Demographic data were collected, together with severity parameters (APACHE II SCALE), level of organ dysfunction upon admission (SOFA score) and comorbidities. Clinical variables (mean blood pressure [MBP], heart rate [HR] and respiratory frequency [RF]) and variables related to ventilatory support and oxygenation (need for invasive mechanical ventilation [IMV], high-flow oxygen therapy [HFO], arterial oxygen pressure [PaO2], arterial carbon dioxide pressure [PaCO2], fractional inspired oxygen [FiO2], arterial pH, PaO2/FiO2 ratio, peak pressure [Pmax], plateau pressure (PPl), positive end-expiratory pressure [PEEP]) were also recorded. In addition, we considered laboratory test parameters such as hemoglobin concentration, leukocyte count, lactate, C-reactive protein (CRP) and procalcitonin (PCT). All variables were recorded upon admission and after 7 days for comparative purposes.

The indication of orotracheal intubation (OTI), HFO or mechanical ventilation (MV) was established by the physician in charge of the patient. There was no specific respiratory management protocol for COVID-19. Due to the recommendation not to use noninvasive ventilation (NIV) or HFO because of the risk of aerosol generation,12 NIV was disadvised on the basis of internal consensus, and HFO was indicated in the context of a limited availability of respirators or as a strategy to delay and avoid MV. All the patients with failed HFO were intubated and subsequently received MV. No patients with limitations of life support were considered. The assessment of patient admission to the ICU was carried out in abidance with the ethical recommendations of the SEMICYUC.13

The patients were divided into three groups according to outcome after 28 days: survivors (group 1); non-survivors (group 2), and patients still in the ICU after 28 days (group 3).

• •

  Severe SARS-CoV-2 pneumonia: defined according to the criteria of the Spanish Ministry of Health14 as a consistent clinical condition, with fever (body temperature > 38 °C), cough, sore throat, muscle pain and flu symptoms plus RT-PCR testing positive for SARS-CoV-2 and the presence of acute respiratory failure, with lung infiltrations on the chest X-rays, requiring admission to the ICU.

• •

  Bacterial co-infection: considered in patients with confirmed COVID-19 in which culture of the lower airway sample (bronchoaspirate [BAS] or bronchoalveolar lavage [BAL]) obtained within the first 24 hours of admission reveals the presence of a pathogenic microorganism at concentrations above the cut-off points defined for each technique.15

• •

  Ventilator-associated pneumonia (VAP): defined according to the new ATS/IDSA guidelines,16 and corresponding to patients who develop a clinical condition characterized by increased radiological infiltration, changes in appearance of the secretions, and culture positivity (BAS or BAL sample) for pathogenic microorganisms at concentrations above the cut-off points defined for the technique, in a sample obtained 48 hours after the start of MV.

• •

  Ventilator-associated pneumonia incidence rate per 1000 ventilator days, expressed as density (number of cases per day/person/exposure) and calculated using the formula: incidence rate = number of cases of VAP during the study period/total days/person/exposure to MV in the population during the study period * 1000.

• •

  Failure of HFO: defined as the need for immediate intubation and subsequent mechanical ventilation. The need for OTI was based on clinical and blood gas criteria, and was left to the criterion of the physician in charge.

• •

  Shock upon admission: defined as the need for any dose of noradrenalin within the first 6 hours of admission in order to maintain mean blood pressure, and once the required volume replacement measures have been adopted, based on dynamic parameters or echocardiography.

• •

  Acute respiratory distress syndrome: classified according to the definition of Berlin17 into two groups: severe/moderate and mild.

• •

  Acute renal failure was established based on the RIFLE classification,18 with three defined categories.

Previously published criteria were used in reference to the comorbidities and the rest of definitions.19

In view of the characteristics of the study, no sample size calculation was made. The sample size therefore was the same as the number of patients admitted during the study period. Continuous variables were reported as the median and interquartile range (IQR), while categorical variables were reported as frequencies and percentages. Differences in distribution of the variables between groups of patients were explored using the chi-square test or the Fisher exact test (categorical variables). Differences between medians were evaluated using the nonparametric Wilcoxon test or Mann-Whitney U-test. Due to the limited number of cases, no multivariate analyses were made. Statistical significance was considered for p < 0.05. The SPSS® version 25.0 statistical package (IBM) was used throughout.

From 1 February to 14 April 2020, a total of 380 cases of COVID-19 were diagnosed in our hospital. Of these, 43 (11.3%) required admission to the ICU due to acute respiratory failure, and constituted the subject of the present analysis. Seventeen patients (39.5%) were admitted from Internal Medicine, 15 (34.9%) were transferred from other hospitals, and 11 (25.6%) came from the Emergency Department of our hospital. Although the patients received conventional oxygen therapy and in some cases HFO before admission to our ICU, these data were not available for analysis. The median time from arrival in hospital to admission to the ICU was one day (IQR: 0.0-3.0).

The baseline characteristics of the patients are shown in Table 1. At the time of the analysis, 28 patients were discharged live (group 1 = 65.1%), 10 patients died (group 2 = 23.3%) and 5 remained in admission (group 3 = 11.6%). The median number of days to death was 23 (IQR: 11-40), and only one patient (2.3%) died within the first 24 hours of admission. The mean duration of stay in the ICU for the patients in group 1 was 27 days (IQR: 13-34).

In general, the patients were young (65 years), predominantly males, and with great disease severity as evidenced by the median APACHE (18) and SOFA scores (6). A significant rise (p = 0.02) in mortality was observed over the increasing age intervals: 18-45 years (0%), 46-64 years (13.3%), 65-75 years (35.3%) and > 75 years (66.7%). The patients over 75 years of age died early, within the first week (Fig. 1).

Figure 1.

Distribution of survival (Kaplan-Meier) corresponding to the different age intervals established for the patients included in the study.

Forty-four percent of the patients registered no comorbidities. None of the healthcare staff members of our hospital were admitted to the ICU due to COVID-19 during the observation period. The main comorbidities were arterial hypertension (30.2%) and obesity (25.6%), followed by diabetes mellitus (18.6%), chronic obstructive pulmonary disease (COPD)(9.3%) and ischemic heart disease (9.3%). Of the 13 patients with arterial hypertension, 7 (53.8%) had a history of drug treatment in the form of angiotensin converting enzyme inhibitors (ACEIs) (n = 2) or angiotensin receptor antagonists (ARAs)(n = 5). Only the presence of shock upon admission was more frequent among the non-survivors (80%) than among the survivors (group 1 = 42.9%) (Table 1). The other variables considered showed no statistically significant differences between the groups.

The time from symptoms onset to the first antiviral drug dose was considerable, with a median of 8 days (IQR: 6-10) – no differences being observed between the groups. Lopinavir/ritonavir was administered in 40 cases (93%), with a median duration of only 5 days (IQR: 3.5-10) due to complications related to important transaminase elevation requiring treatment suspension. Hydroxychloroquine was used in 39 patients (90.7%), with a median duration of 5 days. Interferon ß-1b was only administered to 15 patients (34.9%), due to drug availability problems, and only two patients (4.7%) received tocilizumab in other centers, before transfer to our ICU (Table 1). The administration of corticosteroids was not considered among the treatments for COVID-19. Such medication was only used upon admission to the ICU in a single patient (2.3%) as continuation of pre-existing chronic treatment. Five patients (11.6%) received methylprednisolone as rescue therapy due to persistent lung infiltrations after the second week of stay, and four patients (9.3%) received the medication due to other indications such as thrombocytopenia (n = 1), hemolytic anemia (n = 1) and skin rash (n = 2).

All the patients required some type of ventilatory support during the first hours of admission to the ICU. High-flow oxygen was used as initial treatment for respiratory failure due to COVID-19 in 27 patients (62.7%). However, after 24 hours, only four patients responded favorably – this representing a failure rate of 85.2% (4/27). On day 7, only two patients maintained HFO, and these were discharged live, with no need for any other type of ventilatory support. The median time to HFO failure was 8 hours (IQR: 6-20). All the patients in which HFO failed were subsequently intubated and ventilated. Invasive mechanical ventilation was used as first treatment option for acute respiratory failure in 16 patients (37.2%). However, during the first 24 hours, 37 (86%) of the 43 patients were ventilated, and finally after 7 days, 41 (95%) required IMV. Of the patients subjected to IMV, 34 (82%) required at least one ventilation episode in prone decubitus, with a median of three maneuvers (IQR: 1-5) per patient – though some underwent over 10 ventilation periods in prone decubitus. Only one patient (2.3%) was referred to the reference hospital for ECMO, and this individual finally died. The median number of days of IMV was 27 (IQR: 15-38), and proved similar in the survivors (23 days [IQR: 12-30]) and non-survivors (27 days [IQR: 13-38]; p = 0.37). Of the 37 patients that required IMV, 32 (86.5%) had completed their course in the ICU at the time of the analysis, with a crude mortality rate of 28.1% (9/32), which was similar to that recorded among the patients who did not require IMV in the first 24 hours (16.6%; p = 0.55).

A total of 93% of the patients (40/43) met criteria of ARDS, which proved moderate/severe in 32 cases (74.4%) and mild in 8 (18.6%). Of the 32 patients with moderate/severe ARDS, 27 (72.9%) received IMV after 24 hours, but four (14.8%) remained with HFO, and one patient with a face mask and reservoir. All of the patients with mild ARDS required IMV after 24 hours.

Independently of the ventilatory support received, the median PaO2/FiO2 upon admission was 165 (IQR: 125-210), and the ratio increased significantly up until day 7, reaching 194 (IQR: 153-285; p = 0.05). A similar PaO2/FiO2 behavior was recorded in the survivors, though without reaching statistical significance (p = 0.07), while the non-survivors showed a discrete decrease in PaO2/FiO2 after 7 days despite the treatment provided (Table 2).

A total of 11.6% of the patients (n = 5) received antimicrobial treatment (ATB) upon admission to the ICU due to clinically suspected bacterial co-infection. Four patients received ceftriaxone plus a macrolide, and one patient was treated with piperacillin / tazobactam plus a macrolide. No pathogenic microorganisms were isolated from the lower airway samples (BAS = 5) of these patients. In all subjects, antigen testing in urine for S. pneumoniae and Legionella spp. proved negative, in the same way as the bronchoaspirate and blood cultures performed upon admission to the ICU. No case of bacterial co-infection was recorded in our series. The median PCT concentration upon admission was 0.45 ng/ml (IQR: 0.04-2.45), while the median C-reactive protein concentration was 23 mg/dl (IQR: 10-28). On the other hand, 13 of the 43 patients (30.2%) developed VAP, representing an incidence of 13.3 cases/1000 days of IMV. The microorganisms isolated were S. anginosus (n = 3), P. aeruginosa (n = 3), methicillin-sensitive S. aureus (MSSA)(n = 2), E. coli (n = 1), S. oralis (n = 1), K. pneumoniae (n = 1), E. faecalis (n = 1) and Corynebacterium spp. (n = 1). None of the respiratory sample analyzed proved positive for Aspergillus spp.