Martin-Loeches et al. Critical Care (2019) 23:152 https://doi.org/10.1186/s13054-019-2425-6 RESEARCH Open Access Influenza and associated co-infections in critically ill immunosuppressed patients Ignacio Martin-Loeches9,10,35*† , Virginie Lemiale1†, Pierce Geoghegan9,10, Mary Aisling McMahon9,10, Peter Pickkers2, Marcio Soares3, Anders Perner4, Tine Sylvest Meyhoff4, Ramin Brandt Bukan22, Jordi Rello5, Philippe R. Bauer6, Andry van de Louw7, Fabio Silvio Taccone8, Jorge Salluh3, Pleun Hemelaar2, Peter Schellongowski11, Katerina Rusinova12, Nicolas Terzi14, Sangeeta Mehta15, Massimo Antonelli16, Achille Kouatchet17, Pål Klepstad13, Miia Valkonen19, Precious Pearl Landburg20, Andreas Barratt-Due18, Fabrice Bruneel21, Frédéric Pène23, Victoria Metaxa24, Anne Sophie Moreau25, Virginie Souppart1, Gaston Burghi26, Christophe Girault27, Ulysses V. A. Silva28, Luca Montini16, Francois Barbier29, Lene B. Nielsen30,31, Benjamin Gaborit32, Djamel Mokart33, Sylvie Chevret34, Elie Azoulay1 and For the Efraim investigators and the Nine-I study group Abstract Background: It is unclear whether influenza infection and associated co-infection are associated with patient- important outcomes in critically ill immunocompromised patients with acute respiratory failure. Methods: Preplanned secondary analysis of EFRAIM, a prospective cohort study of 68 hospitals in 16 countries. We included 1611 patients aged 18 years or older with non-AIDS-related immunocompromise, who were admitted to the ICU with acute hypoxemic respiratory failure. The main exposure of interest was influenza infection status. The primary outcome of interest was all-cause hospital mortality, and secondary outcomes ICU length of stay (LOS) and 90-day mortality. Results: Influenza infection status was categorized into four groups: patients with influenza alone (n = 95, 5.8%), patients with influenza plus pulmonary co-infection (n = 58, 3.6%), patients with non-influenza pulmonary infection (n = 820, 50.9%), and patients without pulmonary infection (n = 638, 39.6%). Influenza infection status was associated with a requirement for intubation and with LOS in ICU (P < 0.001). Patients with influenza plus co-infection had the highest rates of intubation and longest ICU LOS. On crude analysis, influenza infection status was associated with ICU mortality (P < 0.001) but not hospital mortality (P = 0.09). Patients with influenza plus co-infection and patients with non-influenza infection alone had similar ICU mortality (41% and 37% respectively) that was higher than patients with influenza alone or those without infection (33% and 26% respectively). A propensity score-matched analysis did not show a difference in hospital mortality attributable to influenza infection (OR = 1.01, 95%CI 0.90–1. 13, P = 0.85). Age, severity scores, ARDS, and performance status were all associated with ICU, hospital, and 90-day mortality. (Continued on next page) * Correspondence: drmartinloeches@gmail.com Ignacio Martin-Loeches and Virginie Lemiale contributed equally. 9 Department of Intensive Care Medicine, Multidisciplinary Intensive Care Research Organization (MICRO), St. James’s Hospital, Dublin, Ireland 10 Department of Clinical Medicine, Wellcome Trust-HRB Clinical Research Facility, St. James Hospital, Trinity College, Dublin, Ireland Full list of author information is available at the end of the article © The Author(s). 2019 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Martin-Loeches et al. Critical Care (2019) 23:152 Page 2 of 10 (Continued from previous page) Conclusions: Category of infectious etiology of respiratory failure (influenza, non-influenza, influenza plus co- infection, and non-infectious) was associated with ICU but not hospital mortality. In a propensity score-matched analysis, influenza infection was not associated with the primary outcome of hospital mortality. Overall, influenza infection alone may not be an independent risk factor for hospital mortality in immunosuppressed patients. Keywords: Influenza, Respiratory failure, Sepsis, Critical illness, Immunosuppression, Introduction co-infection were associated with patient-important out- Immunosuppressed patients admitted to an intensive comes in this group. care unit (ICU) with acute respiratory failure have a very high risk of mortality [1]. Acute respiratory failure can Methods have various etiologies, but pulmonary infection and its Study design and setting sequelae remain the most frequent precipitants in those The current study was a preplanned secondary analysis that require ICU admission [2, 3]. Among the different of the EFRAIM study, a multinational prospective cohort infectious agents which cause pulmonary infection in study in 68 centers in 16 countries. EFRAIM was per- immunocompromised patients, pneumonia caused by formed by the Nine-I (Caring for Critically Ill Immuno- influenza viruses has been associated with a particularly compromised Patients) study group [13]. The Nine-I high mortality rate [4]. group includes critical care physicians from 16 countries Influenza infection can affect patients during pan- who have extensive experience in the management of demic periods (such as the H1N1 pandemic of 2008/ various groups of critically ill immunocompromised pa- 2009) or during seasonal epidemics. Factors associated tients. Physician participation was voluntary, without fi- with the risk of a severe influenza infection during and nancial incentive. Participating investigators obtained after pandemic periods have differed [5]. After the first local institutional review board approval in accordance pandemic period in 2008 and 2009, influenza affected with local ethics regulations. particular subgroups, particularly obese and pregnant patients [6, 7]. During the post-pandemic period, im- munosuppression was a risk factor for both influenza in- Inclusion and exclusion criteria fection and ICU mortality [8]. Other factors associated Eligibility criteria were age ≥ 18 years, acute hypoxemic with greater severity of influenza infection are age, med- respiratory failure (PaO2 < 60 mmHg or SpO2 < 90% on ical comorbidities, and possibly co-infection. Bacterial room air, or tachypnea > 30/min, or labored breathing or pulmonary co-infection has long been described in pa- respiratory distress or dyspnea at rest or cyanosis), need tients with influenza pneumonia, most commonly with for more than 6 L/min oxygen, respiratory symptom Staphylococcus aureus, Streptococcus pneumoniae, and duration less than 72 h, and non-AIDS-related immune Haemophilus influenzae. Recognition of influenza infec- deficiency defined as hematologic malignancy or solid tion is important because it allows the implementation tumor (active or in remission for less than 5 years), solid of appropriate infection control measures and specific organ transplant, long-term (> 30 days, any dose) or antiviral therapy. Furthermore, it might reduce inappro- high-dose (> 1 mg/kg/day) steroids, or any immunosup- priate antibacterial administration. pressive drug taken in a high dosage or for more than Although bacterial co-infection has been associated 30 days. Exclusion criteria were postoperative acute re- with increased mortality during the 2008/2009 pandemic spiratory failure, admission after a cardiac arrest, ICU period [9], the impact of the combination of influenza admission exclusively to secure bronchoscopy, or refusal infection and bacterial or fungal co-infection on the out- of the patient or family to participate in the study. come of critically ill patients has been a matter of debate [10]. Although influenza seems to be associated with Enrolment, data collection, and patient treatment higher mortality rates in immunocompromised patients Participating ICUs enrolled patients from Nov. 5, 2015 to [11, 12], the fraction of mortality attributable to either Jul. 1, 2016. Prospective data were collected on patient influenza infection alone or influenza plus co-infection and disease characteristics, initial oxygenation strategy, has not been well defined. Our aim in the current study acute respiratory failure (ARF) etiology, associated organ was to examine the prevalence of influenza infection and dysfunction, and patient outcomes at hospital discharge co-infection in critically ill immunocompromised pa- and at day 90. The case report forms were sent to the co- tients admitted to the ICU with respiratory failure and ordinating center in Paris for data entry by trained techni- to determine whether influenza and associated cians. The study was funded by the Groupe de Recherche Martin-Loeches et al. Critical Care (2019) 23:152 Page 3 of 10 en Réanimation Onco-Hématologique (GRRR-OH), an proportions. Data management allowed checking for academic non-profit French organization. data inconsistencies that were solved by consensus. All management decisions were performed according Comparisons of proportions between the groups were to standard local practice in each ICU. Diagnostic strat- made using the χ2 test. Comparisons of continuous vari- egies to identify the etiology of respiratory failure were ables between the groups were made using the Wilcoxon based on previous studies by the GRRR-OH rank-sum test. (11,18-20,23). ARF etiologies were based on pre-defined A propensity score (PS)-based approach was used to criteria in each participating ICU (11,18-20,23). All diag- limit the effect of bias on the between-group compari- noses were reviewed by two study investigators (from in- sons of hospital mortality. The propensity score was de- dependent institutions) for coherence and for alignment fined as the probability that a patient with specific with accepted definitions. Oxygenation modalities, the baseline characteristics had influenza infection. We de- use of non-invasive ventilation, high-flow nasal oxygen, veloped the PS using a logistic regression model that in- or intubation was documented daily. Management of as- cluded all baseline characteristics associated with illness sociated organ dysfunction, handling of immunosuppres- severity [2]: mechanical ventilation, age, Eastern Co- sive drugs, or chemotherapy was decided by a physician operative Oncology Group (ECOG) performance status, according to local and recommended practices. Intub- SOFA score at ICU admission, admission within first ation decisions were left at the discretion of the care hours of hospital admission, tobacco use, and underlying team and based on the therapeutic response, clinical sta- disease (hematological or solid tumor or immune dis- tus (including SpO2, respiratory rate, signs of respiratory ease). To handle the missing values in these con- distress, and bronchial secretion volume), and patient’s founders, multiple imputations with chained equation adherence to other oxygenation modalities. were used, where PS for each patient was averaged across 30 completed datasets while PS matching used Exposures, outcomes, and important covariates these averaged scores. We matched individuals based on The exposure of interest in this prespecified secondary their PS using a 1:1 matching algorithm without replace- analysis of the EFRAIM study was influenza infection sta- ment within a caliper of 0.15 standard deviation of the tus. Patients were divided into four groups for the pur- logit of the PS. Final analyses on the matched dataset poses of analysis: (1) influenza respiratory tract infection were then performed using a logistic regression with a alone, (2) influenza respiratory tract infection plus random effects model on the paired observations, except co-infection, (3) non-influenza respiratory tract infection, for the length of stay which we analyzed with a Cox ran- and (4) no suspected or confirmed respiratory tract infec- dom effects model. Results are presented as odds ratio tion. Influenza was diagnosed by the presence of positive (OR) with their 95%CI. Primary analyses were performed reverse transcription polymerase chain reaction (RT-PCR) on the complete cases, assuming missing completely at in immunosuppressed patients admitted to intensive care random covariates. Sensitivity analyses for such assump- units (ICUs) by a nasopharyngeal swab as it is the optimal tions were performed, based on multiple imputation upper respiratory tract specimen collection method for in- with chained equation. Details of the sample size calcu- fluenza recommended by the CDC [14]. RT-PCR was not lation for the original EFRAIM study can be found else- performed in all patients but mainly in those in whom in- where [13]. A post hoc power analysis was not fluenza infection was suspected. considered appropriate for the current secondary ana- Pulmonary co-infection was defined as either clinically lysis. All tests were two sided at the 0.05 significance or microbiologically confirmed bacterial or invasive fun- level. Analyses were performed using R statistical pack- gal respiratory infection in patients with influenza age (online at http://www.R-project.org). RT-PCR-positive respiratory tract infections [15]. The primary study outcome was all-cause hospital Results mortality. Secondary outcomes included ICU length of Baseline characteristics stay and 90-day mortality. Out of 1611 patients (60% men, median age 63 (IQR Data on important covariates were collected prospect- 54–71)) enrolled in the 68 participating ICUs, 4 expos- ively. SOFA score was recorded at ICU admission [16]. ure groups were defined: patients with influenza respira- Shock was defined as a need for vasopressors; acute kid- tory tract infection alone (n = 95, 5.8%), patients with ney injury (AKI) was defined as a need for renal replace- influenza plus co-infection (n = 58, 3.6%), patients with ment therapy as decided by the treating physicians. non-influenza respiratory tract infection (n = 820, 50.9%), and patient without suspected or confirmed re- Statistical analysis spiratory tract infection (n = 638, 39.6%). We also per- Continuous variables are reported as medians (inter- formed additional analysis on 448 patients negative for quartile ranges [IQRs]) and categorical variables as influenza and testing not done. Martin-Loeches et al. Critical Care (2019) 23:152 Page 4 of 10 Characteristics of each group are summarized in Outcomes on crude, propensity score-matched, and Table 1. There were no statistically significant differences multivariate analysis between the groups regarding general clinical character- Outcomes in the different exposure groups in the crude istics including age and comorbidities. However, patients analysis are summarized in Table 2. ICU mortality dif- with influenza tended to have a higher body mass index fered between the four groups (P < 0.001), with the high- and were more frequently admitted to the ICU directly est mortality in patients with influenza plus co-infection from the emergency department. Patients without any (41%) and non-influenza infection (37%) and slightly respiratory tract infection were less likely to have lower mortality in influenza infection alone (33%) and in hematological disease. SOFA score was higher in pa- those without infection (26%). When the analyses tients with influenza and co-infection, mainly driven by were performed, after excluding those patients nega- higher SOFA respiratory subscores in this group. When tive for influenza but with no testing done, similar re- excluding patients negative for influenza but with no sults were found (Additional file 2: Table S2). testing done, only patients with influenza alone were Hospital and day 90 mortality showed a trend with more likely to have solid tumors (Additional file 1: Table the highest mortality in patients with influenza plus S1). Intubation during the ICU stay was higher and co-infection (52%) and non-influenza infection (46%) shock lower in patients with influenza and co-infection respectively (Table 2 and Additional file 2: Table S2). compared to the other groups (Table 2). Survival curves for ICU and hospital stay for the four Table 1 Influenza infection status and baseline characteristics at ICU admission Baseline characteristics No infection (n Infection other than Influenza alone Influenza co-infection P = 638) influenza (n = 820) (n = 95) (n = 58) valueb Age (years), median [IQR] 63 [54–71] 63 [55–71] 65 [54–72] 64 [52–70] 0.80 Gender, male 351 (55) 512 (63) 59 (63) 32 (56) 0.04 Obesitya 108 (17) 151 (18) 21 (22) 12 (21) 0.21 Underlying disease Hematological disease 311 (49) 436 (53) 60 (63) 30 (51) 0.05 Solid tumor 252 (39) 285 (35) 18 (19) 12 (21) < 0.001 Solid organ transplantation 51 (9) 75 (10) 7 (8) 9 (16) 0.4 Systemic disease or other ID 102 (16) 133 (16) 25 (26) 18 (31) 0.002 Disease status at ICU admission Newly diagnosed 161 (36) 154 (27) 12 (17) 7 (20) 0.0004 Remission 67 (15) 82 (14) 15 (21) 8 (23) No remission 62 (14) 68 (12) 3 (4) 6 (17) Allogeneic stem cell transplant 56 (9) 82 (10) 9 (9) 5 (9) 0.03 ECOGc ≥ 2 (severely disabled or bedridden) 210 (33) 299 (36) 36 (38) 23 (40) 0.15 Comorbidities Cardiac 141 (24) 167 (22) 16 (18) 15 (27) 0.43 COPD 103 (17) 123 (15) 14 (15) 7 (12) 0.80 Kidney 88 (14) 117 (15) 16 (18) 10 (17) 0.75 Diabetes 108 (17) 161 (20) 21 (22) 14 (25) 0.30 Alcohol use disorder 63 (10) 76 (10) 5 (5) 4 (7) 0.48 Tobacco use 199 (33) 228 (29) 21 (23) 12 (21) 0.08 Duration of symptoms before ICU admission 1 [0–4] 1 [0–3] 2 [1–7] 1.5 [1–4] < (days), median [IQR] 0.001 Admission from emergency department 208 (33) 256 (32) 41 (43) 19 (33) 0.17 Neutropenia at admission 66 (11) 153 (20) 20 (21) 12 (21) < 0.001 Data are presented as median [IQR] or N (%) a Obesity grade I, II and extreme obesity b Chi-squared test of association with three degrees of freedom c Eastern Cooperative Oncology Group (ECOG) performance status score Martin-Loeches et al. Critical Care (2019) 23:152 Page 5 of 10 Table 2 Association between influenza infection status, clinical characteristics at day 1, and outcomes Variables No infection (n = Infection other than influenza (n Influenza alone (n Influenza & co-infection (n P 638) = 820) = 95) = 58) valuea At day 1 Maximum respiratory rate 30 [24–36] 31 [25–37] 32 [28–36] 32 [26–38] 0.01 (breaths/min) Liters/min O2 7 [3–15] 8 [5–15] 10 [4–15] 15 [2–15] 0.04 FiO2 50 [40–70] 50 [40–80] 50 [50–72] 59 [51–75] 0.03 PaO2/FiO2 ratio 173 [115–215] 110 [79–173] 113 [110–204] 127 [87–170] < 0.001 ARDS at day 1 481 (75) 737 (90) 92 (97) 57 (98) < 0.001 SOFA at ICU admission 6 [4–9] 7 [4–11] 7 [4–10] 8 [6–10] < 0.001 Respiratory SOFA = 0 103 (17) 94 (12) 9 (10) 1 (2) < 0.001 Cardiovascular SOFA = 0 334 (53) 341 (42) 34 (37) 25 (43) < 0.001 Outcome Intubation during the ICU stay 357 (56) 57 (60) 530 (65) 47 (81) < 0.001 Shock 171 (27) 429 (52) 47 (49) 32 (36) < 0.001 Renal replacement therapy 93 (15) 140 (17) 17 (17) 17 (29) 0.04 b Steroids 187 (33) 272 (36) 27 (31) 27 (49) 0.09 ICU-acquired pneumonia 47 (7) 96 (12) 14 (15) 6 (10) 0.01 ICU length of stay (days) 5 [2–10] 7 [3–15] 8 [4–21] 10.5 [5–20] < 0.001 ICU mortality 165 (26) 302 (37) 31 (33) 24 (41) < 0.001 Hospital mortality 251 (41) 365 (46) 36 (38) 30 (52) 0.09 Day 90 mortality 291 (45) 410 (50) 38 (40) 32 (55) 0.06 Data are presented as median, IQR, or N (%) a Chi-squared test of association with three degrees of freedom b Received steroids in ICU groups are presented in Fig. 1. Hospital survival did mortality. When the analysis was performed in patients not differ by group (P = 0.11). negative for influenza and the test not done, similar re- For the propensity score-matched analysis, 152 pa- sults were found (Table 3). tients with influenza were matched. One patient with in- fluenza could not be matched and was excluded from Discussion the analysis. Imbalances in confounders were reduced In summary, our multinational observational study ana- after matching (Fig. 2). In the matched sample, there lyzed 1611 immunosuppressed patients from 68 centers was no difference in hospital mortality attributable to in- and found that if a critically ill immunosuppressed pa- fluenza infection (OR = 1.01, 95%CI 0.90–1.13, P = 0.85). tient is infected with influenza, the outcome depends on Table 3 shows the results of multivariate analysis after the immunosuppression (independently of the mechan- multiple imputation by chained equations. The following ism and/or type of immunosuppression) rather than in- factors were associated with hospital mortality: age, dir- fluenza infection. We found that independent risk ect ICU admission, severity manifested by SOFA score, factors for hospital mortality were age, organ dysfunc- diagnosis of ARDS, and performance status, with the lat- tion severity, direct admission to the ICU, and especially ter two demonstrating the strongest association with OR a diagnosis of ARDS and performance status. Influenza of hospital mortality of 1.53 and 1.44 respectively. None plus co-infection with bacterial or fungal pathogens was of the mechanism and/or type of immunosuppression associated with the highest ICU mortality rate in our was found as an independent risk factor for hospital study. We did not observe a statistically significant Martin-Loeches et al. Critical Care (2019) 23:152 Page 6 of 10 Fig. 1 Hospital mortality and influenza infection status. Hospital mortality in the whole cohort according to influenza infection status categorized by four groups: (1) patients with influenza alone, (2) patients with influenza plus co-infections (clinically or microbiologically confirmed bacterial or fungal infection), (3) patients with infections other than influenza infection, and (4) patients without infection. Survival curves were compared using Cox regression association between influenza infection status and hos- In our cohort, two important factors stood out as in- pital mortality, our primary outcome, in either crude or dependent risk factors for death: the need for intubation propensity score-matched analyses. during ICU stay and ARDS. While both features were Influenza is a risk factor for acute respiratory failure in associated with increased mortality in any immunosup- immunosuppressed patients. However, its role as an in- pressed patient, the mortality rate approached 100% dependent risk factor for mortality in such a population when this occurred in patients with influenza and has been questioned [9]. The number of immunosup- co-infection. Due to a known protective effect on mor- pressed patients hospitalized with influenza has in- tality of direct admission from the emergency depart- creased in recent years, and we showed that while ment to the ICU, it might be hypothesized that earlier influenza alone may not increase mortality, influenza assessment for severity and therefore earlier ICU admis- plus co-infection may be associated with higher ICU sion may improve the outcomes [18]. In a large popula- mortality. It might therefore be argued that empiric anti- tion of patients with influenza, Álvarez-Lerma et al. [19] biotic treatment for co-infection in such patients should observed that ICU mortality was significantly higher be considered until the possibility of co-infection has among patients with late diagnosis as compared with been confidently ruled out. To facilitate earlier detection early diagnosis (26.9% vs 17.1%, P < 0.001). Diagnostic of co-infection, Rodriguez et al. [17] recently described delay was one independent risk factor for mortality (OR that a low level of procalcitonin (PCT) has a high nega- = 1.36, 95%CI 1.03–1.81, P < 0.001). tive predictive value (94%). However, clinicians may not A common diagnostic challenge in immunosuppressed be willing to tolerate even a low probability of untreated patients is the lack of clinical symptoms when develop- pulmonary co-infection in light of our observation that ing infections. In other words, an immunocompromised this category was associated with higher ICU mortality host is a patient who does not have the ability to re- and length of stay. spond normally to an infection due to an impaired or Martin-Loeches et al. Critical Care (2019) 23:152 Page 7 of 10 Fig. 2 Imbalances in confounders of mortality by influenza infection status before and after propensity score matching. Based on the matched sample, there was no evidence of any difference in hospital mortality across groups (OR = 1.01, 95%CI 0.90–1.13, p = 0.85). We developed a propensity score (PS) logistic model to have flu then matched the individuals on the basis of their PS using a 1:1 matching algorithm without replacement within a caliper of 0.15 standard deviation of the logit of the propensity score. To handle missing values in confounders, multiple imputation with chained equation was used for the PS model, where propensity score for each patient was averaged across 30 completed datasets while propensity score matching used these averaged scores to estimate the treatment effect. Only 1 patient with influenza could not be matched. Imbalances in confounders were reduced after matching Table 3 Multivariate analysis of factors associated with hospital mortality after multiple imputations Variable Assuming non-tested = negative Excluding non-tested OR P value OR P value No infection 1.00 1.00 Influenza alone 0.79 (0.49–1.27) 0.33 0.96 (0.51–1.78) 0.89 Infection other than influenza 1.02 (0.81–1.29) 0.85 1.28 (0.85–1.93) 0.23 Influenza co-infection 1.21 (0.68–2.15) 0.51 1.94 (0.84–3.72) 0.09 Age 1.01 (1.003–1.019) 0.0031 1.01 (1.003–1.019) 0.0051 Direct admission 0.72 (0.57–0.91) 0.0061 0.73 (0.53–0.99) 0.0042 SOFA score 1.14 (1.107–1.171) < 0.0001 1.15 (1.11–1.19) < 0.0001 ARDS 1.53 (1.12–2.10) 0.0084 1.57 (0.97–2.53) 0.065 ECOG 1.44 (1.29–1.61) < 0.0001 1.46 (1.27–1.67) < 0.0001 Mechanism and/or type of immunosuppression was not found as an independent risk factor/s for hospital mortality Martin-Loeches et al. Critical Care (2019) 23:152 Page 8 of 10 weakened immune system. This has been well reported immunosuppressive regimen in the adjusted analysis. in cases of bacterial or fungal infections, but little is The propensity score analysis aims at controlling for known in the case of viral infections [20]. In our cohort, confounders, including those variables associated with we found that patients with influenza were actually more the immunodeficiency that may affect the outcome. likely to have a longer duration of symptoms prior to Nevertheless, one cannot assume that all confounders— ICU admission. possibly even not observed—have been taken into ac- Hypoxemia is a common clinical feature of patients with count and that there may be residual confounders. influenza, especially in immunosuppressed hosts. In a re- cent report from two cancer centers describing the out- Conclusion comes in patients with hematological malignancies and In summary age, severity score, ARDS, and performance influenza infection, severe hypoxemia was an independent status were all independent risk factors for ICU, hospital, risk factor (OR 5.87, 1.12–30.77) for 60-day mortality [21]. and 90-day mortality in immunosuppressed patients ad- Similarly, hypoxemia was clearly a signal of illness severity mitted to the ICU for acute hypoxemic respiratory failure. in our study. In patients not intubated at admission to the The main aim in this paper was to determine if influenza ICU, oxygen requirements and ICU mortality rates were alone or co-infection played a role in the mortality in ICU greatest in those with influenza plus co-infection. patients. Category of infectious etiology of respiratory fail- Co-infection has previously been reported as an inde- ure (influenza, non-influenza, influenza plus co-infection, pendent risk factor for poor outcome in patients with in- and non-infectious) was associated with ICU but not hos- fluenza [9]. In our cohort, patients with co-infection pital mortality. In a propensity score-matched analysis, in- were less likely to be cancer patients (have a fluenza infection was not associated with the primary hematological disease or solid tumor) but were more outcome of hospital mortality. Overall, influenza infection likely to have newly diagnosed immunosuppressive sys- alone is not an independent risk factor for hospital mor- temic disease or have poorer functional capacity. In this tality in immunosuppressed patients. population, the criteria that suggest co-infection and therefore higher severity may be higher oxygen require- Additional files ments, greater tachypnea and work of breathing, and higher rates of mechanical ventilation. Additional file 1: Table S1. Influenza infection status and baseline Systemic immune mechanisms play a key role in the characteristics at ICU admission. Group no infection performed excluding 448 patients negative for influenza and testing not done. (DOCX 18 kb) development of co-infection based on the complexity of Additional file 2: Table S2. Association between influenza infection the interaction of the host and the viral and bacterial status, clinical characteristics at day 1, and outcomes. (DOCX 36 kb) pathogens. Several studies have been performed to de- termine the point prevalence of bacterial co-infection in Acknowledgements influenza patients [9, 22–24]. In our cohort, almost half The authors acknowledge the importance of the Nine-I (Caring for Critically of the patients with co-infection received steroids. The Ill Immunocompromised Patients) group in executing this project, the GRRR- OH for funding the project, and all individuals involved in the data collection use of steroids has been controversial and is currently for the EFRAIM study. not recommended in patients with influenza [25]. This is particularly relevant to our studied population because Funding many patients were already receiving corticosteroid ther- The study was funded by the GRRR-OH (Groupe de Recherche en Réanima- tion Onco-Hématologique), an academic non-profit French organization. apy for their primary disease. It appears plausible that steroids were given as a stress response treatment in pa- Availability of data and materials tients that were using longer-term steroids and not as a Requests for data will be considered by the principal study investigators, based on the nature of the request and legal and ethical regulations. treatment for influenza per se. Importantly, we did not find steroids to be a risk factor for hospital mortality. Authors’ contributions Some limitations should be mentioned. Vaccination IML, VL, EA, and PG were involved in the conception, design, and analysis status and information on antiviral regimen, dose, dur- phases and wrote the paper. MAM, PP, MS, AP, RBB, JR, PB, AL, FST, JS, PS, KR, NT, SM, MA, AK, PK, MV, PPL, FB, FP, VM, ASM, VS, GB, CG, UVAS, LM, FB, LBN, ation, and delay in the start of therapy were not col- BG, and DM collected the data and provided significant scientific comments lected. Similar limitations apply to the determination of for the final manuscript. SC conducted the statistical analysis. EA and IML co-infection, which also could have led to misclassifica- acted as guarantors of the project. All authors read and approved the final manuscript. tion error and bias. The sample contained primarily pa- tients with underlying hematological disease. Other Ethics approval and consent to participate subgroups of immunocompromised patients (particularly Participating investigators obtained local institutional review board approval in accordance with local ethics regulations. patients with lung transplant) may be underrepresented which may limit generalizability. Additionally, we did Consent for publication not completely account for the effect of the type of All authors have reviewed the manuscript and approved the publication. Martin-Loeches et al. Critical Care (2019) 23:152 Page 9 of 10 Competing interests multidrug-resistant bacteria: a case-control study. Crit Care Med. 2007;35: The authors declare that they have no competing interest. 1318–23. 3. Boomer JS, To K, Chang KC, Takasu O, Osborne DF, Walton AH, Bricker TL, Jarman SD, Kreisel D, Krupnick AS, Srivastava A, Swanson PE, Green JM, Publisher’s Note Hotchkiss RS. Immunosuppression in patients who die of sepsis and Springer Nature remains neutral with regard to jurisdictional claims in multiple organ failure. JAMA. 2011;306:2594–605. published maps and institutional affiliations. 4. Kash JC, Taubenberger JK. The role of viral, host, and secondary bacterial factors in influenza pathogenesis. Am J Pathol. 2015;185:1528–36. Author details 1 5. Fowlkes A, Steffens A, Temte J, Di Lonardo S, McHugh L, Martin K, Rubino Medical Intensive Care Unit, Hôpital Saint-Louis and Paris Diderot Sorbonne H, Feist M, Davis C, Selzer C, Lojo J, Oni O, Kurkjian K, Thomas A, Boulton R, University, Paris, France. 2Department of Intensive Care Medicine (710), Bryan N, Lynfield R, Biggerstaff M, Finelli L, Influenza Incidence Surveillance Radboud University Medical Centre, Nijmegen, The Netherlands. 3 Project Working Group. Incidence of medically attended influenza during Department of Critical Care and Graduate Program in Translational pandemic and post-pandemic seasons through the influenza incidence Medicine, Programa de Pós-Graduação em Clínica Médica, D’Or Institute for surveillance project, 2009-13. Lancet Respir Med. 2015;3:709–18. Research and Education, Rio de Janeiro, Brazil. 4Department of Intensive 6. Díaz E, Rodríguez A, Martin-Loeches I, Lorente L, del Mar Martín M, Pozo JC, Care, Rigshospitalet, University of Copenhagen, Copenhagen, Denmark. 5 Montejo JC, Estella A, Arenzana A, Rello J. Impact of obesity in patients CIBERES, Universitat Autonòma de Barcelona, European Study Group of infected with 2009 influenza A(H1N1). Chest. 2011;139:382–6. Infections in Critically Ill Patients (ESGCIP), Barcelona, Spain. 6Pulmonary and 7. Napolitano LM, Angus DC, Uyeki TM. Critically ill patients with influenza Critical Care Medicine, Mayo Clinic, Rochester, MN, USA. 7Division of A(H1N1)pdm09 virus infection in 2014. JAMA. 2014;311(13):1289-90. https:// Pulmonary and Critical Care, Penn State University College of Medicine, doi.org/10.1001/jama.2014.2116. Hershey, PA, USA. 8Department of Intensive Care, Hôpital Erasme, Université 8. Martin-Loeches I, Díaz E, Vidaur L, Torres A, Laborda C, Granada R, Bonastre Libre de Bruxelles (ULB), Brussels, Belgium. 9Department of Intensive Care J, Martín M, Insausti J, Arenzana A, Guerrero JE, Navarrete I, Bermejo-Martin Medicine, Multidisciplinary Intensive Care Research Organization (MICRO), St. J, Suarez D, Rodriguez A. Pandemic and post-pandemic influenza A (H1N1) James’s Hospital, Dublin, Ireland. 10Department of Clinical Medicine, infection in critically ill patients. Crit Care. 2011;15:R286. Wellcome Trust-HRB Clinical Research Facility, St. James Hospital, Trinity 9. Martin-Loeches I, Schultz JM, Vincent J-L, Alvarez-Lerma F, Bos LD, Solé- College, Dublin, Ireland. 11Department of Medicine I, Medical University of Violán J, Torres A, Rodriguez A, Sole-Violan J, Torres A, Rodriguez A. Vienna, Vienna, Austria. 12Department of Anesthesiology and Intensive Care Increased incidence of co-infection in critically ill patients with influenza. Medicine and Institute for Medical Humanities, 1st Faculty of Medicine, Intensive Care Med. 2017;43(1):48-58. https://doi.org/10.1007/s00134-016- Charles University in Prague and General University Hospital, Prague, Czech 4578-y. Epub 2016 Oct 5. Republic. 13Norwegian University of Science and Technology, Trondheim, Norway. 14CHU Grenoble Alpes, Service de Réanimation Médicale, Faculté de 10. Cawcutt K, Kalil AC. Pneumonia with bacterial and viral coinfection. Curr Médecine de Grenoble, INSERM U1042, Université Grenoble-Alpes, Grenoble, Opin Crit Care. 2017;23:385–90. France. 15Department of Medicine and Interdepartmental Division of Critical 11. White DB, Angus DC. Preparing for the sickest patients with 2009 influenza Care Medicine, Sinai Health System, University of Toronto, Toronto, Ontario, A(H1N1). JAMA. 2009;302:1905–6. Canada. 16Agostino Gemelli University Hospital, Università Cattolica del Sacro 12. Visseaux B, Burdet C, Voiriot G, Lescure F-X, Chougar T, Brugière O, Crestani Cuore, Rome, Italy. 17Department of Medical Intensive Care Medicine, B, Casalino E, Charpentier C, Descamps D, Timsit J-F, Yazdanpanah Y, University Hospital of Angers, Angers, France. 18Department of Immunology– Houhou-Fidouh N. Prevalence of respiratory viruses among adults, by Department of Emergencies and Critical Care, University of Oslo, Oslo, season, age, respiratory tract region and type of medical unit in Paris, Norway. 19Division of Intensive Care Medicine, Department of France, from 2011 to 2016. PLoS One. 2017;12:e0180888. Anesthesiology, Intensive Care and Pain Medicine, Helsinki University 13. Azoulay E, Pickkers P, Soares M, Perner A, Rello J, Bauer PR, van de Louw A, Hospital, University of Helsinki, Helsinki, Finland. 20Department of Critical Hemelaar P, Lemiale V, Taccone FS, Martin Loeches I, Meyhoff TS, Salluh J, Care, University Medical Center Groningen, Groningen, The Netherlands. Schellongowski P, Rusinova K, Terzi N, Mehta S, Antonelli M, Kouatchet A, 21 Medical-Surgical Intensive Care Unit, Centre Hospitalier de Versailles, Le Barratt-Due A, Valkonen M, Landburg PP, Bruneel F, Bukan RB, Pène F, Chesnay, France. 22Department of Anesthesiology I, Herlev University Metaxa V, Moreau AS, Souppart V, Burghi G, Girault C, et al. Acute Hospital, Herlev, Denmark. 23Medical ICU, Cochin Hospital, Assistance hypoxemic respiratory failure in immunocompromised patients: the Efraim Publique-Hôpitaux de Paris and University Paris Descartes, Paris, France. multinational prospective cohort study. Intensive Care Med. 2017;43(12): 24 Critical Care Department, King’s College Hospital NHS Foundation Trust, 1808-19. https://doi.org/10.1007/s00134-017-4947-1. Epub 2017 Sep 25. London SE5 9RS, UK. 25Critical Care Center, CHU Lille, School of Medicine, 14. https://www.cdc.gov/flu/pdf/freeresources/healthcare/flu-specimen- University of Lille, Lille, France. 26Terapia Intensiva, Hospital Maciel, collection-guide.pdf Acccesed 14 Mar 12, 2019. Montevideo, Uruguay. 27Department of Medical Intensive Care, Normandie 15. Contejean A, Lemiale V, Resche-Rigon M, Mokart D, Pène F, Kouatchet A, Univ, UNIROUEN, EA-3830, Rouen University Hospital, F-76000 Rouen, France. Mayaux J, Vincent F, Nyunga M, Bruneel F, Rabbat A, Perez P, Meert A-P, 28 ICU, Fundação Pio XII - Barretos Cancer Hospital, Barretos, Brazil. 29Medical Benoit D, Hamidfar R, Darmon M, Jourdain M, Renault A, Schlemmer B, Intensive Care Unit, La Source Hospital - CHR Orléans, Orléans, France. Azoulay E. Increased mortality in hematological malignancy patients with 30 Intensive Care Department, University of Southern Denmark, Sønderborg, acute respiratory failure from undetermined etiology: a Groupe de Denmark. 31Department of Anaesthesia and Intensive Care, Odense Recherche en Réanimation Respiratoire en Onco-Hématologie (GRRR-OH) University Hospital, Odense, Denmark. 32Medical Intensive Care Unit, Hôtel study. Ann Intensive Care. 2016;6:102. Dieu-HME-University Hospital of Nantes, Nantes, France. 33Réanimation 16. J T, Vincent JL, Moreno R, Takala J, Willatts S, De Mendonça A, Bruining H, Polyvalente et Département d’Anesthésie et de Réanimation, Institut Reinhart CK, Suter PM, Thijs LG. The SOFA (Sepsis-related Organ Failure Paoli-Calmettes, Marseille, France. 34ECSTRA Team, Biostatistics and Clinical Assessment) score to describe organ dysfunction/failure. On behalf of the Epidemiology, UMR 1153, INSERM, Paris Diderot Sorbonne University and working group on Sepsis-Related Problems of the European Society of Service de Biostatistique et Information Médicale AP-HP, Hôpital Saint-Louis, Intensive Care Medicine. Intensive Care Med. 1996;22:707–10. Saint-Louis, France. 35Department of Intensive Care Medicine, St. James’s 17. Rodríguez AH, Avilés-Jurado FX, Díaz E, Schuetz P, Trefler SI, Solé-Violán J, Hospital, St. James’s St, Dublin, Dublin 8, Ireland. Cordero L, Vidaur L, Estella Á, Pozo Laderas JC, Socias L, Vergara JC, Zaragoza R, Bonastre J, Guerrero JE, Suberviola B, Cilloniz C, Restrepo MI, Received: 24 January 2019 Accepted: 9 April 2019 Martín-Loeches I, Cobo P, Martins J, Carbayo C, Robles-Musso E, Cárdenas A, Fierro J, Fernández DO, Sierra R, Huertos MJ, Carmona Pérez ML, Pozo Laderas JC, et al. Procalcitonin (PCT) levels for ruling-out bacterial References coinfection in ICU patients with influenza: a CHAID decision-tree analysis. J 1. Azoulay E, Afessa B. The intensive care support of patients with malignancy: Inf Secur. 2016;72:143–51. do everything that can be done. Intensive Care Med. 2006;32:3–5. 18. Martin-Loeches I, Levy MMMM, Artigas A. Management of severe sepsis: 2. Nseir S, Di Pompeo C, Diarra M, Brisson H, Tissier S, Boulo M, Durocher A. advances, challenges, and current status. Drug Des Devel Ther. 2015;9: Relationship between immunosuppression and intensive care unit-acquired 2079–88. Martin-Loeches et al. Critical Care (2019) 23:152 Page 10 of 10 19. Álvarez-Lerma F, Marín-Corral J, Vilà C, Masclans JR, Loeches IM, Barbadillo S, González de Molina FJ, Rodríguez A, H1N1 GETGAG/SEMICYUC Study Group. Characteristics of patients with hospital-acquired influenza A (H1N1)pdm09 virus admitted to the intensive care unit. J Hosp Infect. 2017; 95:200–6. 20. Zafrani L, Azoulay E. How to treat severe infections in critically ill neutropenic patients? BMC Infect Dis. 2014;14:512. 21. Vilar-Compte D, Shah DP, Vanichanan J, Cornejo-Juarez P, Garcia-Horton A, Volkow P, Chemaly RF. Influenza in patients with hematological malignancies: experience at two comprehensive cancer centers. J Med Virol. 2018;90:50–60. 22. Martin-Loeches I, Rodriguez A, Sanchez-Corral A, Granada R, Zaragoza R, Albaya A, Cerda E, Catalan R, Luque P, Paredes A, et al.: Bacterial co- infection in critically ill patients infected with pandemic (H1N1) v influenza A infection. In Intensive Care Med. Volume 36; 2010:S369--S369. 23. Matos RG, Moreno RP, Diogo AC, Pereira JM, Martin-Loeches I, Cecconi M, Lisboa T, Rhodes A, Rello J. Bacterial pneumonia complicating influenza A (H1N1) v viral pneumonia: results of the ESICM influenza A (H1N1) v registry. In: Intensive care med. Volume 36; 2010. p. S371. 24. Muscedere J, Ofner M, Kumar A, Long J, Lamontagne F, Cook D, McGeer A, Chant C, Marshall J, Jouvet P, Fowler R, ICU-FLU Group, Canadian Critical Care Trials Group. The occurrence and impact of bacterial organisms complicating critical care illness associated with 2009 influenza A(H1N1) infection. Chest. 2013;144:39–47. 25. Mc Mahon A, Martin-Loeches I. The pharmacological management of severe influenza infection - 'existing and emerging therapies'. Expert Rev Clin Pharmacol. 2017;10(1):81-95. https://doi.org/10.1080/17512433.2017. 1255550. Epub 2016 Nov 25.