We retrospectively analysed patients from the CIBERESUCICOVID study (NCT04457505),8,9 which had prospectively included patients aged ≥18 years with laboratory-confirmed SARS-CoV-2 infection from across 55 Spanish hospitals between 5 February 2020 and 7 October 2021 (participating sites are listed in the S-Table 1 in the Supplementary Material). All consecutive patients admitted to ICU were enrolled if the reason for admission was COVID-19. Exclusion criteria for patients included: (1) unconfirmed SARS-CoV-2 infection; (2) lack of data at baseline or hospital discharge; (3) lack of information about age; (4) lack of data about ventilation requirement or conventional oxygen therapy at hospital and ICU admission. The study received first approval by Hospital Clínic of Barcelona, Spain IRB (Comité Ètic d'Investigació Clínica, registry number HCB/2020/0370), and ulterior approval by local IRBs in the rest of participating hospitals. Either patients or their relatives provided informed consent. De-identified data were collected and stored in Research Electronic Data Capture (REDCap). Trained local researchers incorporated data from patients’ medical records into a separate database. Prior to statistical analyses, three independent and experienced data collectors trained in critical care (PC, AM, CS) reviewed the data; in cases of query, site investigators were contacted. Missing analyses were performed, and site investigators were approached to obtain as much reliable and complete data as possible. Results are reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines.10

We recorded data on demographics, comorbidities, illness severity and organ damage (APACHE-II and SOFA scores), and previous treatment. Standard laboratory and clinical data were collected at hospital and ICU admission. Data on pharmacologic treatments and non-pharmacological interventions during index admission were collected. Main complications during hospital stay, including pulmonary complications (acute respiratory distress syndrome-ARDS); septic shock, bacteraemia, hyperglycaemia, nosocomial infections, thromboembolic events, gastrointestinal bleeding, acute kidney injury and acute hepatic failure were also collected.

The primary outcome was in-hospital mortality. Secondary outcomes included length of ventilation, recovery from ICU admission, ICU-mortality, 90-day mortality, lengths of ICU and hospital stay.

Patients were divided in two groups: 1.- Patients that received non-invasive respiratory support (NIRS) which included patients that received non-invasive mechanical ventilation (NIMV) and/or high-flow nasal cannula (HFNC) at the ICU admission, and 2.- Patients that received invasive mechanical ventilation (IMV) at the ICU admission. Patients who received NIRS before but needed intubation at the ICU admission were included in the IMV group. The start dates of the first respiratory support with NIRS or IMV were recorded whether it was provided in the general ward or in the ICU. Length of ICU and hospital stay was calculated from ICU admission and hospitalization, respectively. Nosocomial pneumonia was defined according to international guidelines.11 Hyperglycaemia was defined as a consistent blood glucose level above 126 mg/dL. Hemorrhage referred to any type of clinically significant bleeding. Further details are reported in a previous publication.12 Driving pressure was defined as plateau pressure minus plateau pressure (PEEP). Static compliance of the respiratory system was calculated as tidal volume/ (plateau pressure − PEEP). Ventilatory ratio was calculated as follows: (minute ventilation  ×  PaCO2) − (PBW × 100  ×  37.5).

We report the number and percentage of patients as categorical variables, and the median (first quartile [Q1]; third quartile [Q3]) as continuous variables. Categorical variables were compared using the chi-squared test or Fisher's exact test, whereas continuous variables were compared using the nonparametric Mann-Whitney U test.

First, we compared patients according to age group (<70 years and ≥70 years). Then, a comparison of patients according to study group (i.e., NIRS and IMV) in patients aged ≥70 years was performed. We also explored the clinical characteristics and outcomes in the subgroup of patients aged 80 years and older.

To describe in-hospital mortality, we utilized a competing risk model,13 considering recovery (i.e., discharge from hospital) as competing risk for mortality. First, we obtained the estimate of the cumulative incidence function (CIF) for the marginal probability of in-hospital mortality and recovery. Gray's test was used to compare equality of cumulative incidence curves across groups.14 To explore the risk factors associated with in-hospital mortality, a Fine-Gray competing risks model stratified on the center variable was used. A list of candidate predictors was established a priori based on previous findings and clinical constraints: age, sex, previous 30 days admission, chronic heart disease, chronic lung disease, chronic renal failure, confusion; the following parameters at ICU admission: APACHE-II score, SOFA score, PaO2/FiO2 ratio, pH, lymphocyte count, platelet count, d-dimers, C-reactive protein, serum creatinine, ferritin, septic shock, MV, and vasopressor treatment, continuous neuromuscular blockers, corticosteroids administered during ICU admission, and COVID-19 wave. Single collinearity was evaluated using the Pearson correlation (r) and multicollinearity was examined by means of the variance inflation factor (VIF). Several variables were excluded from the analysis due to collinearity (see Supplementary Material). Sub-distribution hazard ratios (sHRs) and their 95% confidence intervals (CIs) were calculated. The proportional hazards assumption was checked by an evaluation of the Schoenfeld residuals, as shown in Supplementary S-Figure 1. Patients who were transferred to another hospital were censored in the survival analyses. We used the multiple imputation method15 for missing data in the multivariable analysis (S-Table 1).

The level of significance was set at 0.05 (two-tailed). All analyses were performed using IBM SPSS version 26.0 (IBM Corp., Armonk, NY, USA) and R version 4.1.1 (R Foundation for Statistical Computing, Vienna, Austria).

5090 patients requiring ventilation due to COVID-19 were enrolled in the CIBERESUCICOVID dataset (55 Spanish ICUs) from February 2020 to October 2021. The comparison of characteristics and outcomes between patients aged <70 years and those aged ≥70 years are summarized in S-Tables 2-4 and S-Figures 2-3. Remarkably, 3565 (63%) were aged <70 years (1529 [43%] received NIRS and 2036 [57%] received IMV) and 1525 (27%) were aged ≥70 years (554 [36%] received NIRS and 971 [64%] received IMV) (Fig. 1). Overall in-hospital mortality was 31% (23% in patients <70 years and 50% in those ≥70 years; p<0.001).

Fig. 1.

Flow diagram of the study population.

(0.35MB).

The overall baseline characteristics and ventilation features in patients aged ≥70 years and the comparison between the group receiving NIRS and IMV are shown in Table 1. Notably, patients received NIRS presented higher proportion of patients aged ≥80 years old, have higher rate of chronic lung disease, chronic renal disease and immunosuppression that patients received IMV. They also presented longer day from hospital admission to ICU admission, lower rate of septic shock, lower levels of CRP, d-dimer, neutrophils-lymphocytes ratio and lower SOFA score compared with patients who received IMV.

Table 3 shows the complications and outcomes according to the type of MV in patients ≥70 years. Medians for ICU and hospital length of stay were 17 (9; 30) and 26 (16; 44) days for NIRS and IMV respectively. The mortality rate of patients that failed to NIRS and required IMV was 52% (149/288), whereas the mortality rate of patients that only required NIRS was 26% (55/214). ICU, in-hospital and 90-day mortality rates were 46%, 50% and 52% respectively, in all three cases being significantly higher in the IMV subgroup. The main cause of in-hospital mortality in IMV group was multi-organic failure (41%), while, respiratory failure was the main cause of death in NIRS group (51%). The CIF curves for in-hospital mortality and recovery are depicted in Fig. 2A. Furthermore, the CIF curves show that patients with IMV had a higher likelihood of death (p<0.001) than patients with NIRS, and patients with NIRS had a higher likelihood of recovery (p<0.001) than patients with IMV (Fig. 2B).

Fig. 2.

Cumulative incidence plot of in-hospital mortality and recovery in the overall population of patients ≥70 years old (N = 1525) (A) and according to type of respiratory support group (B).

(0.2MB).

The characteristics of patients aged ≥70 years that survived the index admission vs. those of patients who died are shown in S-Tables 5 to 7. In-hospital mortality significantly increased per 5-year blocks age groups (p<0.001) (Fig. 3A). Meanwhile, there was a decreasing trend in in-hospital mortality across COVID-19 waves (p = 0.006) (Fig. 3B).

Fig. 3.

In-hospital mortality per age group (A), and during the five COVID-19 pandemic waves (B). Study population ≥70 years old (N = 1525).

(0.15MB).

There were 136 patients ≥80 years old, of these 84 (62%) patients received NIRS (28 with initial NIRS, required IMV during hospitalization) and 52 (38%) received IMV. Median APACHE II and SOFA scores were 14 (12; 17) and 4 (4; 7), respectively (S-Table 8). Interestingly, prone position was implemented in 35% of patients and renal replacement therapy was used in 7% of patients (4% in NIRS and 12% in IMV patients; p = 0.085) (S-Table 9). The mortality rate of patients that failed to NIRS and required IMV was 61% (17/28), whereas the mortality rate of patients that only required NIRS was 55% (24/44). Remarkably, ICU, in-hospital and 90-day mortality rates were 51%, 61% and 65% respectively; and respiratory failure (52% in the NIRS group vs. 43% in the IMV group) and multi-organic failure (33% in the NIRS group vs. 24% in the IMV group) were the main causes of in-hospital mortality without differences between groups. Medians for ICU and hospital length of stay were 13 (7; 23) and 29 (17; 45) days, respectively (S-Table 10).

Results of the multivariable analysis are reported in Table 4. The following factors were associated with in-hospital mortality: age, previous admission within the last 30 days, chronic heart disease, chronic renal failure, platelet count, MV, and corticosteroids. Firstly, with every year increase in age, the risk of death increased with 7% (sHR 1.07, 95% CI 1.05 to 1.10), and the chances of recovery decreased with 6% (sHR 0.94, 95% CI 0.91 to 0.96). In other words, if in two patients all variables except for age are the same, the patient who is one year older has a 7% higher risk of dying. Furthermore, patients with previous admission within the last 30 days had a 40% increased risk of death (sHR 1.40, 95% CI 1.04 to 1.89). Moreover, patients with chronic heart disease had a 21% increase in risk of death (sHR 1.21, 95% CI 1.01 to 1.44), while patients with chronic renal failure had a 43% increase in risk of death (sHR 1.43, 95% CI 1.12 to 1.82), and 33% decrease in chances of recovery (sHR 0.67, 95% CI 0.49 to 0.92). In terms of arterial blood gasses, a ten-fold increase in APACHE-II score at ICU admission, the risk of death increased 1% (sHR 1.01, 95% CI 1.00 to 1.03). In terms of laboratory parameters, a ten-fold increase in platelet count at ICU admission was associated with a 2% decrease in risk of death (sHR 0.98, 95% CI 0.98 to 0.99), and a 2% increase in chances of recovery (sHR 1.02, 95% CI 1.01 to 1.03). Moreover, patients with IMV at ICU admission had a 41% increase in risk of death (sHR 1.41, 95% CI 1.16 to 1.73), and 42% decrease in chances of recovery (sHR 0.58, 95% CI 0.47 to 0.72). Finally, patients that used corticosteroids had a 39% decrease in the risk of death (sHR 0.61, 95% CI 0.48 to 0.77).