Research

Objectives: To assess long term outcomes for Aboriginal and Torres Strait Islander (Indigenous) Australians admitted non‐electively to intensive care units (ICUs). Design: Data linkage cohort study; analysis of ICU patient data (Australian and New Zealand Intensive Care Society Adult Patient Database), prospectively collected during 2007–2016. Setting: All four university‐affiliated level 3 ICUs in South Australia. Main outcomes: Mortality (in‐hospital, and 12 months and 8 years after admission to ICU), by Indigenous status. Results: 2035 of 39 784 non‐elective index ICU admissions (5.1%) were of Indigenous Australians, including 1461 of 37 661 patients with South Australian residential postcodes. The median age of Indigenous patients (45 years; IQR, 34–57 years) was lower than for non‐Indigenous ICU patients (64 years; IQR, 47–76 years). For patients with South Australian postcodes, unadjusted mortality at discharge and 12 months and 8 years after admission was lower for Indigenous patients; after adjusting for age, sex, diabetes, severity of illness, and diagnostic group, mortality was similar for both groups at discharge (adjusted odds ratio [aOR], 0.95; 95% CI, 0.81–1.10), but greater for Indigenous patients at 12 months (aOR, 1.14; 95% CI, 1.03–1.26) and 8 years (adjusted hazard ratio, 1.23; 95% CI, 1.13–1.35). The number of potential years of life lost was greater for Indigenous patients (median, 24.0; IQR, 15.8–31.8 v 12.5; IQR, 0–22.3), but, referenced to respective population life expectancies, relative survival at 8 years was similar (proportions: Indigenous, 0.78; 95% CI, 0.75–0.80; non‐Indigenous, 0.77; 95% CI, 0.76–0.78). Conclusions: Adjusted long term mortality and median number of potential life years lost are higher for Indigenous than non‐Indigenous patients after intensive care in hospital. These differences reflect underlying population survival patterns rather than the effects of ICU admission.

William G Mitchell · Adam Deane · Alex Brown · Shailesh Bihari · Hao Wong · Rajaram Ramadoss · Mark Finnis

Elevated serum angiotensin‐converting enzyme (ACE) activity, a biomarker for epithelioid granuloma, has a supportive role in the diagnosis and management of sarcoidosis,1 although in population‐based studies its diagnostic usefulness is modest, with positive and negative predictive values of 25.4% and 89.9% respectively.2 Further, elevated ACE activity is non‐specific; it is also found in people with tuberculous and other infectious granulomata, liver disease, lymphoma, diabetes, or hyperthyroidism, and also as a benign familial condition. However, elevated ACE activity can facilitate some clinical decisions, including the diagnosis of Löfgren syndrome or adults with uveitis.1,3 Serum ACE can be assessed by measuring its enzymatic activity or its protein concentration. Most Australian pathology laboratories measure ACE activity, which is predictably inhibited by ACE inhibitor (ACEI) drugs commonly prescribed for people with high blood pressure,4,5 whereas ACE protein level is not affected by these agents. In this study, we investigated the prevalence of ACEI influencing ACE activity results; for cases of markedly elevated ACE, we also evaluated the clinical performance of the two ACE measures with respect to sarcoidosis. In a preliminary evaluation, all discrepant paired results (high mass with low activity) were for patients using ACEIs at the time of sample collection. Between January 2017 and February 2019, we measured ACE activity and protein concentration in parallel; all test requests were initiated by clinicians as part of routine clinical care. Formal ethics approval was not required for collecting and analysing data to assess the quality of routine care. Further details of the study design and laboratory methods are included in the online Supporting Information. A total of 8882 paired test results were retrieved from the Pathology Queensland database for 4206 women (median age, 53.3 years; interquartile range [IQR], 36.0–65.6 years) and 4014 men (median age, 55.2 years; IQR, 42.2–67.3 years). Two discrete populations were evident in the scatterplot of paired results; for 1346 pairs (15.2%; 95% CI, 14.4–15.9%; green in Box 1), ACE activity was low relative to ACE protein, pathognomonic of ACEI interference. The upper reference limits for the two tests and the regression line for samples not affected by ACEIs nearly intersected, suggesting the general biologic equivalence of the two analytic methods and that the discordant results were not attributable to mismatched reference limits (Box 1). The correlation of values for the unaffected samples was moderate (R2 = 0.71) and the differences between the methods greater than predicted by their variances (Supporting Information, figure), indicating that the assays were not interchangeable. The monthly rate of ACEI interference was fairly consistent throughout the study period, despite comments to requesting physicians about the discrepancy between activity and protein levels included in pathology laboratory reports (Box 2). Of the 50 patients with high ACE protein levels (more than 300 μg/L) and ACE activity below the upper reference limit (70 IU/L), 27 (54%; 95% CI, 40–67%) had sarcoidosis (including 16 with ACE activity below the lower reference limit of 20 IU/L). In contrast, four of 16 people (25%; 95% CI, 10–50%) with high ACE activity (greater than 100 IU/L) and ACE protein within the reference interval had sarcoidosis. From a diagnostic perspective, ACEIs erode the negative predictive value of ACE activity, the most useful characteristic of this biomarker (Box 1; Supporting Information, table). Given that ACEI therapy interferes with ACE activity assessment, we recommend measuring ACE protein in routine practice, with the added benefit of convenience and safety of uninterrupted therapy for people taking ACEIs. The lack of influence of laboratory comments on testing behaviour is disappointing, but perhaps unsurprising given the information overload typical of modern medicine.6 Box 1 – Effect of angiotensin‐converting enzyme inhibitor (ACEI) therapy on serum ACE activity: scatterplot of paired ACE activity and protein assay results Pathology test reference intervals are indicated by the dotted lines. The shaded areas indicate result pairs included in the clinical audit (numbers of patients with sarcoidosis/total number audited). Blue: ACE activity not affected by ACEI therapy; 7536 samples, R2 = 0.71. Green: ACE activity affected by ACEI therapy; 1346 samples, R2 = 0.21. Box 2 – Influence of angiotensin‐converting enzyme (ACE) inhibitor (ACEI) therapy on serum ACE activity, by month

Carel J Pretorius · Jacobus PJ Ungerer

Iron deficiency anaemia, which affects 14–22% of women of reproductive age,1 has adverse effects on pregnant women and their infants. Oral iron supplementation is the first‐line treatment, but intravenous iron therapy is sometimes preferred because of gastrointestinal effects, low patient adherence, and the delayed effect of oral iron therapy. Further, guidelines now recommend intravenous iron therapy in certain situations,2 and more rapidly infusible intravenous iron preparations have recently become available in Australia. We investigated the use of intravenous iron by women of reproductive age, analysing dispensing data for a 10% random sample of Australians eligible to receive subsidised medicines under the Pharmaceutical Benefits Scheme (PBS).3 We included data for all women aged 18–44 years with a dispensing claim for intravenous iron during January 2013 to December 2017. Three preparations were available: iron polymaltose and iron sucrose during 2013–2017, and ferric carboxymaltose from June 2014. We calculated the annual number and rate of intravenous iron dispensing claims and iron preparation types by age group, using Australian Bureau of Statistics 2017 population data,4 and estimated overall dispensing rates by extrapolating these numbers to the national level (Supporting Information). The study was approved by the New South Wales Population and Health Services Research Ethics Committee (reference, 2013/11/494) and the federal Department of Human Services External Request Evaluation Committee. An estimated 259 700 intravenous iron dispensing claims were made for 190 490 women of reproductive age during 2013–2017; the annual number of dispensing claims increased from 17 920 in 2013 to 97 040 in 2017, and the annual rate of intravenous iron dispensing rose from 0.4 per 100 women in 2013 to 2.1 claims per 100 women in 2017 (Box). By iron type, 187 800 dispensing claims were for ferric carboxymaltose (72.3%), 71 110 for iron polymaltose (27.4%), and 790 for iron sucrose (0.3%). Most preparations were prescribed by general practitioners (111 870 claims, 43%), specialists (54 640 claims, 21%), and other medical practitioners (50 868 claims, 20%). The number of dispensing claims increased with age (18–24 years, 1.6 per 100 women; 35–44 years, 2.5 per 100 women). In 2017, intravenous iron was dispensed to one in fifty Australian women of reproductive age, five times the proportion in 2013; in 2017, 90% of these women received ferric carboxymaltose. The optimal rate of intravenous iron treatment is unknown, and there are no comparable overseas data. As possible adverse outcomes include permanent skin staining and the risk (albeit rare) of potentially fatal anaphylaxis,5 intravenous iron should be administered in settings where allergic reactions can be treated promptly, but whether this is generally the case is not known. Intravenous iron therapy for women of reproductive age also has considerable financial implications: based on average PBS prices,6 its total annual cost increased 35‐fold, from $0.75 million in 2013 to $26.9 million in 2017. However, we have probably underestimated the use of intravenous iron therapy, as we included only PBS‐subsidised dispensing, which may not include preparations administered to public hospital inpatients. The reasons for the rise in the use of intravenous iron are unclear, but may include increased awareness of patient blood management guidelines, the ease of treatment, and the perception that its side effect profile is more favourable than for oral iron therapy. The rapid growth raises concerns about whether it is being employed appropriately and cost‐effectively, given the potential harms and the lack of strong evidence for its value for improving quality of life and reproductive health outcomes. Box – Pharmaceutical Benefits Scheme dispensing claims for intravenous iron preparations for women aged 18–44 years, Australia, 2013–2017 *The small numbers of dispensing claims for iron sucrose are not separately depicted, but were included when calculating the rates of dispensing.

Antonia W Shand · Jane Bell · Amanda Henry · Luke E Grzeskowiak · Giselle Kidson‐Gerber · Sallie Pearson · Natasha Nassar

The prevalence of food allergy among Victorian children is rising.1 In Victoria, children with suspected food allergies can be on hospital outpatient clinic waiting lists for months before being assessed.2 This may lead families to consider alternative avenues, which can lead to poor allergy management and the need for emergency care. Increasing numbers of Victorian children are presenting to emergency departments,3 but we do not know whether the number visiting with food allergy is also rising. We analysed Victorian Emergency Minimum Dataset (VEMD) data for the period 2005–06 to 2014–15. The VEMD is a statewide administrative dataset that includes non‐identifiable patient‐level data for all Victorian public emergency department encounters. We included all food allergy‐related presentations by children and young people aged 0–19 years, selected according to International Classification of Diseases, tenth revision, Australian modification (ICD‐10‐AM) codes: T78.0 (anaphylactic shock due to a food reaction), T78.1 (other adverse food reactions, not elsewhere classified), T78.4 (allergy, unspecified: includes non‐food‐related allergies), and L27.2 (dermatitis due to ingested food). Presentation rates by age group were calculated using Australian Bureau of Statistics (ABS) age‐stratified population data for Victorians aged 0–19 years;4 rates for regions were calculated using ABS population data for Statistical Areas 2 (SA2).5 The study was deemed exempt from the need for formal ethics approval by the Royal Children's Hospital Human Research Ethics Committee. The number of children presenting to emergency departments with food allergy‐related problems increased from 2368 in 2005–06 to 4263 in 2014–15; the presentation rate increased from 18 to 29 per 10 000 population (Box 1). About half the children who presented with food allergy‐related problems were aged 0–4 years, the rate for this age group increasing from 38 to 55 per 10 000 population (Box 2). The proportion of presentations triaged as being more urgent (triage categories 1–3) also increased, from 51% to 63% (Box 1). The rate of presentations to metropolitan hospitals increased more (from 18 per 10 000 in 2005–06 to 32 per 10 000 in 2014–15; 78% increase) than did the rate for rural hospitals (26 per 10 000 in 2005–06 to 36 per 10 000 in 2014–15; 38% increase) (Box 1). Hospitals in the North‐West Melbourne region received about one‐third of all allergy‐related emergency department visits, and the number in this region doubled over the study period (706 in 2005–06; 1536 in 2014–15) (Box 3). These data indicate that the demand for emergency services associated with food allergy‐related problems in children increased during 2005–15. The increase was particularly marked for children aged 0–4 years and for children and young people in the North‐West Melbourne and Southern Melbourne regions. While the reason for the increased burden is not clear — that is, whether the prevalence of allergy had increased (including because of a change in population composition), management plans had changed, or access to community services was reduced — the consequence is greater demand on emergency services across Melbourne. Box 1 – Presentations to Victorian public emergency departments by childen and young people (0–19 years) with food allergy‐related problems 2005–06 2006–07 2007–08 2008–09 2009–10 2010–11 2011–12 2012–13 2013–14 2014–15 All food allergy presentations Number 2368 2680 2754 2991 3082 3159 3185 3422 3881 4263 Rate (per 10 000 population)* 18 20 21 22 23 23 23 24 27 29 ICD‐10‐AM diagnostic codes T78.0 141 (6.0%) 152 (5.7%) 154 (5.6%) 168 (5.6%) 233 (7.6%) 283 (9.0%) 289 (9.1%) 339 (9.9%) 437 (11.3%) 501 (11.8%) T78.1 800 (33.8%) 948 (35.4%) 962 (34.9%) 1127 (37.7%) 1167 (37.9%) 1152 (36.5%) 1117 (35.1%) 1288 (37.6%) 1464 (37.7%) 1624 (38.1%) T78.4 1234 (52.1%) 1351 (50.4%) 1441 (52.3%) 1488 (49.8%) 1552 (50.4%) 1597 (50.6%) 1633 (51.3%) 1702 (49.7%) 1881 (48.5%) 2031 (47.6%) L27.2 193 (8.2%) 229 (8.5%) 197 (7.2%) 208 (7.0%) 130 (4.2%) 127 (4.0%) 146 (4.6%) 93 (2.7%) 99 (2.6%) 107 (2.5%) Age Number 0–4 years 1202 (50.8%) 1364 (50.9%) 1424 (51.7%) 1537 (51.4%) 1520 (49.3%) 1595 (50.5%) 1593 (50.0%) 1759 (51.4%) 2015 (51.9%) 2152 (50.5%) 5–9 years 466 (19.7%) 515 (19.2%) 521 (18.9%) 573 (19.2%) 673 (21.8%) 608 (19.3%) 660 (20.7%) 702 (20.5%) 813 (21.0%) 967 (22.7%) 10–14 years 305 (12.9%) 335 (12.5%) 355 (12.9%) 405 (13.5%) 403 (13.1%) 426 (13.5%) 383 (12.0%) 433 (12.7%) 515 (13.3%) 561 (13.2%) 15–19 years 395 (16.7%) 466 (17.4%) 454 (16.5%) 476 (15.9%) 486 (15.8%) 530 (16.8%) 549 (17.3%) 528 (15.4%) 538 (13.8%) 583 (13.7%) Rate (per 10 000 population)* 0–4 years 38 42 43 45 43 45 44 47 53 55 5–9 years 15 16 16 18 21 18 19 20 22 26 10–14 years 9 10 11 12 12 13 12 13 15 16 15–19 years 12 13 13 13 14 15 15 15 15 16 Sex Number Boys 1284 (54.2%) 1435 (53.5%) 1476 (53.6%) 1625 (54.3%) 1663 (54.0%) 1723 (54.5%) 1779 (55.9%) 1867 (54.6%) 2158 (55.6%) 2388 (56.0%) Girls 1084 (45.8%) 1245 (46.5%) 1278 (46.4%) 1366 (45.7%) 1419 (46.0%) 1436 (45.5%) 1406 (44.1%) 1555 (45.4%) 1723 (44.4%) 1875 (44.0%) Rate (per 10 000 population)* Boys 19 21 21 23 24 25 25 26 29 32 Girls 17 19 20 21 21 22 21 22 24 26 Hospital region Number Metropolitan† 1560 (65.9%) 1860 (69.4%) 1907 (69.2%) 2008 (67.1%) 2071 (67.2%) 2194 (69.5%) 2186 (68.6%) 2345 (68.5%) 2781 (71.7%) 3149 (73.9%) Rural‡ 808 (34.1%) 820 (30.6%) 847 (30.8%) 983 (32.9%) 1011 (32.8%) 965 (30.5%) 999 (31.4%) 1077 (31.5%) 1100 (28.3%) 1114 (26.1%) Rate (per 10 000 population)* Metropolitan† 18 22 22 23 23 25 24 25 29 32 Rural‡ 26 27 27 32 33 31 32 35 35 36 Triage category Categories 1–3 1198 (50.6%) 1429 (53.3%) 1568 (57.0%) 1707 (57.0%) 1830 (59.3%) 1861 (59.0%) 1807 (56.8%) 2047 (59.8%) 2365 (61.0%) 2694 (63.2%) Categories 4, 5 1170 (49.4%) 1251 (46.7%) 1186 (43.0%) 1284 (43.0%) 1252 (40.7%) 1298 (41.0%) 1378 (43.2%) 1375 (40.2%) 1516 (39.0%) 1569 (36.8%) ICD‐10‐AM = International Classification of Diseases, tenth revision, Australian modification. * All presentation rates are per 10 000 children in Victoria aged 0–19 years in the corresponding category. † Victorian Emergency Minimum Dataset (VEMD) regions: North‐West, Southern, and Eastern Melbourne. ‡ VEMD regions: Loddon Mallee, Gippsland, Barwon South West, Hume, Grampians. Box 2 – Presentations to Victorian public emergency departments by people aged 0–19 years with food allergy‐related problems: rates per 10 000 population, by age group Box 3 – Presentations to Victorian public emergency departments by people aged 0–19 years with food allergy‐related problems, by hospital campus region

Rachel O'Loughlin · Harriet Hiscock

Objectives: To investigate the quality of diagnostic and triage advice provided by free website and mobile application symptom checkers (SCs) accessible in Australia. Design: 36 SCs providing medical diagnosis or triage advice were tested with 48 medical condition vignettes (1170 diagnosis vignette tests, 688 triage vignette tests). Main outcome measures: Correct diagnosis advice (provided in first, the top three or top ten diagnosis results); correct triage advice (appropriate triage category recommended). Results: The 27 diagnostic SCs listed the correct diagnosis first in 421 of 1170 SC vignette tests (36%; 95% CI, 31–42%), among the top three results in 606 tests (52%; 95% CI, 47–59%), and among the top ten results in 681 tests (58%; 95% CI, 53–65%). SCs using artificial intelligence algorithms listed the correct diagnosis first in 46% of tests (95% CI, 40–57%), compared with 32% (95% CI, 26–38%) for other SCs. The mean rate of first correct results for individual SCs ranged between 12% and 61%. The 19 triage SCs provided correct advice for 338 of 688 vignette tests (49%; 95% CI, 44–54%). Appropriate triage advice was more frequent for emergency care (63%; 95% CI, 52–71%) and urgent care vignette tests (56%; 95% CI, 52–75%) than for non‐urgent care (30%; 95% CI, 11–39%) and self‐care tests (40%; 95% CI, 26–49%). Conclusion: The quality of diagnostic advice varied between SCs, and triage advice was generally risk‐averse, often recommending more urgent care than appropriate.

Michella G Hill · Moira Sim · Brennen Mills

Objective: To describe uptake of hospital in the home (HIH) by major Australian hospitals and the characteristics of patients and their HIH admissions; to assess change in HIH admission numbers relative to total hospital activity. Design: Descriptive, retrospective study of HIH activity, analysing previously collected census data for all multi‐day hospital inpatient admissions to included hospitals during the period 1 January 2011 – 31 December 2017. Setting, participants: Nineteen principal referrer hospital members of the Health Roundtable in Australia. Main outcome measures: HIH admissions by diagnosis‐related group (DRG); patient and admission characteristics. Results: 80 167 of 2 185 421 admissions to the 19 hospitals included HIH care, or 3.7% (95% CI, 3.6–3.7%) of all admissions. Median length of stay for admissions including HIH (7.3 days; IQR, 3.1–14 days) was longer than that for those that did not (2.7 days; IQR, 1.6–5.1 days). For HIH admissions, the proportion of men was higher (54.4% v 45.9%), the proportion of patients who died in hospital was lower (0.3% v 1.4%), and re‐admission within 28 days was less frequent (2.3% v 3.6%). The 50 DRGs with greatest HIH activity encompassed 65 811 HIH admissions (82.1%), or 8.4% (95% CI, 8.4–8.5%) of all admissions in these DRGs. HIH admission numbers grew more rapidly than non‐HIH admissions, but the difference was not statistically significant. Conclusions: HIH care is most frequently provided to patients requiring hospital treatment related to infections, venous thromboembolism, or post‐surgical care. Its use could be expanded in clinical areas where it is currently used, and extended to others where it is not. HIH activity is growing. It should be systematically monitored and reported to allow better overview of its use and outcomes.

Michael Montalto · Patrick McElduff · Kristy Hardy

The coronavirus disease 2019 (COVID‐19) pandemic is pushing health systems to, and possibly beyond, their limits.1 In Italy, the exponential rise in case numbers has caused a corresponding rise in demand for intensive care unit (ICU) beds.2 To determine how many ICU beds will be required in Australia, we propose a simple model of an uninterrupted pandemic process based on the local situation in late March 2020, and compare this model with recent data from the Lombardy.3 The uninterrupted exponential growth scenario In queueing theory, Little's law4 describes the relationship between the number of patients in a system (L) and the mean arrival rate (λ) and length of time the patient remains in the system (W) as: L = λW If a tertiary hospital has a steady state rate of 20 new admissions of patients with confirmed COVID‐19 per day, of whom one requires ICU admission5 (λ) for a mean 10 days (W), the hospital ICU will need at least 10 beds to accommodate these patients. If, however, the number of new confirmed cases increases by 20% each day (in late March 2020, the number was increasing in Australia by 23% each day6), and 100 cases are confirmed on one day, about 120 will be confirmed on the next. This increase in the daily rate of 20 new cases will mean one extra ICU admission per day, and the need for at least 10 further ICU beds. That is, the total number of ICU beds needed will be about 10% of the number of confirmed cases, or 50% of the number of new cases during the exponential growth phase of the epidemic. Approximately 2300 ICU beds are available in Australia;7 if public health measures fail to curb the rate of growth in case numbers, the national ICU capacity would be exceeded when the number of COVID‐19 cases reaches 23 000. Other sources8 have estimated that Australia could cope with as many as 44 580 COVID‐19 cases, but this would grant only a 3‐day extension before ICU capacity was exceeded. In our exponential growth scenario, commencing with 100 confirmed cases on day 1, 31 ICU beds would be required by day 7 and 119 by day 14 (Box 1). In sensitivity analyses, ICU bed capacity is sufficient even after 30 days if the ICU admission rate is reduced to 2.5%, but would be exceeded by day 26 were the ICU admission rate as high as 10%. It is important to note that our model describes a particularly serious scenario, and that actual outcomes will be modified by parameters not included in the model, including potential lags between diagnosis, hospital admission, and transfer to intensive care, and the proportion of true positive results among people tested for infecton. Is the modelled scenario plausible? To evaluate how realistic the uninterrupted exponential growth scenario is, we compared exponential and linear growth models with recent data for the Lombardy in Italy.9 Using piecewise regression models, the increase in the number of ICU patients during days 1–14 was exponential (R2 = 0.96); from day 15, ICU admissions continued to rise steeply, but the increase was linear (R2 = 0.99) (Box 2). To determine the reason for the change in growth rate at day 15, we compared the ICU admission and mortality rates for patients hospitalised with COVID‐19. The mortality rate during days 1–14 was fairly constant at about 8.8%, but rose dramatically from day 15 to a mean 23%. Most deaths during the first 14 days were probably of patients in intensive care, but we suspect that from day 15 patients died partly because of the lack of access to ICU beds as demand exceeded the capacity of the system to provide them, as indicated by the fall in ICU admission rate (Box 3). Conclusion While the assumptions of our model can be debated, the exponential increase in Australian cases until late March suggested that it described a realistic clinical scenario consistent with overseas data available at that time. The exponential increase in case numbers and subsequent demand for ICU beds could have overwhelmed the capacity of even the largest Australian hospitals if SARS‐CoV‐2 transmission had not been as drastically reduced as it appears to have been by the successful public health measures enacted by the federal and state governments and the adherence to these measures by the Australian public. The rate of ICU admissions per positive case may be lower in Australia than reported for Italy and China — because of healthier underlying demographic conditions, a greater number of detected milder cases, or both — but this would not change the overall implications of the model. Australia must maintain measures to strictly control the rate of new cases and continue to improve our ICU surge capacity, lest we squander the chance to avoid an Italian fate. Box 1 – Intensive care unit (ICU) bed demand, by time and proportion of patients with confirmed COVID‐19 who require intensive care Box 2 – Intensive care unit (ICU) admission rate in the Lombardy: actual and modelled Box 3 – Intensive care unit (ICU) admission rate and mortality for all patients with COVID‐19 admitted to hospitals in the Lombardy

Hamish DD Meares · Michael P Jones

Coronavirus disease 2019 (COVID‐19) poses extraordinary challenges for health care in Australia. One of the greatest will be the pressure on hospitals to support people with severe disease. Modelling studies can provide valuable insights into the likely course of the epidemic, and can be particularly useful for anticipating resource requirements, including demand for intensive care services at the peak of the epidemic. In this report, we extrapolate the findings of the Imperial College model of the pandemic1 to the New South Wales population. We also developed a simple SEIR (susceptible–exposed/incubating–infected–removed) model to explore the effect of varying the infection reproduction number (R), which can be reduced by effective social distancing measures, on the timing of the peak of the epidemic. The two models are described in the online Supporting Information. Applying the Imperial College model, the peak demand for intensive care in NSW would be at least 6965 beds if mitigation efforts — isolation of people with confirmed COVID‐19, household quarantine of their contacts, social distancing from people over 70 years of age — are implemented, or almost eight times as many as the baseline number; without mitigation, more than three times as many ICU beds (21 283) could be required (Box 1). Applying our SEIR model to a scenario without social distancing measures (R = 2.4), the number of people requiring hospitalisation in NSW would peak at 450 per 100 000 population (35 375 beds), and the number requiring critical care at 150 per 100 000 population (11 792 ICU beds, or 1349% of baseline ICU capacity). In this scenario, viral transmission would peak during late June and ICU bed occupancy in early July. About 16% of people would be potentially infectious at this point, although a smaller proportion was modelled as exhibiting symptoms (Box 2; Supporting Information, table 3). In a scenario of increased social isolation (R = 1.6) and an assumed hospitalisation rate for people with confirmed COVID‐19 of 6.7%, case numbers would peak in early October and ICU occupancy in mid‐November; about 180 people per 100 000 population would require hospitalisation (14 150 beds) and 65 per 100 000 intensive care (5110 ICU beds, or 585% of baseline ICU capacity) (Box 2; Supporting Information, table 3). That is, the peak figures would be about one‐third the size of those in the no mitigation scenario. Sensitivity analyses in which the proportion of hospitalised patients was varied (5–15%) similarly found that increasing social isolation markedly reduced demand (Supporting Information, table 4). We have used two modelling methods to estimate peak demand for critical care services in NSW during the COVID‐19 epidemic. Both approaches identified that COVID‐19 would impose a major burden on the health care system, and the mismatch between the estimated numbers of ICU beds needed and their availability is stark. Our modelling shows the critical importance of effective COVID‐19 containment strategies, as well as the urgent need to invest in resources that support the surge capacity of critical care services in NSW. Box 1 – Estimated number of intensive care unit (ICU) beds required at the peak of the initial wave of COVID‐19 cases, applying the Imperial College model to New South Wales, by Local Health District (LHD) Mitigation strategy Population (2016)2 No mitigation Close schools, universities Case isolation Case isolation, household quarantine Case isolation, household quarantine, social distancing of people over 70 ICU beds needed per 100 000 population1 — 275 240 190 125 90 ICU beds need, by LHD Sydney 656 460 1805 1576 1247 821 591 South Western Sydney 964 342 2652 2314 1832 1205 868 South Eastern Sydney 914 021 514 2194 1737 1143 823 Western Sydney 948 584 2609 2277 1802 1186 854 Northern Sydney 914 233 2514 2194 1737 1143 823 Illawarra Shoalhaven 405 534 1115 973 771 507 365 Central Coast 335 309 922 805 637 419 302 Other LHDs 2 600 791 7152 6242 4942 3251 2341 All NSW (proportion of baseline bed number)* 7 739 274 21 283 (2435%) 18 574 (2125%) 14 705 (1682%) 9674 (1107%) 6965 (797%) * Estimated number of ICU beds prior to COVID‐19 epidemic: 874.3 Box 2 – The estimated number of patients with COVID‐19 admitted to hospital or to intensive care units (ICUs), according to a SEIR model of the epidemic * For main curves, 10% case hospitalisation rate assumed; shaded areas show range for hospitalisation rates between 5% and 15%.

Gregory J Fox · James M Trauer · Emma McBryde

Objectives: To estimate the carbon footprint of five common hospital pathology tests: full blood examination; urea and electrolyte levels; coagulation profile; C‐reactive protein concentration; and arterial blood gases. Design, setting: Prospective life cycle assessment of five pathology tests in two university‐affiliated health services in Melbourne. We included all consumables and associated waste for venepuncture and laboratory analyses, and electricity and water use for laboratory analyses. Main outcome measure: Greenhouse gas footprint, measured in carbon dioxide equivalent (CO2e) emissions. Results: CO2e emissions for haematology tests were 82 g/test (95% CI, 73–91 g/test) for coagulation profile and 116 g/test (95% CI, 101–135 g/test) for full blood examination. CO2e emissions for biochemical tests were 0.5 g/test CO2e (95% CI, 0.4–0.6 g/test) for C‐reactive protein (low because typically ordered with urea and electrolyte assessment), 49 g/test (95% CI, 45–53 g/test) for arterial blood gas assessment, and 99 g/test (95% CI, 84–113 g/test) for urea and electrolyte assessment. Most CO2e emissions were associated with sample collection (range, 60% for full blood examination to 95% for coagulation profile); emissions attributable to laboratory reagents and power use were much smaller. Conclusion: The carbon footprint of common pathology tests was dominated by those of sample collection and phlebotomy. Although the carbon footprints were small, millions of tests are performed each year in Australia, and reducing unnecessary testing will be the most effective approach to reducing the carbon footprint of pathology. Together with the detrimental health and economic effects of unnecessary testing, our environmental findings should further motivate clinicians to test wisely.

Scott McAlister · Alexandra L Barratt · Katy JL Bell · Forbes McGain

In 2014, we reported in the MJA our findings, based on linked data for cervical screening and human papillomavirus (HPV) vaccination of women in Victoria, that participation of young women in cervical screening during 2010 and 2011 was significantly lower among HPV‐vaccinated than among unvaccinated women.1 In 2018, we had the opportunity to repeat the study at the national level as part of a broader data linkage study of cancer outcomes and screening behaviour across the three national cancer screening programs in Australia.2 In the original study (2014), the Australian Institute of Health and Welfare (AIHW) data linkage unit applied probabilistic name‐based linkage to HPV vaccination and cervical screening data. We acknowledged it was likely that some screened women who were vaccinated would be incorrectly identified as unvaccinated because many young women would have changed their names and addresses between vaccination and cervical screening. In the more recent study (2018), the AIHW again used probabilistic name‐based linkage, but first updated HPV vaccination and cervical screening data by obtaining histories of name and address changes from the Medicare Enrolment File. Medicare registrants’ details are updated when new data are provided to Medicare, the national health care scheme, and are recorded in new records with dates of change. The Australian Department of Human Services agreed to provide these data to the AIHW for data linkage purposes for our 2018 study. Our investigation was approved by the AIHW Ethics Committee (reference, EO 2014‐4‐130) and by state and territory human research ethics committees. After incorporating Medicare data, annual cervical screening rates for Victorian women aged 20–24 years or 25–29 years were higher during 2010 and 2011 for vaccinated than unvaccinated women,2 contrary to our 2014 findings.1 For 20–24‐year‐old Victorian women, the difference in rate changed from 10.1% lower to 14.7% higher for vaccinated women, and for 25–29‐year‐old women from 13.5% lower to 10.0% higher (Box). Our updated findings are consistent with findings from other countries of higher cervical screening participation among women who have been vaccinated against HPV.3,4,5 Incorporating the Medicare Enrolment File into the 2018 linkage was a test of proof of concept. Its successful use in this and similar studies has led to the AIHW data linkage unit granting ethics approval and relevant authorisations for employing the Medicare Enrolment File as a tool for improving the quality of other data linkage studies. The key message of our original study, however, remains unchanged. All women, whether vaccinated against HPV or not, should be encouraged to participate in cervical screening: the HPV vaccine does not protect against all HPV types, and many women in Australia were sexually active before they were vaccinated. While it is as yet unclear whether the association between vaccination and screening will persist for women who were routinely vaccinated at school, it is crucial that we focus on strategies that effectively engage women who do not currently participate in screening. Box – Estimated participation of Victorian women in cervical screening during 2010 and 2011, by HPV vaccination status and age group: 2014 and 2018 data linkage studies HPV = human papillomavirus.

Alison C Budd · Andrew Powierski · Theresa Chau · Marion Saville · Julia ML Brotherton

Out‐of‐pocket expenses for cancer care are of growing concern for patients, clinicians, service providers, non‐governmental organisations, private insurers, and politicians. Contrary to popular belief, there is no direct link between the cost and quality of care. Out‐of‐pocket expenses are a particular problem for patients who live further from treatment centres, are younger, or have later stage disease.1 Adults (18 years or older) with pathologically confirmed colorectal, lung, prostate or breast cancer from four rural (Midwest, South West, Great Southern, Goldfields) and two outer metropolitan (Joondalup/Wanneroo and Rockingham/Peel) regions of Western Australia were identified in the WA Cancer Registry. Between 1 April 2014 and 31 April 2017, eligible patients were invited to complete questionnaires requesting demographic, financial, and treatment information, including all costs during treatment, as reported previously.2 We used log‐linked generalised linear models with gamma distribution, adjusted for age and sex, to estimate out‐of‐pocket expenses (with 95% confidence intervals [CIs]) for participant characteristics found to be significantly associated with out‐of‐pocket expenses in univariate analyses (online Supporting Information). The study was approved by the WA Country Health Service Ethics Committee (reference, 2014:10) and the Department of Health WA Human Research Ethics Committee (reference, 2014/26). One hundred and seventeen of the 119 outer metropolitan participants (98%) and 294 of the 308 rural participants (95%) incurred out‐of‐pocket expenses for their cancer care, chiefly for surgery, medical tests, and medical appointments. These costs ranged between $51 and $106 140 for outer metropolitan participants, and between $13 and $20 842 for rural participants. Fifty‐three rural participants (17%) and 39 outer metropolitan participants (33%) spent more than 10% of their household income on cancer care (data not shown). Among rural participants, mean out‐of‐pocket expenses were higher for men ($1988; 95% CI, $1605–$2461 v $1362; 95% CI, $1092–$1699), for people with private health insurance ($2455; 95% CI, $1973–$3053 v $1103; 95% CI, $877–$1386), and for people who were married ($2086; 95% CI, $1749–$2489 v $1297; 95% CI, $975–$1725), had undergone surgery ($1990; 95% CI, $1684–$2351 v $1360, 95% CI, $1005–$1839), or had worked prior to being diagnosed with cancer ($2084; 95% CI, $1643–$2644 v $1298; 95% CI, $1038–$1625) (Box). Among outer metropolitan participants, mean out‐of‐pocket expenses were higher for men ($5217; 95% CI, $3928–$6928 v $2247; 95% CI, $1756–$2875), for people with private health insurance ($4670; 95% CI, $3588–$6078 v $2510; 95% CI, $1853–$3401), and for those who had undergone surgery ($5434; 95% CI, $4260–$6932 v $2157; 95% CI, $1541–$3020), worked prior to being diagnosed with cancer ($5471, 95% CI, $3952–$7573 v $2143; 95% CI, $1643–$2794), resided in areas of high socio‐economic status ($4299; 95% CI, $3235–$5712 v low, $1859; 95% CI, $1374–$2516), or were receiving chemotherapy ($4286; 95% CI, $3162–$5810 v $2735; 95% CI, $2116–$3534) (Box). It is perhaps surprising that out‐of‐pocket expenses were higher for people in outer metropolitan areas, who presumably lived closer to treatment centres than rural residents. However, these findings are consistent with the recent report that out‐of‐pocket spending on non‐hospital Medicare‐subsidised services and specialist services was higher for metropolitan patients than for those in regional areas.4 The higher out‐of‐pocket expenses for people with private health insurance or undergoing surgery indicate the importance of health care funding arrangements and the magnitude of the costs borne by patients. The marked variation in out‐of‐pocket expenses reported here and by others5 highlights the need for easily accessible information about services, medical costs, and gap payments for all health care services. The Informed Financial Consent website coordinated by the Australian Medical Association,6 consumer organisation fact sheets, and professional body initiatives are steps in the right direction, but their impact is yet to be determined. Problems that still need attention in the unregulated private fee‐setting environment in Australia include price discrimination in some specialist sectors.7 Bundles of care for cancer treatment that would allow patients and their families to better understand and plan for expenses should be explored. Box – Estimated mean out‐of‐pocket expenses for cancer‐related health care (with 95% confidence intervals) for outer metropolitan and rural patients, by patient characteristics significantly associated with higher out‐of‐pocket expenses in univariate analyses* * For outer metropolitan patients, marital status, and for rural patients, socio‐economic status and chemotherapy were not significant predictors of out‐of‐pocket expenses, and were therefore not included in the final models. †Index of Relative Socio‐economic Disadvantage (IRSD):3 low (most disadvantaged), deciles 1–4; moderate, deciles 5–6; high (least disadvantaged), deciles 7–10.

Neli S Slavova‐Azmanova · Jade C Newton · Christobel M Saunders

Objectives: To assess the capacity of intensive care units (ICUs) in Australia to respond to the expected increase in demand associated with COVID‐19. Design: Analysis of Australian and New Zealand Intensive Care Society (ANZICS) registry data, supplemented by an ICU surge capability survey and veterinary facilities survey (both March 2020). Settings: All Australian ICUs and veterinary facilities. Main outcome measures: Baseline numbers of ICU beds, ventilators, dialysis machines, extracorporeal membrane oxygenation machines, intravenous infusion pumps, and staff (senior medical staff, registered nurses); incremental capability to increase capacity (surge) by increasing ICU bed numbers; ventilator‐to‐bed ratios; number of ventilators in veterinary facilities. Results: The 191 ICUs in Australia provide 2378 intensive care beds during baseline activity (9.3 ICU beds per 100 000 population). Of the 175 ICUs that responded to the surge survey (with 2228 intensive care beds), a maximal surge would add an additional 4258 intensive care beds (191% increase) and 2631 invasive ventilators (120% increase). This surge would require additional staffing of as many as 4092 senior doctors (245% increase over baseline) and 42 720 registered ICU nurses (269% increase over baseline). An additional 188 ventilators are available in veterinary facilities, including 179 human model ventilators. Conclusions: The directors of Australian ICUs report that intensive care bed capacity could be near tripled in response to the expected increase in demand caused by COVID‐19. But maximal surge in bed numbers could be hampered by a shortfall in invasive ventilators and would also require a large increase in clinician and nursing staff numbers.

Edward Litton · Tamara Bucci · Shaila Chavan · Yvonne Y Ho · Anthony Holley · Gretta Howard · Sue Huckson · Philomena Kwong · Johnny Millar · Nhi Nguyen · Paul Secombe · Marc Ziegenfuss · David Pilcher

Ambient air quality in Australia is regulated by the National Environment Protection Measure (NEPM), which sets a maximum 24‐hour mean concentration of 50 μg/m3 for particulate matter less than 10 μm in diameter (PM10) and 25 μg/m3 for PM2.5. Each state and territory is required by the NEPM to annually report all breaches of this standard, including the sources of pollution.1 We analysed NEPM reports for Western Australia to identify days during 1 January 2002 – 31 December 2017 on which atmospheric particulate matter levels exceeded air quality standard levels, and classified them according to the most frequently reported sources of pollution: prescribed burns, wildfires, and other (crustal particles such as dust, wood smoke, and indeterminate). During 2008–2013, exceedances caused by smoke from prescribed burns, wildfires, and wood smoke were all recorded by the WA Department of Environment Regulation as “smoke haze”. For this period, we therefore applied a random forest algorithm, a machine learning method that uses a random sample of observations for known classifications to predict the classifications for new data.2 We included the variables month, day of the week, temperature, and pollution level as model predictors. To estimate background PM2.5 level, we obtained historical hourly values for PM10 and PM2.5 from the WA Department of Water and Environmental Regulation3 and calculated historical monthly means, excluding days on which particle levels exceeded the air quality standard. We estimated daily PM2.5 concentrations attributable to smoke events by subtracting the background PM2.5 level from measured daily values. Applying standard methods for assessing the health impact of air pollution,4 we estimated the numbers of premature deaths, hospitalisations for cardiovascular and respiratory problems, and emergency department presentations with asthma attributable to elevated PM2.5 levels. We used the value of statistical life (VSL)5 to estimate costs associated with premature mortality. The VSL is based on the willingness to pay for reduced risk of premature mortality, and does not take into account underlying health status, age, or life expectancy of individuals. Deaths associated with acute exposure to increased air pollution are more likely among people at greater risk because of advanced age or chronic illness.6 We estimated hospital service costs according to the mean cost of each episode of care as reported in the Independent Hospital Pricing Authority national cost data collection report7 and the Health Policy Analysis emergency care costing report.8 We also undertook a sensitivity analysis in which we excluded data for 2008–2013, when exceedances caused by smoke from prescribed burns, wildfires, and wood smoke were all recorded in NEPM reports as “smoke haze”. Further details on our methods, including underlying assumptions and limitations, are included in the online Supporting Information. During 2002–2017, particulate air pollution exceeded the national standard on 271 of 5844 days (4.6%), including 197 days (73%) attributable to prescribed burns or wildfires. We estimated that 41 (95% confidence interval [CI], 15–68) premature deaths, 99 (95% CI, 19–182) hospitalisations for cardiovascular problems and 174 (95% CI, 0–373) for respiratory conditions, and 123 (95% CI, 70–179) emergency department visits with asthma were attributable to elevated PM2.5 concentration (Box 1). Total estimated health costs were $188.8 million (95% CI, $68.1–311.1 million); $97.1 million (51%) was attributable to prescribed burns and $77.7 million (41%) to wildfires. Mean estimated health costs were lower on days affected by smoke from prescribed burns ($703 984; 95% CI, $254 064–$1.2 million) than those affected by wildfire smoke ($1.3 million; 95% CI, $475 000–$2.2 million), although more days were affected by prescribed burns (138) than by wildfires (59). The estimated smoke‐related costs of wildfires were highest in 2012 ($24.8 million); in many years, prescribed fires often accounted for most health‐related costs, peaking in 2017 ($24.1 million) (Box 2). In our sensitivity analysis excluding the period 2008–2013, the relative costs by source were similar (prescribed burns, 53% [$58.4 million]; wildfires, 38% [$41.6 million]; Supporting Information). Particulate matter in fire smoke is associated with adverse health outcomes,9 even at relatively low concentrations.10 Landscape fire smoke was the greatest contributor to excessive atmospheric particulate matter levels in WA during 2002–2017 and was associated with substantial health costs. Our estimates of the health impacts may be conservative, as we included only days when PM2.5 concentrations exceeded the national standard, excluding smoky days on which the air quality standard was not breached. Further, our selection of health outcomes did not encompass the total health burden attributable to smoke exposure. Our study highlights the different smoke‐related health effects and costs of infrequent severe wildfire and regular prescribed burning. While prescribed burning reduces the risk of wildfire, better understanding and incorporation into control strategies of the full health impacts of each type of fire are needed for sustainable fire management.11 Box 1 – Estimated health burden attributable to elevated PM2.5 concentrations, Western Australia, 2002–2017, by particulate matter source Outcome Estimated number of cases (95% confidence interval) Prescribed burns Wildfires Other Total Excess deaths (any cause) 21 (8–35) 17 (6–28) 3 (1–5) 41 (15–68) Hospital admissions, cardiovascular 51 (10–94) 41 (8–75) 7 (1–13) 99 (19–182) Hospital admissions, respiratory 89 (0–192) 72 (0–154) 13 (0–27) 174 (0–373) Emergency department attendances, asthma 63 (36–91) 51 (29–75) 9 (5–13) 123 (70–179) Box 2 – Estimated health costs tributable to elevated PM2.5 concentrations, Western Australia, 2002–2017, by particulate matter source

Nicolas Borchers Arriagada · Andrew J Palmer · David MJS Bowman · Fay H Johnston

The conventional treatment for brain metastases is whole brain radiotherapy (WBRT).1 But there has been a gradual move to managing limited brain metastases with stereotactic radiosurgery (SRS),2 and delaying or avoiding WBRT because of its effects on cognition and quality of life. Data on contemporary SRS practice for managing brain metastases in Australia are, however, very limited.3 We performed a population‐based linkage study, analysing data from the Victorian Cancer Registry and the Victorian Radiotherapy Minimum Data Set (VRMDS). We included all patients with solid tumours (ICD‐10 codes C00–C80), but excluding primary central nervous systems malignancies (ICD‐10 codes C69–72), who received brain radiotherapy in Victoria between 1 January 2012 and 31 December 2017. The primary outcome was the proportion of patients treated with SRS. Although SRS refers to large single fraction radiotherapy, patients treated with fractionated “stereotactic radiotherapy” were also classified as receiving SRS. In addition, because of potential coding inconsistencies, patients who had no more than four fractions of radiotherapy and were treated with “volumetric modulated arc therapy” or “intensity modulated radiation therapy” were also classified as receiving SRS. Differences in factors of interest by SRS use were assessed in Pearson χ2 (categorical variables) and Student t or Mann–Whitney U tests (continuous variables). Temporal changes were assessed with the Cochran–Armitage test for trend. Factors associated with SRS use were assessed by logistic regression, with year as an ordinal categorical variable; variables for which P < 0.10 in univariate analyses were included in the multivariate model. The study was approved by the Austin Health Human Research Ethics Committee (reference, LNR/18/Austin/34). A total of 3961 patients who received radiotherapy for brain metastases were included, of whom 1116 (28%) received SRS. The proportion of patients receiving SRS increased from 27% (105 of 388) in 2012 to 35% (287 of 821) in 2017 (for trend: P < 0.001). The mean age of patients who received SRS (63.5 years; standard deviation [SD], 12.5 years) was lower than for those who did not (65.2 years; SD, 12.5 years). Factors that influenced SRS use included socio‐economic status, primary cancer type (about half the patients with melanoma received SRS, and about one‐quarter of patients with other cancer types), treatment institution type (public institutions, 31%; private institutions, 24%), and location (metropolitan centres, 34%; regional centres, 5%). Remoteness of patients’ area of residence was not a significant factor. In multivariate analyses, age, primary cancer type, treatment centre type, and location were significant factors for SRS use (Box). While the VRMDS captures all radiotherapy delivered in Victoria, it does not include data on patients’ performance status, numbers of brain metastases, the extent of extracranial disease, and other factors that would allow evaluation of the appropriateness of SRS for individual patients. Another limitation is potential misclassification of radiotherapy classified as “SRS”, as the VRMDS did not include data on radiotherapy dose. As evidence supporting the use of SRS for managing brain metastases grows, we would expect SRS rates to rise.6,7 While SRS was less frequently used in regional centres, patients living in regional areas were as likely to receive SRS as patients living in metropolitan areas. It is nevertheless important to ensure easy and convenient access to SRS services for all cancer patients in Victoria. Box – Baseline characteristics of 3961 patients who received radiotherapy for brain metastases, Victoria, 2012–2017 Stereotactic radiosurgery Multivariable analysis: odds ratio (95%CI) P Received Not received Number of patients 1116 (28%) 2845 (72%) Age at first treatment for brain metastases (years) < 55 266 (33%) 543 (67%) 1 55–59 157 (32%) 331 (68%) 1.11 (0.86–1.44) 0.42 60–64 161 (28%) 419 (72%) 0.89 (0.69–1.14) 0.35 65–69 177 (26%) 502 (74%) 0.85 (0.67–1.08) 0.19 70–74 153 (25%) 448 (75%) 0.88 (0.68–1.14) 0.33 75 or more 202 (25%) 602 (75%) 0.78 (0.62–0.99) 0.045 Mean (SD) 63.5 (12.5) 65.2 (12.5) — — Sex Men 528 (28%) 1373 (72%) — — Women 588 (29%) 1472 (71%) — — Primary cancer type Lung 419 (24%) 1344 (76%) 1 Breast 203 (28%) 512 (72%) 1.24 (1.00–1.53) 0.05 Melanoma 252 (47%) 277 (52%) 2.89 (2.32–3.59) < 0.001 Gastrointestinal 93 (28%) 235 (72%) 1.37 (1.03–1.80) 0.028 Genitourinary 73 (28%) 189 (72%) 1.33 (0.97–1.80) 0.07 Other 76 (21%) 288 (79%) 0.80 (0.60–1.06) 0.12 Socio‐economic status (quintile) 1st (most disadvantaged) 188 (24%) 612 (77%) 1 2nd 189 (27%) 501 (73%) 1.12 (0.87–1.44) 0.39 3rd 202 (26%) 572 (74%) 1.02 (0.79–1.30) 0.90 4th 220 (26%) 618 (74%) 0.90 (0.70–1.14) 0.38 5th (least disadvantaged) 317 (37%) 542 (63%) 1.19 (0.94–1.50) 0.14 Remoteness classification5 Major city 780 (29%) 1949 (71%) — — Inner regional 261 (26%) 732 (73%) — — Outer regional/remote/very remote 75 (31%) 164 (69%) — — Treatment institution type Public 744 (31%) 1656 (69%) 1 Private 372 (24%) 1189 (76%) 0.10 (0.07–0.14) < 0.001 Treatment institution location Metropolitan 1071 (34%) 2071 (66%) 1 Regional 45 (5%) 774 (95%) 0.58 (0.49–0.68) < 0.001 Year of first brain metastasis treatment 2012 105 (27%) 283 (73%) 1 2013 111 (25%) 342 (76%) 1.01 (0.72–1.41) 0.95 2014 147 (25%) 439 (75%) 0.86 (0.63–1.18) 0.35 2015 207 (25%) 633 (75%) 0.79 (0.59–1.06) 0.12 2016 259 (30%) 614 (70%) 1.10 (0.83–1.47) 0.50 2017 287 (35%) 534 (65%) 1.41 (1.06–1.88) 0.017 CI = confidence interval; SD = standard deviation. * Index of Relative Socio‐Economic Disadvantage.4

Wee Loon Ong · Therese Ming Jung Kang · Gishan Ratnayake · Morikatsu Wada · Jeremy Ruben · Sashendra Senthi · Roger L Milne · Jeremy L Millar · Farshad Foroudi

Herpes zoster ophthalmicus (HZO), a condition that affects the ophthalmic division of the trigeminal nerve, is caused by reactivation of latent varicella zoster virus;1,2 about 10% of people with varicella zoster infections experience HZO.1 Over the past decade, the number of emergency department presentations by people with herpes zoster in Australia has increased by 2–6% per year, and the number of people with herpes zoster managed in general practice has almost doubled.3 The purpose of our study was to develop a contemporary perspective of the clinical presentation, incidence of complications, and treatment practice for patients with HZO referred to an Australian tertiary eye hospital. We performed a retrospective audit of digital health records of the first 100 consecutive patients who presented to the Royal Victorian Eye and Ear Hospital (RVEEH) emergency department with HZO during July 2017 – July 2018. The investigation was approved by the Human Research Ethics Committee of the Hospital as a quality control project (reference, 18/1416HL). The clinical features at the time of presentation of the 100 patients are summarised in the Box. Sixty‐five patients initially presented to their general practitioner, 20 to a hospital emergency department, and 15 directly to the RVEEH. The mean time between rash onset and presentation to a GP or emergency department was 3.3 days (range, 0–14 days). For 51 patients, treatment commenced before presentation to the RVEEH (famciclovir, 27; valaciclovir, 16; acyclovir, 6; two patients had received no topical treatment); treatment had commenced within 72 hours of the rash developing for 36 of these patients (71%). The recommended dose and frequency were prescribed for 16 of the 51 patients: famciclovir (500 mg three times a day), two patients; valaciclovir (1 g three times a day), 12 patients; acyclovir (800 mg five times a day), two patients. For 29 patients, antiviral therapy was prescribed at lower than the recommended dose (famciclovir, 21 patients; valaciclovir, two patients; acyclovir, two patients) or prescribed as a topical treatment (acyclovir, two patients); the prescribing information was not documented for five patients. Nineteen of the 68 patients who attended follow‐up 7–14 days after their initial presentation to the RVEEH presented with ocular symptoms regarded as late complications, including four with more than one complication. Eight of 29 patients (29%) who had not commenced systemic antiviral therapy within 72 hours of rash onset developed late complications, as did 13 of 71 patients (18%) who were treated within 72 hours (Fisher exact test: P = 0.78). We found concerning variations in timing and practice of treating HZO, despite recognised clinical guidelines.4,5 This may be partly explained by diagnostic uncertainty caused by the variability of clinical signs during the early stages of HZO,6 and by an earlier discrepancy between the famciclovir dosing recommended by therapeutic guidelines (250 mg three times a day) and recommendations based upon the results of a clinical trial4 (500 mg three times a day). This discrepancy has since been resolved in the therapeutic guidelines.4 Our findings suggest that education of all health care professionals involved in the care of patients with HZO needs to be improved. Clinical practice guidelines must provide clear and consistent information about managing HZO. Box – Demographic characteristics and clinical features of 100 consecutive people presenting with herpes zoster ophthalmicus to the Royal Victorian Eye and Ear Hospital, July 2017 – July 2018 Characteristic Sex (men) 52 Age at presentation (years), median (IQR) 59 (39–76) Age at presentation (years), range 16–93 Clinical features at presentation Best‐corrected visual acuity ≥ 6/12 62 Intra‐ocular pressure (mmHg), mean (SD) 15.4 (5.9) Rash 92 Pain 63 Conjunctivitis 62 Lid swelling 53 Skin erythema 41 Anterior uveitis 26 Keratitis 20 Other* 6 Late complications 19 Uveitis 11 Keratitis 5 Other† 3 IQR = interquartile range; SD = standard deviation. * Raised intra‐ocular pressure, retinitis/choroiditis, optic neuritis, cranial nerve palsy. † Neuralgia, elevated intra‐ocular pressure.

Rahul Chakrabarti · Grace George · Kristen Wells · Carmel Crock

Primary care is the single most significant contributor to positive health outcomes,1,2 but the number of general practitioners in Australia has been falling, a situation previously described for nations with poorer health outcomes.2 The reasons for the decline are many,3 but this phenomenon has not been described in detail in the peer‐reviewed literature. We have therefore examined the registration categories, as recorded by the Australian Health Practitioner Regulation Agency (AHPRA), of people who graduated from the University of Western Australia (UWA) medical school during 1985–2007. Our study was approved by the UWA Human Research Ethics Committee (reference, RA 4/1/1627). We included all active medical practitioners who graduated (MBBS) from UWA during 1985–2007 and were registered with AHPRA in December 2019. We included all doctors listed by AHPRA as practising GPs, whether vocationally registered or with college membership, in our GP category; 65 of the 93 1985–1987 graduates registered by AHPRA as GPs did not have postgraduate qualifications, but only five of the 56 2004–2007 graduates. AHPRA registration as a GP was about half as likely for 2004–2007 graduates as for 1985–1987 graduates (relative risk [RR], 0.46; 95% confidence interval, 0.35–0.60). This decline in entry into general practice was accompanied by an increase in the proportion of graduates with general registration status alone (in 2004–2007 v 1985–1987: RR, 3.01; 95% CI, 1.97–4.61) (Box). These findings are consistent with the recently reported drop in the proportion of medical graduates who intend to enter general practice,5 which may lead to a further decline in the number of AHPRA GP registrations. We also found an equally concerning increase in the number of doctors practising as generally registered practitioners alone during 2007–2019, presumably waiting for the opportunity to enter their preferred medical specialty. This problem, first discussed without data in the MJA in 2012,6 has not attracted the attention of policymakers. The reduction in size of the primary care workforce is felt most keenly in rural communities, where dependence on primary health care is more pronounced, but urban practices also struggle to recruit new fellows.2 The causes of this problem include the perceived lower status of general practice, the generally lower income provided by Medicare fees, the burden of practice accreditation, and specialist‐focused teaching in medical schools. Further challenges for rural practice include problems of work–life balance and the focus on fly‐in/fly‐out specialist services instead of developing extended scope primary care models for regional and remote communities. One limitation of our study is that the AHPRA general registration data need to be compared with college registries for accuracy. Our findings nevertheless suggest that solutions for the general practice problem are urgently needed if Australia is to continue enjoying some of the best health outcomes in the world. Box – Category of registration for University of Western Australia medical graduates, 1985–2007, as recorded by the Australian Health Practitioner Regulation Agency (AHPRA)* * Data as at 28 November 2019; data shown for categories with at least 25 graduates during 1985–2007. Total number of practising graduates registered with AHPRA, by graduation year: 1985–1987: 237; 1988–1991: 355; 1992–1995: 355; 1996–1999: 370; 2000–2003: 426; 2004–2007: 419. † Includes all medical graduates who have completed an accredited internship in Australia or New Zealand and are not required to complete any additional supervised practice to become eligible for general registration; those who completed a recognised internship elsewhere and have additionally satisfactorily completed 47 weeks of full‐time approved supervised practice in Australia; those who have previously held general registration in Australia; those who have completed the competent authority pathway; and Australian Medical Council certificate holders in the standard pathway who have completed a period of approved supervised practice in Australia.4

Denese Playford · Jennifer A May · Hanh Ngo · Ian B Puddey

Objectives: To evaluate the impact of the Victorian Stroke Telemedicine (VST) program during its first 12 months on the quality of care provided to patients presenting with suspected stroke to hospitals in regional Victoria. Design: Historical controlled cohort study comparing outcomes during a 12‐month control period with those for the initial 12 months of full implementation of the VST program at each hospital. Setting: 16 hospitals in regional Victoria that participated in the VST program between 1 January 2010 and 30 January 2016. Participants: Adult patients with suspected stroke presenting to the emergency departments of the participating hospitals. Main outcome measures: Indicators for key processes of care, including symptom onset‐to‐arrival, door‐to‐first medical review, and door‐to‐CT times; provision and timeliness of provision of thrombolysis to patients with ischaemic stroke. Results: 2887 patients with suspected stroke presented to participating emergency departments during the control period, 3178 during the intervention period; the patient characteristics were similar for both periods. A slightly larger proportion of patients with ischaemic stroke who arrived within 4.5 hours of symptom onset received thrombolysis during the intervention than during the control period (37% v 30%). Door‐to‐CT scan time (median, 25 min [IQR, 13–49 min] v 34 min [IQR, 18–76 min]) and door‐to‐needle time for stroke thrombolysis (73 min [IQR, 56–96 min] v 102 min [IQR, 77–128 min]) were shorter during the intervention. The proportions of patients who received thrombolysis and had a symptomatic intracerebral haemorrhage (4% v 16%) or died in hospital (6% v 20%) were smaller during the intervention period. Conclusions: Telemedicine has provided Victorian regional hospitals access to expert care for emergency department patients with suspected acute stroke. Eligible patients with ischaemic stroke are now receiving stroke thrombolysis more quickly and safely.

Chris F Bladin · Joosup Kim · Kathleen L Bagot · Michelle Vu · Natasha Moloczij · Sonia Denisenko · Chris Price · Nancy Pompeani · Lauren Arthurson · Casey Hair · Justin Rabl · Mick O'Shea · Patrick Groot · Leslie Bolitho · Bruce CV Campbell · Helen M Dewey · Geoffrey A Donnan · Dominique A Cadilhac

Objectives: To describe the first isolation and sequencing of SARS‐CoV‐2 in Australia and rapid sharing of the isolate. Setting: SARS‐CoV‐2 was isolated from a 58‐year‐old man from Wuhan, China who arrived in Melbourne on 19 January 2020 and was admitted to the Monash Medical Centre, Melbourne from the emergency department on 24 January 2020 with fever, cough, and progressive dyspnoea. Major outcomes: Clinical course and laboratory features of the first reported case of COVID‐19 (the illness caused by SARS‐CoV‐2) in Australia; isolation, whole genome sequencing, imaging, and rapid sharing of virus from the patient. Results: A nasopharyngeal swab and sputum collected when the patient presented to hospital were each positive for SARS‐CoV‐2 (reverse transcription polymerase chain reaction). Inoculation of Vero/hSLAM cells with material from the nasopharyngeal swab led to the isolation of SARS‐CoV‐2 virus in culture. Electron microscopy of the supernatant confirmed the presence of virus particles with morphology characteristic of viruses of the family Coronaviridae. Whole genome sequencing of the viral isolate and phylogenetic analysis indicated the isolate exhibited greater than 99.99% sequence identity with other publicly available SARS‐CoV‐2 genomes. Within 24 hours of isolation, the first Australian SARS‐CoV‐2 isolate was shared with local and overseas reference laboratories and major North American and European culture collections. Conclusions: The ability to rapidly identify, propagate, and internationally share our SARS‐CoV‐2 isolate is an important step in collaborative scientific efforts to deal effectively with this international public health emergency by developing better diagnostic procedures, vaccine candidates, and antiviral agents.

Leon Caly · Julian Druce · Jason Roberts · Katherine Bond · Thomas Tran · Renata Kostecki · Yano Yoga · William Naughton · George Taiaroa · Torsten Seemann · Mark B Schultz · Benjamin P Howden · Tony M Korman · Sharon R Lewin · Deborah A Williamson · Mike G Catton

Objectives: To assess the efficacy of corticosteroid treatment of patients with coronavirus disease 2019 (COVID‐19). Design, setting: Observational study in the two COVID‐19‐designated hospitals in Wuhu, Anhui province, China, 24 January – 24 February 2020. Participants: Thirty‐one patients infected with the severe acute respiratory coronavirus 2 (SARS‐CoV‐2) treated at the two designated hospitals. Main outcome measures: Virus clearance time, length of hospital stay, and duration of symptoms, by treatment type (including or not including corticosteroid therapy). Results: Eleven of 31 patients with COVID‐19 received corticosteroid treatment. Cox proportional hazards regression analysis indicated no association between corticosteroid treatment and virus clearance time (hazard ratio [HR], 1.26; 95% CI, 0.58–2.74), hospital length of stay (HR, 0.77; 95% CI, 0.33–1.78), or duration of symptoms (HR, 0.86; 95% CI, 0.40–1.83). Univariate analysis indicated that virus clearance was slower in two patients with chronic hepatitis B infections (mean difference, 10.6 days; 95% CI, 6.2–15.1 days). Conclusions: Corticosteroids are widely used when treating patients with COVID‐19, but we found no association between therapy and outcomes in patients without acute respiratory distress syndrome. An existing HBV infection may delay SARS‐CoV‐2 clearance, and this association should be further investigated.

Lei Zha · Shirong Li · Lingling Pan · Boris Tefsen · Yeshan Li · Neil French · Liyun Chen · Gang Yang · Elmer V Villanueva

The potential of modern mass communication techniques for managing chronic disease should be explored further

Yang Si · Xiaoqiang Xiao · Cai Xia · Jiang Guo · Qiukui Hao · Qianning Mo · Yulong Niu · Hongbin Sun

Objective: To investigate changes in sales to pharmacies of over‐the‐counter (OTC) and prescription analgesics, cold and flu products, and cough suppressants after the rescheduling of codeine as a prescription only medicine in February 2018. Design: Interrupted time series analysis of sales to pharmacies. Setting: Pharmaceutical sales to community pharmacies in Australia, March 2015 – March 2019. The period January 2017 (month after rescheduling was announced) to January 2018 (month before rescheduling was implemented) was excluded from the time series analysis. Main outcome measures: Monthly pack and tablet sales per 10 000 population of OTC and prescription analgesics, cold and flu products, and cough suppressants. Results: During 2016, 7586 packs and 248 127 tablets of OTC codeine per 10 000 population were sold to pharmacies; in the 14 months after rescheduling, a small level increase in monthly prescription codeine sales was evident (2247 tablets/capsules per 10 000 population; 95% CI, 1231–3264 per 10 000 population). Monthly OTC analgesic sales increased by 258 (95% CI, 151–365) packs per 10 000 population and 37 856 (95% CI, 26 143–49 569) tablet/capsules per 10 000 population. Monthly sales of single ingredient paracetamol (41 415 [95% CI, 31 374–51 456] tablets/capsules per 10 000 population), ibuprofen (1392 [95% CI 916–1868] tablets/capsules per 10 000 population), paracetamol/ibuprofen (1618 tablets [95% CI, 1567–1669] tablets/capsules per 10 000 population), and other paracetamol combinations (233 [95% CI, 112–353] tablets/capsules per 10 000 population) all increased, but not those of prescription analgesic products not containing codeine. Rises for OTC cold/flu products containing the opioid derivative dextromethorphan were small; sales of OTC cough suppressants containing opioid derivatives (dextromethorphan, pholcodine, dihydrocodeine) did not change. Conclusions: The rescheduling of codeine was followed by increased sales to pharmacies of paracetamol, ibuprofen, and paracetamol combination products. While these products carry no risk of dependence, their inappropriate use is also associated with harms that warrant adverse event monitoring.

Andrea L Schaffer · Rose Cairns · Jared A Brown · Natasa Gisev · Nicholas A Buckley · Sallie‐Anne Pearson

Objectives: To evaluate whether an alternative model of care in aged care facilities, including in‐house general practitioners, influenced health outcomes for residents. Design: Stepped wedge, cluster randomised controlled trial over 90 weeks (31 December 2012 – 21 September 2014), with a 54‐week pre‐trial retrospective data period (start: 19 December 2011) and a 54‐week post‐trial prospective data collection period (to 4 October 2015). Participants, setting: Fifteen residential aged care facilities operated by Bupa Aged Care in metropolitan and regional cities in four Australian states. Intervention: Residential aged care facilities sought to recruit general practitioners as staff members; care staff roles were redefined to allow registered nurses greater involvement in care plan development. Main (primary) outcome measures: Numbers of falls; numbers of unplanned transfers to hospital; polypharmacy. Results: The new model of care could be implemented in all facilities, but four could not recruit in‐house GPs at any time during the trial period. Intention‐to‐treat analyses found no statistically significant effect of the intervention on the primary outcome measures. Contamination‐adjusted intention‐to‐treat analyses identified that the presence of an in‐house GP was associated with reductions in the numbers of unplanned hospital transfers (incidence rate ratio [IRR], 0.53; 95% CI, 0.43–0.66) and admissions (IRR, 0.52; 95% CI, 0.41–0.64) and of out‐of‐hours GP call‐outs (IRR, 0.54; 95% CI, 0.36–0.80), but also with an increase in the number of reported falls (IRR, 1.37; 95% CI, 1.20–1.58). Conclusions: Recruiting GPs to work directly in residential aged care facilities is difficult, but may reduce the burden of unplanned presentations to hospitals and increase the reporting of adverse events. Trial registration: Australia New Zealand Clinical Trial Registry, ACTRN12613000218796 (25 February 2013).

Terry P Haines · Andrew J Palmer · Petra Tierney · Lei Si · Andrew L Robinson

Objectives: To assess progress in Australia toward the 2030 WHO hepatitis C elimination targets two years after the introduction of highly effective direct‐acting antiviral (DAA) treatments. Design: Analysis of quarterly data on government‐subsidised hepatitis C RNA testing and hepatitis C treatment in Australia, January 2013 – June 2018. Changes in testing and treatment levels associated with DAA availability were assessed in an autoregressive integrated moving average (ARIMA) statistical model, and the impact by 2030 of different levels of testing and treatment were estimated using a mathematical model. Major outcome measures: Hepatitis C prevalence among people who inject drugs; annual hepatitis C incidence relative to 2015 levels; projections for the hepatitis C care cascade in 2030. Results: The mean annual number of treatments initiated for people with hepatitis C increased from 6747 during 2013–2015 (before the introduction of DAAs) to 28 022 during 2016–18; the mean annual number of diagnostic RNA tests increased from 17 385 to 23 819. If current trends in testing and treatment continue (ie, 2018 testing numbers are maintained but treatment numbers decline by 50%), it is projected that by 2030 only 72% of infected people would be treated (by 2025 all people diagnosed with hepatitis C would be treated). The incidence of hepatitis C in 2030 would be 59% lower than in 2015, well short of the WHO target of an 80% reduction. The identification and testing of people exposed to hepatitis C must be increased by at least 50% for Australia to reach the WHO elimination targets. Conclusion: Hepatitis C elimination programs in Australia should focus on increasing testing rates and linkage with care to maintain adequate levels of treatment.

Nick Scott · Rachel Sacks‐Davis · Amanda J Wade · Mark Stoove · Alisa Pedrana · Joseph S Doyle · Alexander J Thompson · David P Wilson · Margaret E Hellard

Objectives: To characterise the community pharmacy supply of naloxone by supply type — individual prescription, prescriber bag, and non‐dispensed (supplied over the counter or expired) — during 2014–2018; to examine whether the 2016 rescheduling of naloxone as an over‐the‐counter drug influenced non‐dispensed naloxone supply volume. Design, setting: Analysis of monthly naloxone prescriptions (Pharmaceutical Benefits Scheme) and sales data (IQVIA), 2014–2018, for Australia and by state and territory; time series analysis of non‐dispensed naloxone supply to assess effect of rescheduling on naloxone supply. Major outcomes: Total naloxone supply to community pharmacies; prescribed and non‐dispensed naloxone supply. Results: During 2014–2018, 372 351 400 μg units of naloxone were sold to community pharmacies: non‐dispensed naloxone accounted for 205 866.5 units (55.3%), prescriber bags for 155 841 units (41.8%), and individual prescriptions for 10 643.5 units (2.9%). Population‐adjusted national naloxone sales to community pharmacies increased between 2014 and 2018 (per year: incidence rate ratio [IRR], 1.15; 95% CI, 1.09–2.22). This increase was primarily attributable to increased volumes of prescriber bag naloxone (IRR, 1.63; 95% CI, 1.50–1.78) and, to a lesser extent, increased individual prescription supply (IRR, 2.04; 95% CI, 1.85–2.26). Non‐dispensed naloxone supply volume was unchanged at the national level (IRR, 0.93; 95% CI, 0.85–1.01); changes in non‐dispensed supply immediately following rescheduling and subsequently were not statistically significant in time series analyses for most jurisdictions. Conclusions: Total naloxone supply to community pharmacies in Australia increased between 2014 and 2018, but rescheduling that enabled over‐the‐counter access did not significantly influence the volume of non‐dispensed naloxone.

Wai Chung Tse · Paul Sanfilippo · Tina Lam · Paul Dietze · Suzanne Nielsen

Weather conditions conducive to extreme bushfires are becoming more frequent as a consequence of climate change.1 Such fires have substantial social, ecological, and economic effects, including the effects on public health associated with smoke, such as premature mortality and exacerbation of cardio‐respiratory conditions.2,3 During the final quarter of 2019 and the first of 2020, bushfires burned in many forested regions of Australia, and smoke affected large numbers of people in New South Wales, Queensland, the Australian Capital Territory and Victoria. The scale and duration of these bushfires was unprecedented in Australia. We undertook a preliminary evaluation of the health burden attributable to air pollution generated by bushfires during this period. Using standard methods for assessing the health impact of air pollution,4 we estimated the numbers of excess deaths, hospitalisations for cardiovascular and respiratory problems, and emergency department presentations with asthma in NSW, Queensland, the ACT and Victoria between 1 October 2019 and 10 February 2020 that could be attributed to bushfire smoke exposure. We estimated population exposure to particulate matter less than 2.5 μm in diameter (PM2.5) for the regions of NSW, Queensland, the ACT and Victoria for which publicly available air quality monitoring data were available (for about 90% of the total population of these states). Data were obtained from the NSW Department of Planning, Industry and Environment,5 the Queensland Department of Science,6 ACT Health,7 and the Environmental Protection Agency Victoria.8 We defined bushfire smoke‐affected days as days on which the 24‐hour mean PM2.5 concentration exceeded the 95th percentile of historical daily mean values for individual air quality stations. We estimated daily mean PM2.5 levels by Statistical Area Level 2 (SA2), using station level data whenever at least one monitoring station was within 100 km of the SA2 centroid, and applying inverse distance weighting.9 Published population and health data from the Australian Bureau of Statistics,10,11 the Australian Institute of Health and Welfare,12,13,14,15 and the NSW Ministry of Health were used.16 We quantified health outcomes by combining baseline incidence rates12,13,14,15 for each health outcome with daily exposure data and applying the relevant exposure–response risk coefficients for each outcome.17,18 We also conducted sensitivity analyses with different PM2.5 thresholds for defining bushfire smoke‐affected days. Further methodological details, including underlying assumptions and limitations, are included in the online Supporting Information. Our analysis of publicly available aggregated data did not require ethics approval. During the study period, PM2.5 concentrations exceeding the 95th percentile of historical daily mean values were recorded by at least one monitoring station in the study area on 125 of 133 days (Box 1). We estimated that bushfire smoke was responsible for 417 (95% CI, 153–680) excess deaths, 1124 (95% CI, 211–2047) hospitalisations for cardiovascular problems and 2027 (95% CI, 0–4252) for respiratory problems, and 1305 (95% CI, 705–1908) presentations to emergency departments with asthma (Box 2). Applying lower thresholds for defining bushfire smoke‐affected days (no threshold, 90th percentile of historical values) did not markedly alter our findings; a higher threshold (99th percentile) reduced the estimates by about 20%. The highest population‐weighted PM2.5 exposure level, 98.5 μg/m3 on 14 January 2020 (Box 1), exceeded the national air quality 24‐hour standard (25 μg/m3)19 and was more than fourteen times the historical population‐weighted mean 24‐hour PM2.5 value of 6.8 μg/m3. We have estimated the excess health burden during 19 weeks’ continuous fire activity in the states most severely affected by smoke. Our estimates are based on air quality data from monitoring stations in the four eastern states — that is, we did not include data for smoke from all extreme fires in Australia during the study period — and we did not attempt to estimate health effects for which exposure–response relationships are less well characterised, such as primary health care attendances and ambulance calls. Detailed epidemiological analysis of more comprehensive exposure estimation and empirical health data will provide more complete information about the harms attributable to the severe air pollution associated with these unprecedented fires, but our findings indicate that the smoke‐related health impact was substantial. Smoke is just one of many problems that will intensify with the increasing frequency and severity of major bushfires associated with climate change. Expanded and diversified approaches to bushfire mitigation and adaptation to living in an increasingly hot and fire‐prone country are urgently needed.20 Box 1 – Population‐weighted PM2.5 levels, New South Wales, Queensland, the Australian Capital Territory and Victoria, 1 October 2019 – 10 February 2020* * Data by state are included in the online Supporting Information. Box 2 – Estimated health burden attributable to bushfire smoke, Queensland, New South Wales, the Australian Capital Territory and Victoria, 1 October 2019 – 10 February 2020 Outcome Estimated number of cases (95% confidence intervals) Queensland New South Wales Australian Capital Territory Victoria Total Excess deaths (any cause) 47 (17–77) 219 (81–357) 31 (12–51) 120 (44–195) 417 (153–680) Hospital admissions, cardiovascular 135 (25–246) 577 (108–1050) 82 (15–149) 331 (62–602) 1124 (211–2047) Hospital admissions, respiratory 245 (0–513) 1050 (0–2204) 147 (0–308) 585 (0–1227) 2027 (0–4252) Emergency department attendances, asthma 113 (61–165) 702 (379–1026) 89 (48–131) 401 (217–586) 1305 (705–1908)

Nicolas Borchers Arriagada · Andrew J Palmer · David MJS Bowman · Geoffrey G Morgan · Bin B Jalaludin · Fay H Johnston