Background : Planning for the treatment of infection with the 2009 pandemic influenza A (H1N1) virus through health care systems in developed countries during winter in the Northern Hemisphere is hampered by a lack of information from similar health care systems.
Methods : We conducted an inception-cohort study in all Australian and New Zealand intensive care units (ICUs) during the winter of 2009 in the Southern Hemisphere. We calculated, per million inhabitants, the numbers of ICU admissions, bed-days, and days of mechanical ventilation due to infection with the 2009 H1N1 virus. We collected data on demographic and clinical characteristics of the patients and on treatments and outcomes.
Results : From June 1 through August 31, 2009, a total of 722 patients with confirmed infection with the 2009 H1N1 virus (28.7 cases per million inhabitants; 95% confidence interval [CI], 26.5 to 30.8) were admitted to an ICU in Australia or New Zealand. Of the 722 patients, 669 (92.7%) were under 65 years of age and 66 (9.1%) were pregnant women; of the 601 adults for whom data were available, 172 (28.6%) had a body-mass index (the weight in kilograms divided by the square of the height in meters) greater than 35. Patients infected with the 2009 H1N1 virus were in the ICU for a total of 8815 bed-days (350 per million inhabitants). The median duration of treatment in the ICU was 7.0 days (interquartile range, 2.7 to 13.4); 456 of 706 patients (64.6%) with available data underwent mechanical ventilation for a median of 8 days (interquartile range, 4 to 16). The maximum daily occupancy of the ICU was 7.4 beds (95% CI, 6.3 to 8.5) per million inhabitants. As of September 7, 2009, a total of 103 of the 722 patients (14.3%; 95% CI, 11.7 to 16.9) had died, and 114 (15.8%) remained in the hospital.
Conclusions : The 2009 H1N1 virus had a substantial effect on ICUs during the winter in Australia and New Zealand. Our data can assist planning for the treatment of patients during the winter in the Northern Hemisphere.
Infection with the
2009 pandemic influenza A (H1N1) virus emerged
in
Mexico toward the end of the 2008–2009 influenza season
in the Northern Hemisphere. As of September 6, 2009, the
World Health Organization had reported over 277,607 laboratory-confirmed
cases of
2009 H1N1 influenza, with at least 3205 deaths.
1 From
June through August 2009, Australia and New Zealand experienced
the combined effect of the pandemic and winter in the Southern
Hemisphere. The reported incidence of infection with the
2009 H1N1 virus during winter in Australia and New Zealand was 8
times that reported for the same period in the United States.
1,2 This resulted in a substantial increase in demand for hospital
services, particularly critical care services.
Reports of critical illness caused by
2009 H1N1 influenza during
summer in the Northern Hemisphere contain insufficient data
to provide reliable estimates of the burden of critical illness
to be expected during winter in the Northern Hemisphere.
3,4,5,6,7,8,9 Although the successful deployment of a
safe and effective vaccine may modify the burden of disease,
10,11 population-based
data from Australia and New Zealand can currently provide a
reasonable estimate of the likely effect of
2009 H1N1 influenza during the Northern Hemisphere winter. In addition, the data
can be used to identify persons who are at
high risk of developing severe disease.
In this report, we describe the
incidence of intensive care unit (ICU) admission, demographic characteristics, treatment,
use of critical care resources, and outcome for all patients
with laboratory-confirmed infection with the
2009 pandemic influenza A (H1N1) virus admitted to ICUs in Australia and New Zealand
during the winter of 2009 in the Southern Hemisphere.
Methods
We performed a
multicenter inception-cohort study involving
187 ICUs in Australia and New Zealand — all the ICUs (adult,
pediatric, or adult and pediatric) in the two countries.
12 The
ICUs had a total of 1879 beds, of which 1449 were equipped for
mechanical ventilation. Each center obtained approval from the
institutional ethics committee. The requirement for written
informed consent from individual patients was waived at all
sites.
From June 1 through August 31, 2009, we identified all patients
admitted to the ICU with confirmed infection with the
2009 pandemic influenza A (H1N1) virus. The 2009 H1N1 influenza was confirmed
by means of a
polymerase-chain-reaction (PCR) assay or serologic analysis. The 2009 pandemic influenza A (H1N1) virus and seasonal
subtypes (preexisting
H1N1 and H3N2 strains) were confirmed
by PCR assay. The PCR assay was conducted initially at reference
laboratories in each region and later, as the pandemic evolved,
at local laboratories. The performance of these laboratories
was accredited by the
National Association of Testing Authorities in Australia or by
International Accreditation New Zealand. In addition, the 2009 H1N1 virus could be confirmed in a single
reference laboratory by means of a
hemagglutination-inhibition assay to detect antibodies specific for the 2009 H1N1 virus.
Population data for Australia and New Zealand and their constituent
regions were obtained from Australian Bureau of Statistics
13 and Statistics New Zealand.
14
We collected several types of data for the patients: the dates
and times of admission to the hospital and the ICU; age; race
or ethnic group, including indigenous group (reported by patients
or their next of kin or, for patients under 18 years of age,
by a parent or guardian)
; sex; pregnancy or childbirth less
than 28 days previously (for women); coexisting conditions,
which for patients 16 years of age or older were any condition
that is defined within the
Chronic Health Evaluation component
of the Acute Physiology, Age, and Chronic Health Evaluation
(
APACHE III, for which scores can range from 0 to 299, with
higher scores indicating a greater severity of illness),
15 and
for patients under 16 years of age, defined as
prematurity, immunodeficiency, cystic fibrosis, congenital heart disease, neuromuscular disorder, or chronic neurological impairment; history of asthma or another chronic pulmonary disease, chronic heart failure, or diabetes; measured or estimated weight and
height (for calculation of the body-mass index [BMI]); date
and time of first symptoms; presence and type of
influenza syndrome, including
viral pneumonitis or the
acute respiratory distress syndrome, secondary bacterial pneumonia, exacerbation of airflow limitation due to either asthma or
chronic obstructive pulmonary disease, or intercurrent illness; and airway status at the time
of ICU admission (presence or absence of endotracheal intubation,
tracheotomy, sealed face mask, and any artificial airway).
We categorized patients according to the age groups used in
a previous report: 0 to 1 year of age, 1 to 4 years, 5 to 24
years, 25 to 49 years, 50 to 64 years, and 65 years of age or
older.
16 Data were collected daily on the use of mechanical
ventilation and
extracorporeal membrane oxygenation. We calculated
the duration of treatment in the ICU and the hospital, as well
as the rates of occupancy of the ICU, for Australia and New
Zealand and their constituent regions. We recorded outcomes
of patients in the ICU and whether the patients had been discharged
or were still in the hospital or the ICU as of September 7,
2009. To compare data from the current year with those from
previous years, we obtained the number of patients who had been
admitted to Australian or New Zealand ICUs with viral pneumonitis
during the winters of 2004 through 2008, from the
Australian and New Zealand Intensive Care (ANZIC) Society's Adult Patient
Database.
17 This source of data does not categorize the cause
of viral pneumonitis and may include some patients who had viral
pneumonitis due to causes other than influenza A. To determine
which groups were at increased risk of admission to an ICU with
2009 H1N1 influenza, we compared the proportions of patients
with such an admission in each group of interest with the proportions
of the general population of Australia
13 and New Zealand
14 that
those admitted patients represented.
Data Management
We collected data by means of electronic case report forms.
The study coordinating center was the
ANZIC Research Centre, Monash University, Melbourne, Australia.18 Infection with the
2009 H1N1 virus is subject to mandatory reporting in both Australia
and New Zealand, and all diagnoses were confirmed with the relevant
state or territory's Department of Health. In addition, to confirm
the completeness of case ascertainment, we contacted the 83
ICUs that had no reported cases at the end of the study period
(August 31, 2009). Patients transferred between ICUs were counted
as a single ICU admission. We made no assumptions regarding
missing data; all proportions were calculated as percentages
of the patients with available data.
Statistical Analysis
We performed statistical analysis using ,
SAS software version
9.1
(SAS Institute). We calculated descriptive statistics for
all study variables. We report data for continuous variables
as medians (with
interquartile ranges) and for categorical variables
as percentages (with 95% confidence intervals, where appropriate).
We estimated the age-based population-admission rates.
13 We
performed a univariate analysis for in-hospital mortality, using
the chi-square test,
Fisher's exact test, or
Wilcoxon rank-sum test, as appropriate. We performed multivariate logistic-regression
analysis to identify factors independently associated with in-hospital
mortality, with the multivariate model constructed by using
both stepwise-selection and backward-elimination techniques.
We first included age, as a continuous variable. We then included
in the model, as categorical variables,
16 the presence or absence
of pregnancy, asthma or another chronic pulmonary disease, and
chronic heart failure; BMI (the weight in kilograms divided
by the square of the height in meters) greater than 35 versus
35 or less; race or ethnic group; and the presence or absence
of any coexisting condition; and the type of influenza syndrome.
Goodness of fit was determined with the use of the
Hosmer–Lemeshow statistic. A two-sided P value of less than 0.05 was considered
to indicate statistical significance, except in the multivariate
model, where a P value of less than 0.01 was considered to indicate
statistical significance.
Results
We identified 856 patients with influenza A infection who were
admitted to an ICU between June 1 and August 31, 2009. Of these,
722 (84.3%) had a confirmed infection with
2009 pandemic influenza A (H1N1) virus (
Figure 1). The 2009 H1N1 virus was diagnosed
by means of
PCR assay in 717 patients and
serologic analysis in 5 patients. Among the 722 patients with 2009 H1N1 influenza,
626 were admitted to an ICU in Australia and 96 to an ICU in
New Zealand. The numbers of patients with
viral pneumonitis admitted to Australian or New Zealand ICUs from June 1 through
August 31 were 57 in the year 2005, 33 in 2006, 69 in 2007,
and 69 in 2008 (mean, 57 patients). During the winter of 2009,
37 patients were admitted to an ICU with confirmed seasonal
subtypes of influenza A (H1N1) virus. The combined population
of Australia and New Zealand was estimated at 25,180,770, giving
an incidence of ICU admission for 2009 H1N1 influenza during
winter 2009 of 28.7 (95% confidence interval [CI], 26.5 to 30.8)
per million inhabitants.
13,14
| Figure 1. Enrollment and Follow-up of Patients with Influenza A Admitted to Intensive Care Units (ICUs) in Australia and New Zealand. Follow-up data are as of September 7, 2009.
|
|
The number of admissions and the age-specific incidences varied
substantially according to the age group
(Figure 2). The highest
age-specific incidence of ICU admission was among infants (0
to 1 year of age)
(Figure 2A), whereas the highest number of
ICU admissions was among patients 25 to 49 years of age
(Figure 2B). Additional demographic data and data on risk factors and
type of critical illness among patients with 2009 H1N1 influenza
are presented in
Table 1.
| Figure 2. Incidence and Number of Admissions to the Intensive Care Unit (ICU), According to Age Group. The incidence of ICU admission, expressed as the number per million residents, is shown in Panel A, and the number of ICU admissions is shown in Panel B. The age-specific incidence was calculated by dividing the number of admissions in each age group by the estimated number of persons in that age group in the population of Australia and New Zealand (which was 350,346 for 0 to 1 years, 1,324,185 for 1 to 4 years, 6,938,498 for 5 to 24 years, 9,111,600 for 25 to 49 years, 4,563,709 for 50 to 64 years, and 3,377,580 for 65 years or older). I bars indicate the 95% confidence intervals.
|
|
Pregnant women represent approximately 1% of the general population
of Australia and New Zealand.
13,14 A total of 66 of the 722
patients (9.1%) admitted to the
ICU with
2009 H1N1 influenza were
pregnant women. Of the 601 adults for whom
BMI data were
available, 172 (28.6%) had a
BMI greater than 35. The proportion
of a representative adult Australian population with a
BMI greater
than 35 was 5.3% in 2003.
19 We estimate the proportion of patients
with asthma or other chronic pulmonary disease in the general
population to be around 13%.
20 Data on
premorbid pulmonary disease were missing for 15 of the 722 patients with 2009 H1N1 influenza
in our study; of the remaining 707 patients, 231 (32.7%) had
asthma or another chronic pulmonary disease. Indigenous groups
were relatively overrepresented in our study:
aboriginal and Torres Strait Islanders account for 2.5% of the Australian population
13 but made up 9.7% of our patients with 2009 H1N1 influenza
who were admitted to Australian ICUs. Maori represent 13.6%
of the New Zealand population
14 but accounted for 25.0% of the
patients with 2009 H1N1 influenza who were admitted to New Zealand
ICUs. Overall, 229 patients (31.7%) had no known predisposing
factor. Almost half of all patients (48.8%) had the acute respiratory
distress syndrome or viral pneumonitis, and 20.3% of patients
were clinically diagnosed with
bacterial pneumonia (i.e., had
unilateral or bilateral asymmetric lung infiltrates consistent
with bacterial pneumonia, with bacterial infection proven or
suspected) in association with confirmed infection with the
2009 H1N1 virus.
Data on the use of
mechanical ventilation in the ICU were available
for 706 patients; of these, 456 (64.6%) underwent mechanical
ventilation for a median of 8 days (interquartile range, 4 to
16). The total number of days of ventilation was 5249, representing
208 days (95% CI, 203 to 214) per million inhabitants. Of the
456 patients undergoing mechanical ventilation, 53 (11.6%) were
subsequently treated with
extracorporeal membrane oxygenation, representing 2.1 patients (95% CI, 1.5 to 2.7) per million inhabitants.
Available data on other cointerventions are given in the
Supplementary Appendix (available with the full text of this article at NEJM.org).
As of September 7, 2009, a total of 114 of the 722 patients
(15.8%) were still in the hospital, of whom 37 (5.1%) were still
in the ICU. Excluding these 114 patients still in the hospital
or ICU and an additional 33 for whom data were not available
(regarding duration of ICU treatment, for 3 patients, and duration
of in-hospital treatment, for 30), we calculated the median
duration of treatment in the ICU as 7.4 days (interquartile
range, 3.0 to 16.0)
(Figure 3) and the median duration of treatment
in the hospital as 12.3 days (interquartile range, 6.4 to 22.1).
| Figure 3. Length of Stay in the Intensive Care Unit (ICU) among Patients with 2009 H1N1 Influenza.
|
|
The number of ICU admissions per million inhabitants varied
over the study period, for Australia and New Zealand overall
and for each of the main regions affected, as did the number
of ICU beds occupied per million inhabitants (
Figure 4). Patients
with 2009 H1N1 influenza occupied the ICU for a total of 8815
ICU bed-days, representing 350 bed-days (95% CI, 342 to 357)
per million inhabitants. Across Australia and New Zealand, the
maximum number of ICU beds occupied per million inhabitants
was 7.4 (95% CI, 6.3 to 8.5) during the week ending July 27,
2009. The maximum number of beds occupied by region in the Australian
states or New Zealand ranged between 6.3 and 10.6 per million
inhabitants (
Figure 4). Over the 3-month study period, 5.2%
of ICU bed-days were accounted for by patients with 2009 H1N1
influenza. The peak percentage of ICU beds occupied by patients
with 2009 H1N1 influenza in Australian states and New Zealand
ranged from 8.9 to 19.0%.
| Figure 4. Numbers of Patients with 2009 H1N1 Influenza Admitted to an ICU and Numbers of ICU Beds Occupied by Those Patients, According to Study Week and Region. The panels on the left show the numbers of patients admitted to an ICU, and the panels on the right show the numbers of ICU beds occupied by those patients. All data are per million inhabitants. Data are shown for the Australian states of Victoria, New South Wales, and Queensland; for New Zealand; and for all regions of Australia and New Zealand combined.
|
|
As of September 7, 2009, a total of 608 patients (84.2%) had
been discharged from the hospital: 103 (16.9%) had died in the
hospital and 505 (83.1%) had been discharged alive. For those
who had died or been discharged alive, three factors were found,
on multivariate logistic-regression analysis, to be independently
associated with death in the hospital: requirement of invasive
ventilation at ICU admission (odds ratio for in-hospital death,
5.51; 95% CI, 3.05 to 9.94; P<0.001), any coexisting condition
(as defined in our study) (odds ratio, 2.56; 95% CI, 1.52 to
4.30; P<0.001), and older age (odds ratio per year of age,
1.02; 95% CI, 1.01 to 1.04; P=0.002). The data were well fitted
by the model (P=0.79 by the Hosmer–Lemeshow test).
Discussion
This cohort study identified all patients with confirmed
2009 H1N1 influenza who were admitted to Australian or New Zealand
ICUs during winter 2009 in the Southern Hemisphere. We identified
722 patients with the infection and estimated the winter population
incidence of ICU admission: 28.7 per million inhabitants. The
number of ICU admissions due to influenza A in 2009 was 15 times
the number due to
viral pneumonitis in recent years. We were
able to document the use of ICU beds and patients' outcomes
and estimate the number of ICU bed-days occupied: 350 per million
inhabitants. We identified infants (0 to 1 year of age) and
adults 25 to 64 years of age to be at particular risk. Pregnant
women, adults with a BMI greater than 35, and indigenous Australian
and New Zealand populations also appeared to have an increased
risk. In-hospital mortality, estimated on the basis of data
available at the time of this report, exceeded 16%.
Previously published reports have highlighted cases of severe
viral pneumonia affecting patients younger than the expected
age of patients affected during a normal influenza season
8,9 and have noted that pregnant women are at increased risk.
21 Our findings are consistent with these reports. The age-specific
incidence rates were highest among infants and adults 25 to
64 years of age. Although the incidence of ICU admission varied
across the age groups and was low for patients 65 years of age
or older, the risk of death increased with increasing age. The
proportion of patients who were admitted to an ICU and were
pregnant, had chronic lung disease, had a BMI greater than 35,
or were indigenous to Australia or New Zealand were all higher
than the corresponding proportions in the general population.
Finally, a third of our patients were young or middle-aged adults
who neither were pregnant nor had a known coexisting condition.
Australia and New Zealand have 75 ICU beds per million inhabitants.
The number of ICU beds varies greatly among developed countries,
22 and the capacity of countries to cope with a surge in demand
for critical care services owing to infection with the 2009
pandemic influenza A (H1N1) virus will depend on the current
numbers of ICU beds and the countries' ability to expand that
capacity or restrain other demands on it.
Our data indicate that the greatest effect on ICU resources
in a given region occurs approximately 4 to 6 weeks after the
first confirmed winter ICU admission and that the extra workload
lasts several weeks. Current recommendations are that patients
with 2009 H1N1 influenza should receive treatment in isolation.
23 The requirement to treat many patients in isolation, combined
with the need for interhospital transfer for optimal care, may
further increase the strain on critical care resources.
The proportion of patients who died in the hospital in our study
is no higher than that previously reported among patients with
seasonal influenza A who were admitted to an ICU.
24 Patients
admitted to an ICU with seasonal influenza A predominantly are
elderly and have coexisting conditions.
24 Among patients admitted
to ICU, older age, the presence of coexisting conditions, and
a requirement for invasive ventilation were independently associated
with increased risk of death, but because there were greater
numbers of younger patients in our cohort, the majority of deaths
occurred in younger patients.
The inferences that can be drawn from our data are subject to
some limitations. First, to make this report available in time
to assist planning in the Northern Hemisphere, we censored the
hospital-outcome data, which may have introduced bias. Second,
our data were gathered early during the pandemic in Australia
and New Zealand. The findings may be different during future
waves, owing to the timely deployment of an effective vaccine,
to viral mutation, and to resistance to antiviral drugs. Third,
the data regarding previous winters come not from an inception-cohort
study but from our Australia–New Zealand database; therefore,
they are not directly comparable to the data in the current
study for winter 2009. Fourth, ascertainment of patients with
2009 H1N1 influenza who were admitted to an ICU may not have
been complete, and we cannot rule out the possibility that a
small number of cases were not reported to the registry. Finally,
false negative diagnostic tests may well have led us to underestimate
the true burden of 2009 H1N1 influenza in our patients. Among
the patients with confirmed influenza A, there were 97 in whom
the influenza virus was not subtyped, some of whom may have
had false negative tests for the 2009 H1N1 virus. Nonetheless,
with these caveats, knowledge of the rate of ICU admission and
occupancy due to 2009 H1N1 influenza during the winter in Australia
and New Zealand can inform the planning and assessment of critical
care needs in countries yet to face the 2009 winter.
Posts
