The British Columbia Asthma Prediction System (BCAPS): A Surveillance System to Forecast the Public Health Impacts of Wildfire Smoke

摘要

Smoke from severe wildfires during the summers of 2015 and 2017 affected most of the population of British Columbia (BC), Canada. Asthma-related health outcomes are consistently elevated when populations are exposed to wildfire smoke in BC. Timely information during smoke events is important to inform the actions of public health authorities. We developed a surveillance system using short-term wildfire smoke forecasts to predict the potential health impacts of smoke over the coming days and provide this information to public health authorities in easy-to-interpret daily reports. The BCAPS framework is modular such that different data, modelling approaches, and visualizations can be applied. We used daily fine particulate matter (PM2.5) measurements, daily counts of asthma-related population health outcomes, and PM2.5 forecasts to predict and visualize smoke exposures and their health impacts across different geographic areas over the coming 48 hours. We applied BCAPS retrospectively during a 2015 smoke event using a Bayesian latent process model to forecast health outcomes. We prospectively applied BCAPS in 2017 using random forest models to predict PM2.5 exposure and health outcomes. Daily PM2.5 measurements ranged from 0.03 μg/m3 to 301.2 μg/m3 during the 2015 event and from 0.05 μg/m3 to 293.8 μg/m3 in 2017. Daily PM2.5 and daily counts of asthma-related physician visits and medication dispensations were increased during smoky periods in 2015 and 2017. In general, BCAPS predicted the smoke-related increases in asthma outcomes with good accuracy, though performance was dependant on the performance of the smoke forecasts. In 2017 there was a marked decrease in population response to smoke towards the end of the season even though PM2.5 concentrations remained high. Integrating data from multiple sources into a modular framework such as BCAPS can usefully predict the health impacts of smoke exposure in a timely manner to inform public health decision-making and action.