Evaluating an Air Quality Health Index (AQHI) amendment for communities impacted by residential woodsmoke in British Columbia, Canada

Figures & data

Figure 1. Sample one-week time series of AQHI, AQHI-Plus, and fine particulate matter (PM_2.5) hourly values from 2015 in Howe Sound, an area with both wildfire and residential wood burning impacts. Vertical x-axis lines represent the 00:00 hour mark of that day in the time series.

Figure 2. Local Health Area (LHA) boundary map of British Columbia displaying LHAs included in the study and locations of the Air Quality Health Index (AQHI) and weather stations.

Table 1. Results from the woodsmoke-impacted detection algorithm sorted by percentage of days with data that were classified as woodsmoke-impacted. The study period included 1,698 days during the months of October to April, beginning January 1, 2010 and ending December 31, 2017.

Local Health Area	Woodsmoke-impacted days	Total days with data	Percentage of woodsmoke- impacted days
1. Courtenay	637	1357	46.9	Woodsmoke-impacted areas
2. Howe Sound	768	1664	46.2
3. Smithers	655	1587	41.3
4. Greater Victoria	455	1676	27.2
5. Quesnel	390	1466	26.6
6. Cowichan	358	1452	24.7
7. Castlegar	111	462	24.0
8. Campbell River	136	621	21.9
9. Sooke	291	1448	20.1
10. Cariboo-Chilcotin	294	1556	18.9
11. Prince George	277	1502	18.4
12. Peace River North	57	475	12.0	Less-impacted areas
13. Central Okanagan	184	1571	11.7
14. Mission	52	593	8.8
15. Vernon	137	1584	8.7
16. Nanaimo	54	689	7.8
17. Chilliwack	84	1141	7.4
18. Hope	83	1149	7.2
19. Abbotsford	81	1168	6.9
20. Agassiz-Harrison	42	809	5.2
21. Kamloops	37	1556	2.4

Figure 3. Mean fine particulate matter (PM_2.5) plotted by Local Health Area (LHA), time of day, and month. LHAs are ordered by the percent of woodsmoke impacted days (Table 1). The PM_2.5 values are standardized within their respective LHA. Months with intense wildfire activity are excluded (July to September in 2010, 2014, 2015, and 2017).

Table 2. Summary descriptions of the woodsmoke-impacted areas compared with the less-impacted areas.

	Woodsmoke-impacted areas	Less-impacted areas
Local Health Area (LHA) information
Number of LHAs	11	10
Number of woodsmoke-impacted days	4,372	811
Mean population	64,242	90,549
Mean daily health outcome rates (per 100,000)
Non-accidental mortality	2.0	2.5
Cardiovascular physician visit rate	115.6	127.3
Respiratory physician visit rate	124.1	136.1
Asthma physician visit rate	11.7	13.0
Salbutamol dispensation rate	46.2	47.8
Mean daily environmental exposures
Mean daily AQHI	3.2	3.0
Mean daily AQHI-Plus	3.5	3.0
Mean hourly PM2.5	9.0 µg/m^3	6.2 µg/m^3
Mean hourly NO2	7.8 ppb	9.1 ppb
Mean hourly O3	18.7 ppb	18.7 ppb
Mean daily apparent temperature	0.0°C	1.8°C

Figure 4. The percentage of hours during which the Air Quality Health Index (AQHI) was overridden by a greater AQHI-Plus value, shown by Local Health Area (LHA), time of day, and month. The LHAs are ordered by the percent of woodsmoke-impacted days (Table 1). Months with intense wildfire activity are excluded (July to September in 2010, 2014, 2015, and 2017).

Table 3. Frequency distribution of Air Quality Health Index (AQHI) and AQHI-Plus daily values used in the study for the woodsmoke-impacted areas.

Value	AQHI	AQHI-Plus	Risk category
1	14	9	Low
2	2,009	1,586
3	8,911	8,059
4	2,691	2,886	Moderate
5	566	976
6	132	413
7	26	230	High
8	6	117
9	3	41
10	2	18
10+	0	25	Very High

Table 4. AIC values for models fit for the AQHI and AQHI-Plus to each health outcome, controlling for both 21-day rolling average apparent temperature and same-day apparent temperature.

	AQHI	AQHI-Plus
Non-Accidental mortality	37,195.0	37,194.0
Cardiovascular physician visits	104,619.4	104,614.1
Respiratory physician visits	108,306.8	108,311.4
Asthma physician visits	46,247.3	46,238.6
Salbutamol dispensations	94,630.0	94,600.9

Figure 5. Relative rates and confidence intervals for 1-unit increases in the AQHI and AQHI-Plus for each health outcome: i) controlling for both 21-day rolling average apparent temperature and same-day apparent temperature; ii) controlling for same-day apparent temperature; and iii) without any temperature control. For models with both long- and short-term temperature controls, model fit was assessed using the Akaike information criterion (AIC), where a lower value indicates a better fit.

Figure 6. Relative rates and confidence intervals for 1-unit increases in the AQHI and AQHI-Plus for each health outcome for the leave-one-out sensitivity analysis of Local Health Areas (LHAs). Models are adjusted for 21-day rolling average apparent temperature and same-day apparent temperature. Removed LHAs are ordered left-to-right by the percent of woodsmoke-impacted days (Table 1).