In this study we identified all patients with confirmed diagnosis of influenza A H1N1 admitted to the Albert Einstein Jewish Hospital during the year of 2009. We identified 1, 401 patients with the infection and who were treated in our hospital. Of these, only 139 (9.9%) patients needed hospitalization. Of these 139 patients, 20 (14.3%) were admitted to the ICU. Our ICU admission incidence was lower than that reported in most studies [3, 9]. The exception was a study done in Canada that reported an even lower incidence (8.1%) .
Cases of acute respiratory failure due to influenza A H1N1 affecting patients younger than the expected age for patients with seasonal influenza have been previously reported [1–8]. Our findings are consistent with these reports.
In regard to the severity of illness, our APACHE II score (7) was lower than that of other studies, which reported a median score around 13 to 19 [1, 2, 8]. Our APACHE II score was probably lower than other studies because the disease was already an endemic disease when it reached Brazil and the population and health-care co-workers were already alert to the disease signs and received early treatment. This is probably also the reason for our low number of patients admitted to the ICU as reported by other studies.
Similar to other studies, our study identified chronic lung disease as the most frequent coexisting medical condition [1–3, 5, 8, 9]. On the other hand, a study by Cao et al. described arterial hypertension as the most frequent coexisting medical condition .
Most of our patients had ARF, as seen in other studies, with the exception of one study from Japan that did not report any case of ARF between May and June of 2009 [1–3, 8, 9]. In our study, we had 85.7% of the patients who used NIV at ICU admission to treat ARF. Our incidence was higher than that of the other two studies that mentioned the use of NIV at ICU admission (around 33%) [2, 8]. In this study we had a success rate with NIV in 41.6% of the patients, showing greater success with NIV when compared with the other two studies (25% in Spain  and14.6% in Canada ).
We believe that this difference in NIV success is due to the disease severity, because our APACHE II score was lower than that described by others [2, 8]. Because of the low NIV success rate and high endotracheal intubation incidence previously reported in several studies, many researchers stated that NIV should not be routinely used in patients with pandemic respiratory infections [1, 2, 8, 17].
Many researchers are concerned about using NIV during pandemic respiratory infections due to the infectious exhaled aerosol that can reach the environment through the exhalation ports in the NIV mask or tubing, which may contaminate health-care co-workers [17, 18]. For instance, in a recent study from Australia  in which a significant correlation with the need for invasive MV and mortality rates was observed (Odds ratio 5.51; 95% CI 3.95-9.94; p < 0.001), NIV was never considered as an option.
In the study by Perez-Padilla et al. , the authors reported 22 (11.5%) cases of health-care co-worker contamination with influenza A H1N1, while treating the first 3 patients who were admitted to the hospital. None of these patients were undergoing NIV. After an infection-control measure that enforced patient isolation in specific hospital areas, use of N95 respirators in addition to goggles, gowns, and gloves, and constant use of gel-alcohol hand sanitizer, no more health-care co-workers were contaminated by the disease.
This was not the case in our hospital, where our NIV success rate was higher than everywhere else, and no health-care co-worker was contaminated through NIV and all health-care co-workers used infection control measures to avoid contamination, such use of goggles, gowns, gloves, and use of gel-alcohol hand sanitizer. All influenza A H1N1 ICU patients stayed in individual and isolated hospital bed, to avoid contamination to other patients. Our success rate with NIV indicates that patients with influenza A H1N1 can benefit from it, thus preventing the need for invasive MV, and minimizing the incidence of complications such as ventilator-associated pneumonia . This was the situation described in Mexico , where 4 (22%) patients underwent invasive MV and had ventilator-associated pneumonia.
Cheung et al.  reported their experience with 20 patients undergoing NIV during severe acute respiratory syndrome (SARS) in 2003 at a Hong Kong hospital. To reduce the risk of contamination the staff used a viral/bacterial filter, N95 respirators, and an oronasal mask to prevent large leaks through the mouth and exhalation valve. They were able to prevent endotracheal intubation in 14 patients (70%) and none of the health-care co-workers were contaminated. In our hospital all the protective measures were used to prevent contamination of health-care co-workers and other ICU patients.
The fact that influenza A H1N1 arrived in Brazil after being already an endemic alert in other countries provided the necessary time for the government to set strategies to control its spread. Additionally, the population was already aware of the symptoms and the health-care system of the treatment. This allowed patients to be properly and promptly treated for influenza A H1N1, while health-care workers took protective measures to avoid contamination.
In this scenario, NIV can be safely used in patients with influenza A H1N1. Nevertheless, the same indication for using NIV in patients requiring MV for ARF should be considered for patients with influenza A H1N1.
In regard to invasive MV, our study had a lower incidence at admission, when compared to other studies [1–3, 8, 9]. This can also be related to our low APACHE II score, as the study by Rello et al.  described that patients who failed NIV had a higher APACHE II score than those who were successful. Our mechanical ventilation time was lower than that described by others [1, 2, 8, 9].
In our study 5 patients developed ARDS and were intubated. Of these patients, all underwent alveolar recruitment maneuver with high end-expiratory pressure levels to improve oxygenation, without any related complications. In the studies on influenza A H1N1 patients currently available [1–9], none mention using this maneuver to improve oxygenation. In one study, Kumar et al.  reported using inhaled nitric oxide, high frequency oscillatory ventilation, and extracorporeal membrane oxygenation and had 8.3% incidence of barotraumas.
Regarding the ICU length of stay, our findings were lower than those reported in previous studies [8, 9]. However, in regard to our hospital length of stay, our findings were similar to those from a study from Australia .
So far, patients with influenza A H1N1 admitted to hospitals worldwide had a high mortality rate (13 to 39%) [1–9]. However, in our study the hospital mortality rate was 2.1%. Beside the low APACHE II score of our patients, it is possible that our low mortality rate is also associated with a lower number of patients undergoing invasive mechanical ventilation, as compared to other studies [1–4, 8, 9].