key: cord-034578-i9rdubix
authors: Vaschetto, Rosanna; Barone-Adesi, Francesco; Racca, Fabrizio; Pissaia, Claudio; Maestrone, Carlo; Colombo, Davide; Olivieri, Carlo; De Vita, Nello; Santangelo, Erminio; Scotti, Lorenza; Castello, Luigi; Cena, Tiziana; Taverna, Martina; Grillenzoni, Luca; Moschella, Maria Adele; Airoldi, Gianluca; Borrè, Silvio; Mojoli, Francesco; Della Corte, Francesco; Navalesi, Paolo; Cammarota, Gianmaria; Baggiani, Marta; Baino, Sara; Balbo, Piero; Bazzano, Simona; Bonato, Valeria; Carbonati, Sara; Crimaldi, Federico; Daffara, Veronica; De Col, Luca; Maestrone, Matteo; Malerba, Mario; Moroni, Federica; Perucca, Raffaella; Pirisi, Mario; Rondi, Valentina; Rosalba, Daniela; Vanni, Letizia; Vigone, Francesca
title: Outcomes of COVID-19 patients treated with continuous positive airway pressure outside ICU
date: 2020-10-30
journal: ERJ Open Res
DOI: 10.1183/23120541.00541-2020
sha: 
doc_id: 34578
cord_uid: i9rdubix

AIM: We aim at characterising a large population of Coronavirus 19 (COVID-19) patients with moderate-to-severe hypoxemic acute respiratory failure (ARF) receiving CPAP outside intensive care unit (ICU), and ascertaining whether the duration of CPAP application increased the risk of mortality for patients requiring intubation. METHODS: In this retrospective, multicentre cohort study, we included COVID-19 adult patients, treated with CPAP outside ICU for hypoxemic ARF from March 1(st) to April 15(th), 2020. We collected demographic and clinical data, including CPAP therapeutic goal, hospital length of stay (LOS), and 60-day in-hospital mortality. RESULTS: The study includes 537 patients with a median age of 69 (IQR, 60–76) years. Males were 391 (73%). According to predefined CPAP therapeutic goal, 397 (74%) patients were included in full treatment subgroup, and 140 (26%) in the do-not intubate (DNI) subgroup. Median CPAP duration was 4 (IQR, 1–8) days, while hospital LOS 16 (IQR, 9–27) days. Sixty-day in-hospital mortality was overall 34% (95%CI, 0.304–0.384), and 21% (95%CI, 0.169–0.249) and 73% (95%CI, 0.648–0.787) for full treatment and DNI subgroups, respectively. In the full treatment subgroup, in-hospital mortality was 42% (95%CI, 0.345–0.488) for 180 (45%) CPAP failures requiring intubation, while 2% (95%CI, 0.008–0.035) for the remaining 217 (55%) patients who succeeded. Delaying intubation was associated with increased mortality [HR, 1.093 (95%CI, 1.010–1.184)]. CONCLUSIONS: We described a large population of COVID-19 patients treated with CPAP outside ICU. Intubation delay represents a risk factor for mortality. Further investigation is needed for early identification of CPAP failures.

Noninvasive ventilation (NIV) administered as bi-level positive airway pressure (BiPAP) or continuous positive airway pressure (CPAP) is commonly used in various critical care settings across a variety of aetiologies of acute respiratory failure (ARF). For hypercapnic ARF, mainly consequent to chronic obstructive pulmonary disease exacerbation, BiPAP can be used at an early stage to prevent intubation, at a later stage as alternative to first-line endotracheal intubation, or as a mean to facilitate weaning [1] .

For hypoxemic ARF, recommendations strongly support the use of both BiPAP and CPAP in patients with episodes of cardiogenic pulmonary edema [2, 1] , while fewer data suggest their use in immunosuppressed [3, 1] and in post-operative [4, 1] patients. In patients with de novo hypoxemic ARF evidences and recommendations on the use of NIV are still to be determined [1] . Moreover, the application of NIV in patients with acute respiratory distress syndrome (ARDS) complicating viral pneumonia is controversial [5] .

During coronavirus disease 2019 pandemic, Piedmont together with Lombardy, Emilia-Romagna and Veneto was one of the most affected Italian regions. Due to the exceptional demand on intensive care unit (ICU) resources, hospitals increased the number of ICU beds and converted many general wards in respiratory intermediate care units (RICU) to treat patients with severe pneumonia and ARDS-needing respiratory support and monitoring. Indeed, NIV in patients with different therapeutic indications i.e., full-treatment and do-not-intubate [6] has been shown to be successfully applicable also outside the ICU [7, 8] , when appropriate monitored setting and trained personnel are employed.

Data on NIV during COVID-19 pandemic, so far, consider predominantly patients admitted to the ICU [9, 10, 11, 12, 13] . The rate of patients receiving NIV at ICU admission ranges from 11%, as reported by an Italian multicentre investigation [10] , to 56% according to a Chinese single centre study [11] .

Exposure to noninvasive forms of respiratory support might have been even more diffuse outside ICU, though only data from two monocentre studies are presently available, accounting overall for 40 patients, 38 treated with CPAP [14] and two with NIV or high flow oxygen therapy [15] .

We designed this retrospective multicentre study to describe the clinical characteristics of patients with laboratory-confirmed COVID-19 treated with CPAP outside ICU, to assess 60-day in-hospital mortality, and hospital length of stay (LOS), and to ascertain whether the duration CPAP application prior to CPAP failure affects outcome in patients requiring endotracheal intubation.

This is a multicentre, retrospective observational study performed in six hospitals from the area of Eastern Piedmont in Northern Italy. All the participating centres obtained ethic committee approval.

More details on study design and ethics approval are provided in the supplementary material.

All patients admitted to one of the participating hospitals from March 1 st to April 15 th 2020 with hypoxemic ARF secondary to confirmed SARS-CoV-2. Inclusion criteria were: 1) age ≥18 years, 2) respiratory distress and partial pressure of arterial oxygen to fraction of inspired oxygen ratio < 200 mmHg during Venturi mask oxygen therapy, 3) CPAP initiation outside ICU. Patients who received post-extubation CPAP were excluded.

Patients were classified according to predefined CPAP therapeutic goal applied by the medical team, in two subgroups [6] : 1) full treatment, i.e. patients scheduled to receive intubation in the case of CPAP failure; and 2) do-not-intubate (DNI), when CPAP was ceiling of treatment. In case patient changed the therapeutic goal during hospital stay, the last CPAP goal has been considered. The therapeutic goal of CPAP was collegially discussed within the multidisciplinary teams, with the patients and with the families, taking into account comorbidities [16] , quality of life and patient wishes. Possible discrepancies between patients and relatives were solved through additional discussions between patient, relatives, and the medical team.

Demographic characteristics, body mass index (BMI), blood sample exams performed at hospital entrance (white blood cell count, lymphocytes count, creatinine, alanine transaminase, aspartate transaminase, lactate dehydrogenase, C-reactive protein, D-dimer, ferritin), arterial blood gas (ABG) values obtained prior to CPAP initiation and 2 to 24 hours after; and coexisting comorbidities were also recorded. Charlson Comorbidity Index (CCI) [17] was also computed on the first day of hospital admission. This index contemplates 17 categories of comorbidity recorded via anamnesis. Age is not included as comorbidity in the CCI version adopted. Finally, we collected data about the clinical outcomes such as duration of CPAP use, hospital length of stay (LOS), intubation and hospital mortality.

For patients still in the hospital on May 15 th (n=32), the outcomes have been censored on that day.

Details on CPAP setting, schedule, RICU organization and criteria for intubation are described in the supplementary material.

Descriptive statistics are used to summarize the main demographic characteristics and the results of laboratory findings of all patients included in the study. Categorical variables are reported as absolute frequencies and percentages, while numerical variables as median and interquartile range (IQR). The frequency and percentage of missing values for all variables is also reported. Mann Whitney U test is used to assess the difference between two independent samples, while Wilcoxon signed-rank test for repeated measurements. Chi-square statistic is used for testing relationships of categorical variables.

Curves of cumulative incidence of in-hospital mortality are drawn to describe mortality along 60 days, either overall and stratified for treatment goal, and in the full treatment subgroup separately for patients succeeding CPAP or receiving intubation.

In order to avoid immortal time bias, in the survival analysis of patients receiving intubation, observation period started at the day of intubation. In the other analyses, observation period started at the day of CPAP initiation. Since discharge must be considered an informative censoring [18], cumulative incidence was calculated using methods accounting for competing risk. To evaluate the cumulative incidence of in-hospital mortality for patients not undergoing intubation, all full treatment subjects are considered, and intubation is treated as a competing event allowing to account for the contribution of the time spent by intubated patients on CPAP. The Gray's test is used to assess the difference between cumulative incidence functions. Fine and Gray multivariate competing risk model is adopted to calculate the sub-distribution hazard ratios (sHR) and the corresponding 95% confidence intervals (95%CI) for the association between CPAP duration and in-hospital mortality risk in intubated patients, considering discharge as competing event. In the main analyses, missing data are managed by listwise deletion. We also carried out a secondary analysis using multiple imputation to evaluate the impact of missing values on the association estimates. Missing imputation is performed using the Expectation-Maximization algorithm (500 imputations) and considering the "missing at random" mechanism. More details about the model are provided in the supplementary material.

All hypothesis tests are two-tailed and a significance level of 0.05 is considered. All statistical analysis was performed using STATA (Stata Statistical Software: Release 15. College Station, TX, USA:

StataCorp LLC), SAS (version 9.4; SAS Institute Cary, NC, USA) and R (version 3.5.1).

From March 1 st to April 15 th , a total of 2845 patients with confirmed COVID-19 were admitted to the six hospitals of the eastern Piedmont ( Figure 1s ). Of these, 326 (11%) patients were treated in ICU, 31 (1%) and 295 (10%) with noninvasive and invasive mechanical ventilation, respectively. CPAP was applied to 537 (22%) patients in RICU. 1 . Median white blood cell count was 6.9 (IQR, 5.1-9.5) x 10 3 /µL, with lymphopenia, i.e., 0.8 (0.6-1.1)

x 10 3 /µL. Median values of creatinine, aspartate-aminotransferase, alanine-aminotransferase were in the normal range, while C-reactive protein, ferritin, lactate dehydrogenase and D-dimer were all above the normal range. CCI median value was 1 (IQR, 0-2), chronic arterial hypertension was present in 278 (52%) patients, diabetes in 138 (26%) patients, and ischaemic heart disease in 66 (12%) patients.

The most common interface was the helmet, in 399 (74%) patients, while face masks were used in 123 patients (23%); 15 patients (3%) alternated both interfaces. Median CPAP duration was 4 (IQR, [1] [2] [3] [4] [5] [6] [7] [8] days. Overall cumulative 60-day in-hospital mortality was 34% (cumulative incidence, 0.344, 95%CI, 0.304-0.384), as depicted in Figure 1A , while hospital LOS was 16 (IQR, [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] days.

Demographic and clinical patients' characteristics stratified by PaO 2 /FiO 2 performed 2 to 24 hours after CPAP initiation and interface applied, are provided in the supplementary material in table 2s and 3s, respectively. Overall cumulative 60-day in-hospital mortality stratified according to PaO 2 /FiO 2 and interface applied is also depicted in Figure 2s A Within the full treatment subgroup, 60-day in-hospital mortality was 42% (cumulative incidence 0.418, 95%CI, 0.345-0.488) for patients receiving intubation (Figure 2 ), and 2% (cumulative incidence 0.018, 95%CI, 0.008-0.035) for patients succeeding CPAP (Figure 3s ). CPAP duration was 2 days (IQR, 1-3 days) in patients who survived and 3 days (IQR. 1-5 days) in patients who died (p=0.061). Table 3 shows that duration of CPAP application was an independent predictor of mortality for patients requiring intubation. The model, adjusting for age, gender, comorbidities, LDH, C-reactive protein values, and lymphocyte count, indicates a 9.3% [HR, 1.094 (95%CI, 1.010-1.184)] increase of the risk of death for each day of treatment. The association between duration of CPAP and mortality does not substantially change in the secondary analysis using multiple imputation [HR, 1.060 (95%CI, 1.001-1.121)], as presented in Table 7s . 60-day in-hospital mortality was significantly higher in patients subjected to CPAP for more than 3 days (cumulative incidence 0.510, 95%CI, 0.393-0.615) as compared to those receiving CPAP for 3 days or less (cumulative incidence 0.350, 95%CI, 0.259-0.441), as shown in Figure 4s .

This multicentre retrospective observational study on 537 patients hypoxemic ARF secondary to laboratory-confirmed COVID-19 infection, shows that CPAP applied to different therapeutic goals i.e., candidate to intubation in the case of CPAP failure and do-not-intubate in which CPAP is considered the ceiling of treatment, is feasible outside ICU. Treatment duration for patients failing CPAP prior to intubation represents a risk factor for mortality.

CPAP can be delivered both in ICU and outside ICU. Grasselli et al. [10] , found that 11% of the patients entering ICU needed NIV, while early data from China revealed a higher percentage i.e., 41.7% [13] , 43.3% [19], 56% [11] , 62% [15] . In keeping with data by Grasselli et al. [10] , as CPAP was delivered to 31 out of 326 patients (9.5%) entering ICU.

Data on the use of CPAP in COVID-19 patients treated outside ICU are scarce. Two over 28 patients (7%) received NIV or HFNC outside ICU in a single centre study in Wuhan [15] . Oranger et al., treated 38 patients with CPAP in a respiratory ward [14] . Although the study included a limited patients' number, CPAP resulted to be feasible and the authors also suggest a potential benefit for both full treatment and DNI patients, as opposed to those treated with oxygen only [14] .

As far as mortality concerns, we showed an overall cumulative 60-day in-hospital mortality of 34% in patients with moderate to severe forms of ARF COVID-19-related needing CPAP. The rate of mortality observed in our study, is not divergent from those reported in several prior studies [9, 10, 15, 12, 13] for ICU patients, predominantly intubated, which varied from 17% [13] to 67% [9] .

Lastly, our study includes 26% of DNI patients, for whom CPAP was considered ceiling of treatment.

Rate of DNI patients reported in our study is similar to 15% observed in a small cohort of patients treated outside ICU during COVID-19 pandemic [14] , i.e., as well as to 30% reported by a large Italian multicentre observational study in patients with pneumonia non-COVID-19 related, treated with NIV outside ICU [20] . In our study, 60-day in-hospital mortality for DNI patients was 73%.

A major concern when treating patients with hypoxemic ARF with NIV, is related to NIV-failure rate that might occur in up 50% of the cases with consequent recourse to intubation [21] . Undue prolongation of NIV may worsen lung injury resulting in the so-called patient self-inflicted lung injury [22] , while the direct consequence of NIV failure is delaying intubation and adequate treatment with invasive ventilation [23, 1] . Our data confirm that intubation delay for those requiring afterward invasive ventilation is associated with increased risk for mortality. In other pandemics, such as influenza, H1N1, and severe acute respiratory syndrome (SARS), NIV failure ranges from 10 to more than 70% [24], reaching 90% with middle east respiratory syndrome [25] . In our study, CPAP failure rate was 45%, which indicates that effective treatment occurred in more than half of patients, who avoided invasive ventilation through an endotracheal tube, which is a life-saving procedure, but it is also prone to several side-effects and complications [26] .

The study has several limitations. First, we were not able to compare our population with an historical control. Second, most of data have been retrospectively derived from the medical records. According to the retrospective nature of the study, formal criteria to start CPAP treatment were not defined a priori, and the time span between CPAP initiation and control ABG was relatively long. Third, definitions of full treatment and DNI patients, although internationally accepted [6] , are influenced by patients, families, and clinicians thinking and might be influenced by cultural, religious and geographical factors.

Fourth, due to the diversity of interfaces and devices used in our study, the actual applied pressure could somewhat differ from the preset value [27] . Fifth, due to the number of missing data among many important variables such as D-Dimer and respiratory rate, we were not able introduce them in the model that explore the correlation between CPAP duration and mortality. Lastly, because of the exceptionality of pandemic outbreak, our results are not generalizable to other conditions.

To the best of our knowledge, this is the largest retrospective cohort study on patients with COVID -19 treated with CPAP outside ICU. We show that CPAP is feasible outside ICU with overall in-hospital mortality similar to that reported in other studies treating critically ill ICU patients. In-hospital mortality is closely related to the therapeutic goal, patients having DNI order being affected by much higher mortality. Intubation delay is confirmed to be an independent risk factor for mortality. Further studies are necessary to ascertain the potential infective risk related to CPAP treatment outside ICU among healthcare workers. Chronic arterial hypertension, n (%) 278 (52) Diabetes, n (%) 138 (26) Ischaemic heart disease, n (%) 66 (12) CPAP, days 4 (1-8) Curves and corresponding 95% confidence intervals (dashed lines).

CPAP, noninvasive continuous positive airway pressure; DNI, do not intubate. ETI, endo-tracheal intubation.

Outcomes of COVID-19 patients treated with continuous positive airway pressure outside ICU

This was a multicentre, retrospective observational study performed in six hospitals of the eastern Piedmont region, Northern Italy i.e., "Maggiore della Carità" University hospital in Novara, "SS.

Antonio Biagio e Cesare Arrigo", hospital in Alessandria, "S. Andrea" hospital in Vercelli, "VCO ASL" in Domodossola, "Nuovo Ospedale degli Infermi" hospital in Biella. One centre i.e., an as-needed basis. When respiratory parameters improved, CPAP support was gradually reduced with a progressive increase of time off CPAP, until discontinuation.

Nurse to patient ratios varied from a maximum of 1:6 both during day and night to a minimum of 1:8 and 1:12, respectively during days and nights. In three hospitals, medical staff treating CPAP COVID-19 patients was an ad-hoc mixed team, mainly internists, pneumologists, emergency physicians, cardiologists, anaesthesiologists/ICU physicians, while in the other three hospitals the medical team was the same as before COVID-19 pandemic. CPAP was prescribed mainly by anaesthesiologists actively working with the ad-hoc COVID-19 ward team, but also by pneumologists and emergency doctors or by consulting anaesthesiologists. Personnel was adequately trained for NIV; those who were not, received a short-organized training during pandemic. Ward monitoring included SpO 2 , non-invasive blood pressure, ECG applied continuously or at a defined time point depending on the severity of the patient. Blood gas analysis was performed when clinically relevant. Patients received daily visit from the consulting physician who prescribed CPAP if not present in the ad-hoc ward team.

Fine and Gray model included as adjustment age, gender, comorbidities i.e., Charlson comorbidity index and hypertension, LDH, C-reactive protein levels and lymphocyte count. The model was further adjusted by centre. The adjustment variables were selected on the base of their clinical relevance. Multiple imputation procedures were applied to account for missing data. 

Ospedale Ss. Trinità, Department of Anesthesia and Critical Care

Azienda Ospedaliera Sant'Andrea, Department of Anesthesia and Critical Care

Maggiore della Carità

Via dei Ponderanesi, 2 -13875 Ponderano

Via Gallucci

Maggiore della Carità

Raoof S Members Of The Task Force. Official ERS/ATS clinical practice guidelines: noninvasive ventilation for acute respiratory failure

Effect of Noninvasive Ventilation vs Oxygen Therapy on Mortality Among Immunocompromised Patients With Acute Respiratory Failure: A Randomized Clinical Trial

Effect of Noninvasive Ventilation on Tracheal Reintubation Among Patients With Hypoxemic Respiratory Failure Following Abdominal Surgery: A Randomized Clinical Trial

Noninvasive positive pressure ventilation in critical and palliative care settings: understanding the goals of therapy

Noninvasive ventilation outside the intensive care unit from the patient point of view: a pilot study

An international survey on noninvasive ventilation use for acute respiratory failure in general non-monitored wards

Characteristics and Outcomes of 21 Critically Ill Patients With COVID-19 in Washington State

Clinical course and outcomes of critically ill patients with SARS-CoV-2 pneumonia in Wuhan, China: a single-centered, retrospective, observational study

Presenting Characteristics, Comorbidities, and Outcomes Among 5700 Patients Hospitalized With COVID-19 in the

Clinical Characteristics of 138 Hospitalized Patients With 2019 Novel Coronavirus-Infected Pneumonia in Wuhan, China

Continuous positive airway pressure to avoid intubation in SARS-CoV-2 pneumonia: a two-period retrospective case-control study

Clinical features of patients infected with 2019 novel coronavirus in Wuhan, China

A new method of classifying prognostic comorbidity in longitudinal studies: development and validation

Clinical Characteristics of Coronavirus Disease 2019 in China

Non-invasive positive pressure ventilation in pneumonia outside Intensive Care Unit: An Italian multicenter observational study

High-flow oxygen through nasal cannula in acute hypoxemic respiratory failure

Mechanical Ventilation to Minimize Progression of Lung Injury in Acute Respiratory Failure

Noninvasive mechanical ventilation in high-risk pulmonary infections: a clinical review

Noninvasive ventilation in critically ill patients with the Middle East respiratory syndrome

Evolution of mortality over time in patients receiving mechanical ventilation

Helmet with specific settings versus facemask for noninvasive ventilation

mg/dL 13

noninvasive continuous positive airway pressure; FEU, fibrinogen-equivalent unit

Table 3S General characteristics of patients according to CPAP interface applied Values are median (interquartile range) or number (percentage)

FEU, fibrinogen-equivalent unit Helmet vs. mask: * p =0.030, † p=0.002, § p=0.001, °p<0.0001, ‡ p=0.032. Helmet (n=399) Mask (n=123) Mask and Helmet (n=15)

We acknowledge all health-care workers involved in the diagnosis and treatment of patients in the eastern Piedmont region. We thank Dr. Davide Crimaldi for database support. The study received no funding. Preliminary data have been sent, in abstract form, to the European Society of Intensive Care Medicine Congress 2020.

Male 537Body mass index 239White blood cell count 533Lymphocyte count 529Creatinine 531Aspartate-aminotransferase 335Alanine-aminotransferase 503 Abbreviations: ABG, arterial blood gas analysis; CPAP, noninvasive continuous positive airway pressure; PaCO 2 , arterial partial pressure of carbon dioxide; PaO 2 , arterial partial pressure of oxygen; SpO 2 , peripheral oxygen saturation; HCO 3-, bicarbonate; FiO 2 , inspired oxygen fraction; PaO 2 /FiO 2 , arterial partial pressure of oxygen to inspired oxygen fraction ratio; SpO 2 /FiO 2 , peripheral oxygen saturation to inspired oxygen fraction ratio a Arterial blood gas analysis performed before CPAP initiation b First arterial blood gas analysis performed 2-24 hours after CPAP outset