Work

Cohort Profile: The Hazelwood Health Study Latrobe Early Life Follow-Up (ELF) Study

Year: 2020

Type: article

Abstract: The Hazelwood coal mine fire in Victoria, Australia, was an unprecedented national outdoor air pollution event that covered the surrounding area in smoke and ash for 6 weeks in February and March 2014 (Figure 1). The severe smoke event caused considerable community concerns within the neighbouring town of Morwell and the broader Latrobe Valley community, situated 150 km east of the capital city of Melbourne. The Latrobe Valley is a regional setting known for mining and agricultural industries, characterised by relative socioeconomic disadvantage compared with the rest of the state. In response to these concerns, and following extensive community consultation, the Hazelwood Health Study was established to examine the long-term impacts of the mine fire (https://hazelwoodhealthstudy.org.au/). The Hazelwood Health Study involves multiple research streams focusing on various health outcomes and vulnerable groups. The Latrobe Early Life Follow-up (ELF) Study is a stream of the Hazelwood Health Study which aims to investigate the potential impacts of exposure to a severe outdoor air pollution event, as caused by the 2014 Hazelwood coal mine fire, on the health and development of children in the Latrobe Valley. Smoke plume resulting from the 2014 Hazelwood coal mine fire over the Latrobe Valley, Victoria (photo credit: Mike Keating) Smoke plume resulting from the 2014 Hazelwood coal mine fire over the Latrobe Valley, Victoria (photo credit: Mike Keating) Specific objectives are to explore: (i) obstetric and perinatal outcomes; (ii) parental reports of minor illnesses in young children; (iii) respiratory, vascular and immune function of children from 3 to 12 years of age; and (iv) medication use, health care attendance, education and developmental outcomes in childhood following in utero and early childhood exposure to smoke from the Hazelwood coal mine fire. We also aim to evaluate whether the persistence of smoke and ash pollutants in homes is a useful additional marker of exposure to mine fire emissions. Ethics statement Ethics approval was obtained from the Tasmania Health and Medical Human Research Ethics Committee (ref H0015033 and H0014875). The Latrobe ELF Study cohort includes children born from 1 March 2012 to 31 December 2015 in the Latrobe Valley, Victoria. This cohort includes those who were either in utero, a child under the age of 2 years, or conceived after the outdoor air pollution event, and residing in the Latrobe Valley region, to allow a gradient of exposures and developmental windows to be examined. Two streams within the ELF cohort exist to answer the study aims. These streams can be broadly classified into the identified and de-identified cohorts (Figure 2): Outline of main subgroups within the cohort and their constituent participant numbers Outline of main subgroups within the cohort and their constituent participant numbers Identified cohort: from a total eligible population of 3371, this cohort consists of 571 children and their families who have consented to participation in the study and completed a comprehensive baseline survey detailing demographic, health, smoke exposure and housing characteristics. This cohort were invited and participated in various additional activities within the study, including: Clinical testing Completion of monthly symptom and illness diaries Extraction of routinely collected health and education data Domestic dust and soil collection De-identified cohort: this consists of 3679 children born between 1 March 2012 and 31 December 2015. The cohort was identified using the Victorian Birth Registry, a compulsory register of all births in the jurisdiction, by selecting live births where the maternal residence at the time of birth was the Latrobe Valley. Data linkage of eight different health and educational datasets was performed jointly by the Australian Institute for Health and Welfare and the Centre for Victorian Data Linkage (Table 1). For this cohort, the eligible study population was the study sample, as data for this cohort were obtained without individual consent; data were de-identified, and linkage ensured that the separation principle was upheld. Birth, health and educational outcomes available through data linkage for the de-identified cohort (N = 3679) ICD-10AM, International Classification of Diseases, Tenth Revision, Australian Modification. Birth, health and educational outcomes available through data linkage for the de-identified cohort (N = 3679) ICD-10AM, International Classification of Diseases, Tenth Revision, Australian Modification. Recruitment was conducted by identifying potential participants through the generation of a nominal roll. The Latrobe City Council’s (LCC) Maternal Child Health Service routinely collects contact details of families with children from birth to 6 years of age, for the purposes of providing maternal and child health care. The council generated a nominal roll of potentially eligible participants (i.e. children born 1 March 2012 to 31 December 2015) known to their service. A total population of 3371 individuals were available to be approached for participation. To achieve the recruitment targets, stratified random sampling across six groups defined by timing of exposure: (in utero exposure, date of birth 1 April 2014-31 December 2014; early childhood exposure aged 0–2 years, date of birth 1 March 2012 to 31 March 2014; or unexposed, born 1 January 2015–31 December 2015) and by residence (Morwell vs rest of Latrobe Valley) was conducted (Figure 3). Recruitment targets were exceeded for all exposure groups (target n = 166 for each group); the in utero group (n = 198), the early childhood group (n = 203) and the unexposed group (n = 170). The identified cohort is not necessarily a complete subset of the de-identified cohort, as the identified cohort may include children who were born elsewhere but resident in the Latrobe Valley at the time of the fire. There is no way of determining the extent of overlap of the identified with the de-identified cohort, due to the de-identified nature of the latter. Timeline showing Hazelwood coal mine fire and participant exposure groups by date of birth Timeline showing Hazelwood coal mine fire and participant exposure groups by date of birth For the identified cohort, repeat clinical testing will occur every 3 years up to age 10. Phase one of clinical testing involved assessment of cardiovascular and respiratory function in 2017. Phase two of clinical testing is planned for late 2020 and involves repeat measures of cardiovascular and respiratory function, as well as skin prick testing to explore immune sensitization. Serial data extraction of health and educational data from various administrative sources (Table 1) will occur approximately every 3 years. Domestic dust and soil collection and symptom diary collection (of minor illnesses) are complete. For the de-identified cohort, serial data linkages (at varying intervals for different datasets) are planned until at least 2024 (possibly longer pending funding acquisition), to coincide with the ageing of the cohort and the availability and likelihood of various health and educational outcomes. The Hazelwood coal mine fire lasted from 9 February to 31 March 2014 (51 days). Average daily concentrations of fine particulate matter with an aerodynamic diameter <2.5 μm (PM2.5) directly attributable to the mine fire was the primary exposure metric of interest. Peak PM2.5 was the secondary exposure metric, which represented the highest 24-h average PM2.5 exposure during the fire. The average and peak daily PM2.5 concentrations for the 51-day period were determined using the Commonwealth Scientific and Industrial Research Organisation (CSIRO) Chemical Transport Model. The model was a high-resolution exposure model that estimated PM2.5 at an hourly time step and a one-by-one kilometre spatial resolution.2 The full model included background PM2.5 from natural sources, vehicular and power station emissions, landscape fires and the mine fire. The difference between the model run with, or without, emissions from the mine fire was used to determine the concentration of PM2.5 specific to the mine fire. Full details and validity of the model are published elsewhere.1 For those in the identified cohort, we obtained detailed time-activity information about the exposed pregnant mother or young child during the fire event on a day-night scale. These time-activity data were applied to the above chemical transport model to assign daily average and peak PM2.5 exposure. For those in the de-identified cohort, daily average and peak PM2.5 exposure was assigned based on residential address at the geographical level of statistical area 1 (SA1), which is one of the smallest geographical classifications defined by the Australian Bureau of Statistics, and contains approximately 400 individuals.3 Outcome data are available for the identified cohort depending on the various study activities to which they have consented or participated. These data span sociodemographic, birth, health, educational and clinical testing data and are outlined in Table 2. Health outcome data available for identified cohort (N = 571) Sociodemographic details of study child and immediate family Child’s health: including birth details, perinatal outcomes, breastfeeding history, medication use, childcare attendance, number of siblings and history of atopy in siblings Child’s residential history: details of the housing in which the pregnant mother and child lived, including year of build, heating/cooling, floor coverings etc Maternal and/or child’s residence during the smoke event: capturing location information on a day-night time scale Medical history of the biological parents Respiratory system resistance (Rrs) Respiratory system reactance (Xrs) Area under the reactance curve (AX) Bronchodilator response in Rrs, Xrs, AX Pulse wave velocity Carotid and abdominal aorta intima-media thickness (ultrasound) Symptoms: runny nose, cough, wheeze, fever, rash Health care provider contact: doctor/nurse telephone advice, pharmacist advice, child health nurse visit, GP visit, hospital doctor visit Medication use: antibiotics, asthma medications, topical steroid/ointment Doctor-diagnosis of upper respiratory tract infection/cold or flu, ear infection, eczema or dermatitis, chest infection, bronchiolitis, wheezing, asthma, croup, pneumonia Perinatal data (VPDC) (as per Table 1) Emergency department presentations (VEMD) (as per Table 1) Hospital Admissions (VAED) (as per Table 1) Consultations with primary healthcare practitioner (MBS) (as per Table 1) Prescription medications dispensed under national subsidy scheme (PBS) (as per Table 1) Educational and development outcomes (AEDC and NAPLAN data) (as per Table 1) Participant-collected vacuum dust sample from entire house Participant-collected soil sample from four locations around their house Soil and dust samples analysed for elemental and polycyclic aromatic hydrocarbon content to characterise gradient of exposures to mine fire smoke Survey data on whether the house was professionally cleaned after the mine fire, typical vacuum routine, vacuum type and brand Sociodemographic details of study child and immediate family Child’s health: including birth details, perinatal outcomes, breastfeeding history, medication use, childcare attendance, number of siblings and history of atopy in siblings Child’s residential history: details of the housing in which the pregnant mother and child lived, including year of build, heating/cooling, floor coverings etc Maternal and/or child’s residence during the smoke event: capturing location information on a day-night time scale Medical history of the biological parents Respiratory system resistance (Rrs) Respiratory system reactance (Xrs) Area under the reactance curve (AX) Bronchodilator response in Rrs, Xrs, AX Pulse wave velocity Carotid and abdominal aorta intima-media thickness (ultrasound) Symptoms: runny nose, cough, wheeze, fever, rash Health care provider contact: doctor/nurse telephone advice, pharmacist advice, child health nurse visit, GP visit, hospital doctor visit Medication use: antibiotics, asthma medications, topical steroid/ointment Doctor-diagnosis of upper respiratory tract infection/cold or flu, ear infection, eczema or dermatitis, chest infection, bronchiolitis, wheezing, asthma, croup, pneumonia Perinatal data (VPDC) (as per Table 1) Emergency department presentations (VEMD) (as per Table 1) Hospital Admissions (VAED) (as per Table 1) Consultations with primary healthcare practitioner (MBS) (as per Table 1) Prescription medications dispensed under national subsidy scheme (PBS) (as per Table 1) Educational and development outcomes (AEDC and NAPLAN data) (as per Table 1) Participant-collected vacuum dust sample from entire house Participant-collected soil sample from four locations around their house Soil and dust samples analysed for elemental and polycyclic aromatic hydrocarbon content to characterise gradient of exposures to mine fire smoke Survey data on whether the house was professionally cleaned after the mine fire, typical vacuum routine, vacuum type and brand AEDC, Australian Early Development Census; PBS, Pharmaceutical Benefits Scheme; MBS, Medicare Benefits Schedule; NAPLAN, National Assessment Program – Literacy and Numeracy; VAED, Victorian Admitted Episodes Data; VEMD, Victorian Emergency Minimum Dataset; VPDC, Victorian Perinatal Data Collection. Serial clinical testing and data extraction beyond 2024 pending funding acquisition. Health outcome data available for identified cohort (N = 571) Sociodemographic details of study child and immediate family Child’s health: including birth details, perinatal outcomes, breastfeeding history, medication use, childcare attendance, number of siblings and history of atopy in siblings Child’s residential history: details of the housing in which the pregnant mother and child lived, including year of build, heating/cooling, floor coverings etc Maternal and/or child’s residence during the smoke event: capturing location information on a day-night time scale Medical history of the biological parents Respiratory system resistance (Rrs) Respiratory system reactance (Xrs) Area under the reactance curve (AX) Bronchodilator response in Rrs, Xrs, AX Pulse wave velocity Carotid and abdominal aorta intima-media thickness (ultrasound) Symptoms: runny nose, cough, wheeze, fever, rash Health care provider contact: doctor/nurse telephone advice, pharmacist advice, child health nurse visit, GP visit, hospital doctor visit Medication use: antibiotics, asthma medications, topical steroid/ointment Doctor-diagnosis of upper respiratory tract infection/cold or flu, ear infection, eczema or dermatitis, chest infection, bronchiolitis, wheezing, asthma, croup, pneumonia Perinatal data (VPDC) (as per Table 1) Emergency department presentations (VEMD) (as per Table 1) Hospital Admissions (VAED) (as per Table 1) Consultations with primary healthcare practitioner (MBS) (as per Table 1) Prescription medications dispensed under national subsidy scheme (PBS) (as per Table 1) Educational and development outcomes (AEDC and NAPLAN data) (as per Table 1) Participant-collected vacuum dust sample from entire house Participant-collected soil sample from four locations around their house Soil and dust samples analysed for elemental and polycyclic aromatic hydrocarbon content to characterise gradient of exposures to mine fire smoke Survey data on whether the house was professionally cleaned after the mine fire, typical vacuum routine, vacuum type and brand Sociodemographic details of study child and immediate family Child’s health: including birth details, perinatal outcomes, breastfeeding history, medication use, childcare attendance, number of siblings and history of atopy in siblings Child’s residential history: details of the housing in which the pregnant mother and child lived, including year of build, heating/cooling, floor coverings etc Maternal and/or child’s residence during the smoke event: capturing location information on a day-night time scale Medical history of the biological parents Respiratory system resistance (Rrs) Respiratory system reactance (Xrs) Area under the reactance curve (AX) Bronchodilator response in Rrs, Xrs, AX Pulse wave velocity Carotid and abdominal aorta intima-media thickness (ultrasound) Symptoms: runny nose, cough, wheeze, fever, rash Health care provider contact: doctor/nurse telephone advice, pharmacist advice, child health nurse visit, GP visit, hospital doctor visit Medication use: antibiotics, asthma medications, topical steroid/ointment Doctor-diagnosis of upper respiratory tract infection/cold or flu, ear infection, eczema or dermatitis, chest infection, bronchiolitis, wheezing, asthma, croup, pneumonia Perinatal data (VPDC) (as per Table 1) Emergency department presentations (VEMD) (as per Table 1) Hospital Admissions (VAED) (as per Table 1) Consultations with primary healthcare practitioner (MBS) (as per Table 1) Prescription medications dispensed under national subsidy scheme (PBS) (as per Table 1) Educational and development outcomes (AEDC and NAPLAN data) (as per Table 1) Participant-collected vacuum dust sample from entire house Participant-collected soil sample from four locations around their house Soil and dust samples analysed for elemental and polycyclic aromatic hydrocarbon content to characterise gradient of exposures to mine fire smoke Survey data on whether the house was professionally cleaned after the mine fire, typical vacuum routine, vacuum type and brand AEDC, Australian Early Development Census; PBS, Pharmaceutical Benefits Scheme; MBS, Medicare Benefits Schedule; NAPLAN, National Assessment Program – Literacy and Numeracy; VAED, Victorian Admitted Episodes Data; VEMD, Victorian Emergency Minimum Dataset; VPDC, Victorian Perinatal Data Collection. Serial clinical testing and data extraction beyond 2024 pending funding acquisition. Linked data from birth, health and educational datasets will be available for the de-identified cohort. Table 1 outlines the datasets, custodians and requested variables available for the de-identified cohort. Findings to date as they relate to the major aims of the research programme include the following. Exposures to both peak and average PM2.5 were associated with increased likelihood of a diagnosis of gestational diabetes mellitus (GDM), where peak exposure was associated with greater risk [peak PM2.5 risk ratio (RR) 1.16; 95% confidence interval (CI) 1.09, 1.22; average PM2.5 RR 1.07; 95% CI 1.03, 1.10 per interquartile range (IQR) increase]. Exposure in the second trimester was of greatest importance. There were 16 additional cases of GDM attributable to smoke from the fire. We found no association between trimester-specific or whole of pregnancy exposure to coal mine fire PM2.5 and risk of hypertensive disorders of pregnancy, including pregnancy-induced hypertension and preeclampsia. Similarly, there was no association observed for conditions characterised by abnormal placentation, including placental abruption, placenta praevia and placenta accreta.4 In both identified and de-identified cohorts which relied on parent-reported birthweight and gestational age and administrative data, respectively, we found no association between maternal exposure to coal mine fire smoke PM2.5 and adverse fetal growth or maturity. However, we did identify that GDM was an effect modifier of the relationship between mine fire smoke exposure and birthweight, where among mothers with a diagnosis of GDM, smoke exposure was associated with increased birthweight. A 10- µg/m3 increase in average PM2.5 exposure was associated with a 97-g increase (95% CI 74, 120 g) in birthweight of babies born to mothers with GDM.5 The analysis of data for children in utero during the fire included 2678 monthly symptom diaries for 160 children (exposed in utero or unexposed). The early childhood analysis included 3290 diaries for 210 children (exposed from birth-2 years old or unexposed). An increase of 10 μg/m3 in average PM2.5 was associated with increased reports of runny nose and/or cough (RR 1.09; 95% CI 1.02, 1.17), wheeze (RR 1.56; 95% CI 1.18, 2.07), seeking health professional advice (RR 1.28; 95% CI 1.09, 1.49) and doctor diagnosis of ‘upper respiratory tract infection, cold or flu’ (RR 1.35; 95% CI 1.14, 1.60) for the in utero exposure group. Results were similar for peak PM2.5 exposure. At contrast with in utero exposure, no association was observed for early childhood exposure. There was no association between mean or peak PM2.5 exposure and parent-reported symptoms of rash or fever, or antibiotic, or topical steroid use.6 Exploring Australian Government Department of Human Services data that captured activity in primary health care (MBS) and pharmaceutical dispensations (PBS), we found children exposed to smoke during their first 2 years of life (early childhood exposure group) were more likely to be prescribed antibiotics in the year after the fire, compared with those not exposed (average PM2.5 RR 1.24; 95% CI 1.02, 1.50 per 10-ug/m3 increase; peak PM2.5 RR 1.14; 95% CI 1.00, 1.31 per 100-µg/m3 increase). There was no association between either smoke exposure metric and visits to a practitioner children mothers were exposed to coal mine fire smoke during pregnancy (in utero exposure there was no association between GP visits or dispensations of prescribed medications in their first year of In 3 years after the smoke event, measures of function were obtained system resistance at a of respiratory system reactance at a of the area under the reactance using in children who respiratory testing were from the early childhood exposure as participants from the in utero or unexposed groups were not old during first of clinical testing to in We observed an association between early childhood exposure to coal mine fire PM2.5 and but not for function increase in average PM2.5 exposure was associated with a in AX of difference 95% CI years after the mine fire we obtained two measures of vascular health, including intima-media thickness and wave velocity in We did not between exposure to PM2.5 during the mine fire period and adverse vascular in the group as a whole or in the group exposed in However, an association between mine fire smoke exposure and by a was found in children in the early childhood exposure group. increase in exposure to mean PM2.5 during the fire period was associated with an increase in 95% CI these two groups were there was an association between mean PM2.5 and increased in those children who smoke exposure 95% CI per or mothers 95% CI per but not in those not exposed to smoke or maternal There was a effect for peak PM2.5 exposure 95% CI per of the study include that the recruitment a gradient of exposure to coal mine fire smoke and developmental to be in to various outcomes of interest. The of time-activity data in the baseline survey exposure that captured during the smoke event, the likelihood of exposure The of data sources within the study for data The baseline survey the of that are not routinely captured in administrative datasets, including maternal in pregnancy and The of data linkage will the study to outcomes to health and education as the cohort relate to data that are not well captured in the administrative including as parental For the de-identified cohort, there is the potential for exposure as a of on residential address to assign exposure. the cohorts are the regional nature and of the population by the Latrobe ELF Study data are by the Institute for Medical Research at the of The Latrobe ELF Study is by The have a range of in and early childhood health and development and about the study may be to The Hazelwood Health Study a number of streams respiratory, and impacts of the smoke event and also Study and can be found at This was by the Victorian Department of Health and Human The the of the and not the of the

Source: International Journal of Epidemiology

DOI: 10.1093/ije/dyaa136