MODELLING EXPOSURE

Erionite has now been detected on every continent (Berry et al., 2021 and Patel et al., 2022). Exposure to erionite is strongly linked to the development of a range of severe respiratory diseases, including asbestosis, lung cancer, and malignant mesothelioma—a rare but aggressive cancer of the lung or abdominal lining. Erionite is considered even more carcinogenic than asbestos (Scarfi et al, 2025), making it critical to identify its locations and potential exposure pathways.

Figure 1. Global geological occurrence of erionite (Patel et al., 2022)

One method to determine the link between exposure to erionite and health outcomes is to look at the spatial patterns between the occurrence of erionite and malignant mesothelioma. This disease is chosen because it is thought to be caused only by exposure to either asbestos or erionite.

Asbestos is found in many commercial products and therefore is much more abundant in our environment and as a result occupational and environmental exposure standards which recommend maximum exposures have been developed to try and reduce harmful effects. However, little is known about the magnitude of exposure to naturally occurring asbestos. It is therefore very difficult to determine whether the above disease outcomes are the result of exposure to erionite or asbestos.

There is also a long lag (many decades) between exposure to erionite or asbestos and the development of disease. This makes it hard to know when and where the exposure occurred, especially when populations are mobile with people living and working in many different locations and occupations over the course of their lives. This combined with limited recording of the occurrence of mesothelioma, and small numbers of people with the disease, makes it difficult to establish strong spatial trends.

Figure 2. Age-standardised Mortality Rates (ASR) from malignant mesothelioma per 100 000 population, by sex, 2021

Nevertheless, some patterns can be observed as shown in Figure 2. Here it can be seen that age-standardised mesothelioma mortality rates tend to be higher in countries where naturally occurring asbestos is present in rock formations (e.g., USA, Italy, Canada, Australia, New Zealand). Notably, New Zealand has one of the highest rates of mesothelioma in the developed world. Global data on the different rates of mesothelioma incidence in men and women point towards occupational exposure as a dominant causal factor (Fig. 2). However, mortality rates are not consistent with spatial patterns in commercial mining of asbestos. For example, New Zealand, unlike Australia and Canada has very little asbestos related extractive industry. Given the absence of asbestos extraction industry in New Zealand the increased proportion of men affected by this disease must be accounted for by exposure to either asbestos or other mineral fibres such as erionite in other occupations such as construction and excavation (Fig 3).

Figure 3. Age-standardised Mortality Rates (ASR) from malignant mesothelioma per 100 000 population, by sex and DHB for all ages in New Zealand.

Figure 3 shows mortality patterns by place of residence in 2018, but these trends are weak and not particularly informative because exposure is not solely linked to where people lived. People relocate over time, and actual exposure is often driven by activities, such as occupational tasks, recreational use, or proximity to contaminated dust sources like quarries, construction sites, gravel roads, and road cuttings—rather than residence alone. Historical asbestos exposure in New Zealand illustrates this complexity. While there were small local sources, such as the asbestos quarry in the upper Tākaka River valley (Nelson region) in the 1940s–50s, most asbestos was imported and extensively used in construction during the building boom from the 1960s, particularly in Auckland and northern Waikato. These historical industrial activities, rather than simple residential proximity, likely account for some of the exposure risk, meaning population-based mortality patterns by residence may only partially reflect where asbestos mines or manufacturing once operated.

To better understand these dynamics, we used data on mesothelioma cases from the Integrated Data Infrastructure (IDI), a secure database that links administrative and survey data to allow population-level research while protecting individual privacy. Although the dataset was limited, it enabled us to construct partial life trajectories for individuals who died from mesothelioma (Sila-Nowicka et al. 2024). We are currently exploring ways to link these locations to potential causal factors, such as those shown in Fig. 3A and Fig. 3B. While location data must remain anonymised, it is possible to create lifelong indices to investigate potential relationships between occupational and environmental exposures in these individuals (Sila-Nowicka et al., 2025).

However, such retrospective analysis based on residence or even partial life histories are not sufficient on their own. We need to move beyond broad-scale trends and static maps that simply illustrate mortality rates at residential scale and instead focus on finer detail and build models that integrate multiple datasets to predict where and when exposure is most likely to occur. While current knowledge largely centres on identifying erionite’s geological presence, the real hazard emerges when erionite-bearing rock is disturbed through activities like tunnelling or quarrying, releasing respirable fibres into the air (Fan et al. 2024). Mapping of the hazard (the geological occurrence of erionite) alone is insufficient—predicting potential exposure is crucial, as urban development, major infrastructure projects, and earthworks can create future risk hotspots. Advanced modelling enables scenario testing (Zelman-Fahm et al., 2023), supports targeted health and safety planning, and addresses the long latency of diseases such as mesothelioma, helping us shift in the future from reactive responses to proactive prevention.

Given the highly localised distribution of mineral fibres in rocks, we focused on modelling at local scales rather than at regional or global levels. We applied several approaches including regression, land-use regression, and XGBoost, to identify areas with higher or lower concentrations of fibres in the air (Zelman-Fahm et al. 2024a). Models incorporated datasets of varying quality and spatial and temporal resolution to evaluate their effects on probability maps of fibre concentration. We also used data fusion techniques to integrate multiple sources and improve model reliability (Zelman-Fahm et al. 2024b).

Figure 4. Maps to show A) the proportion of construction and mining employees usually resident or working in Greater Auckland and B) Locations of unsealed roads and quarries in the same area.

Figure 5. Number of unique mobile phone users passing by a construction site, showing increased activity during morning and afternoon rush hours, with limited variation overnight.

We conducted GIS modelling to identify the locations of at-risk employees (Fig. 4A) and high-risk land uses (Fig. 4B), which were then compared with geological maps from the earlier report. However, a breakdown of disease incidence by employment type is not available. Similarly, there were too few cases of non-occupational mesothelioma to detect any spatial patterns that might coincide with environmental exposure to erionite. The limited availability of high-resolution historical residential and workplace data further adds uncertainty to exposure modelling, making it impossible to quantify risks at broader scales.

Where population-level risk estimation is not feasible, relative intensity of movement through areas with potential exposure such as near construction sites or gravel roads to quarries, can be assessed using mobile phone location data. These data, collected through apps during use or installation, reveal patterns of space usage and temporal variations (Fig. 5). Such analyses support estimation of potential exposure rates at sites with elevated likelihood of airborne mineral fibres.

By combining environmental data, occupational information, and movement patterns, this work can provide a clearer understanding of where mineral fibres in the air may pose a risk. Localised models allow targeted assessment in areas with the greatest potential exposure, supporting informed public health strategies and preventive measures. These methods also establish a foundation for future research that can improve risk evaluation and protection for communities and workers exposed to naturally occurring fibrous minerals.

References:

Berry, Terri-Ann, and Shannon Wallis. “Hazardous mineral wastes and options for bioremediation: New Zealand research.” (2021).

Patel, Janki Prakash, et al. “Global geological occurrence and character of the carcinogenic zeolite mineral, erionite: A review.” Frontiers in Chemistry 10 (2022): 1066565.

Sila-Nowicka, K., Salmond, J. (2024). Life trajectories of exposure to carcinogenic erionite in New Zealand, International Society of Exposure Science, Montreal, Canada. October 20-24.

Sila-Nowicka K. (2025). From seconds to lifetimes: challenges in measuring environmental exposure across spatial and temporal scales, Christchurch, New Zealand, August 21.

Fan, Wenxia Wendy, et al. “Investigating the deposition of fibrous zeolite particles on leaf surfaces: A novel low-cost method for detecting the presence of airborne hazardous mineral fibers.” Journal of Hazardous Materials 480 (2024): 135982.

Zelman-Fahm D. (2023). The Utility of GIS for Assessing Exposure Risk. ESRI Regional User Group, 31 March.

Zelman-Fahm, D., Sila-Nowicka, K., Salmond, J. (2024a). Modelling Exposure to Little-Known Hazardous Mineral Fibres: Iterative Approach with Regression Models and Spatial Variables, International Society of Exposure Science, Montreal, Canada. October 20-24.

Zelman-Fahm, D., Sila-Nowicka, K., Salmond, J. (2024b). Spatial Regression Modelling for Exposure Assessment of Hazardous Mineral Fibers. Hobart, Australia, August.