LAUSR

While bushfires have long been a part of Australia's ecosystem, the 2019– 20 bushfires were an unprecedented event leading to 5.5 million hectares of land being affected in New South Wales (NSW) alone.1 This is approximately the same as the area burnt in the combined bushfire seasons of 1993– 94 and 2018– 19.2 Air pollution has been recognized as a risk factor for asthma development.3 An increase in emergency department visits and hospital admissions has been described in asthmatic children exposed to bushfire smoke.4 Bushfire smoke contains a complex range of gases and particles, including a particulate matter measuring less than 2.5 microns (PM2.5) composition, carbon monoxide, ozone, methane, and nitrogen oxides which is different from the one emitted by fossil fuel combustion.5 The NSW Air Quality Monitoring Network reported that 74% of the spring and summer days in 2019– 20 met poor air quality standards according to the Air Quality Index.6 This study therefore aimed to determine the effects of bushfire smoke exposure during the 2019– 20 bushfire season on lung function in asthmatic children. Our study used routinely collected data from children with asthma who attended the John Hunter Children's Hospital outpatient childhood asthma clinic between January 2019 and April 2020. Data were collected from electronic medical records and clinical lung function databases from children aged 4 to 16 years of age, with a confirmed diagnosis of asthma seen by a Pediatric respiratory specialist. Only children with a valid spirometry meeting ATS/ERS standards during their clinical visit were included.7 Before the appointment, parents and patients were instructed not to take the following medications for the times indicated, unless they are breathless and need to use them: shortacting inhalers (Ventolin, Atrovent, Flixotide, Bricanyl, Asmol, Airomir, Epaq) for 8 h; and longacting inhalers (Symbicort, Seretide, Serevent, Spiriva, Oxis, Foradile) for 12 h. Data included demographic characteristics, lung function (spirometry and fraction of exhaled nitric oxide at a flow rate of 50 ml/s [FeNO50] if available), and symptom control in the past 4 weeks. Global Lung Initiative (GLI) reference equations were used to calculate spirometry measurements (FEV1, FVC, and FEV1/FVC ratio percent of predicted value [% predicted]). Bronchodilator reversibility was calculated after administration of inhaled rapidonset beta2 agonist bronchodilator (SABA).8 For first nation Australian children, the GLI ‘Other/Mixed equation reference’ was used for the data collection. PM2.5 exposure during the period was assessed for participants located within the Sydney Greater Metropolitan Region (GMR) study region in NSW, an area of approximately 1860 square kilometers. Daily 24h mean PM2.5 data from fixedsite government air quality monitoring stations within the study regions were measured9 and interpolated using an inverse distance weighting procedure to estimate daily PM2.5 exposure concentration for participant's residential location.10 Bushfire days were defined when three prerequisites were met: (1) the entire study region's 24h average of PM2.5 concentration exceeded the 95th percentile (based on the period 01/01/2010 to 31/12/2018 for the Sydney GMR), (2) visual confirmation of fire for that day or up to 3 days before or after via satellite imagery, and (3) interpolated PM2.5 reading for each participant's residential address also exceed the 95th percentile for the region, to control for spatial variability in the region.10 This approach has been successfully used to identify bushfiresmokeaffected study days in several previous epidemiological studies.10,11 Categorical measures were described using counts and percentages, while continuous measures were characterized using means with standard deviation (SD). Statistical significances were calculated using twotailed Mann– Whitney U test or Student's t test as appropriate and chisquare test if categorical. As this was a longitudinal study with different time points, we used mixed effect regression models with a random intercept to estimate the total effect of bushfirerelated smoke exposure on lung function of children. Outputs of fixed and random effects were calculated for the presence of bushfire smoke on the day of the test and for PM2.5 levels on the day of the lung function testing. Models were adjusted to account for differences found between the groups in the univariate analysis and to account for factors that usually can influence lung function, including sex, age, weight, height, being a first nation Australian, tobacco smoke exposure at home, and season. A specific sample size was not targeted; instead, a convenience sample was used. Statistical analysis was performed using STATA version 15 (Texas, USA). A pvalue of <.05 was considered statistically significant. This study was approved by the Hunter