Respiratory system compliance at the same PEEP level is similar in COVID and non-COVID ARDS

Background The comparison of respiratory system compliance (Crs) between COVID and non-COVID ARDS patients has been the object of debate, but few studies have evaluated it when considering applied positive end expiratory pressure (PEEP), which is one of the known determinants of Crs itself. The aim of this study was to compare Crs taking into account the applied PEEP. Methods Two cohorts of patients were created: those with COVID-ARDS and those with non-COVID ARDS. In the whole sample the association between Crs and type of ARDS at different PEEP levels was adjusted for anthropometric and clinical variables. As secondary analyses, patients were matched for predicted functional residual capacity and the same association was assessed. Moreover, the association between Crs and type of ARDS was reassessed at predefined PEEP level of 0, 5, 10, and 15 cmH2O with a propensity score-weighted linear model. Results 367 patients were included in the study, 276 patients with COVID-ARDS and 91 with non-COVID ARDS. The association between Crs and type of ARDS was not significant in both the complete cohorts (p = 0.17) and in the matched cohorts (p = 0.92). This was true also for the propensity score weighted association at PEEP 5, 10 and 15 cmH2O, while it was statistically significant at PEEP 0 (with a median difference of 3 ml/cmH2O, which in our opinion is not clinically significant). Conclusions The compliance of the respiratory system is similar between COVID ARDS and non-COVID ARDS when calculated at the same PEEP level and while taking into account patients’ anthropometric characteristics. Supplementary Information The online version contains supplementary material available at 10.1186/s12931-022-01930-0.


Background
Acute respiratory distress syndrome (ARDS) is common in severe novel coronavirus 2019 disease (COVID-19) [1]. Recently, it was suggested [2] that respiratory system compliance (C rs ) might be different in COVID-ARDS when compared to ARDS from other diseases, due to the existence of two different "phenotypes" in COVID ARDS which might not be present in non-COVID ARDS. These phenotypes have been identified as Type L, characterized by Low elastance (i.e., high compliance) and Type H, characterized by High elastance (i.e., low compliance) [2]. Subsequent studies have compared C rs between COVID ARDS and non-COVID ARDS, but the debate concerning differences in C rs between COVID and non-COVID ARDS patients is still ongoing, due to conflicting findings [3][4][5][6].

Open Access
*Correspondence: f.fusina@gmail.com 1 Department of Anesthesia, Intensive Care and Pain Medicine, Fondazione Poliambulanza Hospital, via Bissolati, 57, 25124 Brescia, Italy Full list of author information is available at the end of the article Fusina et al. Respiratory Research (2022) 23:7 C rs is modified by Positive End Expiratory Pressure (PEEP) levels [7][8][9][10][11], but previous studies have compared C rs between COVID ARDS and non-COVID ARDS at different PEEP levels [3,5,12]. Only by knowing C rs at similar PEEP levels can the C rs of COVID and non-COVID ARDS patients be correctly compared, thus avoiding attributing to respiratory system characteristics what could be the effect of PEEP itself on C rs . Therefore, the aim of the study was to assess if C rs in COVID ARDS and non-COVID ARDS is similar or different when the applied PEEP is taken into account.

Methods
In this retrospective cohort study, data was collected from all consecutive adult subjects (over 18 years of age) with a diagnosis of ARDS (according to the Berlin Definition criteria [13]) to the Intensive Care Unit (ICU) of Poliambulanza Foundation Hospital of Brescia (Lombardy, Italy) from January 1st 2015 to May 1st 2021.The referral Ethics Committee (Comitato Etico di Brescia) approved the study (protocol number 4893).
Subjects were excluded from the analysis if: (a) they did not undergo invasive mechanical ventilation with sedation and paralysis; (b) no recordings of C rs at different PEEP levels during invasive ventilation with sedation and paralysis during the ARDS period were available.
Two cohorts of subjects were created: (1) subjects with COVID-19 ARDS, i.e., the ones with ARDS attributable to Severe Acute Respiratory Syndrome-CoronaVirus-2 (SARS-CoV -2) infection. (2) subjects with non-COVID ARDS, i.e., ARDS subjects without SARS-CoV 2 infection. ARDS patients were included in this group if ICU admission occurred before February 18th 2020, the day of the first COVID-19 diagnosis in our region (RT-PCR test for SARS-CoV-2 was not performed before this date), or if they had a negative RT-PCR test for SARS-CoV-2 after 18th February 2021.
Data on respiratory mechanics were collected, on median, on the same day as ARDS diagnosis in both COVID ARDS and non-COVID ARDS (0 [0-1] and 0 [0-1] days after ARDS diagnosis, respectively, p = 0.51). Tracheal intubation was performed, on median, on the same day as ARDS diagnosis in patients with COVID ARDS and on the day before ARDS diagnosis in patients with non-COVID ARDS (0 [0-1] and 1 [0-1.5] days before ARDS diagnosis, respectively, p < 0.001).
All measurements were taken with patients in the semirecumbent position, sedated and paralyzed. Ventilatory settings were adjusted according to our ICU's ventilation protocol [4]: C rs at different PEEP levels was assessed with a "PEEP trial", setting volume controlled ventilation if not in use and maintaining the tidal volume set by the attending clinician in order to ensure a low-tidal volume ventilation. An end-inspiratory pause of 0.5 s was set and the respiratory rate was progressively decreased until complete exhalation was achieved. A stepwise increase (of 2 cmH 2 O) in PEEP level was performed, until an evident pattern of decrease of C rs with increasing PEEP was noticeable. The lowest PEEP level at which C rs was assessed in all patients was 4 cmH 2 O. A PEEP of 0 and 2 cmH 2 O could be applied in patients with SpO 2 higher than 85%. Each PEEP level was maintained for at least 2 min [14]. When the PEEP trial was completed, the previous ventilatory mode and respiratory rate were restored, and PEEP was set to obtain a total PEEP equal to the PEEP associated with the highest C rs obtained during the PEEP trial. The PEEP trial was stopped when a systematic increase in driving pressure was seen with increasing PEEP, or when adverse cardiovascular effects or a decrease in SpO 2 greater than 10% of baseline were noticed.
The "PEEP trial" was completed, on median, on the same day as intubation in all groups.
Data (demographic, clinical, laboratory data and outcome) were extracted from the electronic medical chart of enrolled subjects.

Measurements and calculations
Respiratory system compliance (C rs ) was calculated as the ratio between tidal volume and driving pressure, which is the pressure distending the lungs, and is calculated as end inspiratory pressure (Plateau Pressure) minus PEEP [7]. "Best compliance" was identified as the highest C rs obtained at the tested PEEP levels and "best minimum PEEP" was the lowest PEEP with which the best compliance was obtained in each patient.
Body mass index (BMI) was calculated as kg/m 2 , where kg is the weight in kilograms and m 2 is the square of the height in meters.
Ideal Functional Residual Capacity (FRC) was calculated as 2.34·height in meters + 0.01·age in years − 1.09 in male patients and as 2.24·height in meters + 0.001·age in years − 1.00 in female patients [15]. Predicted FRC was obtained correcting for BMI with the following formula 231.9·e (−0.070·BMI) + 55.2 [16].
Ventilatory ratio was calculated as (tidal volume·respiratory rate·PaCO 2 )/(ideal body weight·100 ml/min·37.5 mmHg), where tidal volume is expressed in milliliters and 37.5 is assumed to be the PaCO 2 during the ideal minute ventilation [17].

Study outcomes
The main study outcome was to assess if the C rs of the respiratory system was independently associated with the type of ARDS (COVID ARDS or non-COVID ARDS) when weighted for PEEP level and patients' baseline characteristics.

Statistical analysis
Sample size analysis was conducted with Montecarlo simulation based on data collected for a previous study [4]. Two hundred and sixty patients, hypothesising a ratio between COVID and non-COVID ARDS of 0.30, would guarantee a power of 0.80 to identify a difference in C rs between the two groups of 5 ml/cmH 2 O, with a type I error frequency of 0.05.
Data are shown as count (percentage) or median (interquartile range) and comparison between non-COVID ARDS and COVID ARDS were performed with χ 2 test and Wilcoxon rank-sum test, respectively.
We planned to analyze the association between C rs and type of ARDS at different PEEP levels, using linear mixed models. The dependent variable was C rs and the following covariates were chosen a priori: type of ARDS (COVID vs non-COVID), PEEP, age, sex, BMI, Ventilatory Ratio, tidal volume, PaO 2 /FIO 2 , with patients as random effect.
Two secondary analyses were planned, with the aim to repeat the analysis on matched samples: (1) two cohorts of non-COVID and COVID ARDS were created, matched for predicted FRC. Matching was performed with the nearest neighbor method. The association between C rs and the type of ARDS (COVID versus non-COVID) was assessed with linear mixed models with interaction, using the type of ARDS and PEEP level at which C rs was measured as covariate, and patients as random effect. (2) The average respiratory system compliance (C rs,avg ) was calculated in each patient at a PEEP level of 0, 5, 10, and 15 cmH 2 O, taking the average C rs obtained at a PEEP of 0, 5, 10 and 15 cmH 2 O ± 2 cmH 2 O. The association between C rs,avg and type of ARDS (COVID versus non-COVID) was reassessed at each of the four PEEP levels with linear model weighted with stabilized Inverse Probability Treatment Weighting obtained by propensity score [18]. Propensity score was calculated using age, sex, BMI, Ventilatory Ratio, tidal volume (used in the PEEP trial) and PaO 2 /FIO 2 as covariates.
The analyses were repeated excluding non-COVID patients with extrapulmonary ARDS.
Missing values were imputed with Multivariate Imputation by Chained Equations (Predictive Mean Matching). A p value lower than 0.05 was considered significant. Statistical analyses were performed with R 3.6.3 (R Core Team, 2021. R Foundation for Statistical Computing, Vienna) [19].

Results
Three hundred sixty seven subjects were included in the study, 276 with COVID-ARDS and 91 with non-COVID ARDS (Fig. 1). Patient characteristics are shown in Table 1. In 24% of non-COVID ARDS patients, the cause of ARDS was extrapulmonary. Mean arterial pressure,  Figure 2 shows C rs at different PEEP levels in non-COVID ARDS and COVID ARDS, in all enrolled patients and in the subset of patients which were matched for predicted FRC. The association between C rs and type of ARDS was not significant in both the complete cohorts (p = 0.17, see Additional file 1: Table S1) nor in the matched cohorts (p = 0.93, see Additional file 1: Table S2). Table 2 shows the results of the propensity score weighted association between C rs,avg and COVID status at different PEEP levels. C rs,avg was not associated with the ARDS type at PEEP 5, 10 and 15 cmH 2 O, showing a significant association at PEEP 0, despite a clinically insignificant difference. Estimates for the other variables are displayed in Additional file 1: Table S3. Figure 3 shows the distribution of C rs,avg at different PEEP levels.
Results of the analyses excluding non-COVID patients with extrapulmonary ARDS are shown in Additional file 1: Tables S4-S7).
Updating the power analysis with collected data, we reached a power of 0.85 in detecting a change in compliance of 4 ml/cmH 2 O between COVID and non-COVID ARDS, with a type I error frequency of 0.05.

Discussion
This study shows that C rs is similar between COVID ARDS and non-COVID ARDS when weighted for PEEP level and patients' baseline characteristics.
In particular, Haudebourg et al. [3] found similar C rs between the COVID and non-COVID ARDS patients, with values close to the ones found in our study (44 ml/ cmH 2 O in COVID e 42 ml/cmH 2 O in non-COVID patients in their study, versus 40 ml/cmH 2 O in COVID e 38 ml/cmH 2 O in non-COVID in our study). However, the authors did not adjust for patient characteristics, the cohorts were small (30 with COVID-ARDS versus 30 with non-COVID ARDS) and patients were not matched. In another work comparing COVID and non-COVID ARDS patients [20], the two cohorts were matched on PaO 2 /FiO 2 , FiO 2 , PEEP and tidal volume, but there was The last column (*) shows the comparison between non-COVID ARDS and COVID ARDS patients matched for predicted FRC. Best compliance is the highest C rs obtained at the tested PEEP levels Best minimum PEEP is the lowest PEEP level associated with the highest C rs PEEP positive end expiratory pressure, BMI body mass index, V T tidal volume, IBW ideal body weight, FRC functional residual capacity, C rs respiratory system compliance, ARDS acute respiratory distress syndrome no adjustment for patient characteristics. Higher (7 ml/ cmH 2 O) average values of C rs were found in COVID-19 patients, but the sample size was small (30 patients) and set and total PEEP were different between COVID and non-COVID patients for both high and low PEEP levels.
Other authors [5] found a higher median C rs in patients with COVID-19 (41 ml/cmH 2 O) than in those with classical ARDS (32 ml/cmH 2 O). The analysis was adjusted for sequential organ failure assessment score at ICU admission, sex, age, and PaO 2 /FiO 2 ratio, but PEEP selection was not protocolized in their patients. Ferrando et al. [12] found a tendency towards the statistical significance (p 0.06) by comparing the C rs of non-COVID ARDS patients from different studies and the authors' own data on COVID ARDS patients. Apart from the intrinsic mechanical properties of the respiratory system, C rs is affected by applied PEEP [7][8][9][10][11]. COVID and non-COVID patients often receive different PEEP levels during mechanical ventilation [4,20] therefore the comparison of C rs at different PEEP levels could reflect not only the impact of the disease on lung parenchyma but also the effects of the ventilatory treatment. Finding a similar C rs with a different applied PEEP in COVID and non-COVID ARDS [4] does not therefore exclude that the elastic properties of the respiratory system would have been different if the same level of positive pressure had been applied. On the other hand, a different C rs in COVID and non-COVID ARDS measured at a different PEEP levels cannot exclude similar mechanical properties of the respiratory system if the conditions of measurement had been the same [20].
Moreover, C rs depends on the FRC: the physiological variable which links compliance to FRC is called specific compliance (the ratio between C rs and FRC) [21][22][23]. C rs is dramatically different in healthy subjects with different FRCs, ranging on average between 5 and 7 ml/cmH 2 O in newborns [20] and 120 ml/cmH 2 O in adults [21]. When C rs is normalized by FRC, specific elastance (i.e., the ratio betweenFRC and C rs ) has similar values in healthy newborns, children and adults [24]. In other words, despite similar elastic properties for each milliliter of lung tissue, the same amount of applied pressure gives a different increase in volume (and therefore of C rs ) if the FRC is different. Matching for predicted FRC is meaningful when C rs is compared in COVID and non-COVID ARDS because FRC depends on height, age, sex and BMI [15,16,25], and some of these characteristics have been  shown to be different in COVID and non-COVID ARDS (mainly sex and BMI) [5]. The effect of a different predicted FRC could explain the findings of Li Bassi et al. [26] who found, in their analysis on COVID-19 ARDS patients, a lower C rs in female patients: on average, females have a smaller stature than males, and this predicts lower FRC. Moreover, female sex by itself gives a lower predicted FRC than male sex for the same height and age, whereas BMI, which was higher for female patients in Li Bassi's cohort, has an inverse relationship with predicted FRC. Therefore a lower C rs in female patients could be better explained by different anthropometric characteristics than by a different severity of the lung disease.
Based on these pathophysiological premises and clinical data, we designed an analysis which accounted for the effect of both PEEP and FRC.
Our analyses showed that C rs is not different between COVID and non-COVID ARDS if assessed at a similar PEEP level when the association was weighted for covariates, when it was carried out in a cohort matched for predicted FRC and when the inverse probability of treatment weighting was used to obtain unbiased estimates of average treatment effects. These results contribute to clarify the debate on the effect of the COVID-19 on alveolar damage, demonstrating that the inflammatory response involving the pulmonary parenchyma in ARDS affects the mechanical properties of the lung regardless of the primary disease. This is consistent with the pattern and definition of ARDS, which is considered as the common final pathway of different pulmonary and extrapulmonary diseases that spread their mediators of inflammation in the lungs [13,27].
The similar pattern of C rs in COVID and non-COVID ARDS is supported by a similar best C rs obtained during the PEEP trial (median 38 versus 40 ml/cmH 2 O in non-COVID and COVID ARDS, respectively; see Table 1 for further details), and by a similar lowest PEEP at which the best C rs was obtained.
As shown in Fig. 3 it is apparent that, for some patients, a high PEEP level worsens respiratory system compliance, while improving it for some patients. We can therefore speculate that the appearance of two "phenotypes" is in fact the effect of a high PEEP level, and not of COVID Primary analysis was also weighted on dead space, as estimated by Ventilatory Ratio [17], since one of the main differences between COVID and non-COVID ARDS appears to be the dead space ventilation, expression of altered perfusion and vascular endothelial injury [4,5,28]. The peculiar increase in dead space ventilation of COVID-19 has been explained as a consequence of pulmonary microvascular thrombosis which locally increases the ventilation/perfusion ratio [5,28,29].
The dead space increase in COVID ARDS compared with non-COVID ARDS is confirmed in our analysis, with a higher ventilatory ratio in COVID ARDS compared to non-COVID ARDS in both unmatched and matched cohorts.
This study has two main limitations. First, it is a single center study and therefore requires further confirmation. Nonetheless, since we analyzed the association of C rs and type of ARDS for a wide range of PEEPs, we believe that our data could be generalized to centers which use a different approach to mechanical ventilation. Second, it is a retrospective study, even if data was prospectively collected. More research is needed on the subject in order to corroborate these findings.

Conclusion
The compliance of the respiratory system is similar between COVID ARDS and non-COVID ARDS when calculated at the same PEEP level and while taking into account patients' anthropometric characteristics.