Global event-based surveillance of chemical incidents

The paper addresses the growing risk of human exposure to hazardous chemicals due to the increasing manufacture, transport, and use of chemicals globally. It frames chemical incidents as uncontrolled releases that can arise from accidents, disasters, or deliberate acts, causing acute and/or chronic health effects via various exposure routes. Because chemical incidents can escalate rapidly compared to other hazards, timely surveillance is critical for public health response. In line with International Health Regulations (2005) and EU Decision 1082/2013/EU, the study explores event-based surveillance (EBS) as a means to collect and assess largely unstructured information from formal and informal sources to heighten situational awareness of chemical incidents and inform cross-border public health preparedness and response.

The EBS strategy began in 2014 to identify chemical incidents within the EU with cross-border potential and was later expanded globally. Searches of specified online sources (news media, aggregation platforms, and surveillance tools) are conducted three times per week using predefined terms (e.g., chemical, toxic, poison, explosion) across sites such as BBC News, MediSYS, ProMedMail, RSOE EDIS, HealthMap, GPHIN, InformationAware, Twitter, Google News, and Google Alerts. Inclusion focuses on incidents with public health impact (injuries or deaths) rather than near misses or solely environmental/property damage. Recording and notification criteria require: (1) an unexpected chemical incident (release/explosion/contamination/fire) and (2) mortality or (3) significant injury; with notification guided by (4) cross-border potential (proximity to borders/ports/rivers) or (5) potential spread leading to additional morbidity/mortality. Stakeholders notified include relevant PHE staff, European poison centres and public health agencies, WHO (International/Europe), and the EU SCHEER committee, depending on incident characteristics. Data management employs a Microsoft Access interface with SharePoint Lists backend, structured into four entities (Incidents, Agents, References, Categories). Incidents can have multiple category tags. Detected items are first circulated by email for internal review after each search cycle, then emails are imported from Outlook and parsed into the database. Geographic coordinates are recorded to enable mapping and spatial analyses.

From November 2014 to June 2020, 1592 chemical incidents were recorded across 121 countries, involving 252 unique chemical agents. Examples of cross-border–relevant incidents shared with RASCHEM include: an October 2015 methanol poisoning outbreak in Turkey linked to adulterated spirits; a July 2016 incorrect formulation of infant vitamin D supplement in Denmark sickening 25 babies; and an August 2017 chemical mist exposure affecting ~150 beachgoers at Birling Gap, UK, with uncertain origin and potential to affect French coasts. The five countries with the most detected incidents were USA 322 (20.2%), India 225 (14.1%), UK 130 (8.2%), China 117 (7.3%), and Russia 55 (3.5%). Although detection likely reflects reporting capacity differences, associated health burdens differed: USA-linked incidents were associated with 117 deaths and 6497 injuries, whereas India-linked incidents had 989 deaths and 6749 injuries, indicating a higher fatality burden in India among detected events. Source performance over the most recent 12 months showed RSOE EDIS 68 (20.0%), InformationAware 63 (18.5%), Google Alerts 48 (14.1%), GPHIN 36 (10.6%), Google News 23 (6.8%), BBC News 15 (4.4%), HealthMap 13 (3.8%), ProMedMail 10 (2.9%), MediSYS 10 (2.9%), Twitter 8 (2.4%), Other 46 (13.5%). Methanol-related incidents emerged as frequent and high-burden events; during the COVID-19 pandemic, misinformation drove increased consumption of adulterated alcohol in some settings (e.g., Iran reported over 700 deaths and more than 5000 injuries due to methanol poisoning). Overall, the majority of health burden stemmed from frequent smaller-scale incidents (e.g., methanol poisonings) rather than the large-scale industrial events that EBS often targets.

The findings demonstrate that multi-source EBS can rapidly detect and characterize chemical incidents with public health impact, supporting early situational awareness and cross-border notification obligations under EU Decision 1082/2013 and IHR (2005). The observed country distribution of detected events likely reflects differences in media/reporting capacity rather than true incidence; nonetheless, the dataset highlights substantial morbidity and mortality, with smaller, recurrent events (e.g., methanol poisonings) contributing disproportionately to health burden compared with infrequent large industrial accidents. The EBS approach, leveraging tools designed primarily for communicable diseases, provides cost-effective coverage for chemical incidents and can inform training, exercises, and preparedness planning. Integration with complementary systems (e.g., syndromic surveillance, poison centre data, hospital data) could improve validation, case ascertainment, and follow-up, enhancing trend analysis and public health actionability. While not designed for longitudinal follow-up, EBS effectively meets its goal of early detection and alerting, enabling stakeholders to assess risk and initiate responses when appropriate.

The study presents a practical, low-cost methodology for global event-based surveillance of chemical incidents, showing its utility in rapidly identifying incidents, informing cross-border notifications, and enhancing situational awareness and preparedness. The compiled database and analyses highlight key trends (notably the significant burden from methanol poisonings) and source performance. Future work should focus on sustainability and scalability through increased automation of source scanning and data ingestion, systematic notification mechanisms, and integration with other surveillance and clinical data systems (poison centres, hospital data, syndromic platforms) to improve validation, follow-up, and comprehensive all-hazards surveillance. Strengthening networks and sustained mechanisms at national and international levels would further enhance chemical incident preparedness and reduce global health impacts.

EBS relies largely on unstructured media and online reports that may be incomplete, unverified, or change as incidents evolve; chemical identities are often not reported, particularly where analytical capacity is limited. There is reporting bias across countries and difficulty establishing primary sources; detection counts may reflect media/reporting capacity rather than true incidence. The system focuses on incidents with reported injuries or deaths, omitting near misses and many chronic/environmental exposures; mental health impacts are out of scope. Routine follow-up and verification are not included due to resource constraints, leading to potential inaccuracies in casualty figures. Manual searching and data entry are time-intensive, with limited automation. Few surveillance tools are dedicated specifically to chemical incidents, necessitating reliance on platforms designed for communicable diseases.