Influenza A (H1N1pdm09)-Related Critical Illness and Mortality in Mexico and Canada, 2014

Transcription

1 Influenza A (H1N1pdm09)-Related Critical Illness and Mortality in Mexico and Canada, 2014 Guillermo Dominguez-Cherit, MD 1,2 ; Alethse De la Torre, MD, MSc, PHDC 3 ; Asgar Rishu, MBBS 4 ; Ruxandra Pinto, PhD 5 ; Silvio A. Ñamendys-Silva, MD, MSc 6 ; Adrián Camacho-Ortiz, MD 7 ; Marco Antonio Silva-Medina, MD 8 ; Carmen Hernández-Cárdenas, MD 9 ; Michel Martínez-Franco, MD 10 ; Alejandro Quesada-Sánchez, MD 11 ; Guadalupe Celia López-Gallegos, MD 12 ; Juan L. Mosqueda-Gómez, MD 13 ; Norma E. Rivera-Martinez, MD 14 ; Fernando Campos-Calderón, MD 15 ; Eduardo Rivero-Sigarroa, MD 1 ; Thierry Hernández-Gilsoul, MD 16 ; Lourdes Espinosa-Pérez, MD 17 ; Alejandro E. Macías, MD, MSc 18 ; Dolores M. Lue-Martínez, MD 7 ; Christian Buelna-Cano, MD 10 ; Ana-Sofía Ramírez-García Luna, MD 11 ; Nestor G. Cruz-Ruiz, MD 19 ; Manuel Poblano-Morales, MD 20 ; Fernando Molinar-Ramos, MD 21 ; Martin Hernandez-Torre, MD 22 ; Marco Antonio León-Gutiérrez, MD 23 Oscar Rosaldo-Abundis, MD 24 ; José Ángel Baltazar-Torres, MD 25 ; Henry T. Stelfox, MD, PhD 26 ; Bruce Light, MD 27 ; Philippe Jouvet, MD, PhD 28 ; Steve Reynolds, MD 29 ; Richard Hall, MD 30 ; Nikki Shindo, MD, PhD 31 ; Nick Daneman, MD, MSc 32 ; Robert A. Fowler, MDCM, MS(Epi) 33 1 Department of Critical Care Medicine, Instituto Nacional de Ciencias Médicas y Nutrición Salvador Zubirán, Mexico City, Mexico. 2 Escuela de Medicina, Tecnologico de Monterrey, Monterrey, Mexico. 3 Department of Hospital Epidemiology, Instituto Nacional de Ciencias Médicas y Nutrición Salvador Zubirán, Mexico City, Mexico. 4 Trauma, Emergency and Critical Care, Sunnybrook Health Sciences Centre, University of Toronto, Toronto, ON, Canada. 5 Department of Critical Care Medicine, Sunnybrook Health Sciences Centre, University of Toronto, Toronto, ON, Canada. 6 Department of Critical Care Medicine, Instituto Nacional de Cancerología, Mexico City, Mexico. 7 Department of Infectious Diseases, Hospital universitario Dr. José Eleuterio González, Monterrey, Nuevo León, Mexico. 8 Department of Critical Care Medicine and Emergency, Instituto de Salud del Estado de México, Toluca, Mexico. 9 Department of Critical Care Medicine, Instituto Nacional de Enfermedades Respiratorias, Mexico City, Mexico. 10 Department of Critical Care Medicine, Hospital General de Tijuana, Tijuana, Baja California, Mexico. 11 Department of Critical Care Medicine, Hospital Central Ignacio Morones Prieto, San Luis Potosi, Mexico. 12 Department of Critical Care Medicine, Hospital General de Uruapan, Uruapan, Michoacan, Mexico. 13 Department of Infectious Diseases, Hospital General Regional de León, León, Guanajuato, Mexico. 14 Department of Infectious Diseases, Hospital Regional de Alta Especialidad de Oaxaca, Oaxaca, Mexico. 15 Department of Critical Care Medicine, Hospital General Dr. Manuel Gea González, Mexico City, Mexico. Copyright 2016 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved. DOI: /CCM Department of Disasters Response, Instituto Nacional de Ciencias Médicas y Nutrición Salvador Zubirán, Mexico City, Mexico. 17 Department of Critical Care Medicine, Hospital San José Monterrey, Monterrey, Nuevo León, Mexico. 18 Faculty of Medicine, Universidad de Guanajuato, León, Guanajuato, Mexico. 19 Department of Critical Care Medicine, Hospital Regional de Alta Especialidad de Oaxaca, Oaxaca, Mexico. 20 Department of Critical Care Medicine, Hospital Juárez de México, Mexico City, Mexico. 21 Department of Critical Care Medicine, Hospital General Dr. Rubén Leñero, Mexico City, Mexico. 22 Department of Critical Care Medicine, Hospital Zambrano Hellion, Monterrey, Nuevo León, Mexico. 23 Department of Critical Care Medicine, Centro Médico Nacional Siglo XXI, Mexico City, Mexico. 24 Department of Critical Care Medicine, Hospital General de Chetumal, Clínica Carranza, Chetumal, Quintana Roo, Mexico. 25 Department of Critical Care Medicine, Hospital de Especialidades, Centro Médico Nacional la Raza, IMSS, Mexico City, Mexico. 26 Foothills Hospital, Department of Critical Care Medicine, University of Calgary, Calgary, Alberta, Canada. 27 Department of Medicine and Critical Care Medicine, St. Boniface General Hospital, University of Manitoba, Winnipeg, MB, Canada. 28 Département de Pédiatrics et Soins Intensifs, Hôpital Ste-Justine, l Université de Montréal, Montréal, QC, Canada. 29 Department of Medicine and Critical Care Medicine, Royal Columbian Hospital, University of British Columbia, New Westminster, BC, Canada. 30 Department of Anesthesia and Critical Care Medicine, Queen Elizabeth II Hospital, Dalhousie University, Halifax, NS, Canada. 31 Department of Pandemic and Epidemic Diseases, World Health Organization, Geneva Switzerland. 32 Department of Medicine, Sunnybrook Health Sciences Centre, University of Toronto, Toronto, ON, Canada. Critical Care Medicine 1

2 Dominguez-Cherit et al 33 Department of Medicine and Department of Critical Care Medicine, Sunnybrook Health Sciences Centre, University of Toronto, Toronto, ON, Canada. Drs, Dominguez-Cherit, Torre, Rishu, Daneman, and Fowler have been involved in the design of the study; Drs. Dominguez-Cherit, Torre, Rishu, Pinto, Ñamendys-Silva, Camacho-Ortiz, Silva-Medina, Hernández- Cárdenas, Martínez-Franco, Quesada-Sánchez, Celia López-Gallegos, Mosqueda-Gómez, Rivera-Martinez, Campos-Calderón, Rivero-Sigarroa, Hernández-Gilsoul, Espinosa-Pérez, Macías, Lue-Martínez, Buelna-Cano, Ramírez-García Luna, Cruz-Ruiz, Poblano-Morales, Molinar-Ramos, Hernandez-Torre, León-Gutiérrez, Rosaldo-Abundis, Baltazar-Torres, Stelfox, Light, Jouvet, Reynolds, Hall, Daneman, and Fowler have been involved in data acquisition and data extraction; Drs. Dominguez-Cherit, Torre, Rishu, Pinto, Ñamendys-Silva, Camacho-Ortiz, Silva-Medina, Hernández- Cárdenas, Martínez-Franco, Quesada-Sánchez, Celia López-Gallegos, Mosqueda-Gómez, Rivera-Martinez, Campos-Calderón, Rivero-Sigarroa, Hernández-Gilsoul, Espinosa-Pérez, Macías, Lue-Martínez, Buelna-Cano, Ramírez-García Luna, Cruz-Ruiz, Poblano-Morales, Molinar-Ramos, Hernandez-Torre, León-Gutiérrez, Rosaldo-Abundis, Baltazar-Torres, Stelfox, Light, Jouvet, Reynolds, Hall, Shindo, Daneman, and Fowler have been involved in analysis and interpretation of the data; all authors have contributed to writing the article prior to submission. Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal s website ( ccmjournal). Supported, in part, by the Public Health Agency of Canada, and the Heart and Stroke Foundation (ON, Canada). Dr. Torre received funding from Pfizer and GSK. Dr. Hernández-Cárdenas disclosed government work. Dr. Martínez-Franco disclosed government work. Dr. Poblano-Morales disclosed government work. Dr. Molinar- Ramos disclosed employment (secretaria de salud del Distrito Federal) and disclosed government work. Dr. Stelfox s institution received funding from Canadian Public Health Agency. Dr. Jouvet disclosed other support (NM3 monitor lent by Philips Medical, Respirator lent by Maquet Medical, Respirator lent by Hamilton medical). Dr. Hall disclosed other support (Canadian Institutes for Health Research Nova Scotia Heart and Stroke Foundation Asahei Kasei Pharma Bristol Myers Squibb Astra Zeneca La Jolla Pharma). His institution received funding from the Canadian Public Service Agency. Dr. Fowler s institution received funding from the Public Health Agency of Canada and the Heart and Stroke Foundation (ON, Canada). The remaining authors have disclosed that they do not have any potential conflicts of interest. Address requests for reprints to: Robert A. Fowler, MDCM, MS(Epi), Department of Critical Care Medicine, 2075 Bayview Avenue, Room D478, Sunnybrook Hospital, Toronto, ON, Canada. rob.fowler@ sunnybrook.ca Objectives: The influenza A (H1N1pdm09) pandemic caused substantial morbidity and mortality among young patients; however, mortality estimates have been confounded by regional differences in eligibility criteria and inclusion of selected populations. In , H1N1pdm09 became North America s dominant seasonal influenza strain. Our objective was to compare the baseline characteristics, resources, and treatments with outcomes among critically ill patients with influenza A (H1N1pdm09) in Mexican and Canadian hospitals in 2014 using consistent eligibility criteria. Design: Observational study and a survey of available healthcare setting resources. Setting: Twenty-one hospitals, 13 in Mexico and eight in Canada. Patients: Critically ill patients with confirmed H1N1pdm09 during influenza season. Interventions: None. Measurements and Main Results: The main outcome measures were 90-day mortality and independent predictors of mortality. Among 165 adult patients with H1N1pdm09-related critical illness between September 2013 and March 2014, mean age was 48.3 years, 64% were males, and nearly all influenza was community acquired. Patients were severely hypoxic (median Pao 2 ratio, 83 mm Hg), 97% received mechanical ventilation, with mean positive end-expiratory pressure of 14 cm H 2 O at the onset of critical illness and 26.7% received rescue oxygenation therapy with prone ventilation, extracorporeal life support, high-frequency oscillatory ventilation, or inhaled nitric oxide. At 90 days, mortality was 34.6% (13.9% in Canada vs 50.5% in Mexico, p < ). Independent predictors of mortality included lower presenting Pao 2 ratio (odds ratio, 0.89 per 10-point increase [95% CI, ]), age (odds ratio, 1.49 per 10 yr increment [95% CI, ]), and requiring critical care in Mexico (odds ratio, 7.76 [95% CI, ]). ICUs in Canada generally had more beds, ventilators, healthcare personnel, and rescue oxygenation therapies. Conclusions: Influenza A (H1N1pdm09)-related critical illness still predominantly affects relatively young to middle-aged patients and is associated with severe hypoxemic respiratory failure. The local critical care system and available resources may be influential determinants of patient outcome. (Crit Care Med 2016; XX:00 00) Key Words: Canada; critical; influenza; intensive; Mexico In , a new strain of influenza A was discovered as the cause of severe respiratory failure among unprecedented numbers of young patients in Mexico and the southwestern United States (1). The H1N1pdm09 influenza strain subsequently resulted in the century s first pandemic (2). In 2013, H1N1pdm09 became North America s dominant seasonal influenza strain (3) and many jurisdictions experienced a new wave of H1N1pdm09-related critical illness (4, 5). During the Influenza A (H1N1) pandemic, it is estimated that up to 3% of the world s population may have died (6 8). Now, however, patients who develop severe influenza-related illness are admitted to hospital, receive antiviral medications, and receive treatment for secondary bacterial infections. Affected critically ill can be admitted to an ICU for organ support, mechanical ventilation, and rescue extracorporeal oxygenation therapy when necessary (9). There is substantial prior evidence that outcomes of critical illness are dependent not only upon the precipitating cause, patient characteristics, and severity of illness, but also upon the setting in which patients receive care (10 13). During the early pandemic, critical illness related mortality was substantially higher in Mexico than in Canada; however, this may have been confounded by inclusion of selected populations with more severe illness as commonly occurs early in emerging outbreaks, and by an excess in demand relative to the availability of intensive care in Mexico (11 13). To clarify these issues, we performed an observational study using uniform definitions of H1N1pdm09-related critical illness in Canadian and Mexican hospitals during the influenza season to determine outcomes and investigate predictors of these outcomes. 2 XXX 2016 Volume XX Number XXX

3 Clinical Investigation MATERIALS AND METHODS Study Design and Eligibility Criteria We designed a prospective surveillance study of all hospitalized patients for H1N1pdm09-related severe acute respiratory infection. Patients were included if they were critically ill and met eligibility criteria, defined by 1) critical illness (hypoxemia [Pao 2 ratio, 300 mm Hg irrespective of application of positive airway pressure] and radiographic opacities consistent with pneumonia [not fully explained by effusions, lobar/ lung collapse, nodules, or cardiogenic pulmonary edema]); or, receipt of invasive or noninvasive ventilation; or, admission to an ICU; or, receipt of IV vasoactive medication for blood pressure homeostasis; and 2) diagnosis of influenza A (H1N1pdm09) by real-time polymerase chain reaction (PCR), viral culture, or a four-fold rise in virus-specific neutralizing antibodies. Participant Identification Research coordinators and site study investigators at participating centers in Canada and Mexico screened patients in ICU and other areas of the hospital that may provide care for patients with critical illness (e.g., high dependency wards, areas of common overflow when ICU occupancy is high), prospectively, and when necessary, retrospectively for severe acute respiratory infection. Although research coordinators or site investigators could prompt treating clinical teams to perform diagnostic testing for viral pathogens, ultimately, influenza A (H1N1pdm09) testing was at the discretion of the clinical team and under the guidance of site-specific laboratory PCR, viral culture, or neutralizing antibody serology techniques. Data Collection Data were collected on a paper-based case report form that was then transcribed into an electronic case report form maintained by the coordinating center. The case report form was adapted and modified from prior studies used in Canada and Mexico during the pandemic (11, 12). The new case report form underwent multiple rounds of trialing and feedback by research coordinators before completion (supplemental data, Supplemental Digital Content 1, com/ccm/b843). Data included an anonymized identifier, eligibility criteria, possible exposures (contact, travel, occupational, and animal), demographics and characteristics (sex, age, height, weight, ethnicity, and occupation), vaccination history, comorbidities, date of onset and nature of symptoms, pregnancy or postpartum state, copresenting illnesses, severity of illness by the Sequential Organ Failure Assessment (SOFA) score components, mode of ventilation and degree of oxygenation assistance on the day that eligibility criteria were fulfilled, and chest radiograph findings (14). We also recorded dates of hospital and ICU admission and discharge or death (and primary cause of death), initiation(s) and liberation(s) dates from noninvasive and invasive mechanical ventilation, receipt of high-frequency oscillatory ventilation, extracorporeal life support, inhaled nitric oxide, prone ventilation, dialysis; systemic oral or IV corticosteroids; neuraminidase inhibitor, other antiviral medication, antibiotic medication, and the type, dose, route, and duration of each, in addition to microbiologic and virologic culture results, date, and source. Study sites in Canada and Mexico were invited to participate through collaborations within the Canadian Critical Care Trials Group ( and prior collaborations with the Mexico pandemic influenza study group (12). Study site (hospital and ICU) characteristics was also collected, including location; type of patient population (medical, surgical, trauma, pediatric, adult, and so on); teaching or nonteaching status; number of hospital and ICU beds; number of noninvasive and invasive mechanical ventilators theoretically accessible to critically ill patients; presence of extracorporeal life support, prone ventilation, high-frequency oscillatory ventilation, inhaled nitric oxide, dialysis; availability of neuraminidase inhibitors, antibiotics, vasopressors, IV sedation, infusion pumps, urinary catheters, electronic vital sign monitors, hand hygiene assistance, nonsterile gloves, medical or N-95 equivalent face masks; number of doctors, nurses, and respiratory therapists routinely present during day and night shifts; estimates of daily ICU occupancy, proportion of patients in ICU receiving mechanical ventilation, and average severity of illness scores, where available. All data underwent automated electronic error checks using prespecified plausible ranges, and then manual inspection and data queries were performed for all uncertain or missing fields by the coordinating center. Our objective was to compare the baseline characteristics, resources, and treatments with outcomes among critically ill patients with influenza A (H1N1pdm09) in Mexican and Canadian hospitals in 2014 using consistent eligibility criteria. The primary outcome of the study was mortality at 90 days and the secondary outcome included independent predictors of 90-day mortality. Statistical Analyses Descriptive data are presented as counts and frequencies (%) for discrete variables and means (sd) or medians (interquartile range [IQR]) for continuous variables. To determine if there were differences in baseline characteristics between survivors and nonsurvivors, we used a two-sample t test or the Wilcoxon signed rank test for continuous variables and a chisquare test or Fisher exact test for the discrete variables. We conducted simple logistic regression using a priori defined clinical variables (body mass index [BMI], time from symptoms to ICU admission, SOFA score at admission, Pao 2 ratio) to explore associations with mortality. We used a collinearity matrix and examined Pearson correlation coefficients to ensure no variables had a correlation larger than 0.8. By visually inspecting the linearity between hospital mortality and linear predictors, we did not detect any major departure from this assumption. We identified all variables with p value less than 0.20 and entered them into the multiple logistic regression model for the outcome of mortality at 90 days, including a country variable (Mexico vs Canada) and ICU clinician staffing (per 10 ICU beds), and finally added age to the model. We assessed model fit using the Hosmer-Lemeshow test. We Critical Care Medicine 3

4 Dominguez-Cherit et al performed a sensitivity analysis to account for clustering of patients within individual hospitals using generalized estimating equations, with similar and different correlation structures for each country. The Kaplan-Meier method, where all people discharged from hospital alive were censored at 90 days, was used to determine and compare the probability of survival over the duration of follow-up in Canada versus Mexico and to generate survival curves. Between-group differences were evaluated with the log-rank test. All statistical tests were two tailed and variables were considered statistically significant at α less than The SAS System for Windows version 9.3 (SAS Institute, Cary, NC) and R (The R Project for Statistical Computing, were used for all analyses. Ethics, Funding, and Oversight Research ethics board approval was obtained from all participating sites. The need for a priori informed consent was waived given the noninterventional study design. This study was assisted through funding from the Public Health Agency of Canada and the Heart and Stroke Foundation (ON, Canada). Funders had no input into the study design, data collection, analysis, interpretation of the findings, or decision to submit for peer review and publication. RESULTS Patients There were 165 patients (93 in Mexico and 72 in Canada) with H1N1pdm09-related critical illness admitted to the 21 participating hospitals (13 in Mexico and eight in Canada) between September 1, 2013, and March 31, The mean age was 48.3 years (sd, 14.3), 64% were males, the mean BMI was 32.0 kg/m 2 (8.7), and nearly all infections were community acquired (Table 1). Patients had a median of 2 (IQR, 1 3) comorbid conditions, most commonly obesity, hypertension, diabetes, or a history of smoking (Table 1). Patients with H1N1pdm09-related critical illness in Mexico (compared to Canada) were slightly younger, more frequently men, and of Latin-American heritage (Table 1). Patients in Mexico were less likely to have hypertension and pulmonary or gastrointestinal disease (Table 1). At presentation, critically ill patients in Mexico had lower arterial oxygen tension and higher applied positive end-expiratory pressure (PEEP) levels but similar resulting Pao 2 ratios and SOFA score. Presentation and Illness Course The most common presenting symptoms included cough (84.2%), fever (81.2%), and shortness of breath (72.1%). From Table 1. Characteristics of Critically Ill Patients With Confirmed Influenza A (H1N1pdm09) Infection Variable All (n = 165) Canada (n = 72) Mexico (n = 93) p Age, yr, mean (sd) 48.3 (14.3) 51.5 (15.2) 45.8 (13.1) Male, n (%) 106 (64.2) 38 (52.8) 68 (73.1) BMI, kg/m 2, mean (sd) 32.0 (8.7) 32.0 (9.0) 31.9 (8.5) Health care worker, n (%) Yes 1 (0.61) 1 (1.4) 0 (0) No 154 (93.9) 63 (88.7) 91 (97.9) Unknown 9 (5.5) 7 (9.9) 2 (2.2) Nosocomial acquisition, n (%) 4 (2.4) 3 (4.2) 1 (1.1) Race/ethnicity, n (%) White/Caucasian 20 (12.1) 20 (27.8) 0 (0) < Black 2 (1.2) 2 (2.8) 0 (0) Asian 4 (2.4) 4 (5.6) 0 (0) South Asian 2 (1.2) 2 (2.8) 0 (0) Arab/West Asian 1 (0.61) 1 (1.4) 0 (0) Latin American 94 (57) 1 (1.4) 93 (100) First Nations 5 (3.0) 5 (6.9) 0 (0) Unknown/mixed 37 (22.4) 37 (51.4) 0 (0) Overall comorbidities, median (IQR) 2.0 ( ) 2.0 ( ) 2.0 ( ) a (Continued) 4 XXX 2016 Volume XX Number XXX

5 Clinical Investigation Table 1. (Continued). Characteristics of Critically Ill Patients With Confirmed Influenza A (H1N1pdm09) Infection Variable All (n = 165) Canada (n = 72) Mexico (n = 93) p Comorbidities, n (%) None 14 (8.6) 9 (12.5) 5 (5.6) BMI (51.6) 35 (53.9) 45 (50) Hypertension 53 (32.7) 30 (41.7) 23 (25.6) Ever smoker 49 (30.3) 25 (34.7) 24 (26.7) Diabetes mellitus 42 (25.9) 21 (29.2) 21 (23.3) Any chronic lung disease 28 (17.3) 21 (29.2) 7 (7.8) < Chronic obstructive pulmonary disease 19 (11.7) 13 (18.1) 6 (6.7) Asthma 12 (7.4) 11 (15.3) 1 (1.1) Obstructive sleep apnea 3 (1.9) 3 (4.2) 0 (0) Other chronic lung disease 2 (1.2) 2 (2.8) 0 (0) Cardiac disease 20 (12.4) 11 (15.3) 9 (10) Renal insufficiency/dialysis dependency 17 (10.5) 7 (9.7) 10 (11.1) Immune suppression 16 (9.9) 10 (13.9) 6 (6.7) Corticosteroid use 9 (5.6) 5 (6.9) 4 (4.4) Organ transplantation 5 (3.2) 2 (2.8) 3 (3.3) Chemotherapy 2 (1.2) 1 (1.4) 1 (1.1) HIV 2 (1.2) 1 (1.4) 1 (1.1) Other immunosuppression 4 (2.5) 2 (2.8) 2 (2.2) Cerebrovascular disease 4 (2.5) 2 (2.8) 2 (2.2) Seizure disorder 3 (1.9) 3 (4.2) 0 (0) Other neurologic disease 6 (3.7) 5 (6.9) 1 (1.1) Any malignancy 12 (7.4) 7 (9.7) 5 (5.7) Lymphoma 5 (3.1) 2 (2.8) 3 (3.3) Leukemia 4 (2.5) 2 (2.8) 2 (2.2) Cancer, nonmetastatic 3 (1.9) 3 (4.2) 0 (0) Cancer, metastatic 1 (0.6) 0 (0) 1 (1.1) Hyperlipidemia 11 (6.8) 7 (9.7) 4 (4.4) Gastrointestinal disease 6 (3.7) 6 (8.3) 0 (0) Liver disease 4 (2.5) 2 (2.8) 2 (2.2) Systemic autoimmune disorder 3 (1.9) 1 (1.4) 2 (2.2) Rheumatologic disease 3 (1.9) 3 (4.2) 0 (0) Peripheral vascular disease 2 (1.2) 1 (1.4) 1 (1.1) Pregnancy 2 (1.2) 2 (2.8) 0 (0) Dementia 1 (0.6) 0 (0) 1 (1.1) Scoliosis 1 (0.6) 1 (1.4) 0 (0) Cerebral palsy/development delay 1 (0.6) 1 (1.4) 0 (0) Other comorbidity 26 (16.1) 19 (26.4) 7 (7.8) (Continued) Critical Care Medicine 5

6 Dominguez-Cherit et al Table 1. (Continued). Characteristics of Critically Ill Patients With Confirmed Influenza A (H1N1pdm09) Infection Variable All (n = 165) Canada (n = 72) Mexico (n = 93) p Sequential Organ Failure Assessment score, mean (sd) 7.8 (3.2) 7.6 (2.9) 8 (3.4) Respiratory parameters, mean (sd) b Pao 2, mm Hg, median (IQR) 83 (59 119) 90 (60 130) 80 (58 113) Positive end-expiratory pressure, cm H 2 O 13.9 (4.4) 12.7 (3.9) 14.8 (4.6) Pao 2, mm Hg 66 (21) 71 (23) 62.2 (19) Mechanical ventilation, n (%) 160 (97) 72 (100) 88 (94.6) Vasopressor/inotropic, n (%) 99 (60) 44 (61.1) 55 (59.1) Laboratory findings, median (IQR) b Creatinine, μmol/l 97.2 (62 167) 89 ( ) 103 ( ) Platelets 10 3 /μl 156 ( ) 158 ( ) 152 ( ) WBC 10 9 /μl 7.9 ( ) 8.1 ( ) 7.9 (5 12.4) Chest radiograph, n (%) 161 (97.6) 69 (95.8) 92 (98.92) Unilateral opacities 10 (6.5) 5 (7.9) 5 (5.4) Bilateral opacities 145 (93.6) 58 (92.1) 87 (94.6) BMI = body mass index, IQR = interquartile range. a Mann-Whitney U tests compares distributions of the two groups and their quartiles, which despite similar median values are statistically different with greater comorbidity among Canadian patients. b Unless specified. symptom onset, patients were admitted to hospital after a median of 4 days (2 7) in Canada versus 7 days (4 7) in Mexico (p = ) and to ICU within 5 days (3 7) in Canada versus 7 days (5 10) in Mexico (p < ; Table 2). The most common presentations included pneumonia (77%) and bilateral pulmonary infiltrates and septic shock syndromes (24.2%), renal insufficiency (28.5%), fluid and electrolyte abnormalities (11.5%), and chronic obstructive pulmonary disease exacerbations (9.1%). Critical Illness Treatments Critically ill patients were severely hypoxic, with a median Pao 2 ratio of 83 mm Hg (59 119) using 14 cm H 2 O PEEP on the first study day (Table 1). Virtually all (97%) were treated with mechanical ventilation. Rescue oxygenation therapy either with prone ventilation, extracorporeal life support, high-frequency oscillatory ventilation, or inhaled nitric oxide was received by 44 patients (26.7%). Patients were more likely to receive ventilation and oxygenation support with extracorporeal life support, nitric oxide, and high-frequency oscillatory ventilation in Canadian ICUs (Table 2). Medical therapy included IV vasoactive medications (60%; Table 1), neuraminidase inhibitors (98.8%), corticosteroids (34.2%), and antibiotics (96.4%). Time to antiviral treatment was 6 days (4 8) (5 d [3 7] in Canada and 7 d [5 8] in Mexico; p = ; Table 2). Outcomes Among adult survivors, duration of mechanical ventilation was 14 days (7 24) (11 vs 21 d in Canada vs Mexico; p = 0.001), length of stay in ICU was 14 days (6 26), and length of stay in hospital was 26 days (12 43) (Fig. 1 and Table 3). At 90 days, after the onset of H1N1pdm09-related critical illness, 34.6% of all adult patients had died; however, mortality in Canadian ICUs was 13.9% compared to 50.5% in Mexico ICUs (p < ). A comparison of surviving and nonsurviving patients is shown in etable 1 (Supplemental Digital Content 2, and etable 2 (Supplemental Digital Content 2, Independent predictors of mortality included presenting Pao 2 ratio (odds ratio [OR], 0.89 per 10 cm H 2 O increment [95% CI, ]), age (OR, 1.49 per 10 yr increment [95% CI, ]), and country (OR, 7.76 [95% CI, ]; Table 4). In a sensitivity analysis, country remained a significant predictor of outcome after accounting for clustering at the level of the hospital using the same (4.55 [ ]; p = 0.012) or unique (7.78 [ ]; p = 0.013) intraclass correlation coefficients for each country. The most common primary cause of death included acute lung injury (17 patients, 29.8%), multi-organ dysfunction syndrome (14 patients, 24.6%), septic shock (13 patients, 22.8%), pneumonia (eight patients, 14%), myocardial infarction (one patient, 1.8%), malignancy (one patient, 1.8%), and other causes (three patients, 5.3%). Comparison of Healthcare Resources in Mexico and Canadian ICUs ICUs in Mexico and Canada were predominantly teaching institutions that employed an intensivist-led model of care and 6 XXX 2016 Volume XX Number XXX

7 Clinical Investigation Table 2. Treatments and Copresenting Illnesses of Critically Ill Patients With Confirmed Influenza A (H1N1pdm09) Infection Variable All (n = 165) Canada (n = 72) Mexico (n = 93) p Symptoms to hospital admission, median days (IQR) (n = 161) 6 (3 7) 4 (2 7) 7 (4 7) Symptoms to ICU admission, median days (IQR) (n = 161) a 7 (4 9) 5 (3 7) 7 (5 9.5) < Any antiviral treatment, n (%) 163 (98.8) 70 (97.2) 93 (100) Oseltamivir 163 (98.8) 70 (97.2) 93 (100) Amantadine 2 (1.2) 0 (0) 2 (2.2) Other antivirals 12 (7.3) 6 (8.3) 6 (6.5) Symptoms to antiviral treatment, median days (IQR) 6 (4 8) 5 (3 7) 7 (5 8) Corticosteroids, ever in ICU, n (%) 56 (34.2) 26 (36.6) 30 (32.3) Respiratory supportive interventions, n (%) Invasive mechanical ventilation 146 (88.5) 66 (91.7) 80 (86) Noninvasive mechanical ventilation 56 (33.9) 29 (40.3) 27 (29) Prone ventilation 36 (21.8) 12 (16.7) 24 (25.8) Extracorporeal life support 7 (4.2) 6 (8.3) 1 (1.1) Inhaled nitric oxide 7 (4.2) 7 (9.7) 0 (0) High-frequency oscillation 6 (3.6) 6 (8.3) 0 (0) Antibiotic treatment in ICU, n (%) 159 (96.4) 68 (94.4) 91 (97.9) Respiratory organisms isolated other than H1N1, n (%) Staphylococcus aureus (including methicillin-resistant S. aureus) 10 (6.1) 8 (11.1) 2 (2.2) Acinetobacter baumannii 8 (4.8) 0 (0) 8 (8.6) Escherichia coli 5 (3) 1 (1.4) 4 (4.3) Klebsiella pneumonia 5 (3) 0 (0) 5 (5.4) Pseudomonas aeruginosa 5 (3) 2 (2.8) 3 (3.2) IQR = interquartile range. a Excluding hospital-acquired cases. cared predominantly for adult patients with similar severity of illness at presentation (etable 3, Supplemental Digital Content 2, Canadian ICUs were more likely to have extracorporeal life support and inhaled nitric oxide as available treatment options but similar proportions could perform prone ventilation. Although the number of ICU nurses, doctors, or respiratory therapists (per 10 ICU beds) was not significantly associated with individual patient mortality (Table 4), Canadian hospitals had a significantly greater number of total beds, ICU beds, and ventilators and a greater number of critical care personnel, particularly during night shifts. DISCUSSION In this comparison of consecutive patients with H1N1pdm09- related critical illness in 21 hospitals in Mexico and Canada during the influenza season, we found that patients were young, severely hypoxic, and required substantial oxygenation support, as experienced in the pandemic (11, 12). Mortality at 90 days after critical illness was 34.6%; however, this was more than three-fold higher in Mexico (50.5%) versus Canada (13.9%). Country of care was the strongest predictor of mortality, but other independent predictors of worse outcomes included degree of hypoxemia and older age. These findings are important. First, they provide contemporary mortality estimates of H1N1pdm09-related critical illness and demonstrate that the most common seasonal influenza strain in North America continues to cause substantial morbidity and mortality among relatively young patients. Second, these findings emphasize a potential large mortality difference in influenza-related critical illness between different regions of the world. The potential reasons underlying differences in 90-day mortality rates between Canada and Mexico deserve careful attention. First, differences in outcome may reflect a selection bias of patients in the two countries leading to inclusion of Critical Care Medicine 7

8 Dominguez-Cherit et al Figure 1. Survival curve from onset of critical illness among critically ill patients with confirmed influenza A (H1N1pdm09) infection. systematically sicker patients in Mexico. However, we undertook this study using prospectively defined criteria in order to capture all critically ill patients admitted to any area of a hospital that might care for severely ill patients. Most patients were admitted to ICU, and in fact, baseline severity of illness, Pao 2 ratio, and vasopressor requirements were similar between groups. It is also possible that outcome differences reflect different preexisting health states; however, Canadian and not Mexican patients were marginally older and with a slightly greater number of comorbid conditions. It is also possible that there may be differential biologic responses and outcomes to influenza-related critical illness based upon ethnicity (15). In Mexico, all patients ethnicity was classified as Latin American, whereas in Canada there was a range of ethnicities, the most common being mixed or uncertain. Prior investigations have hinted at high mortality among Aboriginal and First Nations communities in North America during past pandemics (16, 17). However, recent studies using more detailed patient-level data indicate that differences in outcome are not strongly related to ethnicity, but are often confounded by other factors, many of which we collected and considered in our analyses (17, 18). There may also exist differences in other or copresenting infections. For example, Staphylococcus aureus was the most commonly isolated bacteria from respiratory sampling in Canada, whereas Gram negatives predominated in Mexico, which may have influenced mortality (19). We did not have systematic mechanisms to diagnose noninfluenza viral pathogens that may also differ between countries (20). One important potential emerging explanation is that patients have different mechanisms to access and receive care, and that this may influence outcomes. As we observed in (11, 12), patients in Mexico presented to hospital, on average, 3 days later in their illness course; received antiviral medications, on average 2 days later; and were admitted to ICU, on average 2 days later than critically ill patients in Canada. Although potentially important, location of ultimate ICU care had a much larger and stronger association with survival. Prior observational studies indicated that outcomes from influenza-related critical illness in Mexico during the pandemic were worse than that in other jurisdictions and some have speculated that this was because of a mismatch in demand and capacity early on in the pandemic. Therefore, in addition to recording clinical outcomes, we simultaneously recorded resources in all hospitals and ICUs participating in this study. We found that although treatments were generally similar between groups, participating hospitals in Mexico generally had fewer total beds, ICU beds, mechanical ventilators, personnel, and certain advanced forms of oxygenation support. Accordingly, patients in Mexico were less likely to receive these modalities of support. The greatest separation in mortality curves occurred during the first 2 weeks of critical illness. Of note, nonsurvivors had substantially shorter durations of mechanical ventilation, ICU and hospital stay in Mexico compared to Canada. Because withdrawal of mechanical support is a frequent occurrence among patients in ICU (21), these findings suggest that criteria for deciding whether patients with high severity of illness continue to receive life support may vary among different jurisdictions. We found that between-country differences in outcome remained after adjusting for potential differences at the hospital level using generalized estimating equations. Comparative endof-life care is an important area for future investigation. Influenza A (H1N1pdm09) can cause severe acute respiratory distress syndrome (ARDS); however, these patients 8 XXX 2016 Volume XX Number XXX

9 Clinical Investigation Table 3. Outcomes of Critically Ill Patients With Confirmed Influenza A (H1N1pdm09) Infection Variable All (n = 165) Canada (n = 72) Mexico (n = 93) p Mortality from onset of critical illness (d), n (%) (23.6) 4 (5.6) 35 (37.6) < (32.1) 9 (12.5) 44 (47.3) < (34.6) 10 (13.9) 47 (50.5) < Survivors, median (IQR) ICU length of stay, d (n = 108) 14 (6 25.5) 12.5 (6 22) 17 (4 28) Hospital length of stay, d (n = 108) 25.5 ( ) 26 (13 37) 24.5 (8 43) Mechanical ventilation duration, d (n = 85) 14 (7 24) 11 (5 18) 21 (15 31) Nonsurvivors, median (IQR) ICU length of stay, d (n = 57) 8 (5 16) 16.5 (5 26) 8 (4 16) Hospital length of stay, d (n = 57) 10 (5 19) 18.5 (10 27) 9 (5 17) Mechanical ventilation duration, d (n = 56) 9 (5 16) 14 (5 20) 8 (5 16) IQR = interquartile range. Table 4. Multivariate Model for the Outcome of Mortality Among Critically Ill Patients With Confirmed Influenza A (H1N1pdm09) Infection Variable OR a 95% CI Country, Mexico vs Canada Age, yr (per 10 yr increment) Sex, female vs male Sequential Organ Failure Assessment score (per one-point increment) Pao 2 (per 10 cm H 2 O increment) Time from symptoms to ICU admission, d No. of nurses on duty per day in ICU (per 10 beds) No. of physicians on duty per day in ICU (per 10 beds) No. of respiratory therapists on duty per day in ICU (per 10 beds) OR = odds ratio. a Adjusted in multivariable model. are often relatively young. Follow-up studies of ARDS have shown substantially better long-term outcomes among younger patients (22). In other areas of critical care, mortality differences have been observed to correlate with differences in availability and application of resources (10, 23, 24). As observed in this study, this has important implications regarding potential benefits of the provision of organ-supportive intensive care for specific conditions in jurisdictions with resource challenges. The strengths of our study include its prospectively defined criteria and surveillance for severe respiratory infection. We included a relatively large number of hospitals, improving the generalizability of our comparisons. We followed all patients for up to 90 days to ensure that differences in timing of death or withdrawal of care did not lead to spurious conclusions. Importantly, we sought to explore potential explanations underlying differences in outcome by concurrently measuring not only patient-level predictors, but also hospital and ICU resources and healthcare system access variables. The potential limitations of this study are also important to highlight. Differential selection of patients is a common form of bias among comparative studies in different jurisdictions; however, we employed a common definition and study procedures for surveillance and inclusion. Even if there were residual differences in application of eligibility criteria across the population of patients with H1N1pdm09-related critical illness, we Critical Care Medicine 9

10 Dominguez-Cherit et al attempted to adjust for such known characteristics, including severity of illness, between Canadian and Mexican patients at the onset of critical illness. As well, although we explored underlying differences in access to and receipt of critical care, the associations with mortality and healthcare system characteristics we present cannot be considered causative. CONCLUSIONS We have found that influenza A (H1N1pdm09), the dominant North American seasonal strain in , continues to be an important cause of severe hypoxemic respiratory failure among relatively young to middle-aged adults. The degree of hypoxemia at presentation and older age are associated with mortality; however, other country differences and access to critical care may be among the strongest independent predictors of outcome. This has important global implications for the provision of healthcare to critically ill patients. REFERENCES 1. Perez-Padilla R, de la Rosa-Zamboni D, Ponce de Leon S, et al; INER Working Group on Influenza: Pneumonia and respiratory failure from swine-origin influenza A (H1N1) in Mexico. N Engl J Med 2009; 361: Pandemic (H1N1) Update Available at: who.int/csr/don/2009_11_13/en/index.html. Accessed July 18, Influenza Laboratory Surveillance Information. Available at: gamapserver.who.int/gareports/default.aspx?reportno=6. Accessed July 18, Napolitano LM, Angus DC, Uyeki TM: Critically ill patients with influenza A(H1N1)pdm09 virus infection in JAMA 2014; 311: Clinical Management of Critically Ill Patients With Influenza A(H1N1) pdm09, Geneva, Switzerland, World Health Organization. Available at: WHOGIPReport_A4_WEB_FA.pdf. Accessed July 18, Crosby A: America s Forgotten Pandemic. Cambridge (UK), Cambridge University Press, Patterson KD, Pyle GF: The geography and mortality of the 1918 influenza pandemic. Bull Hist Med 1991; 65: Johnson NP, Mueller J: Updating the accounts: Global mortality of the Spanish influenza pandemic. Bull Hist Med 2002; 76: Davies A, Jones D, Bailey M, et al; The Australia and New Zealand Extracorporeal Membrane Oxygenation (ANZ ECMO) Influenza Investigators: Extracorporeal membrane oxygenation for 2009 influenza A(H1N1) acute respiratory distress syndrome. JAMA 2009; 302: Adhikari NK, Fowler RA, Bhagwanjee S, et al: Critical care and the global burden of critical illness in adults. Lancet 2010; 376: Kumar A, Zarychanski R, Pinto R, et al; Canadian Critical Care Trials Group H1N1 Collaborative: Critically ill patients with 2009 influenza A(H1N1) infection in Canada. JAMA 2009; 302: Domínguez-Cherit G, Lapinsky SE, Macias AE, et al: Critically ill patients with 2009 influenza A(H1N1) in Mexico. JAMA 2009; 302: Fowler RA, Adhikari NK, Bhagwanjee S: Clinical review: Critical care in the global context disparities in burden of illness, access, and economics. Crit Care 2008; 12: Vincent JL, Moreno R, Takala J, et al: The SOFA (Sepsis-related Organ Failure Assessment) score to describe organ dysfunction/failure. On behalf of the Working Group on Sepsis-Related Problems of the European Society of Intensive Care Medicine. Intensive Care Med 1996; 22: La Ruche G, Tarantola A, Barboza P, et al; Epidemic Intelligence Team at InVS: The 2009 pandemic H1N1 influenza and indigenous populations of the Americas and the Pacific. Euro Surveill 2009; 14: Humphries M: Lessons From the 1918 Pandemic: Focus on Treatment, Not Prevention Available at: news/opinions/lesson-from-the-1918-pandemic-focus-on-treatmentnot-prevention/article /. Accessed July 18, Statistics Canada: 2006 Census: Aboriginal Peoples in Canada in 2006: Inuit, Métis and First Nations, 2006 Census: Highlights Available at: as-sa/97 558/p1-eng.cfm. Accessed January 15, Jung J, Long J, Pinto R, et al: influenza A(H1N1)-related critical illness among Aboriginal and Non-Aboriginal Canadians. Am J Resp Crit Care Med 2010:A Zheng YL, Wan YF, Zhou LY, et al: Risk factors and mortality of patients with nosocomial carbapenem-resistant Acinetobacter baumannii pneumonia. Am J Infect Control 2013; 41:e59 e Galindo-Fraga A, Ortiz-Hernández AA, Ramírez-Venegas A, et al; La Red ILI 002 Study Group: Clinical characteristics and outcomes of influenza and other influenza-like illnesses in Mexico City. Int J Infect Dis 2013; 17:e510 e Cook D, Rocker G, Marshall J, et al; Level of Care Study Investigators and the Canadian Critical Care Trials Group: Withdrawal of mechanical ventilation in anticipation of death in the intensive care unit. N Engl J Med 2003; 349: Herridge MS, Tansey CM, Matté A, et al; Canadian Critical Care Trials Group: Functional disability 5 years after acute respiratory distress syndrome. N Engl J Med 2011; 364: Phua J, Koh Y, Du B, et al; MOSAICS Study Group: Management of severe sepsis in patients admitted to Asian intensive care units: Prospective cohort study. BMJ 2011; 342:d Wunsch H, Angus DC, Harrison DA, et al: Variation in critical care services across North America and Western Europe. Crit Care Med 2008; 36: , e1 e XXX 2016 Volume XX Number XXX