Patients with acute respiratory distress syndrome (ARDS) are particularly exposed to ventilator-induced lung injury (VILI)1,2. In these patients, only a small proportion of the lungs is aerated and these ventilated areas might be exposed to excessive strain and stress, leading to local inflammation due to mechanical transduction3,4. The use of low tidal volume (VT=6 to 8mL/kg of predicted body weight [PBW]) and the maintenance of plateau pressure (Pplat) <30cm H2O might decrease mortality in ARDS patients by reducing VILI occurrence5,6.

However, limiting the VT based on PBW normalizes the VT to the lung size but might not guarantee protective mechanical ventilation (MV), particularly in patients with severe ARDS. In these patients, alveolar overdistension can occur at the end of inspiration even with low VT7. Moreover, protective MV might not be guaranteed by limiting the Pplat, especially in patients ventilated with low levels of positive end-expiratory pressure (PEEP)7. Facing these limitations, Amato et al. proposed the driving pressure (DP) as an important parameter to guide the ventilatory strategy in ARDS patients. DP allows to adjust VT based on the amount of preserved and aerated lung tissue, which can be better estimated by the compliance of the respiratory system (CRS) than by the PBW8. Observational studies have confirmed the association between lower DP levels during MV and lower mortality in ARDS patients9,10.

Patients with severe COVID-19 may develop acute respiratory failure and frequently fulfill the criteria for ARDS according to the Berlin definition11,12. For those patients, the same principles of protective MV, including VT of 6−8mL/kg PBW, Pplat lower than 30cm H2O, and DP lower than 15cm H2O, are recommended for both groups of patients13.

As the COVID-19 pandemic occurred years after the description of the concepts of protective MV, we hypothesized that adherence to protective MV might be greater during the pandemic, compared with the MV applied to ARDS patients before COVID-19. To test this hypothesis, we compared the ventilatory parameters applied to ARDS patients from two cohorts, one of ARDS caused by COVID-19 (C-ARDS) and the other of ARDS with other etiologies (NC-ARDS).

Overall 120 of 902 patients admitted to 37 Brazilian ICUs that participated in the original cohort study met the Berlin criteria for ARDS diagnosis at admission or on the first day of MV and constituted the NC-ARDS cohort. The C-ARDS cohort was constituted of 228 patients admitted to two COVID-19 dedicated ICUs (Fig. 1).

Figure 1.

Study participant flow chart.

(0.28MB).

Table 1 shows demographic and clinical characteristics, and respiratory parameters on day 1 of MV in both cohorts. Patients with C-ARDS were younger and presented severe form of ARDS less frequently. At baseline, a higher proportion of patients with C-ARDS presented cardiovascular dysfunction and a lower proportion presented renal dysfunction, compared with patients with NC-ARDS. C-ARDS patients received higher VT and higher PEEP levels, they presented higher Pplat and higher PaO2/FiO2 compared with patients with NC-ARDS. C-ARDS patients presented lower DP compared with patients with NC-ARDS. Similar results were found on day 2 of MV and they are shown in Table 2 in Supplementary Appendix.

Protective MV, as defined by the three parameters (VT ≤8mL/kg PBW; Pplat ≤30cm H2O; and DP≤15cm H2O), was applied to 65.8% of C-ARDS patients and to 50.0% of the NC-ARDS patients (p=0.005) on day 1 of MV. VT was lower than 8.0mL/kg PBW in 86.8% in C-ARDS and in 80.2% in NC-ARDS (p=0.12) on day 1 of MV. The proportion of patients with Pplat ≤30cm H2O was 91.7% in C-ARDS and 94.9% in NC-ARDS (p=0.28), and the proportion of patients with DP≤15cm H2O was 75.0% in C-ARDS and 62.4% in NC-ARDS (p=0.02) on day 1 of MV (Table 1 and Fig. 2). Fig. 3 shows the distributions of DP and Pplat in different VT and PEEP levels. Multivariable logistic regression showed that the C-ARDS cohort was independently associated with protective MV on day 1 (Table 2).

Figure 2.

Ventilatory parameters on the first day of mechanical ventilation in C-ARDS and NC-ARDS patients.

Cummulative frequency distribution of tidal volume, mL/kg PBW (A), plateau pressure, cm H2O (B); driving pressure, cm H2O (C); positive end expiratory pressure, cm H2O (D). Vertical dotted lines represent the respective cutoffs for tidal volume, plateau pressure and driving pressure. Horizontal dotted line represents the median of applied PEEP.

(0.27MB).

Figure 3.

Distribution of ventilatory parameters on the first day of mechanical ventilation in C-ARDS and NC-ARDS patients.

Distribution of tidal volume, mL/kg PBW against plateau pressure, cm H2O (A) and driving pressure, cm H2O (B). Distribution of PEEP, cm H2O against plateau pressure, cm H2O (C) and driving pressure, cm H2O (D). Dotted lines represent the respective cutoffs for each variable.

ARDS: acute respiratory distress syndrome; C-ARDS: COVID-19 acute respiratory distress syndrome; NC-ARDS: acute respiratory distress syndrome from other etiologies; PEEP: positive end expiratory pressure; PBW: predicted body weight.

(0.36MB).

On day 2 of MV, protective MV was applied to 67.1% of C-ARDS patients and to 42.0% of the NC-ARDS patients (p<0.001). On day 2, the proportions of patients ventilated with VT > 8mL/kg PBW or with Pplat > 30 cm H2O were similar between the two cohorts. However, 75.4% of C-ARDS patients presented DP≤15cm H2O compared to 53.9% of NC-ARDS patients (p<0.001) (Table 3, Figs. 3 and 4 in Supplementary Appendix).

There was neither significant difference in ICU mortality between C-ARDS and NC-ARDS cohorts (48.3% vs. 52.5%, respectively, p=0.45), nor in hospital mortality (52.2% vs. 60.5%, respectively, p=0.14) (Table 1). Protective MV (VT ≤8mL/kg PBW; Pplat ≤30cm H2O; and DP≤15cm H2O) was not independently associated with lower ICU mortality (Table 3) and hospital mortality (Table 5 in Supplementary Appendix) in both cohorts. Among the individual components of the protective MV definition, there was a significant association between DP≤15cm H2O and ICU mortality (Table 3). However, there were no significant associations between VT ≤8mL/kg PBW or Pplat ≤30cm H2O and ICU mortality (Table 4 in Supplementary Appendix).

In this observational study, which compared two prospective Brazilian cohorts of ARDS patients (C-ARDS and NC-ARDS), we found that a greater proportion of patients with C-ARDS received protective MV, compared with NC-ARDS patients. Multivariable analysis showed that the C-ARDS cohort was independently associated with protective MV. The higher adherence to protective MV in the C-ARDS cohort occurred mainly due to a higher proportion of patients with DP equal to or lower than 15cm H2O. Among the three individual components of the protective MV, only driving pressure was independently associated with ICU mortality.

The first clinical studies demonstrating that a protective MV could reduce mortality in ARDS patients were based on limiting VT and Pplat5,6. In our study, the proportion of patients with these two parameters on safe levels was similar between the two cohorts, and in accordance with recent studies in C-ADRS and NC-ARDS patients9,18,19. These findings might reflect progressive incorporation of ventilatory strategies aimed at reducing VILI throughout the last two decades, as demonstrated by Peñuelas et al. 14. These authors evaluated four cohorts of mechanically ventilated patients (1998, 2004, 2010, and 2016) and showed a significant increase in the use of protective MV over time. Moreover, they showed that 28-day mortality decreased significantly and that a lung protective strategy mediated an important fraction of the effect of temporal evolution on mortality.

More recently, DP has been considered an important ventilatory variable in the protective ventilatory strategy. Firstly, Amato et al. analyzed data from ARDS patients enrolled in previous randomized trials and demonstrated that DP was the ventilation variable best associated with mortality8. Since then, other authors have demonstrated an association between DP and mortality in ARDS patients of different etiologies10,20,21, including ARDS due to COVID-1918,22. In our study, C-ARDS patients presented lower DP than NC-ARDS patients, and a higher proportion of patients in the C-ARDS cohort presented DP equal to or lower than 15cm H2O. This difference was responsible for the higher adherence to protective MV in the C-ARDS cohort. It might have occurred due to progressive incorporation of the DP as a protective ventilatory parameter over the years that separate the two cohorts. Moreover, the C-ARDS cohort was formed with patients admitted to two COVID-19 dedicated ICUs over 18 months, i.e. patients with a single clinical condition, a fact that might have improved compliance to protective MV.

ICU and hospital mortality were similar between the two cohorts and in line with mortality rates observed in previous studies of ARDS patients due to COVID-19 or other causes9,22–25. In our analysis, the association between protective MV and ICU mortality did not reach statistical significance. When each component of the protective MV was analyzed separately, DP but not VT and Pplat was independently associated with ICU mortality. This may have occurred because most patients were ventilated with VT lower than 8mL/kg PBW and Pplat lower than 30cm H2O, leading to insufficient statistical power to detect an association between these variables and mortality. DP presented a broader distribution, especially in the NC-ARDS cohort, which may have happened because the role of DP in protective MV has been described more recently. Another possible explanation for these findings is that once DP is kept under safe levels, the importance of VT and Pplat as protective variables decreases26.

Our study has several relevant limitations. 1. The C-ARDS cohort was formed in only two COVID-19 dedicated ICUs, whereas the NC-ARDS cohort was formed in 37 ICUs from different Brazilian states. Therefore, the NC-ARDS cohort better represents the ventilatory practices in Brazil, compared to the C-ARDS cohort. 2. NC-ARDS was formed over only one month, and C-ARDS cohort was formed over 18 months, during which time adherence to protective MV may have progressively increased. 3. Ventilatory parameters were collected only on the first two days of MV and may not represent those applied during the following days. We cannot exclude the possibility that ventilatory parameters applied beyond day 2 influenced the mortality. 4. Not all clinical variables possibly associated with protective MV or mortality were collected, a fact that might have limited the precision of the DAGs that were drawn for the multivariable analyses. 5. Different disease severity scores were applied in each cohort (SAPS-3 in C-ARDS, SAPS-II in NC-ARDS), consequently they could not be compared.

In conclusion, we observed a higher adherence to protective MV among C-ARDS patients, mainly secondary to higher adherence to limiting DP. Incorporating DP in the protective MV seems to be important, since, among the three components of the protective MV, DP was the one best associated with mortality. However, to ascertain a causal relationship between DP and mortality in ARDS patients, randomized clinical trials are necessary.

• -

  Erich V. Carvalho: conception and design of the study, acquisition of data, analysis and interpretation of data, literature search, drafting the article, final approval of the version to be published.

• -

  Sérgio P. S. Pinto: conception and design of the study, acquisition of data, analysis and interpretation of data, literature search, drafting the article, final approval of the version to be published.

• -

  Edimar P. Gomes: conception and design of the study, acquisition of data, analysis and interpretation of data, literature search, drafting the article, final approval of the version to be published.

• -

  Óscar Peñuelas: conception and design of the study, analysis and interpretation of data, revising the article providing intellectual content of critical importance to the work described, final approval of the version to be published.

• -

  Caio G. Stohler: acquisition of data, analysis and interpretation of data, revising the article providing intellectual content of critical importance to the work described, final approval of the version to be published.

• -

  Gustavo C. Arantes: acquisition of data, analysis and interpretation of data, revising the article providing intellectual content of critical importance to the work described, final approval of the version to be published.

• -

  Lucas L. Carvalho: acquisition of data, analysis and interpretation of data, revising the article providing intellectual content of critical importance to the work described, final approval of the version to be published.

• -

  Roberta M. F. Oliveira: acquisition of data, analysis and interpretation of data, revising the article providing intellectual content of critical importance to the work described, final approval of the version to be published.

• -

  Pedro N. Martins: acquisition of data, analysis and interpretation of data, revising the article providing intellectual content of critical importance to the work described, final approval of the version to be published.

• -

  Maycon M. Reboredo: conception and design of the study, analysis and interpretation of data, drafting the article, final approval of the version to be published.

• -

  Bruno V. Pinheiro: conception and design of the study, analysis and interpretation of data, drafting the article, final approval of the version to be published.

This study was supported by a National Council for Scientific and Technological Development – CNPq.

None.