Introduction

The first formal description of acute respiratory distress syndrome (ARDS) dates back to 1967 [1]; however, it was only in 1994 that a broad consensus to define this complex syndrome was achieved [2]. These definitions were widely adopted by clinicians and researchers over the subsequent two decades. In 2012, however, a new definition of ARDS, the Berlin definition, was developed to address some of the limitations of the earlier definition [3].

Several aspects related to the management of patients with ARDS have changed over the last few decades, including use of lung protective ventilation [4], prone positioning [5], and extracorporeal membrane oxygenation (ECMO) [6, 7]. Despite these changes in patient management and respiratory support, ARDS is still associated with mortality rates between 40 and 60% and represents a high burden on intensive care resources [8]. Although several studies have assessed the epidemiology of, outcome from, and patterns of respiratory support in patients with ARDS [8,9,10,11,12,13], temporal changes have not been widely reported [14, 15] because of the use of different definitions and the considerable heterogeneity among cohorts. However, assessment of these changes is important to understand the evolution of the burden of the disease overtime and to trace the effects of possible changes in clinical practice.

Importantly, mechanical ventilation, the main pillar in the management of patients with ARDS, has been recognized as a possible cause of lung damage or ventilator-induced lung injury (VILI), which may have a negative impact on outcome [16, 17]. Accumulating evidence suggests that adopting a lung-protective strategy [4], by implementing low tidal volume, low plateau pressure, and titrated positive end-expiratory pressure (PEEP), does not per se preclude the development of VILI [18,19,20,21]. Assessment of the possible impact of ventilatory parameters on outcome may help in developing new approaches that may minimize VILI and improve survival.

In this post-hoc analysis, we tested the hypothesis that management of ARDS would change over time, especially with respect to ventilator settings, including driving pressure, which would have an impact on outcome in patients with ARDS. We therefore first assessed temporal changes in the epidemiology and management of ARDS requiring mechanical ventilation in European intensive care units (ICUs) included in two large observational studies, performed in 2002 (SOAP study) [22] and 2012 (ICON audit) [23], and second investigated the possible association between ventilatory settings on the first day of ARDS and outcome.

Methods

This was a post hoc analysis of two multicenter European cohorts. The SOAP study was conducted in 24 European countries and included 3147 patients [22]. The ICON audit included 10,069 patients from 82 countries worldwide [23]. For the purposes of this comparison, we considered only the 4601 ICON patients who were admitted to ICUs in the same 24 European countries as in the SOAP study and had physiologic and ventilation data recorded in the ICU (Fig. 1, Additional file 1: Table S1). For both studies, recruitment for participation was by open invitation and participation was voluntary. Institutional review board approval for both studies was obtained by the participating institutions according to local ethical regulations.

Fig. 1
figure 1

Flow diagram showing patient inclusion

Full size image

Participating ICUs (see Additional file 1: e-Appendix) were asked to prospectively collect data on all adult patients admitted between May 1 and 15, 2002, for the SOAP study and between May 8 and 18, 2012, for the ICON audit. In both studies, patients who stayed in the ICU for < 24 h for routine postoperative surveillance were not included. Re-admissions of previously included patients were also not included.

Data collection

Data were collected daily during the ICU stay for a maximum of 28 days. Data collection on admission included demographic data and comorbid diseases as well as source and reason for admission. Clinical and laboratory data for the Simplified Acute Physiology Score II (SAPS II) [24] score were recorded as the worst values within 24 h after admission. A daily evaluation of organ dysfunction/failure (cardiovascular, respiratory, renal, hepatic, coagulation, and central nervous systems) was performed using the sequential organ failure assessment (SOFA) score [25].

Values of tidal volume, PEEP, plateau pressure (Pplat) and fraction of inspired oxygen (FiO2) corresponding to the most abnormal value of arterial PO2 (PaO2) or arterial O2 saturation (SaO2) were recorded every 24 h; the mode of mechanical ventilation was not recorded. Patients were followed up for outcome data until death, hospital discharge or for 60 days.

Definitions

Patients were retrospectively identified as having ARDS requiring mechanical ventilation if they presented all the following: (a) severe hypoxemia, as defined by a PaO2/FiO2 ratio < 300 mmHg with a minimum of 5 cmH2O PEEP; (b) presence of bilateral lung infiltrates on the chest radiograph; (c) no evidence of pre-existing heart failure; (d) absence of chronic obstructive pulmonary disease (COPD) or other chronic pulmonary disorders; (e) invasive mechanical ventilation. The severity of ARDS was categorized according to the Berlin definitions into mild, moderate, and severe [6].

For calculation of tidal volume per predicted body weight (PBW), the average PBW of male patients was calculated as equal to 50 + [0.91 (height in centimeters—152.4)]; and that of female patients as equal to 45.5 + [0.91 (height in centimeters—152.4)] [4]. We calculated driving pressure as the difference between Pplat and PEEP. Due to the observational nature of the original studies [22, 23], the management of ARDS did not follow a predefined protocol.

Non-respiratory organ failure was defined as a SOFA score > 2 for the organ in question.

Outcome parameters

The primary outcome parameter was in-hospital mortality within 60 days of admission to the ICU. Secondary outcome parameters included death in the ICU, ICU and hospital lengths of stay, and organ failure as assessed by the SOFA score.

Statistical analysis

All data were processed and analyzed in the Department of Intensive Care of Erasme Hospital, University of Brussels, in collaboration with Jena University Hospital, Jena, Germany. Data were analyzed using IBM® SPSS® Statistics software, v.21 for Windows (IBM, Somers, NY, USA). Data were reviewed for plausibility and availability of the outcome parameter, and any doubts were clarified with the center in question. There was no on-site monitoring. Missing data represented < 6% of the data collected for SOAP and 6.1% of the ICON data.

Data are summarized using means with standard deviation, medians and interquartile ranges, or numbers and percentages. Difference testing between groups was performed using Student’s t test, Mann–Whitney test, Chi–square test or Fisher’s exact test, as appropriate. The Kolmogorov–Smirnov test was used, and histograms and quantile–quantile plots were examined to verify whether there were significant deviations from the normality assumption of continuous variables.

To evaluate the possible association between ventilatory parameters and outcome in patients with ARDS, we grouped the patients with ARDS from the SOAP study and ICON audit and performed a multivariable logistic regression analysis, with in-hospital death as the dependent variable. Covariates to be included in the final model were based on a univariate logistic regression analysis (p < 0.2) of demographic variables (age and sex), comorbid conditions, severity scores on admission to the ICU (SAPS II and SOFA scores), and severity of respiratory failure according to the PaO2/FiO2 ratio on the first day of mechanical ventilation. Colinearity between variables was ruled out before covariates were introduced in the model. Goodness of fit was tested using a Hosmer and Lemeshow test, and odds ratios (OR) with 95% confidence interval (CI) were computed. As driving pressure, Pplat, and PEEP are mathematically linked and were confirmed to be colinear (r2 > 0.6), we constructed separate logistic regression models for each parameter including the previously mentioned parameters. The multivariable models were adjusted for tidal volume > 8 ml/kg PBW, respiratory rate, the country of origin and the study period (ICON audit vs. SOAP study).

No statistical adjustments were used for multiple testing. All reported p values are two-sided and a p value < 0.05 was considered to indicate statistical significance.

Results

Temporal differences in the characteristics of patients with ARDS

The characteristics of the patients with ARDS included in the two cohorts are given in Table 1. The frequency of ARDS on admission to the ICU (5.1 vs. 5.0%, p = 0.866) and at any time during the ICU stay (10.4 vs. 10.7%, p = 0.793) was similar in the SOAP and ICON patients (Fig. 1). The diagnosis of ARDS was established at a median of 3 (IQ: 1–7) days after admission in the SOAP and 2 (IQ: 1–6) days in the ICON audit. Within 24 h of diagnosis, ARDS was mild in 244 (29.7%), moderate in 388 (47.3%), and severe in 189 (23.0%) patients.

Table 1 Characteristic of patients with acute respiratory distress syndrome (ARDS) on admission to the ICU
Full size table

Patients with ARDS in the later period (ICON audit) were more commonly admitted to the ICU for medical reasons than after surgical interventions (Table 1) and had slightly higher SAPS II and SOFA scores on admission to the ICU than those with ARDS included in the earlier study (SOAP) (Table 1).

Mechanical ventilation

Ventilator settings in ARDS patients who required mechanical ventilation in the SOAP study and ICON audit are shown in Table 2. Respiratory rates were similar in the two cohorts. Tidal volumes were set at lower levels in the later (ICON) than in the earlier (SOAP) cohort (Fig. 2). Although the proportion of patients ventilated with protective tidal volumes (≤ 8 ml/kg) was higher in ICON than in SOAP (35.5% vs 18.0%, p < 0.001) and of patients ventilated with tidal volumes associated with VILI (i.e., > 10 ml/kg) was lower (117/465 [25.2%] vs 151/322 [46.9%], p < 0.001), after 10 years, more than 60% of patients with ARDS were still ventilated with tidal volumes greater than 8 ml/kg. PEEP was set at a slightly lower level in the ICON compared to the SOAP, and Pplat and driving pressure were also lower in the ICON audit than in the SOAP study (Table 2 and Fig. 2).

Table 2 Ventilatory parameters within 24 h of meeting ARDS criteria on mechanical ventilation in the two cohorts
Full size table
Fig. 2
figure 2

Histograms with normality curves representing the tidal volumes, plateau pressures, and positive end-expiratory pressures (PEEP) during mechanical ventilation in patients with acute respiratory distress syndrome (ARDS) in the SOAP and ICON studies

Full size image

Morbidity and mortality

The incidence of hepatic failure on admission to the ICU was higher and the incidence of renal failure lower in the ICON audit than in the SOAP study; the overall prevalence of hepatic, renal, and cardiovascular organ failure during the ICU stay was higher in the ICON audit than the SOAP study (Additional file 1: Table S2). ICU lengths of stay were similar in patients with ARDS in the two cohorts [median (IQ: 10 (5–21) vs. 9 (4–18) days, p = 0.257], whereas, hospital lengths of stay were longer in the SOAP study than ICON audit [median (IQ: 27 (11–55) vs. 16 (7–34) days, p < 0.001]. Hospital mortality rates in patients with mild, moderate and severe ARDS were not significantly different between the two studies (Additional file 1: Figure S1). Patients with severe ARDS within 24 h of diagnosis or at any time during the ICU stay had higher hospital mortality rates than those with mild and moderate ARDS. However, hospital mortality rates were similar in patients with mild and moderate ARDS during the ICU stay (Additional file 1: Figure S1).

Predictors of worse outcome in patients with ARDS

In logistic regression analysis in all patients with ARDS from the two cohorts, with in-hospital death as the dependent variable, older age, greater SAPS II score, metastatic cancer, the presence of coagulation, renal and neurological system failures on admission to the ICU, and lower PaO2/FiO2 were independently associated with a greater risk of in-hospital death. Pplat > 29 cmH2O and driving pressure > 14 cmH2O on the first day of mechanical ventilation after establishing a diagnosis of ARDS, but not tidal volume > 8 ml/kg PBW or respiratory rate, were independently associated with a greater risk of death in these patients (Table 3).

Table 3 Logistic regression analysis with in-hospital death as the dependent variable in patients with ARDS
Full size table

Discussion

The main findings of our study are: (1) the frequency of ARDS in European ICUs did not change significantly from 2002 to 2012 and morbidity and mortality rates were similarly high; (2) ventilation with lower tidal volumes and lower airway pressures (Pplat and driving pressure) increased over time; and (3) Pplat > 29 cmH2O and driving pressure > 14 cmH2O on the first day of mechanical ventilation but not tidal volume > 8 ml/kg PBW were independently associated with a higher risk of death in these patients.

In these two large European ICU cohorts [22, 23], performed 10 years apart, the frequency of ARDS at any time during the ICU stay remained relatively constant over time at just over 10%. Bellani et al. [8] reported that 10.4% of patients admitted to ICUs in 50 countries had ARDS during the ICU stay using the Berlin definitions [3]. Other studies [26,27,28,29,30] have reported a frequency of ARDS between 3 and 29%, varying according to the studied population and the definition used. Indeed, we previously reported that the frequency of ARDS was 12.6% from the SOAP study database [9] using the earlier European American Consensus criteria [2], which may overestimate the actual frequency of ARDS by including mild cases of respiratory dysfunction. Although we used the Berlin definitions to define ARDS [3], only patients requiring invasive mechanical ventilation were considered in our analysis due to the absence of precise data on non-invasive mechanical ventilation. Therefore, the overall frequency of ARDS in our study may have been slightly underestimated. Nonetheless, the same set of data were collected using similar protocols for the two cohorts [22, 23] and we only included data from patients admitted to ICUs in the same 24 countries.

Our data confirm the persistently high morbidity and mortality rates in patients with ARDS. Other studies have similarly reported mortality rates ranging from 40 to 60% in these patients [8, 31]. ARDS represents a major burden to the healthcare system, making it an important target for research into how best to manage these patients so as to improve outcomes. Despite increased adherence to a lung-protective strategy in mechanically ventilated patients with ARDS observed in the more recent ICON audit [23] compared to the earlier SOAP study [22], mortality rates did not seem to have improved. Other factors may, therefore, have played a role in determining the outcome in these patients. Indeed, we identified several factors, such as older age, greater SAPS II score, metastatic cancer, and the presence of coagulation, renal, and neurologic organ failures on admission to the ICU as being independently associated with a greater risk of in-hospital death. These factors, reflecting the severity of illness and the degree of organ dysfunction in these patients on admission to the ICU, have been reported in previous studies [9, 10].

Although the proportion of ARDS patients ventilated with low tidal volume (≤ 8 ml/kg PBW) and low Pplat (≤ 29 cmH2O) was higher in the later ICON audit than in the early SOAP study, a considerable proportion of ARDS patients in both studies were not mechanically ventilated using lung protective settings. One possible explanation for this gap between best evidence and practice is that ARDS may not have been adequately recognized by the clinicians in the ICUs contributing to the SOAP study and ICON audit. Indeed, a large observational study in ICU patients in 50 countries reported that only 34% of clinicians recognized ARDS at the time of actual fulfillment of ARDS criteria as assessed by a computer algorithm from raw data, suggesting that diagnosis of ARDS is frequently delayed [8]. These authors [8] also reported that ARDS was underdiagnosed, with only 60% of all patients with ARDS being recognized by the clinician. We may also assume that use of pressure-controlled mechanical ventilation may lead to inevitable fluctuations in tidal volume with possible transitory increases above the required limit of 8 ml/kg PBW. Calculation of tidal volume according to the actual weight rather than the PBW may also lead to erronously high tidal volume levels, especially in obese patients. Tidal volume > 8 ml/kg PBW on the first day of mechanical ventilation was not associated with the risk of death in patients with ARDS. This is perhaps not so surprising because the increased use of lower tidal volumes in the ICON audit decreased the median tidal volume in patients with ARDS included in the analysis (9 ml/kg PBW), which may have masked the potentially deleterious effects of high tidal volume observed in our previous analysis on the SOAP study database [9]. Airway pressures were also generally low, which may have outweighed the possible deleterious effects of high tidal volume. Low tidal volume remains, therefore, a main stay of ventilator management for these patients as supported by the best available evidence [2].

Pplat > 29 cmH2O on the first day of mechanical ventilation after establishing the diagnosis of ARDS was independently associated with the risk of death in these patients. Indeed, Pplat is an important determinant of lung overdistention [32] and a good indicator of lung stress [17], and higher levels are well correlated to the risk of barotrauma [33]. Therefore, limiting Pplat is a crucial component of lung-protective ventilation.

We also observed a potentially deleterious influence of driving pressure > 14 cmH2O on outcome. Amato et al. reported that driving pressure was the variable most strongly associated with mortality in a post-hoc analysis of data from nine randomized controlled trials of mechanically ventilated patients with ARDS [21]. Driving pressure > 14 cmH2O was also reported to be associated with an increased risk of hospital mortality in patients with moderate and severe ARDS [8]. Another study showed that driving pressure was associated with risk of death in hypoxemic patients regardless of the results of the chest radiograph or the presence of ARDS [34].

Our study has some limitations. First, this was a post hoc analysis and ARDS was not a primary or secondary outcome in either of our cohorts, and was not predefined in the SOAP or ICON surveys. The ventilation and physiologic parameters were recorded as the worst values during the day, so that baseline values at the onset of ARDS cannot be precisely determined. Nevertheless, the variables needed to define ARDS were collected prospectively by the two studies. In addition, although the Berlin definition of ARDS [3] addressed some limitations of the earlier AECC definition [2], poor reliability of some criteria may contribute to underrecognition by clinicians [35]. Second, the multivariable analysis is limited by the variables included and the effects of other non-reported variables cannot be excluded. However, we adjusted for a large number of factors that are known to influence outcomes in patients with ARDS. Third, a cause-effect relationship between the risk factors we reported and outcome cannot be ascertained due to the observational nature of the study. In this context, our data can be considered as hypothesis-generating to help guide future RCTs on the subject. Fourth, colinearity between the various airway pressure parameters due to a mathematical link between these parameters precluded their inclusion in the same multivariable model. Finally, ventilatory parameters were recorded at a fixed time point and may have changed during the day. Reporting of these parameters also did not follow specific instructions to standardize the timing of measurements within the respiratory cycle and the possible effect of spontaneous breathing cannot be fully excluded due to the observational nature of the study.

Conclusion

The frequency of and outcome from ARDS remained unchanged between 2002 and 2012. The adoption of lower tidal volume in ARDS increased overtime and lower driving pressure and Pplat were observed in patients with ARDS included in the more recent ICON audit than in the earlier SOAP study. Pplat > 29 cmH2O and driving pressure > 14 cmH2O on the first day of mechanical ventilation, but not tidal volume > 8 ml/kg PBW, were independently associated with a higher risk of death.