Exposure to air pollution during the first 1000 days of life and subsequent health service and medication usage in children

Journal Pre-proof Exposure to air pollution during the first 1000 days of life and subsequent health service and medication usage in children Jingyi Shao, Graeme R. Zosky, Amanda J. Wheeler, Shyamali Dharmage, Marita Dalton, Grant J. Williamson, Tierney O'Sullivan, Katherine Chappell, Luke D. Knibbs, Fay H. Johnston PII:

S0269-7491(19)33477-3

DOI:

https://doi.org/10.1016/j.envpol.2019.113340

Reference:

ENPO 113340

To appear in:

Environmental Pollution

Received Date: 24 July 2019 Revised Date:

1 October 2019

Accepted Date: 1 October 2019

Please cite this article as: Shao, J., Zosky, G.R., Wheeler, A.J., Dharmage, S., Dalton, M., Williamson, G.J., O'Sullivan, T., Chappell, K., Knibbs, L.D., Johnston, F.H., Exposure to air pollution during the first 1000 days of life and subsequent health service and medication usage in children, Environmental Pollution (2019), doi: https://doi.org/10.1016/j.envpol.2019.113340. This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2019 Published by Elsevier Ltd.

Relative risk (95% CI) of each health outcome per 10 µg m-3 increase in infant exposure to mine fire related PM2.5

• Background: Hazelwood coal mine fire 2014 in the Latrobe Valley, Victoria,

1.8 1.6 1.4

Australia

1.2 1

• Participants: 286 children (age by

0.8 0.6

31/12/2016: 2.8±1.0 years)

0.4 0.2 0 GP attendances Asthma inhaler dispensation

Steroid skin cream dispensation

Antibiotic dispensation

• Exposure estimate: Chemical transport modelling

Relative risk (95%CI) of each health outcome per 10 µg m-3 increase in intrauterine exposure to mine fire related PM2.5

• Key findings: Exposure to coal mine fire emissions during infancy was associated

3 2.5

with increased dispensing of antibiotics.

2 1.5

This could reflect increased childhood

1 0.5

infections or increased prescriptions of

0 GP attendances Asthma inhaler Steroid skin dispensation cream dispensation

Antibiotic dispensation

antibiotics in the year following the fire.

1

Exposure to air pollution during the first 1000 days of life and

2

subsequent health service and medication usage in children

3 4

Jingyi Shao1, Graeme R. Zosky1,2, Amanda J. Wheeler1,3, Shyamali Dharmage4, Marita Dalton1, Grant

5

J. Williamson5, Tierney O’Sullivan1, Katherine Chappell1, Luke D. Knibbs6, Fay H. Johnston1

6

1

7

Medicine, Faculty of Health, University of Tasmania, Hobart, Tasmania 7000, Australia; 3Behaviour,

8

Environment and Cognition Research Program, Mary MacKillop Institute for Health Research, Australian

9

Catholic University, Melbourne, Victoria 3000, Australia; 4Allergy and Lung Health Unit, Melbourne School of

Menzies Institute for Medical Research, University of Tasmania, Hobart, Tasmania 7000, Australia; 2School of

10

Population and Global Health, University of Melbourne, Carlton, Victoria 3052, Australia; 5School of Natural

11

Sciences, University of Tasmania, Hobart, Tasmania 7000, Australia; 6School of Public Health, Faculty of

12

Medicine, The University of Queensland, Herston, Queensland 4006, Australia.

13 14

Correspondence: A/Prof Fay H. Johnston, Menzies Institute for Medical Research, University of Tasmania, 17

15

Liverpool Street, Hobart, Tasmania 7000, Australia. E-mail: [email protected]

16 17

18 19 20 21 22 23 24 25 26

27

Abstract

28

Background: Evidence of health effects following early life exposure to short-to-medium

29

duration of high pollution levels is extremely limited.

30

Objectives: We aimed to evaluate the associations between: 1. intrauterine exposure to fine

31

particulate matter (PM2.5) from coal mine fire emissions and the frequencies of general

32

practitioner attendances and dispensations of prescribed asthma inhalers, steroid skin creams,

33

and antibiotics during the first year of life; 2. infant exposure and those outcomes during the

34

year following the fire.

35

Methods: All participants were recruited from the Latrobe Valley of Victoria, Australia.

36

Participants’ 24-hour average and hourly peak mine fire-specific PM2.5 exposures from

37

09/02/2014 to 31/03/2014 were estimated using chemical transport modeling. Outcome data

38

were obtained from the Australian Medicare Benefits Schedule and Pharmaceutical Benefits

39

Scheme from each child’s birth to 31/12/2016. We used negative binomial and logistic

40

regression models to independently assess risks of the outcomes associated with every 10 and

41

100 µg/m3 increase in average or peak PM2.5 exposure, respectively, while adjusting for

42

potential confounders.

43

Results: We included 286 of 311 children whose parents consented to be linked, comprising

44

77 with no exposure, 88 with intrauterine exposure and 121 with exposure in infancy. 10- and

45

100- µg m-3 increases in average and peak PM2.5 exposure during infancy were associated

46

with greater incidence of antibiotics being dispensed during the year following the fire: the

47

adjusted incidence rate ratios were 1.24 (95% CI 1.02, 1.50, p=0.036) and 1.14 (1.00, 1.31,

48

p=0.048) respectively. No other significant associations were observed.

49

Conclusion: Exposure to coal mine fire emissions during infancy was associated with

50

increased dispensing of antibiotics. This could reflect increased childhood infections or

51

increased prescriptions of antibiotics in the year following the fire.

52

Keywords: particulate matter, infant, prenatal exposure, infection, allergy and immunology

53 54

Main findings: Infant exposure to air pollution from coal mine fire emissions might be

55

associated with increased childhood infections.

56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73

74

Introduction

75

The first thousand days of life includes the periods in utero which usually lasts for about 9

76

months, and the first two years after birth (i.e. 9 ∗ 30 + 365 ∗ 2 = 1000 days). It is

77

recognised as a critical window for the development and growth of the respiratory and

78

immune systems (Dietert et al., 2000). There is emerging evidence that air pollution exposure

79

during this period could result in long-term adverse immunological or respiratory outcomes.

80

For example, previous studies have demonstrated that early life exposure to industrial and

81

traffic-related air pollution is associated with the development of childhood asthma and

82

allergic diseases (Hehua et al., 2017, Bowatte et al., 2015, Deng et al., 2015). Intrauterine

83

exposure to both particulate matter with an aerodynamic diameter < 2.5 micrometers (PM2.5)

84

and second-hand smoke (SHS) has been associated with increased risk of infantile eczema

85

(Jedrychowski et al., 2011). Epidemiological studies have also shown significant associations

86

between air pollution exposure in utero or during the first year of life and childhood

87

pneumonia, bronchiolitis and ear infections (Soh et al., 2018, Kennedy et al., 2018, Rice et al.,

88

2015), further highlighting potential susceptibility during this period. Exposure to air

89

pollution prompts immediate immune responses (Carlsen et al., 2016, Adetona et al., 2013)

90

and can modulate later immune expression (Yi et al., 2017, Rice et al., 2015). It is therefore

91

plausible that short-term exposure to air pollution in the critical first 1000 days of life, from

92

conception to age 2 years, could affect later immunological function (Wopereis et al., 2014).

93

However very few studies have evaluated this.

94 95

Smoke from outdoor landscape fires including burning forest, grass and peat makes a

96

significant contribution to air pollution (Johnston et al., 2012) and is an increasing global

97

concern due to the rising frequency and severity of fires resulting from climate change (Liu et

98

al., 2016). Epidemiological studies suggest that smoke exposure is associated with short-term

99

increases in medication usage, physician/emergency department visits, hospitalisations and

100

death (Reid et al., 2016, Black et al., 2017b). However, evidence of long-term health

101

outcomes following exposure to short-to-medium duration smoke events (i.e. weeks) is

102

extremely limited (Melody and Johnston, 2015, Black et al., 2017b).

103 104

Embers from a bushfire in the Latrobe Valley region of Victoria, Australia, ignited a fire in

105

an open cast coal mine located close to several rural towns in February 2014 that lasted for

106

about 45 days. The episode resulted in dramatically increased concentrations of PM2.5. The

107

peak daily average PM2.5 concentration reached 731 µg/m3 in the closest town, Morwell,

108

which is substantially higher than the national daily air quality standard of 25 µg m-3 (Reisen

109

et al., 2017, Department of the Environment and Heritage). One of the main concerns of the

110

community during this period was the possible risks to their long-term health. As there was

111

little existing evidence to draw on, the state government initiated a long-term study, the

112

Hazelwood Health Study, to investigate the health and wellbeing of adults and children

113

affected by the smoke episode (Melody et al., 2018).

114 115

We hypothesised that exposure to air pollution from the coal mine fire during the intrauterine

116

or infant periods would increase the risk of common allergic or infective illnesses in the year

117

following exposure. The aim of this study was to test if exposure to smoke from the coal

118

mine fire during the first 1000 days of life was associated with increased physician visits or

119

dispensing of medications used to treat infections, asthma or atopic skin conditions.

120 121

Materials and Methods

122

1.1 Study design

123

We linked data from a cohort of children recruited to the Latrobe Early Life Follow-up (ELF)

124

Study (Melody et al., 2018) to two national Australian administrative health datasets: the

125

Medicare Benefits Schedule (MBS) and the Pharmaceutical Benefits Scheme (PBS). Data

126

were extracted by the Australian Department of Human Services for the period from each

127

child’s date of birth to 31/12/2016. The MBS dataset contained de-identified information on

128

claims to Medicare, the national insurance system, for out-of-hospital health services such as

129

visits to GPs and specialists. The PBS dataset contained de-identified information on

130

prescription medications dispensed to patients. It captured medication dispensations that were

131

subsidised by the Australian government.

132 133

The Latrobe ELF cohort comprises 571 children born between 01/03/2012 and 31/12/2015,

134

who were recruited from the Latrobe Valley, Victoria, during 2016 as part of a long-term

135

follow-up study of the health impacts of the 2014 Hazelwood coal mine fire (Shao et al.).

136

Details of this cohort are described elsewhere (Melody et al., 2018). Sociodemographic,

137

health and residential characteristics of the participants (n=571) were obtained from a

138

baseline questionnaire completed by parents/carers at enrolment. Parental consent for linkage

139

with MBS and PBS datasets was obtained from 311 participants. We recruited four groups of

140

participants according to their dates of birth and gestational age at delivery. These were: (1)

141

the intrauterine exposure group, which comprised children whose mothers were pregnant

142

with them during the period of the mine fire; (2) the infant exposure group which comprised

143

children who were aged between 0-2 years during the entire fire period; (3) the mixed

144

exposure group, who were born during the fire period; and (4) an unexposed group, who were

145

conceived and born in the year following the fire. Children in the mixed exposure group

146

(n=25) were not included in our primary analysis. The annual average PM2.5 concentration

147

during the year 2015 when most of the unexposed children were born was 6.7 µg m-

148

3

149

been exposed to very low levels of environmental PM2.5 during their perinatal periods.

150

The Tasmanian Health and Medical Human Research Ethics Committee (reference H14875)

151

approved this study. Additional approval was received from the Human Research Ethics

152

Committees of Monash University, Monash Health, and the University of Melbourne. All

153

parents or caregivers of the studied participants provided signed consent forms for accessing

154

data from the MBS/PBS datasets.

155

1.2 Exposure estimates

156

Hourly coal mine fire-specific PM2.5 concentrations during the time of the fire (09/02/2014-

157

31/03/2014) were estimated using meteorological, dispersion and chemical transport

158

modelling at a spatial resolution of 1×1 km. Details of the methods used in generating the

159

modelled exposure estimates have been previously reported (Emmerson et al., 2016). The full

160

model included PM2.5 from natural sources, traffic, power stations, landscape fires and mine

161

fire emissions. The differences between the model run with, and without, mine fire emissions

162

were used to estimate the concentration of mine fire-specific PM2.5. The magnitude of the

163

modelled PM2.5 matched reasonably well with the observed PM2.5 concentrations, but the

164

exact timing of the peak values was less accurate on an hourly basis (Emmerson et al., 2016).

165

Therefore, we calculated individual 24-hour average and the peak hourly value of 24-hour

166

PM2.5 concentrations during the exposure period, based on air pollution concentrations at

167

participants’ day and night locations from baseline questionnaires. Those children conceived

168

after the mine fire were allocated a mine fire-specific PM2.5 concentration of zero.

169

We also assessed each child’s exposure to annual average nitrogen dioxide (NO2)

170

concentrations in order to adjust for the effects from longer-term exposures to background

171

non-fire sources of air pollution particularly from motor vehicles and power stations. Annual

(Environment Protection Authority Victoria, 2019), therefore, the unexposed children had

172

ambient nitrogen dioxide (NO2) concentrations for the years 2011 to 2015 were estimated

173

using a national satellite-based land-use regression (LUR) model (Knibbs et al., 2014) at

174

‘mesh blocks’, the smallest spatial unit in the Australian census (n = ~347,000 nationally)

175

(Australian Bureau of Statistics, 2011b). In an external validation, the LUR model explained

176

66% of spatial variation in NO2 at traffic-influenced and background sites (RMSE: 2 ppb

177

[25%]) (Knibbs et al., 2016). We assigned birth year NO2 estimates to the participants

178

according to their home addresses at birth.

179

1.3 Outcome definition

180

Health outcomes of interest were decided a priori, including GP attendances, dispensations of

181

prescribed asthma inhalers, steroid skin creams and antibiotics during the first year of life, or

182

the year following the fire. We analysed all MBS claims relating to consultations with a GP

183

and PBS records of dispensations of prescribed medications used to treat asthma, atopic

184

dermatitis, and bacterial infections (Table S1-S4).

185

Evaluation of outcomes in intrauterine exposure analysis

186

For intrauterine exposure analysis, we included children in the intrauterine exposure group

187

(birthdate: 01/04/2014-31/12/2014) and all children who were not exposed to mine fire

188

smoke in their first year of life. Unexposed children included the unexposed group who were

189

conceived and born after the fire (birthdate: 01/01/2015-31/12/2015), and also those from the

190

infant exposure group who were not exposed to mine fire smoke until their second year of life

191

(birthdate: 01/03/2012-09/02/2013). Our main outcome measures for intrauterine exposure

192

analysis were restricted to the first year of life.

193

Evaluation of outcomes in infant exposure analysis

194

This analysis included children aged 0-2 years at the time of the fire, and the unexposed

195

group of children born during 2015. For the infant exposure group, we evaluated outcomes in

196

the year following the fire from 01/04/2014 to 31/03/2015 and for the comparison group we

197

evaluated outcomes in the year from 01/01/2016 to 31/12/2016.

198

1.4 Covariates

199

We selected a list of potential confounders and effect modifiers a priori using a directed

200

acyclic graph in DAGitty (Textor et al., 2011, Williamson et al., 2014). Potential covariates

201

were selected according to the existing literature on air pollution and child health (Vanker et

202

al., 2017, Feleszko et al., 2014, Lee et al., 2018, Deng et al., 2018, Uphoff et al., 2014,

203

Golenko et al., 2015). We included age (months), sex, maternal tobacco smoking status

204

during pregnancy (yes vs. no), SHS exposure (yes vs. no), maternal prenatal stress

205

(frequently stressed vs. not/infrequently stressed), birth year nitrogen dioxide (NO2) exposure

206

and socio-economic status (SES) indicated by both maternal education (≤year 12 vs. post-

207

secondary) and the Socio-economic Index (IRSD) deciles within Victoria (Australian Bureau

208

of Statistics, 2011a). The IRSD measures the relative socio-economic disadvantage of people

209

and households within an area. A low score indicates greater disadvantage or lower SES.

210

SHS exposure status was determined by whether there was a regular smoker in the child’s

211

house at baseline.

212

1.5 Statistical analysis

213

Intrauterine and infant exposure analysis were conducted separately. Negative binomial

214

regression models were used to assess the associations between 10 or 100 µg m-3 increases in

215

average and peak PM2.5 exposure, respectively, prenatally or postnatally, and GP attendances,

216

and dispensations of prescribed asthma inhalers and antibiotics, with and without adjustment

217

for covariates. The association between mine fire PM2.5 exposure and dispensations of steroid

218

skin creams was assessed using logistic regression models by defining the outcome as a

219

binary variable due to the low frequency (0.2 per child per year during the first year of life

220

and the year following the fire) in the participants. Maternal prenatal stress was excluded

221

from these models, as the models failed to converge because of complete or quasi-complete

222

separation (i.e. low or no maternal prenatal stress perfectly predicted the outcomes). Possible

223

effect modification by sex was evaluated by adding an interaction term in the multivariable

224

models. Multiple imputation by chained equations was employed to estimate missing

225

covariates values (n=4 for both intrauterine and infant exposure analysis) by generating 20

226

independent datasets (Azur et al., 2011). Imputation models included exposure, all covariates,

227

maternal stress during the fire and outcome variables. All statistical analyses were performed

228

in R 3.5.3 (the R Foundation) (R Core Team, 2019) via RStudio, and a p value <0.05 was

229

considered statistically significant.

230

231

Results

232

Participant characteristics

233

Parents/carers of 311 (54.5%) children from the full Latrobe ELF cohort (n=571) consented

234

to be linked to the MBS/PBS datasets. There were 88 children in the intrauterine exposure

235

group, 77 in the no exposure group, 121 in the infant exposure group, and 25 children born

236

during the fire period. Therefore, 218 children were included in the intrauterine exposure

237

analysis, while 198 were included in the infant exposure analysis.

238

In the intrauterine exposure analysis, no statistically significant differences were observed

239

between exposed and non-exposed children for ambient NO2 exposure, or across sex, tobacco

240

smoke exposure, SES and maternal prenatal stress (Table 1; p>0.050 for all comparisons). In

241

contrast, children in the infant exposure group were, on average, older by approximately 4.6

242

months (Table 2; p<0.050) than those in the no exposure group. The other covariates were

243

approximately equally distributed across different groups (Table 2; p>0.050 for all

244

comparisons). Exposure to mine fire PM2.5 was higher in the infant exposure group than in

245

the intrauterine exposure group (Table 1-2).

246

Overall, a higher proportion of well-educated (i.e. post-secondary) (67.8%) and non-smoking

247

(87.1%) primary carers of the children were included in this study compared with the full

248

ELF cohort (Table S5).

249

250

GP visits and medication use by exposure groups

251

The frequencies of GP attendances, and dispensations of prescribed asthma inhalers, steroid

252

skin creams and antibiotics were generally low among all participants (Table 3). No

253

significant differences were observed between exposed and non-exposed children in the

254

intrauterine exposure analysis (Table 3; p>0.050 for all comparisons). In the infant exposure

255

analysis, the average rate of antibiotic prescribing was approximately double in the group

256

exposed compared with those not exposed (1.5 vs. 0.8, p<0.050), but there was a lower

257

frequency of prescribed steroid cream dispensations (0.1 vs. 0.4, p<0.050) in the exposed

258

children during the one year follow up period (Table 3).

259

260

Associations between mine fire smoke exposure and health outcomes

261

For intrauterine exposure analysis, univariable and multivariable regression analyses did not

262

show any significant associations between intrauterine mine fire PM2.5 exposure and any of

263

the outcomes (Table 4-5).

264

For infant exposure analysis, univariable analyses suggested that mine fire PM2.5 exposure

265

(continuous variable) was associated with increased antibiotic dispensations during the

266

follow-up year (Table 6). After adjusting for potential confounders, every 10 µg m-3 increase

267

in average PM2.5 exposure during infancy were associated with increased incidence of

268

antibiotics being dispensed during the year following the fire: adjusted incidence rate ratio

269

(IRR) 1.24 (95%CI, 1.02, 1.50; p=0.036). Every 100 µg m-3 increase in peak PM2.5 during

270

infancy was also associated with an increase in antibiotic dispensations (IRR 1.14, 95%CI

271

1.00, 1.31; p=0.048). Similar associations were not found for other outcomes (Table 7).

272

There was no evidence of effect modification by sex in either the intrauterine or infant

273

exposure analyses (Table S6; interaction p>0.050 for all analyses).

274

275

Discussion

276

To our knowledge, this study provides the first evidence that infant exposure to increased

277

PM2.5 derived from coal mine fire emissions over a medium duration was associated with

278

increased dispensations of antibiotics during the year following the fire. The association was

279

independent of potential confounders including age of the child, tobacco smoke exposure,

280

socio-economic status and background air pollution exposure. In contrast, we did not observe

281

significant associations for other outcomes among infants (frequency of GP attendances or

282

the usage of medications for asthma or allergic skin conditions), nor did we observe effects of

283

in utero exposure to fire smoke with any of the outcomes during the first year of life.

284

Our finding of an increase in antibiotic use after mine fire smoke exposure during infancy is

285

similar to an American study evaluating the associations between short-term increases in

286

ambient PM2.5 concentrations and respiratory infections in young children aged 0-2 years

287

(Horne et al., 2018). The authors of the American study suggested that every 10 µg m-3

288

increase in PM2.5 concentration was associated with a 15% (95% CI, 12%, 19%) greater odds

289

of healthcare encounters for acute lower respiratory infections one month following the

290

exposure. We are not aware of previous epidemiological studies investigating long-term

291

immune effects of perinatal exposure to fire smoke. However, animal and cell line studies

292

have shown that perinatal exposure to particles from landscape fire emissions induces

293

oxidative stress and inflammation, resulting in immune dysregulation and increased

294

susceptibly to respiratory infections (Black et al., 2017a, Roscioli et al., 2017, Wegesser et al.,

295

2010).

296

Our study did not observe any significant associations between in utero fire smoke exposure

297

and antibiotic usage during the first year of life. This is inconsistent with current evidence

298

regarding intrauterine air pollution exposure and childhood respiratory infections. For

299

example, a Polish study suggested a dose-response relationship between in utero PM2.5

300

exposure levels and the incidence of recurrent respiratory infections (≥5 episodes of

301

bronchitis and/or pneumonia) from birth to age 7 (OR 2.44; 95%CI, 1.12, 5.36)

302

(Jedrychowski et al., 2013). Another study suggested that intrauterine exposure to traffic-

303

related air pollution, estimated by proximity to a major roadway and traffic density, was

304

associated with increased risk of childhood respiratory infection (Rice et al., 2015). A

305

Chinese study suggested that intrauterine exposure to industrial air pollution was associated

306

with increased risk of otitis media in children aged 3-4 years (Deng et al., 2017). This

307

inconsistency might be due to the relatively short duration of exposure in our study compared

308

with the Polish study, the different ages of the children at the time of follow-up, and the

309

different chemical composition and toxicological properties of PM2.5 from the fire emissions

310

and urban sources (Verma et al., 2009). It is also worth noting that although a proportion of

311

our study participants were exposed to very high concentrations of mine fire PM2.5, the

312

average PM2.5 concentrations in our study were much lower than the cut-off points (2.8 vs.

313

26.6 and 45.9 µg m-3) used in the Polish study.

314

Our study did not observe any significant associations between either intrauterine or infant

315

fire smoke exposures and asthma inhaler dispensations by age 3. There is very limited

316

evidence regarding the long-term risk of childhood wheezing or asthma after perinatal

317

exposure to severe, medium duration air pollution events. The only comparable study

318

investigated the association between early life exposure to the Great Smog of 1952 in London

319

and childhood asthma assessed by self-reported diagnosis from birth to age 15. That study

320

suggested that children exposed to the Great Smog during infancy had increased risk of

321

childhood asthma by 19.87 percent (95%CI, 3.37, 36.38) compared with those conceived

322

before or after the event and those living beyond the affected area. In utero smog exposure

323

was not associated with asthma development (Bharadwaj et al., 2016). The inconsistent

324

results for infant exposure between the Great Smog study and ours could be due to the

325

different data collection methods, the challenges of asthma diagnoses in preschool children

326

(Cave and Atkinson, 2014) and our participants who were exposed during infancy had a

327

mean age of 2.0 years during the year of followed up. A harvesting effect might also exist

328

due to increased deaths from the Great Smog.

329

Using asthma medication prescription as an indicator of asthma diagnosis might

330

underestimate asthma incidence as many asthma inhalers can be purchased without a

331

prescription. Further, while prescription data can be a good proxy for the diagnosis of many

332

diseases such as asthma (Furu et al., 2007), we were not able to directly evaluate diagnoses

333

among the cohort. It will be important to continue to monitor these outcomes in our

334

participants to further explore any potential associations.

335

The observed increase in the dispensing of antibiotics might represent an increase in

336

infections commonly managed with antibiotics, or it could reflect a lower threshold for

337

prescribing antibiotic by doctors in the year following the fire, or an increase in parental

338

concern associated with a greater number of requests for antibiotics. However, the unchanged

339

rate of doctor attendances in the year following the fire, and the lack of association with

340

antibiotic prescribing in the first year of life in children who were exposed in utero, both

341

suggest that doctor or parental health seeking behaviour did not appreciably change and these

342

factors are unlikely to explain the association we observed.

343

A strength of the study is that we estimated individual PM2.5 exposure adjusting for

344

residential histories and activity patterns for each participant during the fire period, and we

345

were able to adjust for exposure to background air pollution using modelled estimates of

346

annual non-fire related NO2. This could reduce exposure measurement error. A previous

347

study reported that ignoring residential mobility when estimating traffic-related air pollution

348

exposure caused a modest bias of the associations towards the null (Pennington et al., 2017).

349

In addition, we used multiple imputations to minimise the bias from missing data, and loss of

350

power associated with reductions in sample size (Sterne et al., 2009).

351

However, we acknowledge some limitations in this study. First, our sample of 286

352

participants was relatively small and this limited the power of our analyses to detect

353

significant associations, especially those of small magnitude. A small sample may also affect

354

the generalisability of our study because it was not completely representative of the wider

355

population. Relative to the local population, a higher proportion of children with well-

356

educated and non-smoking parents were recruited (Melody et al., 2018) and included in our

357

study. Our results could be an underestimate of the impacts which might be expected in a

358

population with a higher prevalence of smoking and social disadvantage. Second, exposure

359

misclassification and recall error may have occurred due to the subjective measurement of

360

participants’ locations during the fire period for which we relied on parental reports. However,

361

most respondents reported that they were confident of their recall and eyewitness studies

362

have suggested a strong correlation between measures of confidence and accuracy of recall

363

(Wixted et al., 2016). In addition, the exposure estimate modelling we used could not capture

364

the impact of home air conditioning systems on personal exposures. Personalized monitoring

365

devices can be more accurate but not feasible to deploy during a public health emergency

366

such as this coal mine fire. Third, the use of medication dispensation data from PBS datasets

367

as indicators of childhood illnesses may introduce measurement error. The MBS/PBS

368

datasets only recorded the histories of medical service usage and medication dispensations

369

that were covered by the Australian government, so asthma inhalers purchased without a

370

medical prescription are not included in this analysis (Australian Government Department of

371

Health, 2018a, Australian Government Department of Health, 2018b). Furthermore, seasonal

372

variations in circulating pathogens may impact on antibiotic prescription. However, the effect

373

size (i.e. around 24% increase) was large enough to suggest a possible association between

374

infant coal mine fire smoke exposure and increased childhood infections. In addition, our

375

results might be influenced by residual confounding. However, we adjusted for the most

376

important factors including tobacco smoke exposure, SES and background air pollution

377

exposure.

378

379

Conclusions

380

In conclusion, our study suggested that infant exposure to a short-term severe air pollution

381

event was associated with increased childhood antibiotic dispensations, which might reflect

382

increased childhood infections. Future follow-up of the participants will be necessary to

383

confirm these findings and evaluate long-term effects.

384

385

Acknowledgements

386

The Latrobe Early Life Follow-up (ELF) Study constitutes the child health and development

387

stream of the Hazelwood Health Study. The Latrobe ELF Study forms part of the wider

388

research program of the Hazelwood Health Study (HHS) and is run by a multidisciplinary

389

group of researchers and administrative staff from the University of Tasmania, Monash

390

University, the University of Melbourne, the University of Sydney and CSIRO. We would

391

like to acknowledge all of these staff for their important contributions. Most of all, the study

392

team would like to acknowledge the contribution of all families and community members

393

who have participated in the study to date.

394

This work was supported by the Victorian Department of Health and Human Services

395

(Australia). The paper presents the views of the authors and does not represent the views of

396

the Department. Dr Wheeler’s fellowship was funded by the Centre for Air pollution, energy

397

and health Research.

398

Conflict of interest

399

The authors declare no conflict of interest with this study. Fay Johnston received payment for

400

expert testimony from Environment Protection Authority Victoria (Australia). Amanda

401

Wheeler’s fellowship was funded by the Centre for Air pollution, energy and health Research.

402

403

404

405

406

407

408

409

410

411

Appendix

412

Supplemental material

413

Table S1. Medicare Benefit Schedule items for general practitioner attendances

414

Table S2. Pharmaceutical Benefits Scheme items for the dispensations of prescribed asthma

415

medications

416

Table S3. Pharmaceutical Benefits Scheme items for the dispensations of steroid skin creams

417

Table S4. Pharmaceutical Benefits Scheme items for the dispensations of antibiotics

418

Table S5. Comparisons between participants in the study and the full cohort

419

Table S6. Effect modification by sex in intrauterine and infant exposure analysis

420

421

422

423

424

425

426

427

428

429

430

References

431

ADETONA, O., DIAZ-SANCHEZ, D., CARTER, J. D., COMMODORE, A. A., RATHBUN,

432

S. L. & NAEHER, L. P. 2013. Inflammatory Effects of Woodsmoke Exposure

433

Among Wildland Firefighters Working at Prescribed Burns at the Savannah River

434

Site, SC AU - Hejl, Anna M. Journal of Occupational and Environmental Hygiene,

435

10, 173-180.

436

AUSTRALIAN BUREAU OF STATISTICS. 2011a. 2033.0.55.001 - Census of Population

437

and Housing: Socio-Economic Indexes for Areas (SEIFA), Australia [Online].

438

Available:

439

http://www.abs.gov.au/AUSSTATS/[email protected]/DetailsPage/2033.0.55.0012011?Open

440

Document [Accessed Feb 8 2019].

441

AUSTRALIAN BUREAU OF STATISTICS. 2011b. Australian Statistical Geography

442

Standard (ASGS): Volume 1 - Main Structure and Greater Capital City Statistical

443

Areas [Online]. Available:

444

http://www.ausstats.abs.gov.au/ausstats/subscriber.nsf/0/D3DC26F35A8AF579CA25

445

7801000DCD7D/$File/1270055001_july%202011.pdf [Accessed Oct 4 2018].

446

AUSTRALIAN GOVERNMENT DEPARTMENT OF HEALTH. 2018a. Medicare Benefits

447

Schedule Book Operating from 1 November 2018 [Online]. Available:

448

http://www.mbsonline.gov.au/internet/mbsonline/publishing.nsf/Content/1BC94358D

449

4F276D3CA257CCF0000AA73/$File/201811-MBS.pdf [Accessed Nov 15 2018].

450

AUSTRALIAN GOVERNMENT DEPARTMENT OF HEALTH. 2018b. Schedule of

451

Pharmaceutical Benefits Effective 1 November 2018 [Online]. Available:

452

http://www.pbs.gov.au/publication/schedule/2018/11/2018-11-01-general-

453

schedule.pdf [Accessed Nov 15 2018].

454

AZUR, M. J., STUART, E. A., FRANGAKIS, C. & LEAF, P. J. 2011. Multiple Imputation

455

by Chained Equations: What is it and how does it work? International journal of

456

methods in psychiatric research, 20, 40-49.

457

BHARADWAJ, P., ZIVIN, J. G., MULLINS, J. T. & NEIDELL, M. 2016. Early-Life

458

Exposure to the Great Smog of 1952 and the Development of Asthma. Am J Respir

459

Crit Care Med, 194, 1475-1482.

460

BLACK, C., GERRIETS, J. E., FONTAINE, J. H., HARPER, R. W., KENYON, N. J.,

461

TABLIN, F., SCHELEGLE, E. S. & MILLER, L. A. 2017a. Early Life Wildfire

462

Smoke Exposure Is Associated with Immune Dysregulation and Lung Function

463

Decrements in Adolescence. Am J Respir Cell Mol Biol, 56, 657-666.

464

BLACK, C., TESFAIGZI, Y., BASSEIN, J. A. & MILLER, L. A. 2017b. Wildfire smoke

465

exposure and human health: Significant gaps in research for a growing public health

466

issue. Environmental Toxicology and Pharmacology, 55, 186-195.

467

BOWATTE, G., LODGE, C., LOWE, A. J., ERBAS, B., PERRET, J., ABRAMSON, M. J.,

468

MATHESON, M. & DHARMAGE, S. C. 2015. The influence of childhood traffic-

469

related air pollution exposure on asthma, allergy and sensitization: a systematic

470

review and a meta-analysis of birth cohort studies. Allergy, 70, 245-56.

471

CARLSEN, H. K., BOMAN, P., BJOR, B., OLIN, A. C. & FORSBERG, B. 2016. Coarse

472

Fraction Particle Matter and Exhaled Nitric Oxide in Non-Asthmatic Children. Int J

473

Environ Res Public Health, 13.

474 475

CAVE, A. J. & ATKINSON, L. L. 2014. Asthma in preschool children: a review of the diagnostic challenges. J Am Board Fam Med, 27, 538-48.

476

DENG, Q., DENG, L., LU, C., LI, Y. & NORBACK, D. 2018. Parental stress and air

477

pollution increase childhood asthma in China. Environ Res, 165, 23-31.

478

DENG, Q., LU, C., DAN, N., BORNEHAG, C. G., ZHANG, Y., LIU, W., YUAN, H. &

479

SUNDELL, J. 2015. Early life exposure to ambient air pollution and childhood

480

asthma in China. Environmental Research, 143, 83-92.

481

DENG, Q., LU, C., LI, Y., CHEN, L. & DAN, N. 2017. Association between prenatal

482

exposure to industrial air pollution and onset of early childhood ear infection in China.

483

Atmospheric Environment, 157, 18-26.

484

DEPARTMENT OF THE ENVIRONMENT AND HERITAGE. National standards for

485

criteria air pollutants 1 in Australia 2005 [Online]. Available:

486

http://www.environment.gov.au/protection/publications/factsheet-national-standards-

487

criteria-air-pollutants-australia [Accessed Oct 30 2018].

488

DIETERT, R. R., ETZEL, R. A., CHEN, D., HALONEN, M., HOLLADAY, S. D.,

489

JARABEK, A. M., LANDRETH, K., PEDEN, D. B., PINKERTON, K.,

490

SMIALOWICZ, R. J. & ZOETIS, T. 2000. Workshop to identify critical windows of

491

exposure for children's health: immune and respiratory systems work group summary.

492

Environmental health perspectives, 108 Suppl 3, 483-490.

493

EMMERSON, K. M., REISEN, F., LUHAR, A., WILLIAMSON, G. & COPE, M. E. 2016.

494

Air quality modelling of smoke exposure from the Hazelwood mine fire. CSIRO

495

Australia. [Online]. Available: http://hazelwoodhealthstudy.org.au/wp-

496

content/uploads/2017/01/Hazelwood_AirQualityModelling_December2016_Final.pdf

497

[Accessed Sep 17 2018].

498

ENVIRONMENT PROTECTION AUTHORITY VICTORIA. 2019. Air monitoring results

499

around Victoria [Online]. Available: https://www.epa.vic.gov.au/our-

500

work/monitoring-the-environment/monitoring-victorias-air/monitoring-results

501

[Accessed Aug 21 2019].

502

FELESZKO, W., RUSZCZYŃSKI, M., JAWORSKA, J., STRZELAK, A., ZALEWSKI, B.

503

M. & KULUS, M. 2014. Environmental tobacco smoke exposure and risk of allergic

504

sensitisation in children: a systematic review and meta-analysis. Archives of Disease

505

in Childhood, 99, 985-992.

506

FURU, K., SKURTVEIT, S., LANGHAMMER, A. & NAFSTAD, P. 2007. Use of anti-

507

asthmatic medications as a proxy for prevalence of asthma in children and adolescents

508

in Norway:a nationwide prescription database analysis. European Journal of Clinical

509

Pharmacology, 63, 693-698.

510

GOLENKO, X. A., SHIBL, R., SCUFFHAM, P. A. & CAMERON, C. M. 2015.

511

Relationship between socioeconomic status and general practitioner visits for children

512

in the first 12 months of life: an Australian study. Aust Health Rev, 39, 136-145.

513

HEHUA, Z., QING, C., SHANYAN, G., QIJUN, W. & YUHONG, Z. 2017. The impact of

514

prenatal exposure to air pollution on childhood wheezing and asthma: A systematic

515

review. Environ Res, 159, 519-530.

516

HORNE, B. D., JOY, E. A., HOFMANN, M. G., GESTELAND, P. H., CANNON, J. B.,

517

LEFLER, J. S., BLAGEV, D. P., KORGENSKI, E. K., TOROSYAN, N., HANSEN,

518

G. I., KARTCHNER, D. & POPE III, C. A. 2018. Short-term Elevation of Fine

519

Particulate Matter Air Pollution and Acute Lower Respiratory Infection. Am J Respir

520

Crit Care Med.

521

JEDRYCHOWSKI, W., PERERA, F., MAUGERI, U., MROZEK-BUDZYN, D., MILLER,

522

R. L., FLAK, E., MROZ, E., JACEK, R. & SPENGLER, J. D. 2011. Effects of

523

prenatal and perinatal exposure to fine air pollutants and maternal fish consumption

524

on the occurrence of infantile eczema. Int Arch Allergy Immunol, 155, 275-81.

525 526

JEDRYCHOWSKI, W. A., PERERA, F. P., SPENGLER, J. D., MROZ, E., STIGTER, L., FLAK, E., MAJEWSKA, R., KLIMASZEWSKA-REMBIASZ, M. & JACEK, R.

527

2013. Intrauterine exposure to fine particulate matter as a risk factor for increased

528

susceptibility to acute broncho-pulmonary infections in early childhood. International

529

journal of hygiene and environmental health, 216, 395-401.

530

JOHNSTON, F. H., HENDERSON, S. B., CHEN, Y., RANDERSON, J. T., MARLIER, M.,

531

DEFRIES, R. S., KINNEY, P., BOWMAN, D. M. & BRAUER, M. 2012. Estimated

532

global mortality attributable to smoke from landscape fires. Environ Health Perspect,

533

120, 695-701.

534

KENNEDY, C. M., PENNINGTON, A. F., DARROW, L. A., KLEIN, M., ZHAI, X.,

535

BATES, J. T., RUSSELL, A. G., HANSEN, C., TOLBERT, P. E. & STRICKLAND,

536

M. J. 2018. Associations of mobile source air pollution during the first year of life

537

with childhood pneumonia, bronchiolitis, and otitis media. Environmental

538

Epidemiology, 2, e007.

539

KNIBBS, L. D., COOREY, C. P., BECHLE, M. J., COWIE, C. T., DIRGAWATI, M.,

540

HEYWORTH, J. S., MARKS, G. B., MARSHALL, J. D., MORAWSKA, L.,

541

PEREIRA, G. & HEWSON, M. G. 2016. Independent Validation of National

542

Satellite-Based Land-Use Regression Models for Nitrogen Dioxide Using Passive

543

Samplers. Environmental Science & Technology, 50, 12331-12338.

544

KNIBBS, L. D., HEWSON, M. G., BECHLE, M. J., MARSHALL, J. D. & BARNETT, A. G.

545

2014. A national satellite-based land-use regression model for air pollution exposure

546

assessment in Australia. Environmental Research, 135, 204-211.

547

LEE, A., LEON HSU, H. H., MATHILDA CHIU, Y. H., BOSE, S., ROSA, M. J., KLOOG,

548

I., WILSON, A., SCHWARTZ, J., COHEN, S., COULL, B. A., WRIGHT, R. O. &

549

WRIGHT, R. J. 2018. Prenatal fine particulate exposure and early childhood asthma:

550

Effect of maternal stress and fetal sex. J Allergy Clin Immunol, 141, 1880-1886.

551

LIU, J. C., MICKLEY, L. J., SULPRIZIO, M. P., DOMINICI, F., YUE, X., EBISU, K.,

552

ANDERSON, G. B., KHAN, R. F. A., BRAVO, M. A. & BELL, M. L. 2016.

553

Particulate air pollution from wildfires in the Western US under climate change.

554

Climatic Change, 138, 655-666.

555

MELODY, S., DALTON, M., DENNEKAMP, M., WHEELER, A., DHARMAGE, S.,

556

WILLS, K., REEVES, M., FORD, J., O’SULLIVAN, T., WILLIAMSON, G., VENN,

557

A., ROBERTS, C. & JOHNSTON, F. 2018. The Latrobe Early Life Follow-up (ELF)

558

Cohort Study Volume 1 Description of the cohort and preliminary assessment of

559

possible associations between mine fire emissions and parent-reported perinatal

560

outcomes [Online]. Available: http://hazelwoodhealthstudy.org.au/wp-

561

content/uploads/2018/08/ELF-Vol-1_-CohortDesciption_ParentReportedOutcomes-

562

v1.2.pdf [Accessed Sep 17 2018].

563 564

MELODY, S. M. & JOHNSTON, F. H. 2015. Coal mine fires and human health: What do we know? International Journal of Coal Geology, 152, 1-14.

565

PENNINGTON, A. F., STRICKLAND, M. J., KLEIN, M., ZHAI, X., RUSSELL, A. G.,

566

HANSEN, C. & DARROW, L. A. 2017. Measurement error in mobile source air

567

pollution exposure estimates due to residential mobility during pregnancy. J Expo Sci

568

Environ Epidemiol, 27, 513-520.

569

R CORE TEAM. 2019. R: A language and environment for statistical computing. R

570

Foundation for Statistical Computing, Vienna, Austria. [Online]. R Foundation for

571

Statistical Computing, Vienna, Austria. Available: https://www.R-project.org/

572

[Accessed Apr 10 2019].

573

REID, C. E., BRAUER, M., JOHNSTON, F. H., JERRETT, M., BALMES, J. R. &

574

ELLIOTT, C. T. 2016. Critical Review of Health Impacts of Wildfire Smoke

575

Exposure. Environmental health perspectives, 124, 1334-1343.

576 577 578

REISEN, F., GILLETT, R., CHOI, J., FISHER, G. & TORRE, P. 2017. Characteristics of an open-cut coal mine fire pollution event. Atmospheric Environment, 151, 140-151. RICE, M. B., RIFAS-SHIMAN, S. L., OKEN, E., GILLMAN, M. W., LJUNGMAN, P. L.,

579

LITONJUA, A. A., SCHWARTZ, J., COULL, B. A., ZANOBETTI, A.,

580

KOUTRAKIS, P., MELLY, S. J., MITTLEMAN, M. A. & GOLD, D. R. 2015.

581

Exposure to traffic and early life respiratory infection: A cohort study. Pediatr

582

Pulmonol, 50, 252-259.

583

ROSCIOLI, E., HAMON, R., JERSMANN, H., REYNOLDS, P. & HODGE, S. 2017.

584

Airway cells exposed to wildfire smoke extract exhibit dysregulation of autophagy

585

and changes characteristic of inflammatory airway disease. European Respiratory

586

Journal, 50.

587

SHAO, J., ZOSKY, G. R., HALL, G. L., WHEELER, A. J., DHARMAGE, S., MELODY, S.,

588

DALTON, M., FOONG, R. E., O'SULLIVAN, T., WILLIAMSON, G. J.,

589

CHAPPELL, K., ABRAMSON, M. J. & JOHNSTON, F. H. 2019. Early life exposure

590

to coal mine fire smoke emissions and altered lung function in young children.

591

Respirology [Online]. Available:

592

https://onlinelibrary.wiley.com/doi/abs/10.1111/resp.13617.

593

SOH, S. E., GOH, A., TEOH, O. H., GODFREY, K. M., GLUCKMAN, P. D., SHEK, L. P.

594

& CHONG, Y. S. 2018. Pregnancy Trimester-Specific Exposure to Ambient Air

595

Pollution and Child Respiratory Health Outcomes in the First 2 Years of Life: Effect

596

Modification by Maternal Pre-Pregnancy BMI. Int J Environ Res Public Health, 15.

597

STERNE, J. A., WHITE, I. R., CARLIN, J. B., SPRATT, M., ROYSTON, P., KENWARD,

598

M. G., WOOD, A. M. & CARPENTER, J. R. 2009. Multiple imputation for missing

599

data in epidemiological and clinical research: potential and pitfalls. Bmj, 338, b2393.

600 601 602

TEXTOR, J., HARDT, J. & KNÜPPEL, S. 2011. DAGitty: A Graphical Tool for Analyzing Causal Diagrams. Epidemiology, 22, 745. UPHOFF, E., CABIESES, B., PINART, M., VALDÉS, M., ANTÓ, J. M. & WRIGHT, J.

603

2014. A systematic review of socioeconomic position in relation to asthma and

604

allergic diseases. European Respiratory Journal.

605

VANKER, A., GIE, R. P. & ZAR, H. J. 2017. The association between environmental

606

tobacco smoke exposure and childhood respiratory disease: a review. Expert Review

607

of Respiratory Medicine, 11, 661-673.

608

VERMA, V., POLIDORI, A., SCHAUER, J. J., SHAFER, M. M., CASSEE, F. R. &

609

SIOUTAS, C. 2009. Physicochemical and toxicological profiles of particulate matter

610

in Los Angeles during the October 2007 southern California wildfires. Environ Sci

611

Technol, 43, 954-60.

612

WEGESSER, T. C., FRANZI, L. M., MITLOEHNER, F. M., EIGUREN-FERNANDEZ, A.

613

& LAST, J. A. 2010. Lung antioxidant and cytokine responses to coarse and fine

614

particulate matter from the great California wildfires of 2008. Inhal Toxicol, 22, 561-

615

70.

616

WILLIAMSON, E. J., AITKEN, Z., LAWRIE, J., DHARMAGE, S. C., BURGESS, J. A. &

617

FORBES, A. B. 2014. Introduction to causal diagrams for confounder selection.

618

Respirology, 19, 303-11.

619

WIXTED, J. T., MICKES, L., DUNN, J. C., CLARK, S. E. & WELLS, W. 2016. Estimating

620

the reliability of eyewitness identifications from police lineups. Proc Natl Acad Sci U

621

S A, 113, 304-9.

622

WOPEREIS, H., OOZEER, R., KNIPPING, K., BELZER, C. & KNOL, J. 2014. The first

623

thousand days - intestinal microbiology of early life: establishing a symbiosis. Pediatr

624

Allergy Immunol, 25, 428-38.

625

YI, S. J., SHON, C., MIN, K. D., KIM, H. C., LEEM, J. H., KWON, H. J., HONG, S., KIM,

626

K. & KIM, S. Y. 2017. Association between Exposure to Traffic-Related Air

627

Pollution and Prevalence of Allergic Diseases in Children, Seoul, Korea. Biomed Res

628

Int, 2017, 4216107.

629

630

631

632

633

634

635

636

637

638

639

640

641

642

643

644

645

Table 1. Comparison of participant characteristics between groups in the intrauterine

646

exposure analysis Characteristics

Intrauterine exposure

No exposure group

group

(n=130)

(n=88) Median [IQR] IRSD deciles within Victoria

P

a

3 [1, 5]

3 [1, 7]

0.842

Birth year NO2 exposure (ppb)

3.9 [3.4, 4.4]

4.0 [3.3, 4.6]

0.884

Average mine fire PM2.5 (µg m-3)

2.8 [1.6, 7.8]

0.0 [0.0, 0.0]

-

76.7 [49.7, 162.3]

0 [0.0, 0.0]

-

Peak mine fire PM2.5 (µg m-3)

n (%)

P

b

Sex: male

39 (44.3%)

64 (49.2%)

0.476

Maternal smoking during pregnancy: yes

9 (10.2%)

23 (17.7%)

0.127

16 (18.2%)

27 (20.9%)

0.618

16 (18.4%)

20 (15.5%)

0.577

28 (31.8%)

47 (36.4%)

0.483

c

Secondhand smoke exposure : yes d

Maternal prenatal stress : frequently stressed c

Maternal education : ≤year 12

647

Note: IQR, interquartile range; IRSD, Index of Relative Socio-economic Disadvantage; NO2, nitrogen dioxide.

648

a

649

b

650

c

651

d

652 653 654 655 656

Mann-Whitney U test. Pearson’s chi-square test.

Having missing values (n=1). Having missing values (n=2).

657

Table 2. Comparison of participant characteristics between groups in the infant exposure

658

analysis Characteristics

Infant exposure group

No exposure group

(n=121)

(n=77) Median [IQR]

Age at the start of outcome year (months)

P

a

11.6 [7.1, 17.0]

7.0 [3.4, 9.3]

0.000

2 [1, 5]

3 [1, 8]

0.071

Birth year NO2 exposure (ppb)

4.3 [3.6, 4.9]

3.8 [3.2, 4.3]

0.001

Average mine fire PM2.5 (µg m-3)

6.8 [2.0, 13.6]

0.0 [0.0, 0.0]

-

106.5 [53.1, 195.8]

0 [0.0, 0.0]

-

IRSD deciles within Victoria

Peak mine fire PM2.5 (µg m-3)

n (%)

P

b

Sex: male

63 (52.1%)

36 (46.8%)

0.466

Maternal smoking during pregnancy: yes

21 (17.4%)

11 (14.3%)

0.567

29 (24.2%)

14 (18.4%)

0.344

18 (14.9%)

9 (11.8%)

0.547

43 (35.8%)

28 (36.4%)

0.940

c

Secondhand smoke exposure : yes d

Maternal prenatal stress : frequently stressed d

Maternal education : ≤year 12

659

Note: IQR, interquartile range; IRSD, Index of Relative Socio-economic Disadvantage; NO2, nitrogen dioxide.

660

a

661

b

662

c

663

d

664

665

666

667

Mann-Whitney U test. Pearson’s chi-square test.

Having missing values (n=2). Having missing values (n=1).

668

Table 3. Frequency of health services and medication usage in exposed and non-exposed

669

children Intrauterine exposure group

No exposure group

(during the first year of life)

(during the first year of life)

GP attendances

8.0

7.8

0.944

Prescribed asthma inhalers

0.4

0.3

0.598

Steroid skin creams

0.1

0.3

0.187

Antibiotics

0.8

0.7

0.248

Exposure group

No exposure group

(01/04/2014-01/04/2015)

(01/01/2016-31/12/2016)

GP attendances

6.9

7.8

0.179

Prescribed asthma inhalers

0.7

0.4

0.198

Steroid skin creams

0.1

0.4

0.005

Antibiotics

1.5

0.8

0.001

670

Note: GP, general practitioner.

671

a

673

674

675

676

677

678

a

Exposure group

Postnatal exposure group

672

Mean (per child/year)

Mann-Whitney U test.

P

679

Table 4. Univariable analysis of intrauterine mine fire PM2.5 exposure, risk factors and health

680

outcomes during the first year of life

Univariable analysis

GP attendances

(n=218)

Dispensations of

Dispensations of

Dispensations of

prescribed asthma

steroid skin creams

antibiotics

inhalers IRR

P

(95%CI) Average PM2.5 (per

1.00

10 µg m-3 increase)

(0.85, 1.18)

Peak PM2.5 (per 100

1.00

µg m-3 increase) Maternal education: ≤year 12 Maternal tobacco smoking status during

0.997

0.85

0.934

0.99

0.635

0.99

0.953

(0.80, 1.31)

1.20

1.10

0.98

0.982

0.94

0.804

(0.47, 3.07)

0.70

P

1.20

0.361

(0.81, 1.78) 0.925

1.07

0.498

(0.88,1.29) 0.894

(0.40, 2.22) 0.700

IRR (95%CI)

(0.67, 1.43)

(0.48, 2.03) 0.848

P

(0.53, 2.27)

(0.72, 1.36) 0.198

OR (95%CI)

(0.42, 1.69)

(0.94, 1.35) 1.02

P

(95%CI)

(0.92, 1.08) 1.13

IRR

1.68

0.023

(1.08, 2.62) 0.578

(0.20, 2.47)

1.78

0.047

(1.01, 3.12)

pregnancy: yes Second hand smoke exposure: yes

0.89

0.320

(0.72, 1.11)

Maternal prenatal

1.13

stress: frequently

(0.90, 1.42)

0.75

0.537

(0.31, 1.84) 0.292

1.73

0.91

0.857

(0.32, 2.56) 0.209

(0.74, 4.04)

0.37

1.28

0.364

(0.75, 2.19) 0.193

(0.08, 1.64)

1.52

0.142

(0.87, 2.64)

stressed IRSD

0.97

0.075

(0.95, 1.00) Background NO2 exposure

1.02 (0.97, 1.07)

0.99

0.910

(0.88, 1.12) 0.504

0.96 (0.76, 1.20)

1.12

0.091

(0.98, 1.27) 0.691

0.68 (0.42, 1.08)

0.91 (0.84, 0.98)

0.105

1.16 (1.04, 1.30)

681

Note: GP, general practitioner; IRR, incidence rate ratio; CI, confidence interval; OR, odds ratio; PM2.5,

682

particulate matter with an aerodynamic diameter less than 2.5 micrometers; IRSD, Index of Relative Socio-

683

economic Disadvantage; NO2, nitrogen dioxide.

684

0.016

0.009

685

Table 5. Mine fire smoke exposure during pregnancy and health outcomes during the first

686

year of life

Multivariable

GP attendances

analysis (n=218)

Dispensations of

Dispensations of

Dispensations of

prescribed asthma

steroid skin creams

antibiotics

inhalers IRRa

P

(95%CI) Average PM2.5 (per

1.00

10 µg m-3 increase)

(0.85, 1.18)

Peak PM2.5 (per 100

1.00

µg m-3 increase)

(0.93, 1.08)

IRRa

P

(95%CI) 0.987

0.87

1.01 (0.74, 1.37)

P

0.691

1.26

0.565

(0.57, 2.77) 0.969

1.00 (0.68, 1.46)

IRRa

1.16

0.991

1.08 (0.90,1.31)

Note: GP, general practitioner; IRR, incidence rate ratio; CI, confidence interval; OR, odds ratio; PM2.5,

688

particulate matter with an aerodynamic diameter less than 2.5 micrometers.

689

a

690

during pregnancy, secondhand smoke exposure, maternal prenatal stress and background nitrogen dioxide

691

exposure.

693

694

695

696

697

698

699

0.433

(0.80, 1.68)

687

692

P

(95%CI)

(95%CI)

(0.45, 1.71) 0.953

a

OR

Models adjusted for maternal education, index of relative socio-economic disadvantage, maternal smoking

0.393

700

Table 6. Univariable analysis of infant mine fire PM2.5 exposure, risk factors and health

701

outcomes during the year following the fire

Univariable analysis

GP attendances

(n=198)

Dispensations of

Dispensations of

Dispensations of

prescribed asthma

steroid skin creams

antibiotics

inhalers IRR

P

(95%CI) Average PM2.5 (per

0.99

10 µg m-3 increase)

(0.89, 1.10)

Peak PM2.5 (per 100

0.98

µg m-3 increase)

(0.91, 1.06)

Age (per month)

0.98

≤year 12 Maternal tobacco smoking status during

1.11

0.789

1.19

0.578

1.12

0.196

1.02

0.219

1.36

0.334

(0.68, 1.17)

1.36

0.277

0.67

0.88

0.260

0.38

0.150

(0.69, 2.66)

0.24

1.22

0.024

1.14

0.034

(1.01,1.28) 0.015

1.01

0.556

(0.98, 1.04) 0.100

(0.12, 1.19) 0.377

P

(1.03, 1.44)

(0.80, 0.98)

(0.80, 2.31) 0.421

0.70

IRR (95%CI)

(0.40, 1.15)

(0.98, 1.06) 0.312

P

(0.36, 1.34)

(0.94, 1.34) 0.003

OR (95%CI)

(0.92, 1.54)

(0.91, 1.37) 0.89

P

(95%CI)

(0.96, 0.99) Maternal education:

IRR

1.44

0.043

(1.01, 2.05) 0.167

(0.03, 1.82)

1.02

0.944

(0.63, 1.63)

pregnancy: yes Second hand smoke exposure: yes

0.82

0.121

(0.65 1.05)

Maternal prenatal

0.97

stress: frequently

(0.73, 1.30)

0.81

0.564

(0.41, 1.63) 0.837

1.48

0.34

0.164

(0.08, 1.54) 0.276

-

1.21

0.358

(0.80, 1.83) -

(0.73, 3.01)

0.92

0.753

(0.55, 1.54)

stressed IRSD

0.98

0.172

(0.95, 1.01) Birth year NO2 exposure (per ppb)

1.02 (0.97, 1.08)

0.95

0.215

(0.86, 1.03) 0.397

1.06 (0.92, 1.21)

1.01

0.908

(0.87, 1.17) 0.420

0.99 (0.76, 1.29)

0.98

0.497

(0.92, 1.04) 0.948

1.05 (0.95, 1.15)

702

Note: GP, general practitioner; IRR, incidence rate ratio; CI, confidence interval; OR, odds ratio; PM2.5,

703

particulate matter with an aerodynamic diameter less than 2.5 micrometers; IRSD, Index of Relative Socio-

704

economic Disadvantage; NO2, nitrogen dioxide.

0.339

705

Table 7. Mine fire smoke exposure in infancy and health outcomes during a one-year period

706

after the fire

Multivariable

GP attendances

analysis (n=198)

Dispensations of

Dispensations of

Dispensations of

prescribed asthma

steroid skin creams

antibiotics

inhalers IRRa

P

(95%CI) Average PM2.5 (per

0.96

10 µg m-3 increase)

(0.85, 1.09)

Peak PM2.5 (per 100

0.96

µg m-3 increase)

(0.89, 1.05)

IRRa

P

(95%CI) 0.550

1.16

1.08 (0.88, 1.33)

P

0.339

0.66

0.270

(0.31, 1.38) 0.459

0.65

IRRa

1.24

0.135

(0.37, 1.14)

1.14

708

particulate matter with an aerodynamic diameter less than 2.5 micrometers.

709

a

710

during pregnancy, secondhand smoke exposure, maternal prenatal stress and background nitrogen dioxide

711

exposure.

712

b

713

during pregnancy, secondhand smoke exposure and background nitrogen dioxide exposure.

716

717

718

719

0.048

(1.00,1.31)

Note: GP, general practitioner; IRR, incidence rate ratio; CI, confidence interval; OR, odds ratio; PM2.5,

715

0.036

(1.02, 1.50)

707

714

P

(95%CI)

(95%CI)

(0.86, 1.57) 0.380

b

OR

Models adjusted for age, maternal education, index of relative socio-economic disadvantage, maternal smoking

Models adjusted for age, maternal education, index of relative socio-economic disadvantage, maternal smoking

1

Highlights

2

•

Evidence on long-term health effects following early life air pollution is scarce.

3

•

We assessed the health impact of perinatal fire smoke exposure on 286 children.

4

•

Infant exposure was associated with increased antibiotic dispensation.

5

Declaration of interests ☒ The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. ☐The authors declare the following financial interests/personal relationships which may be considered as potential competing interests:

Fay Johnston received payment for expert testimony from Environment Protection Authority Victoria (Australia). Amanda Wheeler’s fellowship was funded by the Centre for Air pollution, energy and health Research.