Skip booklet index and go to page content

Data Sources and Methods: Air Health Indicator – Ozone and Fine Particulate Matter


Canadian communities for which the ground-level ozone and fine particulate matter (PM2.5) concentrations were used for the National Air Quality Indicators of CESI were considered. The AHI is based on the criteria of having a reasonably complete time series of pollution and weather measurements, and enough daily mortality.

For each community there were three types of data used for the AHI: daily numbers of cause-specific deaths, air pollution concentrations, and potential confounders to the mortality-air pollution association.

Data source

Daily numbers of cause-specific deaths

The daily numbers of cause-specific deaths (non-accidental mortality data) were obtained from the national mortality database (Vital Statistics Database–Death 2004) maintained by Statistics Canada. Based on the International Classification of Diseases (ICD), the mortality data included only deaths from internal causes (ICD-9 code < 800 and ICD-10 code A00-R00), excluding external causes such as injuries. Regarding cause-specific deaths, in particular, we were interested in cardiopulmonary mortality related to the circulatory or respiratory system. For this specification, our mortality data were categorized into a cardiopulmonary group (ICD-10 code between I20–I50 and J10–J67). The cardiopulmonary mortality data were extracted by Statistics Canada for a specified census division only where the census division of residence was the same as the census division of death occurrence.

Air pollution concentrations

The daily ozone and PM2.5 (the latter measured by the tapered element oscillating microbalance method or TEOM) concentration data were obtained from the National Air Pollution Surveillance (NAPS) Network operated by Environment Canada. Established in 1969, NAPS provides accurate and long-term air quality data of a uniform standard across Canada to monitor the quality of ambient (outdoor) air in populated regions by specific procedures for the selection and positioning of monitoring stations. For each NAPSmonitoring station, the daily average concentration for a certain day was calculated only if at least 75% of 24 hourly concentrations for that day (i.e. at least 18 hourly concentrations) were available. Otherwise, it was recorded as missing. For each census division, the daily average concentration was averaged over monitoring stations if there were 2 or more stations located in that census division. For the metric of air pollutions, the daily 8-hour maximum was selected for ozone and the daily mean for PM2.5

Potential confounders to the mortality-air pollution association

As for potential confounding variables to the exposure-mortality association, three factors were considered: time; temperature; and indicators for days of the week. Calendar time is included to control both temporal and seasonal variations. Daily temperature controls for the short-term effect of weather on daily mortality; and day of the week accounts for mortality that varies by day of the week. Specifically, to account for the weather effect, daily mean temperature data were obtained from the National Climate Data and Information Archive of Environment Canada. As for lifestyle factors such as smoking or cholesterol in the community, they do not vary meaningfully from day to day and thus can be ignored as confounders.

Spatial coverage

Twenty Canadian communities (Saint John, Québec, Montréal, Ottawa, York, Toronto, Peel, Oakville, Hamilton, Niagara Falls, Kitchener, Windsor, Sarnia, Sault Ste. Marie, Winnipeg, Regina, Saskatoon, Calgary, Edmonton, and Vancouver) were selected for ozone. Eighteen communities (Saint John, Québec, Montréal, Ottawa, Toronto, Peel, Oakville, Hamilton, Niagara Falls, London, Windsor, Sarnia, Waterloo, Winnipeg, Regina, Calgary, Edmonton, and Vancouver) were selected for PM2.5.

Each community’s geographic boundaries were defined by the census division associated with the city.

Temporal coverage

Yearly data for the years 1990 to 2008 were used for ozone and yearly data for the years 2000 to 2008 were used for PM2.5.

Data timeliness

Mortality data are difficult to obtain and are a few years behind the other data. Raw 2007 data are now available but only the 2004 data were available in the correct format and details for use with the AHI. Consequently, the years 2005 to 2008 were approximated from the average of national annual risk (mortality data).