The fall of 2009 heralded the influx of patients suffering from severe hypoxic respiratory complications secondary to the pandemic H1N1 influenza to ICUs across the country. Due to the severity of pulmonary disease that many of these patients experienced, perception among treating clinicians was that these patients would have worse outcomes and consume more resources, as measured by hospital charges, than patients who developed ALI from other etiologies. We demonstrated that, contrary to what was perceived, pandemic influenza A ALI/ARDS was associated with a lower acuity and, consequently, lower hospital mortality that ALI/ARDS from other etiologies, and had a similar ICU and hospital LOS. ICU and total hospital charges reflected a trend toward higher overall charges for room and board, blood products, pharmacy, and overall charge per patient in the noninfluenza group.
In accordance with other descriptive reports of pandemic influenza [2
], patients who tested positive for H1N1 infection, tended to be young (no patients >64
years old), obese (15 had BMI >30
), and in relatively good health (30% with no comorbid medical conditions). There were no pregnant patients in either group. Compared with other studies of pandemic influenza patients who required mechanical ventilation, SOFA scores (mean 8.3) were similar, although APACHE II (25
9) scores were higher [5
]. The degree of respiratory compromise in our patients was more severe than other reports judging by the higher PEEP requirements and longer duration of mechanical ventilation, which was roughly double that reported in other studies [4
]. Plateau pressures in these studies were not consistently reported. However, despite significantly longer ventilation duration and prolonged ICU and hospital stays, the mortality in our cohort was not higher than that seen in other studies, which ranged from 22–41% in patients who required mechanical ventilation [4
Looking at the different patient characteristics between groups, it may be tempting to postulate that the higher rate of patients with pulmonary ARDS in the H1N1 group, in contrast to prevalent nonpulmonary ARDS in the noninfluenza group, would correlate with a higher PEEP response among the latter [24
]. Our findings suggest the contrary. Patients in the H1N1 group had higher mean plateau pressure, likely indicative of lower compliance. The similarity of PaO2
ratios in the two groups may be a reflection of higher PEEP values used in the H1N1 group for lung recruitment, rather than being indicative of comparable degrees of lung injury. Although assessing recruitability from this retrospective analysis is difficult and may be inaccurate, the higher PEEP used and the implication of lower compliance observed are predictors of potentially recruitable lung [24
]. These observations support the recent call for a reevaluation of the ALI and ARDS criteria to account for this heterogeneity in the patient population [25
A number of important differences between the two cohorts emerged as well. As expected, the noninfluenza group was older, had more comorbid medical conditions, and less often presented to the ICU with respiratory failure. The degree of ventilator support was significantly higher in the H1N1 group on days 1, 3, and 14, and there was a trend to more severe hypoxemia during that time as well. Nevertheless, the use of use of APRV and rescue therapies was comparable in both groups. Despite more severe respiratory compromise, H1N1 patients did not have longer time on the ventilator, longer ICU or hospital stays, or higher mortality. Although SOFA scores were similar, the noninfluenza group had significantly higher APACHE III scores, likely secondary to points assigned to comorbid medical conditions. The high acuity of illness, as well as the presence of severe comorbidities, such as solid and hematologic oncologic conditions (7 patients), chronic renal insufficiency (6 patients), and cirrhosis of the liver (4 patients), likely contributed to the poor outcomes in the noninfluenza group. Conversely, despite more severe respiratory compromise, patients in the H1N1 group were more likely to recover due to their younger age and better overall health histories.
The 77% mortality in the noninfluenza group was much higher than typically reported in clinical trials, with one notable exception [26
]. However, reports from tertiary care centers involving patient cohorts with similar underlying comorbid conditions have reported equally high mortality rates [27
]. Our observation brings up an interesting point, namely the difference between the reported mortality in clinical trials and the observed mortality in a similar clinical condition affecting patients that would have been excluded from such trials due to coexisting comorbidities. A Kaplan-Meier plot of ICU mortality (Figure ) indicates that although patients in the H1N1 group were less likely to survive the first 14
days of ICU care, those that did survive past day 25 were more likely to be discharged alive from the hospital. Patients in the noninfluenza group were unlikely to survive if their ICU length of stay exceeded 3
ARDS is among the most expensive conditions encountered in the ICU [28
]. In 1984, Bellamy and Oye described the charges of patients with ARDS, with the most expensive being room and board (30%), clinical laboratory (24%), pharmacy (14%), and inhalation therapy and ventilation (8%) [27
]. Twenty-five years later, our study indicates that the aforementioned categories continue to represent the most expensive charges incurred by ARDS patients in the ICU.
The overall similarity of charges in room and board and respiratory therapy between the two groups is likely indicative of the comparative durations of hospitalization and mechanical ventilation. Interestingly, despite higher ventilatory requirements and more severe hypoxemia in the H1N1 group, respiratory charges were similar between the two groups, suggesting that the high cost of maintaining a patient on mechanical ventilation is independent of the degree of ventilator support necessary. Thus, respiratory charges are more likely a reflection of duration of mechanical ventilation rather than the degree of ventilator support necessary. Absolute ICU charges for room and board, blood products, pharmacy, radiology, average daily charge, and overall charge per patient were larger in the noninfluenza group. ICU charges for blood products in the noninfluenza group were greater by a factor of four, and pharmacy charges double that of the H1N1 group. This finding is likely a reflection of the higher prevalence of underlying comorbid medical conditions in the noninfluenza group, such as malignancy and cirrhosis, which require expensive medications and predispose to anemia. Moreover, the high mortality in this cohort likely precluded even higher hospital charges. Nevertheless, the H1N1 cohort amassed charges of similar magnitude to the most ill and expensive patients in the ICU, indicating the abundant health care resources consumed by severe pandemic influenza infection.
There are a number of limitations to our study. As a retrospective chart review rather than a prospective investigation, the information was culled from sources that were at times incomplete. Second, the study contained a relatively small number of patients, and measures taken to ensure internal validity of each group, such as limiting the influenza group to confirmed H1N1 infection and the noninfluenza group to the duration of the influenza season, further limited its size. Additionally, whereas our study provides descriptive information relevant to the patient population of our institution and tertiary referral centers with similar acuity, other ICUs may be exposed to a different cohort of patients. On the other hand, as a single-center study, potential differences in clinical and billing practices could be minimized. Although a comprehensive charge profile of each patient was generated, trends in the timing of charges could not be obtained. Finally, the hospital charge data were mined from an extensive database divided by charge coding, and therefore, some charges may have been mislabeled or inappropriately categorized.