February 12 | 2018

Effect of air quality alerts on human health: a regression discontinuity analysis in Toronto, Canada

Hong Chen, PhD, Qiongsi Li, MMath, Jay S Kaufman, PhD, Jun Wang, MSc, Ray Copes, MD, Yushan Su, PhD, Tarik Benmarhnia, PhD

The Lancet Planetary Health, Volume 2, No. 1, e19–e26, January 2018  DOI: https://doi.org/10.1016/S2542-5196(17)30185-7



Ambient air pollution is a major health risk globally. To reduce adverse health effects on days when air pollution is high, government agencies worldwide have implemented air quality alert programmes. Despite their widespread use, little is known about whether these programmes produce any observable public-health benefits. We assessed the effectiveness of such programmes using a quasi-experimental approach.


We assembled a population-based cohort comprising all individuals who resided in the city of Toronto (Ontario, Canada) from 2003 to 2012 (about 2·6 million people). We ascertained seven health outcomes known to be affected by short-term elevation of air pollution, using provincial health administrative databases. These health outcomes were cardiovascular-related mortality, respiratory-related mortality, and hospital admissions or emergency-department visits for acute myocardial infarction, heart failure, stroke, asthma, and chronic obstructive pulmonary disease (COPD). We applied a regression discontinuity design to assess the effectiveness of an intervention (ie, the air quality alert programme). To quantify the effect of the air quality alert programme, we estimated for each outcome both the absolute rate difference and the rate ratio attributable to programme eligibility (by intention-to-treat analysis) and the alerts themselves (by two-stage regression approach), respectively.


Between Jan 1, 2003, and Dec 31, 2012, on average between three and 27 daily cardiovascular or respiratory events were reported in Toronto (depending on the outcome). Alert announcements reduced asthma-related emergency-department visits by 4·73 cases per 1 000 000 people per day (95% CI 0·55–9·38), or in relative terms by 25% (95% CI 1–47). Programme eligibility also led to 2·05 (95% CI 0·07–4·00) fewer daily emergency-department visits for asthma. We did not detect a significant reduction in any other health outcome as a result of alert announcements or programme eligibility. However, a non-significant trend was noted towards decreased asthma-related and COPD-related admissions.


In this population-based cohort, the air quality alert programme was related to some reductions in respiratory morbidity, but not any other health outcome examined. This finding suggests that issuing air quality alerts alone has a limited effect on public health and that implementing enforced public actions to reduce air pollution on high pollution days could be warranted. Together with accumulating evidence of substantial burden from long-term air pollution exposure, this study underscores the need for further strengthening of global efforts that can lead to long-term improvement of overall air quality.


Public Health Ontario, Canadian Institutes for Health Research.