraged data from 2001 at a spatial unit containing ∼11,000 electrical customers updated every 30min, but such studies have so far been conducted using data from New York State.³⁹ Spatial or temporal misalignment among individual climate change-related exposures in a climate mixture due to data limitations may make it impossible to chronicle trends for long-term climate exposure mixtures and characterize their chronic health impacts.

An additional consideration is that individuals and communities can adapt to certain climate change-related exposures over time. Adaptation can act as an effect modifier between climate mixtures and health outcomes and be in the form of changes in individual behaviors, residential mobility, infrastructure changes, or other avenues. Individuals could purchase indoor air filters or use facemasks to protect against wildfire smoke.⁵⁷ Cities have implemented programs to provide residents access to public temperature-controlled areas during extreme heat.⁵⁸ As climate change poses threats to public health, individuals and communities will respond to reduce adverse outcomes to climate change-related exposures, but adaptation efforts may vary. However, incorporating such adaptations poses challenges as data related to adaptation are limited. Simulations and prediction models have been used to integrate adaptation in understanding its role in climate mixture exposures,^59,60 but the dearth of available data limits our ability to analyze the possible effects of adaptation on the relationship between climate mixtures and adverse health.

As climate change exacerbates weather events and downstream exposures, the exposome framework can be used as a guide to integrate exposure mixtures to accurately capture their health impacts. Using an example climate mixture, we presented several possible research questions of interest as well as examples of the appropriate methodologies to evaluate mixtures. Methods exist to analyze simultaneous exposures, but studying climate mixtures necessitates additional considerations such as violations of the positivity assumption, the complexity of defining long-term exposure, and limited data. Future research directions should aim to address these gaps and account for the nuances of climate mixtures and more generally of climate change-related exposures.

One limitation when integrating climate change-related exposure data with other cohort data—ideally those with -omics data available—for exposome research and analyses is that if cohort participants live close to each other, there will not be enough variability in the climate exposures to allow for enough statistical power to discover relationships. To effectively incorporate climate change-related exposures in exposome research, therefore, a wide geographical distribution of study participants may be key. The UK Biobank,⁶¹ for example, has -omics data from individuals across the United Kingdom, and given the country’s wide geographic range, there is likely high variability in exposure to climate change-related exposures within the cohort (eg, temperatures, storms, snow, flooding). Researchers could then identify representative subgroups and analyze whether exposure to climate change-related events is linked to molecular or other -omics outcomes of interest. Variability in climate change-related exposures is key to connecting these external macro-level exposures to internal omics-level outcomes.

One future direction is to use new ways to gather data on climate change-related exposures to supplement limited data in some domains like flooding. Researchers should consider leveraging novel sources of data such as social media, news outlets, images (eg, from satellites), or videos. In the case of climate change-related exposures such as flooding, these data sources can provide information about the event’s spatial and temporal extent, allowing researchers to validate, update, or improve their exposure measures.

Relevant to studies linking climate mixtures and health, data should be collected to account for climate adaptation and differences in the ability to adapt. Societies and individuals will likely adapt to climate change, building new infrastructure or modifying behaviors. As the threat of physically destructive disasters is growing, governments may invest in fortifying electrical infrastructure to prevent or minimize subsequent power outages. In the face of rising temperatures and increasing likelihood of extreme heat events, individuals may rely more on air conditioners, installing them if they did not have one previously. However, the ability to adapt differs by subpopulation, and will have limits. For example, low-income individuals may not be able to afford air conditioners, while those facing energy insecurity may limit air conditioning use to minimize electricity bills, resulting in cooling hardship.^62,63 Another adaptation technique is relocating from high- to lower-risk climate change-related disaster areas, which includes evacuations in response to immediate wildfire disaster or tropical cyclones and pre-emptive moves from flood-prone neighborhoods. However, health status (eg, reliance on electricity-dependent medical equipment) or socioeconomic status may limit mobility. Obtaining data about adaptation across subpopulations and then accounting for them is necessary to accurately estimate the relationship between climate change-related exposures and adverse health outcomes.

As the exposome framework emphasizes cumulative environmental exposures, future exposome research can examine the possible joint impact of climate and non-climate environmental exposures on health. Climate mixtures and related exposures are external, and studies examining the associated health impacts have predominantly relied on administrative data (eg, hospitalizations). On the other hand, exposomic studies, such as metabolomics, center on internal biological processes using biospecimens for analysis^64,65 and have been predominantly used cohort or panel data. Although there are challenges in synthesizing these different exposome aspects,⁶⁶ it is a necessary step to accurately estimate cumulative environmental exposure. Additionally, exposome research in the context of climate change may uncover unexpected combinations of factors, both related and not related to climate change, that are particularly detrimental to population health.

Climate change-related exposures will continue to pose risks to population health, which will likely increase over time. Understanding the current tools and gaps in evaluating the comprehensive impact of these exposures on health can explicate their role as part of the exposome. By more explicitly considering climate change-related exposures in the field of exposomics, exposome research can move towards the direction of discovering unexpected pathways linking climate change and health.