PANDEMIC 2009 INFLUENZA A(H1N1) IN ARGENTINA: A STUDY OF 337
PATIENTS ON MECHANICAL VENTILATION.
Elisa Estenssoro1, Fernando G. Ríos2, Carlos Apezteguía2, Rosa Reina1, Jorge Neira3, Daniel
H. Ceraso4,5, Cristina Orlandi6, Ricardo Valentini7, Norberto Tiribelli8, Matías Brizuela9,
Carina Balasini10, Sebastián Mare11, Gustavo Domeniconi12, Santiago Ilutovich13, Alejandro
Gómez14, Javiera Giuliani15, Cecilia Barrios16; and Pascual Valdez17.
For The Registry of the Argentinian Society of Intensive Care (SATI).
1 Hospital Interzonal San Martin de La Plata, Buenos Aires; 2 Hospital Nacional Alejandro
Posadas, El Palomar, Buenos Aires; 3 Sanatorio de la Trinidad-Palermo, Buenos Aires; 4.
Hospital Fernández, Buenos Aires, 5 Sanatorio San Lucas, Provincia de Buenos Aires; 6 Sanatorio
Lopez Lima, Gral. Roca, Río Negro; 7 CEMIC; Buenos Aires; 8 Hospital Churruca, Buenos
Aires; 9 Hospital Tránsito Cáceres de Allende, Córdoba; 10 Hospital Pirovano, Buenos Aires; 11
Sanatorio Julio Mëndez; Buenos Aires; 12 Sanatorio de la Trinidad-San Isidro, Provincia de
Buenos Aires; 13 Sanatorio de la Trinidad-Mitre, Buenos Aires; 14 Clínica de Los Arcos, Buenos
Aires;15 Hospital Italiano Garibaldi, Rosario; 16 Sanatorio Franchin, Buenos Aires, 17 Hospital
Velez Sarsfield, Buenos Aires.
CORRESPONDING AUTHOR and REQUESTS FOR REPRINTS:
Elisa Estenssoro, MD,
Page 1 of 55
Media embargo until 2 weeks after above posting date; see thoracic.org/go/embargo
AJRCCM Articles in Press. Published on March 4, 2010 as doi:10.1164/201001-0037OC
Copyright (C) 2010 by the American Thoracic Society.
Servicio de Terapia Intensiva, Hospital Interzonal de Agudos San Martín de La Plata,
42 No. 577; 1900 La Plata; Buenos Aires, Argentina.
SHORT RUNNING HEAD: Pandemic Influenza in Argentina.
DESCRIPTOR: 10.14 Pneumonia: Viral Infections
WORD COUNT OF THE MANUSCRIPT: 3751 words
This study was supported by the Argentinian Society of Intensive Care Medicine (SATI).
AT A GLANCE COMMENTARY:
Scientific Knowledge on the Subject.
Pandemic influenza A(H1N1) emerged in April of 2009 and rapidly spread throughout the world.
Though the majority of the patients undergo a benign course of disease, some present with acute
respiratory failure requiring ICU admission for mechanical ventilation, with a subsequent high
What This Study Adds to the Field.
Page 2 of 55
This study, one of the largest cohorts of the sickest patients with 2009 Influenza A(H1N1),
characterized these patients as middle-aged, and predominantly male, with frequent comorbidities
and severe ARDS. Most patients died primarily of refractory hypoxemia; but nonpulmonary
organ failure, expressed as shock and acute kidney failure, sometimes requiring hemodyalisis,
were also very common. Late ICU admission and coexistent infection with S.pneumoniae on
admission worsened patient outcome
This article has an online data supplement which is available from this issue’s table of content
online at www.atsjournals.org
Page 3 of 55
Rationale: The rapid spread of the 2009 Influenza A(H1N1) around the world underscores the
need for a better knowledge of epidemiology, clinical features, outcomes, and mortality
predictors, especially in the most severe presentations.
Objectives: To describe these characteristics in patients with confirmed, probable, and
suspected viral pneumonia caused by 2009 influenza A(H1N1) admitted to 35 intensive care
units with acute respiratory failure requiring mechanical ventilation in Argentina, between June
3 and September 7.
Methods: Inception-cohort study including 337 consecutive adult patients. Data were collected
in a form posted on the Argentinian Society of Intensive Care website.
Measurements and main results: Proportions of confirmed, probable, or suspected cases
were 39%, 8%, and 53% and had similar outcomes. APACHE II was 18±7; age 47±17 years;
56% were male; and 64% had underlying conditions, with obesity (24%), chronic obstructive
respiratory disease (18%), and immunosupression (15%) being the commonest. Seven percent
were pregnant. On admission, patients had severe hypoxemia (PaO2/FIO2140[87-200]),
extensive lung radiologic infiltrates (2.87±1.03 quadrants) and bacterial coinfection, (25%;
mostly with S.pneumoniae). Use of adjuvants such as recruitment maneuvers (40%) and prone
positioning (13%); and shock (72%) and acute kidney injury requiring hemodialysis (17%),
were frequent. Mortality was 46%, and was similar across all ages. APACHE II, lowest
PaO2/FIO2, shock, hemodialysis, prone positioning, and S.pneumoniae coinfection
independently predicted death.
Conclusions: Patients with 2009 influenza A(H1N1) requiring mechanical ventilation were
mostly middle-aged adults, often with comorbidities, and frequently developed severe ARDS
Page 4 of 55
and multiorgan failure requiring advanced organ support. Case-fatality rate was accordingly
WORD COUNT: 250
KEYWORDS: acute lung injury; ARDS; virus; mechanical ventilation; refractory hypoxemia;
multiple organ dysfunction.
Page 5 of 55
New diseases, or diseases that acquire distinctive characteristics on their
presentation and evolution, pose a challenge to the clinician. When an emergent
infectious disease becomes pandemic and causes severe illness and significant mortality
rates, the situation evolves into a major public health problem. By January 2, 2010,
more than 208 countries worldwide had reported laboratory-confirmed cases of
pandemic 2009 influenza A(H1N1), including at least 12,220 deaths (1). The seasonal
behavior of influenza offers the opportunity to assess the outbreaks occurring during
winters in the Southern and Northern hemispheres sequentially. In Argentina, the first
case occurred on April 25, 2009; the virus started to circulate rapidly after May 17 and
peaked between June 20 and July 3, with dissemination over the entire country (2). As
of January 2, 2010, there had been 1,390,566 cases of Influenza-Like Illness (ILI; 3);
14,034 were admitted to the hospital, with 617 deaths ensuing among them [4.4%] (2).
A massive admission of patients with severe pneumonia, many of them young
and in previous good health, crowded the hospitals—within a scenario involving a lack
of epidemiological and clinical data—and generated uncertainty and stress in the
Intensive Care Unit (ICU) staff, until the early reports from Mexico were published
showing some of the distinctive features of the illness (4-6). As did other intensive-care
societies, the Society of Intensive Care of Argentina (SATI) foresaw the risks and
challenges of the situation (7) and on June 27, 2009 uploaded to the society website a
voluntary Registry of Cases in order to answer the following questions:
Page 6 of 55
(1) What was the frequency of adult patients admitted to ICUs during the 2009
Influenza A(H1N1) pandemic with acute respiratory failure with ILI and viral
pneumonia necessitating mechanical ventilation (MV)?.
(2) What were the main risk factors, the clinical and physiological
characteristics, and the complications?
(3) What was the hospital mortality, and what were the conditions independently
associated with that outcome?
Design of the study and of the Registry
This was an inception-cohort study that included patients aged 15 years or more
admitted to the ICU with ILI and acute respiratory failure requiring MV, during the
winter season in the Southern Hemisphere. Data were collected online in a form
designed by experts of the SATI that, after pilot testing, was finally posted at the
Society website on June 27. Also included was an instruction form containing
operational definitions. Information was recorded both prospectively and
retrospectively. All this information was also available, on request, in paper form. Each
participating center filled out a form describing the characteristics of the hospital and of
its ICU. Records were controlled for errors, and local researchers were contacted by the
study authors (EE and FGR), if needed.
Page 7 of 55
On ICU admission, respiratory specimens were collected from each patient and
tested for the 2009 influenza A(H1N1) virus. Most samples nationwide were submitted
initially to a central reference laboratory to perform a real-time polymerase-chain-
reaction (RT-PCR) analysis. Many could not be analyzed, however, because diagnostic
laboratories soon became overwhelmed. As of September 25, 2009 the national health
authorities had made the announcement that the novel 2009 influenza A(H1N1) virus
had displaced other respiratory viruses in patients 5 years or older and, together with
other unidentified influenza A viruses, constituted 93.4% of the samples processed (2).
A seasonal influenza A virus was found in fewer than 2% of the samples. In view of
this information, included in the study were both probable and suspected cases (8) that
fulfilled the criteria of ILI and acute respiratory failure necessitating MV. Samples were
also analyzed for the diagnosis of concomitant bacterial pneumonia.
The following data were also recorded: severity of illness by the APACHE II
score, age, gender, underlying diseases (defined as: immunosuppression, COPD;
asthma, diabetes, chronic heart failure, chronic renal failure, cirrhosis), vaccination for
seasonal influenza A within the current year, pregnancy or childbirth, habitual smoking,
height and estimated or measured body weight for body-mass-index [BMI] calculation;
or the absence of any risk factor. Obesity was defined as a BMI >30. We recorded the
time in days from symptom onset to hospital admission and from hospital admission to
MV initiation; the place in which MV was started (e. g., the ICU; the Emergency
Department (ED), or the Coronary Care Unit (CCU); and, finally, the time from
hospital admission to ICU admission. The extension of lung infiltrates on chest X-ray
Page 8 of 55
was registered as the number of quadrants involved and the Lung Injury Score
On a daily basis, we collected the results of arterial blood gases, oxygenation
variables, progression to ARDS (10), and data of MV, which included the use of non-
invasive ventilation (NIV), the concurrent use of MV adjuvants (recruitment
maneuvers, prone positioning, or tracheal gas insufflation (TGI), the occurrence of
ventilator-associated pneumonia (VAP), and the need for inotropic drugs. The use and
dosage of oseltamivir—the only neuraminidase inhibitor available in Argentina—along
with the treatment of possible concurrent bacterial pneumonia were recorded as well.
Newly developed acute kidney injury requiring hemodialysis and measurement of
creatine-kinase levels (IU/L) were also registered.
The main measurement with respect to outcome was hospital mortality. The
lengths of MV, of ICU, and of hospital stay were calculated.
The descriptive statistics used were the means plus or minus standard deviations,
or the medians and interquartile ranges (IQR) for continuous variables of normal and
nonnormal distributions, and frequency analysis (as percentages) for categorical data.
The main comparisons performed were between survivors and nonsurvivors by means
of unpaired t test, Wilcoxon ranksum test, and either Fisher’s exact test or Chi-square
test, as appropriate. A P value of <.05 was considered significant. The proportions of
confirmed, probable, and suspected cases between survivors and nonsurvivors were
also explored. The incidence of bacterial pneumonia, and especially that of
Page 9 of 55
Streptococcus pneumoniae, was analyzed in confirmed vs. nonconfirmed (probable +
A bivariate analysis for hospital mortality was performed and variables showing
a P value of <.20 included in a multivariable logistic-regression analysis in search of
independent predictors of hospital mortality. A predictive model was built, and the
goodness-of-fit assessed with the Hosmer-Lemeshow test. Discrimination of the model
was evaluated by the area under a receiver-operating characteristic (ROC) curve. A
Kaplan-Meier curve was constructed to evaluate survival over the follow-up period.
All analyses were performed with STATA 9 software.
Since no intervention was performed, informed consent was waived by
institutional review boards.
Characteristics of the hospitals.
A brief description of the SATI and of some characteristics of the participating
centers is displayed in the Online Supplemental Data (Tables E1 and E2). Thirty-five
medical-surgical ICUs participated in the study; all constituting centers of high-acuity
care for critically ill patients. Fourteen ICUs (40%) belonged to university or
university-affiliated hospitals. The mean numbers of hospital and ICU beds were
216±143 and 13±10, respectively. The annual admissions to the ICUs were 709±539,
while 45%±20 of the patients usually require MV.
Clinical characteristics of the patients.
Page 10 of 55
Between June 6 and August 28, 2009, 337 adult patients with confirmed,
probable, or suspected cases of 2009 influenza A(H1N1) with acute respiratory failure
requiring MV were admitted to 35 ICUs. Most initiated MV in the ICU itself. The
collection of all data performed in 214 patients was prospective (60%, admitted after
June 29), while in the remaining the registry was partially prospective. Hospital and
ICU admissions were more frequent between June 21 and July 12, peaking on June 28
and then gradually decreasing (Fig. E1).
Since no differences in mortality were found among confirmed, probable, and
suspected cases (Table 1) the population was analyzed as a single group.
In all but two patients, respiratory samples were obtained with nasopharyngeal
swabs or tracheal aspirates for detection of viruses and other microorganisms.
Treatment with oseltamivir was given to 98% of patients, with 60% receiving 300
mg/day. The frequency of use and doses were similar in both survivors and
nonsurvivors. In all cases, antibiotic treatment for possible associated bacterial and
atypical pneumonia was started. Prior seasonal influenza vaccination was infrequent.
The epidemiological characteristics, severity of illness, and usual risk factors for
2009 influenza A (H1N1) for the whole population, as well as the comparisons between
survivors and nonsurvivors, are displayed in Table 1. Of note, patients were middle-
aged and predominantly male, especially in the nonsurvivor group. The most frequent
previous conditions found in 64% of the patients were habitual smoking, obesity,
COPD, and immunosuppression.
Pregnancy was common (N = 22; 7%), with 17 of the pregnant women being in
the third trimester, 4 in the second, and 1 in the first (ending in a spontaneous abortion).
Page 11 of 55
The most frequent comorbidities were asthma (N = 3) and diabetes (N = 1), and none of
these patients had received a prior seasonal influenza vaccination.
There were no differences in the distribution between survivors and nonsurvivors
with respect to underlying conditions. A risk factor could not be identified in 36% of
the patients. These latter were younger (41±16 vs. 50±17 years, P <.001) and had lower
APACHE II scores (16±7 vs. 19±8, P <.001). In the nonsurvivors, the time from
hospital admission to ICU admission was significantly longer (Table 2).
Respiratory compromise and ventilation support.
Most of the physiological variables were greatly compromised, and the whole
cohort displayed a high incidence of ARDS, extensive infiltrates on chest x-ray, and a
marked alteration in oxygenation, with the need for intensive ventilation support and
the use of MV adjuvants (Table 3). In nonsurvivors the incidence of ARDS was higher
than in survivors (96% vs. 82%, P <.001). Death was associated with a more profound
hypoxemia upon admission (PaO2/FIO2114 [70-188] vs. 152 [109-210], nonsurvivors
vs. survivors, P <.001); lower worst PaO2/FIO2values (80 [61-121] vs. 126 [98-164], P
<.001), a higher maximal PEEP (14±5 vs. 12±4 cm H2O, P <.001), and a more frequent
use of salvage therapies to reverse refractory hypoxemia; such as prone positioning,
recruitment maneuvers, and TGI (Table 3). One hundred fifty patients (45%) showed
PaO2/FIO2≤ ≤100, and this characteristic was more frequent in nonsurvivors (65% vs.
28%, RR 2.26, P <.001). This subgroup of severely compromised patients is further
described in the Online Data Supplement (Table E 3).
Page 12 of 55
NIV was used in 64 patients (19%) and was associated with a better outcome.
None of the patients received extracorporeal membrane oxygenation (ECMO), high-
frequency oscillatory ventilation (HFOV), or inhaled nitric oxide.
Nonpulmonary organ involvement.
The high incidence of shock was remarkable (72% of patients were on
inotropics) especially in the nonsurvivors (83% vs. 62%; P <.001). Renal failure
requiring hemodialysis occurred in 17% of patients, more commonly in the
nonsurvivors (25% vs. 9%, P <.001). Age, shock, and creatine-kinase levels upon
admission were significantly higher in patients undergoing hemodialysis (Table 4).
Coexistent bacterial pneumonia on admission was diagnosed in 80/325 patients
(25%), with the proportions being similar between confirmed and nonconfirmed cases
(Table 5). Within the entire group, 28 patients (9%) had pneumonia caused by
Streptococcus pneumoniae (6% in the survivors vs. 11% in the nonsurvivors, P = .11).
Ventilator-associated pneumonia developed in 84/325 patients (26%); Acinetobacter
baumanii was the most frequently isolated microorganism (N=35), followed by
Pseudomona aeruginosa (N = 22).
Outcomes and predictors of mortality.
One hundred fifty-six patients died (46%; Table 1 and Fig. 1); 62% were male,
and 67% had a previous medical condition, of which habitual smoking and obesity were
Page 13 of 55
the most frequent. Patients with immunosuppression died earlier (Fig. 1). Mortality was
distributed evenly across all age categories, without significant differences among them
(Fig. E2). Logistic-regression analysis identified the APACHE II score, the lowest
PaO2/FIO2, the use of inotropics, hemodialysis, prone positioning, and concomitant
pneumococcal pneumonia as independent predictors of hospital mortality (Table 6).
The predictive model showed good calibration (Hosmer-Lemeshow test = 10.85; P =
.21) and discrimination (area under ROC curve = .81).
Comparison with other studies
A systematic comparison of epidemiological, clinical, and outcome data between
this study and others (11-17) is shown in Table E4 of the Online Data Supplement. The
relationship between outcome and the period elapsing from symptom onset to hospital
admission for our study and other studies (11-17) is included in Table E5 of the Online
We report on a large cohort of critically ill patients admitted to 35 ICUs of
Argentina with suspected, probable, or confirmed 2009 influenza A(H1N1) and with
acute respiratory failure requiring MV. These severely compromised patients were
typically middle-aged adults, predominantly male, and presented with great
physiological deterioration, as evidenced by high APACHE II score, bilateral lung
infiltrates on chest x-ray, and deep hypoxemia. Nonpulmonary organ dysfunctions
Page 14 of 55
12 Download full-text
requiring extracorporeal support, such as shock and renal failure, were frequent; and
mortality was correspondingly high. This evolution occurred rapidly after about one
week of ILI symptoms, a pattern that seems to be a hallmark of severe disease observed
also in other studies (11, 12).
With respect to the patients’ characteristics, 74% were between 25 and 64 years
old, but 15% were older than 65. In contrast to the usual target population of seasonal
influenza, in which children and adults aged >65 years are preferentially affected (18), a
lower mean age has been a consistent finding in populations affected by this 2009
influenza A(H1N1) virus (4, 5, 11-17). Still, the mortality was comparable across all
age groups (Figure E2).
Similar to what has been described in other reports, nearly two-thirds of the
patients had previous medical conditions (13, 14, 16); and, as in seasonal influenza,
habitual smoking and chronic lung disease were the most frequent The COPD
prevalence of 18%, however, was not higher than in the general population for the
region (20). Obesity, a novel risk factor for influenza A described during the 2009
pandemic (21), occurred in 24% of the patients and was comparable to the prevalence
of this condition in Argentina (22). Thus, obesity and chronic respiratory disease were
the two main risk factors for this novel influenza virus, which has been a consistent
report (11-17; Table E5). Immunosuppression was also frequent, and mortality occurred
earlier in this subgroup. Finally, 36% of the patients were in a previous state of good
health. The prevalence of pregnancy (7%) (11-17, 23) was higher than in the general
population (1.7% for Argentina; 24). Pregnancy is a well known risk factor for seasonal
influenza and had caused significant morbidity and mortality during past epidemics and
Page 15 of 55