Critical Care and Resuscitation • Volume 20 Number 4 • December 2018 ORIGINAL ARTICLES 294 Intensive care implications of epidemic thunderstorm asthma Jai N Darvall, Matthew Durie, David Pilcher, Geoffrey Wigmore, Craig French, Dharshi Karalapillai, Forbes McGain, Edward Newbigin, Timothy Byrne, Vineet Sarode, Ben Gelbart, Andrew Casamento, John Dyett, Ashley Crosswell, Joseph Vetro, Joseph McCaffrey, Gopal Taori, Ashwin Subramaniam, Christopher MacIsaac, Anthony Cross, David Ku and Rinaldo Bellomo Thunderstorm asthma is a rare and incompletely understood phenomenon characterised by bronchospasm in association with thunderstorms. A unique confluence of meteorological conditions generates aeroallergens, which are liberated by storms (commonly rye grass pollen, with sporulating fungi also implicated). 1,2 Thunderstorm asthma is described worldwide, with reports from Canada, Australia, the Unites States, Greece, the United Kingdom, Italy and Iran. 2 Despite the potential to affect thousands of patients simultaneously, only one prior death due to previous events has been reported. 3 There is thus little prior description of the intensive care implications, and the demographics, incidence and disease outcomes of affected patients. Whether patients differ from other critically ill asthma admissions is also not understood. The evening of 21 November, 2016, in Victoria, Australia, saw the most severe outbreak of thunderstorm asthma in recorded history. Over 3400 affected patients were seen in public hospital emergency departments (EDs), overwhelming ambulance services and causing activation of hospital disaster plans. 4 Nine deaths are thought to have eventuated, with many additional patients requiring intensive care. 5 Having previously described the multidisciplinary aspects, 6 our aim is to describe in detail the intensive care implications of this largest and most lethal thunderstorm asthma event. Understanding the characteristics, resource implications, and outcome of affected patients may help inform risk stratification, disaster planning and resource allocation for future events. Methods Study design and population We performed a retrospective multicentre observational study using data from individual intensive care units (ICUs), the Australian and New Zealand Intensive Care Society (ANZICS) Adult Patient Database (APD), the Bureau of Meteorology (weather data), Environmental Protection Authority Victoria (airborne particulate data) and the Department of Botany, University of Melbourne (pollen data). Ethics approval was obtained from the Melbourne Health Human Research Ethics Committee (LNR/16/MH/406). All public hospital ABSTRACT Objective: To investigate the environmental precipitants, treatment and outcome of critically ill patients affected by the largest and most lethal reported epidemic of thunderstorm asthma. Design, setting and participants: Retrospective multicentre observational study. Meteorological, airborne particulate and pollen data, and a case series of 35 patients admitted to 15 intensive care units (ICUs) due to the thunderstorm asthma event of 21–22 November 2016, in Victoria, Australia, were analysed and compared with 1062 total ICU-admitted Australian patients with asthma in 2016. Main outcome measures: Characteristics and outcomes of total ICU versus patients with thunderstorm asthma, the association between airborne particulate counts and storm arrival, and ICU resource utilisation. Results: All 35 patients had an asthma diagnosis; 13 (37%) had a cardiac or respiratory arrest, five (14%) died. Compared with total Australian ICU-admitted patients with asthma in 2016, patients with thunderstorm asthma had a higher mortality (15% v 1.3%, P < 0.001), were more likely to be male (63% v 34%, P < 0.001), to be mechanically ventilated, and had shorter ICU length of stay in survivors (median, 31.8 hours [interquartile range (IQR), 14.8–43.6 hours] v 40.7 hours [IQR, 22.3–75.1 hours]; P = 0.025). Patients with cardiac arrest were more likely to be born in Asian or subcontinental countries (5/10 [50%] v 4/25 [16%]; relative risk, 3.13; 95% CI, 1.05–9.31). A temporal link was demonstrated between airborne particulate counts and arrival of the storm. The event used 15% of the public ICU beds in the region. Conclusion: Arrival of a triggering storm is associated with an increase in respirable airborne particles. Affected critically ill patients are young, have a high mortality, a short duration of bronchospasm, and a prior diagnosis of asthma is common. Crit Care Resusc 2018; 20 (4): 294-303
Transcript
Critical Care and Resuscitation • Volume 20 Number 4 • December 2018
ORIGINAL ARTICLES
294
Intensive care implications of epidemic thunderstorm asthma
Jai N Darvall, Matthew Durie, David Pilcher, Geoffrey Wigmore, Craig French, Dharshi Karalapillai, Forbes McGain, Edward Newbigin, Timothy Byrne, Vineet Sarode, Ben Gelbart, Andrew Casamento, John Dyett,
Ashley Crosswell, Joseph Vetro, Joseph McCaffrey, Gopal Taori, Ashwin Subramaniam, Christopher MacIsaac, Anthony Cross, David Ku and Rinaldo Bellomo
Thunderstorm asthma is a rare and incompletely understood phenomenon characterised by bronchospasm in association with thunderstorms. A unique confluence of meteorological conditions generates aeroallergens, which are liberated by storms (commonly rye grass pollen, with sporulating fungi also implicated).1,2 Thunderstorm asthma is described worldwide, with reports from Canada, Australia, the Unites States, Greece, the United Kingdom, Italy and Iran.2 Despite the potential to affect thousands of patients simultaneously, only one prior death due to previous events has been reported.3 There is thus little prior description of the intensive care implications, and the demographics, incidence and disease outcomes of affected patients. Whether patients differ from other critically ill asthma admissions is also not understood.
The evening of 21 November, 2016, in Victoria, Australia, saw the most severe outbreak of thunderstorm asthma in recorded history. Over 3400 affected patients were seen in public hospital emergency departments (EDs), overwhelming ambulance services and causing activation of hospital disaster plans.4 Nine deaths are thought to have eventuated, with many additional patients requiring intensive care.5
Having previously described the multidisciplinary aspects,6 our aim is to describe in detail the intensive care implications of this largest and most lethal thunderstorm asthma event. Understanding the characteristics, resource implications, and outcome of affected patients may help inform risk stratification, disaster planning and resource allocation for future events.
Methods
Study design and population
We performed a retrospective multicentre observational study using data from individual intensive care units (ICUs), the Australian and New Zealand Intensive Care Society (ANZICS) Adult Patient Database (APD), the Bureau of Meteorology (weather data), Environmental Protection Authority Victoria (airborne particulate data) and the Department of Botany, University of Melbourne (pollen data). Ethics approval was obtained from the Melbourne Health Human Research Ethics Committee (LNR/16/MH/406). All public hospital
ABSTRACT
Objective: To investigate the environmental precipitants, treatment and outcome of critically ill patients affected by the largest and most lethal reported epidemic of thunderstorm asthma.Design, setting and participants: Retrospective multicentre observational study. Meteorological, airborne particulate and pollen data, and a case series of 35 patients admitted to 15 intensive care units (ICUs) due to the thunderstorm asthma event of 21–22 November 2016, in Victoria, Australia, were analysed and compared with 1062 total ICU-admitted Australian patients with asthma in 2016.Main outcome measures: Characteristics and outcomes of total ICU versus patients with thunderstorm asthma, the association between airborne particulate counts and storm arrival, and ICU resource utilisation.Results: All 35 patients had an asthma diagnosis; 13 (37%) had a cardiac or respiratory arrest, five (14%) died. Compared with total Australian ICU-admitted patients with asthma in 2016, patients with thunderstorm asthma had a higher mortality (15% v 1.3%, P < 0.001), were more likely to be male (63% v 34%, P < 0.001), to be mechanically ventilated, and had shorter ICU length of stay in survivors (median, 31.8 hours [interquartile range (IQR), 14.8–43.6 hours] v 40.7 hours [IQR, 22.3–75.1 hours]; P = 0.025). Patients with cardiac arrest were more likely to be born in Asian or subcontinental countries (5/10 [50%] v 4/25 [16%]; relative risk, 3.13; 95% CI, 1.05–9.31). A temporal link was demonstrated between airborne particulate counts and arrival of the storm. The event used 15% of the public ICU beds in the region.Conclusion: Arrival of a triggering storm is associated with an increase in respirable airborne particles. Affected critically ill patients are young, have a high mortality, a short duration of bronchospasm, and a prior diagnosis of asthma is common.
Crit Care Resusc 2018; 20 (4): 294-303
Critical Care and Resuscitation • Volume 20 Number 4 • December 2018
ORIGINAL ARTICLES
295
ICUs in the state of Victoria were contacted, as well as private ICUs attached to a hospital with an ED. The patients included had an admission diagnosis of asthma (APACHE [Acute Physiology and Chronic Health Evaluation] III-j code 209, Australian and New Zealand Paediatric Intensive Care code 431), or an admission due to acute bronchospasm but with alternative coding (eg, cardiac or respiratory arrest). An ICU specialist at each site examined all admissions on 21–22 November to identify patients meeting the case definition. Asthma admissions for the whole of 2016 were extracted from the ANZICS APD.
Study variables
Weather, airborne pollen and particulate data (particles ≤ 2.5 µm [PM2.5] and ≤ 10 µm [PM10] in size) was aggregated from geographical sites around Victoria. We collected demographic characteristics (age, height, weight, gender, country of birth), chronic disease status as defined by the ANZICS APD, and prior comorbid respiratory disease (asthma, chronic obstructive pulmonary disease, allergic rhinitis or known pollen or rye grass allergy). We recorded pre-admission respiratory medications (inhaled bronchodilator, corticosteroid or other preventer, and antihistamines). Complications incurred either before or during hospitalisation were recorded (pneumothorax, pneumonia, acute respiratory distress syndrome, other organ failure, cardiac or respiratory arrest). In addition,
daily ICU data were collected including mode of ventilation, airway pressures, arterial blood gas analysis, and specific asthma therapies.
Our primary aim was to examine whether the characteristics, treatment course and outcomes of critically ill patients with thunderstorm asthma differ from other ICU patients admitted with asthma in Australia. Secondary aims were to describe the regional ICU-resource implications and the association between meteorological phenomena and airborne particulate and pollen counts. Outcomes of interest were change in humidity and airborne particle counts at different time points along the storm front, ICU bed utilisation, duration of mechanical ventilation, asthma therapies administered, ICU and hospital length of stay (LOS) and mortality.
Statistical analysis
All statistical analyses were performed using Stata 14.1 (StataCorp, College Station, TX, USA). Normally distributed data were summarised using the mean (standard deviation [SD]) and compared using Student unpaired two-tailed t tests. Skewed data were summarised using the median (interquartile range [IQR]) and compared using rank sum tests. Categorical data were summarised using the number (%) and compared using the c2 test or Fisher’s exact test, where applicable. A value of P < 0.05 was considered statistically significant.
Figure 1. Pollen counts and Victorian intensive care unit admissions from emergency departments with respiratory and asthma diagnostic codes, 16–27 November 2016
ED = emergency department. ICU = intensive care unit.
Critical Care and Resuscitation • Volume 20 Number 4 • December 2018
ORIGINAL ARTICLES
296
Figure 2. Greater Melbourne, 21 November 2016: location and timing of storm front arrival (purple line), weather and air monitoring stations (grouped as A–E), relationship between rate of change in relative humidity (∆ RH) and atmospheric particle pollution (< 10 µm, PM10) at five sites (A–E, westward-eastward) and affected intensive care units (ICUs) with admitted patient numbers
Results
Weather and pollen data
The 21 November 2016 followed a week of dry weather and occasional high grass pollen count days (daily average ≥ 50 grains/m3) (Figure 1). Temperatures rose to around 35°C (95°F), with northerly winds of 20–30 km/h and humidity falling to 20–30%. Between 17:00 and 18:30, a north-south line of thunderstorms passed from west to
east over the cities of Geelong and Melbourne, Victoria, reaching Melbourne’s western suburbs around 17:40 and outer eastern suburbs after 18:30 (Figure 2). A sharp fall in temperature and rise in relative humidity to 80–90% was observed with the storm front, coinciding with a spike in atmospheric PM10 levels (particles ≤ 10 µm) at Melbourne air quality monitoring stations (Figure 1). Total rainfall was minimal (≤ 1 mm). Airborne grass pollen levels were
Critical Care and Resuscitation • Volume 20 Number 4 • December 2018
ORIGINAL ARTICLES
297
extreme (daily average, ≥ 100 grains/m3); other pollen types and fungal spores were also present (Figure 1 and Figure 3). Although PM10 counts were high (> 97th percentile for November days), these levels had been exceeded on at least four occasions within the last five Novembers. Similarly, although grass pollen counts were in the extreme range on 21 November, between 31 October 2016 and 31 December 2016, there were 6 days with average grass pollen count readings ≥ 100 grains/m3.
Patient characteristics
Compared with the 3-yearly average, on 21–22 November, there was a 450% increase (3460 patients) of respiratory presentations to public hospital EDs in the region.5 Compared with the Monday and Tuesday average, there were 210 additional private hospital respiratory presentations.4
As previously described,6 a total of 35 patients (33 adults and two children, aged 9 and 12 years) with thunderstorm asthma were admitted to 15 ICUs (14 public and one private hospital ICU) (Figure 2). Baseline demographics and prior respiratory disease status are summarised in Table 1. Median age was 39 years (IQR, 9–69 years); 22 patients (63%) were male. Nineteen patients (54%) were non-Australian born, including 10 (29%) born in Asian or subcontinental countries. All patients had a prior diagnosis of asthma. Pre-existing respiratory medications are listed in Table 1. Only 12 patients (34%) were taking an inhaled corticosteroid preventer; 34 (97%) had been prescribed inhaled bronchodilator therapy.
Outcomes
Thirteen patients (37%) had a respiratory arrest, of whom nine (69%) also had a cardiac arrest. Eight respiratory arrests occurred pre-hospital, five in the ED. Five patients (14%)
died — all patients who had had a cardiac arrest before ICU admission. All other patients were discharged home without neurological injury. Patients who had a cardiac arrest were more likely to be of Asian or subcontinental origin (5/10 [50%] compared with 4/25 [16%]; relative risk, 3.13; 95% CI, 1.05–9.31). Twenty-six patients (74%) required tracheal intubation and mechanical ventilation, nine patients (35%) had this instituted pre-hospital. No patients received extra-corporeal membrane oxygenation support.
LOS, duration of mechanical ventilation and other outcome data are summarised in Table 2. Duration of mechanical ventilation was significantly shorter in the 22 patients who did not have a respiratory arrest (17 hours [IQR, 12–24 hours] v 94 hours [IQR, 13–139 hours]; P < 0.04). For these patients, the median length of ICU stay was also shorter (29 hours [IQR, 18–43 hours] v 93 hours [IQR, 35–162 hours]; P = 0.004), and duration of bronchospasm was short, with only one of the 22 patients (4.5%) having a pH < 7.3 or arterial partial pressure of carbon dioxide (Paco2) > 45 mmHg persisting after 48 hours, compared with seven of 13 patients after respiratory arrest (54%; P = 0.002).
Pharmacological therapies are listed in Table 2. Adrenaline infusions were more common in the patients who had had a respiratory arrest (21 [60%] v 9 [41%]; P = 0.004). Aminophylline was used in only two adults (6%), although both children received it, and sevoflurane was only used in one patient. Peak and plateau airway pressures were similar between groups (Table 2). Only one patient had a peak airway pressure > 35 cmH2O beyond 48 hours (a patient with respiratory arrest).
Thunderstorm asthma versus all asthma admissions in 2016
There were more asthma admissions to Victorian ICUs on 21–22 November than any other day in 2016 (Figure 4). Patients with thunderstorm asthma were of similar age to all 1062 patients with asthma admitted to Australian ICUs in 2016; however, they were more likely to be male (63% v 34%; P < 0.001), to be mechanically ventilated, and had higher severity of illness and mortality (14% v 1.3%; P < 0.001) (Table 3). Patients with thunderstorm asthma were also ten times and 20 times more likely to have a respiratory or cardiac arrest, respectively. Overall, ICU and hospital LOS was similar, with ICU LOS significantly shorter in survivors of thunderstorm asthma compared with all patients admitted with asthma in 2016 (median, 31.8 hours [IQR, 14.8–43.6 hours] v 40.7 hours [IQR, 22.3–75.1 hours]; P = 0.025).
Intensive care resource utilisation
Thirty-five patients occupied a public ICU bed out of a total of 228 available beds (15.4%) in the affected Melbourne and Geelong regions. When rural or regional ICU bed capacity
Figure 3. Representative glass slide from a pollen trap in Melbourne on 21 November 2016 showing intact and ruptured pollen grains
Critical Care and Resuscitation • Volume 20 Number 4 • December 2018
ORIGINAL ARTICLES
298
Table 1. Baseline patient characteristics
Respiratory arrest
Patient characteristic Total Yes No P
Total number of patients 35 (100%) 13 (37%) 22 (63%)
Age (years), median (range) 39 (9–69) 47 (22–69) 37.5 (9–67) 0.32
was included, 12% of all statewide Victorian public ICU beds were required for patients with thunderstorm asthma. In total, 2177 hours (91 ICU bed-days) were occupied as a result of the epidemic. Total statewide ICU cost attributed to this event was estimated at around $330 000 (median 2016 tertiary ICU bed-day cost $3600) (David Pilcher, Professor, Department of Epidemiology and Preventative Medicine, Monash University; personal communication, March, 2017; based on Australian and New Zealand Intensive Care Society Critical Care Resources Registry data).
Discussion
Key findings
We have described the demographics, outcomes and resource implications of critically ill patients affected by this largest ever reported outbreak of thunderstorm asthma. Patients admitted to an ICU were young, but in contrast to previous literature, all had a prior diagnosis of asthma.7 Five patients (14%) died, but the remaining patients all survived without neurological impairment. Asian and subcontinental patients were over-represented in having cardiac arrest. Compared with all other patients admitted to the ICU with asthma in 2016, those with thunderstorm asthma had rapid resolution of illness and a shorter ICU stay. This event had significant ICU resource implications,
using 15% of all public ICU beds in the affected region. Overall, pollen and particulate counts were not particularly remarkable, but a unique finding was the relationship of spikes in airborne particle counts with the arrival of the storm front.
Relationship to previous studies
Our study confirms past literature showing an association between weather events and an increase in particles of respirable size. Thunderstorm asthma events are associated with increased humidity, and it is hypothesised that moisture absorption leads to lysis of pollen.8 Although the association between humidity and airborne particulate count levels has been previously reported,9,10 this study is unique in demonstrating a temporal association at multiple sites between storm front arrival and particulate count increase. The comparatively low precipitation
observed also suggests it is this rise in humidity, not precipitation, that is important for pollen granule lysis.
In isolation, high pollen counts, high PM10 counts or thunderstorms are not uncommon for Melbourne in November; however, thunderstorm asthma events remain infrequent, with five previous events reported within the medical literature, all in November (1984, 1987, 1989, 2003, 2010 and 2016).4,11,12 More minor events have also occurred between these periods.13 Our data confirm the hypothesis that an event is more likely to occur with the confluence of a high burden of airborne allergens, a thunderstorm outflow concentrating airborne particles at ground level, and conditions that promote particle breakdown.14 What remains unclear is why the storms of 21 November 2016 triggered a much greater asthmatic response.
Prior literature suggests that susceptible patients need not have a prior diagnosis of asthma; indeed, for many, the thunderstorm triggers their first episode of bronchospasm.7 Conversely, a history of atopy or hay fever is more commonly associated, in keeping with the likely allergic pathophysiology.15,16 Of 640 patients in a 1994 London outbreak, 44% had no prior asthma diagnosis, yet, 63% were patients with hay fever.17 In 1997 at Wagga Wagga, Australia, 36% of 148 patients with thunderstorm asthma were non-asthmatics, yet 90% had experienced recent hay fever.7 As we have previously reported,6 follow-up
Critical Care and Resuscitation • Volume 20 Number 4 • December 2018
ORIGINAL ARTICLES
299
Table 2. Intensive care unit (ICU) outcomes, complications and therapies after hospitalisation
APACHE = Acute Physiology and Chronic Health Evaluation. Fio2 = fraction of inspired oxygen. IQR = interquartile range. IV = intravenous. LOS = length of stay. NIV = non-invasive ventilation. Paco2 = arterial partial pressure of carbon dioxide. Pao2 = arterial partial pressure of oxygen. * Arterial blood gas with lowest pH. † Relative risk, 0.62 (95% CI, 0.40–0.95). ‡ Within first 48 hours.
Critical Care and Resuscitation • Volume 20 Number 4 • December 2018
ORIGINAL ARTICLES
300
of 1435 affected patients from this 2016 event indicated that 57% had no previous asthma diagnosis. In contrast to this literature, all 35 patients in our study had asthma, suggesting that bronchospasm of a severity requiring ICU admission preferentially affects patients with asthma. Evidence also supports specific allergen sensitisation: in Wagga Wagga, 96% and 61% of affected patients were found to be sensitised to rye grass pollen and Cladosporium, respectively.7 A protective effect from inhaled corticosteroid use is also likely, first observed in Melbourne, Australia, in 1987 and 1989.11 Affected patients in Wagga Wagga in 1997 were less than half as likely than controls to be taking an inhaled corticosteroid medication. Our study supports this association, with only 34% (and only 15% of patients with respiratory arrest) taking an inhaled corticosteroid. This suggests suboptimal management of this cohort of patients diagnosed with asthma.
Compared with all other admissions of patients with asthma to Australian ICUs in 2016 (of whom 66% were female), we observed a male predominance in critically ill patients with thunderstorm asthma, similar to the 57% male presentations to EDs.5 Female patients with asthma have higher rates of hospitalisation, severity of attacks, and mortality — perhaps related to hormonal, immunological or environmental gender differences.18-20 The reasons for the opposite gender skew in thunderstorm asthma remain unclear, they are perhaps related to the underlying allergic pathophysiology and warrant further research.
Figure 4. 2016 Intensive care unit (ICU) asthma admissions per million population (in Victoria and in Australia)
Although patients with thunderstorm asthma fared worse overall, survivors had a significantly shorter duration of ICU LOS than overall patients admitted with asthma to ICUs in 2016. This indicates rapid bronchospasm recovery, also suggested by the rapid resolution in pH, Paco2 and peak inspiratory pressures. This rapid resolution supports the underlying biological process, consisting of an allergic precipitant with rapid onset, of comparatively short duration. Mortality due to thunderstorm asthma was very high in comparison to overall asthma admissions to the ICU, which declined from 4.7% in 1997 to 1.1% in 2003 in an 8-year study of Australian asthma ICU admissions.21 This high mortality is likely due to the high number of patients presenting after cardiac and/or respiratory arrest in our cohort. Respiratory arrest appears to be a major predictor of mortality in thunderstorm asthma; although two-thirds of patients who did not have a respiratory arrest also required mechanical ventilation, all of these patients were extubated within 24 hours and survived to discharge.
We observed an interesting association with Asian and subcontinental country of birth and severity of critical illness and outcomes due to thunderstorm asthma. This parallels our previous report, which showed a similarly increased risk of ED presentation, hospital admission and mortality in this group.6 Although aspects such as suboptimal asthma education, prevention and management may in part explain this observation, it is likely that we have identified a particular at-risk population. Previous studies have revealed
Critical Care and Resuscitation • Volume 20 Number 4 • December 2018
ORIGINAL ARTICLES
301
an increased prevalence of asthma, hay fever and allergic rhinitis among Asian migrants to Australia compared with non-migrant populations, a risk that increases with duration of stay.22,23 Taken together, these findings suggest a particular at-risk group from epidemic thunderstorm asthma.
Thunderstorm asthma events have significant potential for major disruption of health services. Before this event, the two largest reported outbreaks were in Iran in 2013 (2000 patients) and in greater London in 1994 (640 patients).17,24 In this latter event, six of the 12 EDs affected exhausted their β-agonist inhaler supply. Very little information exists, however, regarding the intensive care implications of past thunderstorm asthma events. In London in 1994, five patients were admitted to an ICU, in Wagga Wagga in 1997, two patients, and in Cambridge, 2002, three patients.3,7,17 Our study is thus the first to illustrate the demands on ICU services across a broad metropolitan region of nearly 5 million people, with 15.4% of all available public ICU beds occupied by patients with thunderstorm asthma. At the individual hospital level, this impact was even greater; one ICU, for example, admitted five patients into a total of ten available beds.
Study strengths and limitations
Strengths of this study are the inclusion of all critically ill patients affected by this event, complete patient data, and the size of the cohort studied — the largest number of patients affected by an outbreak of thunderstorm asthma on record. A further strength is the linkage of various data sources to confirm a temporal association with meteorological and airborne particulate changes.
A limitation of our study is the lack of follow-up of affected patients; the low rates of grass or pollen allergy observed in our cohort may simply reflect undiagnosed patients.
It would also have been instructive to know whether asthma action plans were instituted for surviving patients on hospital discharge. A further limitation is our inability to compare critically ill patients with thunderstorm asthma with the total population of about 3400 affected patients. The availability of demographic and outcome data on the whole cohort may have enabled better description of which patients are likely to progress to life-threatening asthma. Finally, it is possible that some patients with asthma who presented to Australian ICUs in 2016 may not have been identified through coding (eg, if presenting with a cardiac or respiratory arrest). Insights into overall bronchospasm duration in survivors of thunderstorm asthma versus other critically ill patients with asthma, however, remain valid.
Study implications
This study adds support to the hypothesis that aeroallergens in thunderstorm asthma play an important role in the development of bronchospasm. Our findings also imply that interventions to improve preventer use in patients with asthma who reside in thunderstorm asthma-prone areas may have the potential to ameliorate the human impact of a future event. We also observed rapid resolution of bronchospasm, with a relatively short LOS. This may help guide ICU-resource planning, in which future events will be characterised by high initial demand that rapidly decreases over the following 24 hours.
Conclusion
We have described the largest and most lethal episode of thunderstorm asthma and its impact on intensive care services in a major urban region. The relatively low prior use of preventer medication in those patients who became
Table 3. Thunderstorm asthma compared with overall 2016 Australian asthma
ICU LOS (hours), median (IQR) 39 (18–90) 41 (22–78) 0.324
Survivors 32 (15–44) 41 (22–75) 0.025
Non-survivors 180 (118–181) 104 (62–133) 0.139
Hospital LOS (hours), median (IQR) 64 (46–120) 106 (64–183) 0.010
APACHE = Acute Physiology and Chronic Health Evaluation. ICU = intensive care unit. IQR = interquartile range. LOS = length of stay.
Critical Care and Resuscitation • Volume 20 Number 4 • December 2018
ORIGINAL ARTICLES
302
critically ill is of concern. More research is required to identify which patients are at risk of life-threatening thunderstorm asthma, to investigate aspects of gender and ethnicity that may confer risk, and to develop interventions to reduce its severity.
Acknowledgements: We thank Ruth Huang of the Environmental
Protection Agency, Victoria, and OpenStreetMap contributors
(openstreetmap.org) for their assistance.
This work was performed in the Intensive Care Unit, Royal
Melbourne Hospital, Melbourne, Australia.
Competing interestsNone declared.
Author detailsJai N Darvall1,2,3
Matthew Durie1
David Pilcher4,5,6
Geoffrey Wigmore1
Craig French7
Dharshi Karalapillai8
Forbes McGain7
Edward Newbigin9
Timothy Byrne4
Vineet Sarode10,11
Ben Gelbart2,12,13
Andrew Casamento8,14
John Dyett15,16
Ashley Crosswell17
Joseph Vetro15,16
Joseph McCaffrey18
Gopal Taori19
Ashwin Subramaniam20
Christopher MacIsaac1,21
Anthony Cross14
David Ku22
Rinaldo Bellomo1,2,8,21,23
1 Department of Intensive Care, Royal Melbourne Hospital, Melbourne, Vic, Australia.2 Centre for Integrated Critical Care, University of Melbourne, Melbourne, Vic, Australia.3 Department of Anaesthesia and Pain Medicine, Royal Melbourne Hospital, Melbourne, Vic, Australia.4 Department of Intensive Care, Alfred Hospital, Melbourne, Vic, australia.5 Department of Epidemiology and Preventative Medicine, Monash University, Melbourne, Vic, Australia.6 Australian and New Zealand Intensive Care Society (ANZICS) Centre for Outcome and Resource Evaluation, Melbourne, Vic, Australia.7 Department of Intensive Care, Western Health, Melbourne, Vic, Australia.8 Department of Intensive Care, Austin Hospital, Melbourne, Vic, Australia.9 School of BioSciences, University of Melbourne, Melbourne, Vic, Australia.10 Department of Intensive Care, Cabrini Hospital, Melbourne, Vic, Australia.11 Department of Medicine, Monash University, Melbourne, Vic, Australia.12 Department of Intensive Care, Royal Children’s Hospital, Melbourne, Vic, Australia.13 Murdoch Children’s Research Institute, Melbourne, Vic, Australia.14 Department of Intensive Care, Northern Hospital, Melbourne, Vic, Australia.
15 Intensive Care Service, Box Hill Hospital, Eastern Health, Melbourne, Vic, Australia.16 Intensive Care Service, Maroondah Hospital, Eastern Health, Melbourne, Vic, Australia.17 Department of Intensive Care, St Vincent’s Hospital, Melbourne, Vic, Australia.18 Department of Intensive Care, University Hospital Geelong, Geelong, Vic, Australia.19 Department of Intensive Care, Monash Hospital, Melbourne, Vic, Australia.20 Department of Intensive Care, Frankston Hospital, Melbourne, Vic, Australia.21 School of Medicine, University of Melbourne, Melbourne, Vic, Australia.22 Department of Intensive Care, Dandenong Hospital, Melbourne, Vic, Australia.23 Australian and New Zealand Intensive Care Research Centre, School of Public Health and Preventive Medicine, Monash University, Melbourne, Vic, Australia.
References1 Suphioglu C, Singh MB, Taylor P, et al. Mechanism of grass-
