Lung-protective ventilation in intensive care unit and operation room Serpa Neto, A.

Transcription

1 UvA-DARE (Digital Academic Repository) Lung-protective ventilation in intensive care unit and operation room Serpa Neto, A. Link to publication Citation for published version (APA): Serpa Neto, A. (2017). Lung-protective ventilation in intensive care unit and operation room: Tidal volume size, level of positive end-expiratory pressure and driving pressure General rights It is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), other than for strictly personal, individual use, unless the work is under an open content license (like Creative Commons). Disclaimer/Complaints regulations If you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, stating your reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please Ask the Library: or a letter to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam, The Netherlands. You will be contacted as soon as possible. UvA-DARE is a service provided by the library of the University of Amsterdam ( Download date: 29 Jan 2018

2 305 Chapter 11 Epidemiological characteristics, practice of ventilation, and clinical outcome in patients at risk of acute respiratory distress syndrome in intensive care units from 16 countries (PRoVENT): an international, multicentre, prospective study Neto AS, Barbas CS, Simonis FD, Artigas-Raventós A, Canet J, Determann RM, Anstey J, Hedenstierna G, Hemmes SN, Hermans G, Hiesmayr M, Hollmann MW, Jaber S, Martin-Loeches I, Mills GH, Pearse RM, Putensen C, Schmid W, Severgnini P, Smith R, Treschan TA, Tschernko EM, Melo MF, Wrigge H, de Abreu MG, Pelosi P, Schultz MJ; PRoVENT; for the PROVE Network investigators Lancet Respir Med 2016; 4:882-93

3 306 Abstract Background: Scant information exists about the epidemiological characteristics and outcome of patients in the intensive care unit (ICU) at risk of acute respiratory distress syndrome (ARDS) and how ventilation is managed in these individuals. We aimed to establish the epidemiological characteristics of patients at risk of ARDS, describe ventilation management in this population, and assess outcomes compared with people at no risk of ARDS. Methods: PRoVENT (PRactice of VENTilation in critically ill patients without ARDS at onset of ventilation) is an international, multicentre, prospective study undertaken at 119 ICUs in 16 countries worldwide. All patients aged 18 years or older who were receiving mechanical ventilation in participating ICUs during a 1-week period between January, 2014, and January, 2015, were enrolled into the study. The Lung Injury Prediction Score (LIPS) was used to stratify risk of ARDS, with a score of 4 or higher defining those at risk of ARDS. The primary outcome was the proportion of patients at risk of ARDS. Secondary outcomes included ventilatory management (including tidal volume [V T ] expressed as ml/kg predicted bodyweight [PBW], and positive end-expiratory pressure [PEEP] expressed as cm H 2 O), development of pulmonary complications, and clinical outcomes. The PRoVENT study is registered at ClinicalTrials.gov, NCT The study has been completed. Findings: Of 3023 patients screened for the study, 935 individuals fulfilled the inclusion criteria. Of these critically ill patients, 282 were at risk of ARDS (30%, 95% CI 27 33), representing 0 14 cases per ICU bed over a 1-week period. V T was similar for patients at risk and not at risk of ARDS (median 7 6 ml/kg PBW [IQR ] vs 7 9 ml/kg PBW [ ]; p = 0 346). PEEP was higher in patients at risk of ARDS compared with those not at risk (median 6 0 cm H 2 O [IQR ] vs 5 0 cm H 2 O [ ]; p < ). The prevalence of ARDS in patients at risk of ARDS was higher than in individuals not at risk of ARDS (19/260 [7%] vs 17/556 [3%]; p = 0 004). Compared with individuals not at risk of ARDS, patients at risk of ARDS had higher in-hospital mortality (86/543 [16%] vs 74/232 [32%]; p < ), ICU mortality (62/533 [12%] vs 66/227 [29%]; p < ), and 90-day mortality (109/653

4 307 [17%] vs 88/282 [31%]; p < ). V T did not differ between patients who did and did not develop ARDS (p = for those at risk of ARDS; p = for those not at risk). Interpretation: Around a third of patients receiving mechanical ventilation in the ICU were at risk of ARDS. Pulmonary complications occur frequently in patients at risk of ARDS and their clinical outcome is worse compared with those not at risk of ARDS. There is potential for improvement in the management of patients without ARDS. Further refinements are needed for prediction of ARDS.

5 308 Introduction Invasive mechanical ventilation is a frequently applied intervention in patients in the intensive care unit (ICU). 1,2 Although ventilation is usually seen as a life-saving strategy, it has strong potential to worsen pre-existing lung injury. 3 Ventilation strategies aimed at preventing lung overdistension through use of low tidal volumes (V T 6 ml/kg predicted bodyweight [PBW]) improved the outcome (ie, decreased mortality) of patients in the ICU with acute respiratory distress syndrome (ARDS). 4,5 Thus, low V T is seen as the key element of so-called lungprotective ventilation in patients with this life-threatening complication of critical illness. In a meta-analysis of individual patient data from three randomised controlled trials, 6 ventilation strategies aimed at avoiding repetitive opening and closing of atelectatic lung tissue through use of high positive end-expiratory pressure (PEEP > 10 cm H 2 O) were beneficial in patients with moderate or severe ARDS. Therefore, several guidelines recommend the use of higher rather than lower PEEP in patients with moderate-to-severe ARDS. 7,8 Growing evidence suggests that ventilation not only worsens but also induces lung injury, particularly in patients at risk of ARDS. 3,9 Moreover, meta-analyses of observational studies and randomised controlled trials indicate improved outcomes with use of low V T during ventilation in patients in the ICU who did not have ARDS at the start of ventilation Convincing evidence, however, remains scarce. 14 Associations between ventilation at low V T and increased need for sedation and prolonged use of muscle paralysis are some reasons why clinicians remain reluctant to use low V T in patients without ARDS. 14 Whether PEEP benefits patients without ARDS is even more uncertain. 9,15 17 Risk of overdistension with higher PEEP, potentially inducing additional lung injury, has made clinicians reluctant to use PEEP as liberally in patients with uninjured lungs compared with in those with ARDS. 18,19 Preventing ARDS might be a more effective strategy than treating the disorder, with respect to improving outcomes of critically ill patients. One major obstacle to preventive studies is the inability to predict which patients are likely to develop ARDS. 20 Epidemiological data suggest that the syndrome is present rarely at admission, but develops over a period of hours to days in a subset of patients at risk of ARDS, 20 with considerable negative effects on

6 309 outcome. Although lung protection has the potential to improve outcomes in this subset of patients in particular, we do not know how ventilation is currently managed in these patients and whether management differs from that in patients not at risk of ARDS. As part of the PROtective VEntilation (PROVE) Network, we undertook the PRactice of VENTilation in critically ill patients without ARDS at onset of ventilation (PRoVENT) study to establish the epidemiological characteristics and clinical outcomes of patients at risk of ARDS, to describe and compare ventilation management in patients at risk of ARDS versus individuals not at risk, and to ascertain whether ventilation at higher VT is associated with higher prevalence of ARDS. Methods Study design and participants PRoVENT is an investigator-initiated, international, multicentre, observational, cohort study undertaken at 119 ICUs in 16 countries worldwide (Appendix). Part of the study protocol has been published previously. 21 Study sites were recruited through direct contact made by members of the PRoVENT steering committee and potential national coordinators. Approved national coordinators contacted local coordinators, who sought approval from their respective institutional review boards or research ethics committees. If required, written informed consent was obtained from individual patients. National coordinators assisted local coordinators and monitored the study according to the International Conference on Harmonization Good Clinical Practice guidelines. Local coordinators ensured integrity and timely completion of data collection. Consecutive patients aged 18 years or older were eligible for participation in the PRoVENT study if they were admitted to participating ICUs during a predefined 1-week period, as selected by the national coordinator for each country but within the timeframe for the study (January, 2014, to January, 2015), and received invasive ventilation. Invasive ventilation could have been initiated outside the hospital, in the emergency room, in a general ward, or in the operating room, or it could have been started in the ICU after

7 310 admission. We excluded patients in whom ventilation was started before the study recruitment week, those receiving only non-invasive ventilation, and individuals who were transferred to the ICU from another hospital under mechanical ventilation. Data for patients who fulfilled the Berlin definition for ARDS at the start of ventilation 8 were obtained but not included in the primary analysis. Procedures Site investigators obtained baseline and demographic variables on the day of admission to the ICU for calculation of disease severity scores (the Acute Physiology and Chronic Health Evaluation II [APACHE II] or the Simplified Acute Physiology Score III [SAPS III]) and the Lung Injury Prediction Score (LIPS). 20 We defined day 0 as the first calendar day that the patient received invasive ventilation, irrespective of ICU admission date, and recorded reasons for ventilation. Every day, until ICU discharge or death, site investigators assessed the patient for ventilation and intubation status (including tracheostomy). We counted a ventilation day as any day that the patient received mechanical ventilation, irrespective of the duration of mechanical ventilation for that day and whether or not it was done through an orotracheal tube or tracheostomy. Investigators were prompted on the case report form to provide an expanded dataset until day 7, at discharge from the ICU, and at 90 days after discharge, or after death in the ICU. They recorded ventilator settings and parameters, the patient s vital signs, transfusion requirements, daily fluid balances, sedation scores, and Sequential Organ Failure Assessment (SOFA) scores every day, close to 0800 h, until the end of mechanical ventilation, discharge from the ICU, or death in the ICU, as appropriate. Investigators also recorded rescue therapies for refractory hypoxaemia, including recruitment manoeuvres, inhaled nitric oxide, extracorporeal membrane oxygenation (ECMO) or extracorporeal removal of carbon dioxide (ECCO 2 R), high frequency oscillatory ventilation (HFOV), and prone positioning. We used either APACHE II or SAPS III scores to derive the patient s risk of death.

8 311 Patients data were anonymised before entry into a password-secured, web-based, electronic case record form (Oracle Clinical, Redwood Shores, CA, USA). Furthermore, before analysis, investigators screened all data for potentially erroneous findings and outliers and verified or corrected the information. Outcomes The primary outcome was the proportion of patients in the ICU who were at risk of ARDS, as established by a LIPS of 4 or higher. Secondary outcomes were ventilation management, occurrence of ARDS according to the Berlin definition, 8 and other pulmonary complications such as pneumonia, pneumothorax, pleural effusion, atelectasis, and cardiogenic pulmonary oedema. Site investigators diagnosed pulmonary complications, including ARDS, with chest radiographs and laboratory measures. We provided complete definitions for the pulmonary complication of interest, to increase efficiency and accuracy of pulmonary diagnoses (etable 1). Other endpoints were: the duration of ventilation, expressed as the number of ventilator days; the number of ventilator-free days, calculated as the number of days from weaning off invasive ventilation to day 28 (patients who died before weaning were deemed to have zero ventilator-free days); length of stay in the ICU and hospital; and mortality in the ICU, hospital, and at 90 days. This study is registered at ClinicalTrials.gov, NCT Statistical analysis Part of the analysis plan has been published before. 21 We stratified patients into ARDS risk groups based on LIPS, with those at risk of ARDS having a LIPS of 4 or higher and individuals not at risk of ARDS having a LIPS of less than 4. We have presented ventilation settings for all patients and focused on the first day only. We did not adjust for multiplicity across analyses, therefore, we do not claim confirmatory statistical evidence. The evidence level of our findings, however, is more than exploratory because we prespecified analyses in the protocol. Nevertheless, for analyses of outcomes, we controlled the false discovery rate using the Benjamini-Hochberg procedure, with a false discovery rate of 0 2.

9 312 We calculated the proportion of patients at risk of ARDS by dividing the number at risk of ARDS by the total number of patients. We ascertained the number of patients at risk of ARDS per ICU bed over the study period by dividing the number at risk of ARDS by the number of ICU beds available. We used scatterplots to present distributions of V T versus PEEP, V T versus respiratory rate, and V T versus plateau pressure. We chose cutoffs of 8 ml/kg PBW for V T, 14 breaths per min for respiratory rate, 30 cm H 2 O for plateau pressure, and 5 cm H 2 O for PEEP to form the matrices. We based these cutoffs on widely accepted values for each variable, or according to normal daily practice. We analysed V T and PEEP according to outcome subgroups: development of ARDS (yes vs no); development of other pulmonary complications (yes vs no); and hospital mortality (yes vs no). Because V T did not differ between patients at risk of ARDS and not at risk of ARDS, and between those who developed ARDS and who did not develop ARDS, we deviated from the original study protocol: we did not do association analyses of the relation between V T and the occurrence of ARDS. We did several post-hoc analyses. First, we compared ventilation parameters in patients who did not have ARDS with those in people who fulfilled the Berlin definition for ARDS at the start of ventilation, for whom data had been collected. Second, we analysed the driving pressure defined as plateau pressure minus PEEP with the same analysis plan as for other ventilatory parameters. We plotted combinations of V T versus driving pressure in a scatterplot, for which we used a cutoff of 15 cm H 2 O for the driving pressure to build the matrix. Finally, we ascertained the accuracy of LIPS for predicting development of ARDS using different cutoffs. To do this post-hoc analysis, we calculated the area under the receiver operating characteristic curve (AUC) and corresponding positive and negative predictive values, positive and negative likelihood ratios, and their 95% CIs. We also did a sensitivity analysis to establish the model performance at different cutoff points. We presented variables that were collected daily eg, V T, PEEP, peak and plateau pressure or maximum airway pressure (when available), respiratory rate, fraction of inspired

10 313 oxygen (FiO 2 ), and driving pressure as medians with IQRs. We presented V T as both an absolute volume (ml) and volume normalised for predicted bodyweight (ml/kg PBW). We calculated the predicted bodyweight of male patients as equal to (cm of height 152 4) and for female patients we used the calculation (cm of height 152 4). 8 The amount of missing data was low; therefore, we made no assumptions for missing data. We compared proportions with the χ² test or Fisher s exact test and continuous variables with the t test or Wilcoxon rank sum test, as appropriate. We did a Kaplan-Meier analysis of the cumulative probability of unassisted breathing and survival. We assumed that patients discharged from hospital before the end of follow-up were alive and without complications at this timepoint. We used log-rank tests to compare survival distributions in patients at risk of ARDS with those of individuals not at risk. To assess the effect of baseline imbalances and the association between risk of ARDS and outcomes, we developed a frailty model using centres as the cluster variable. First, we selected variables that showed imbalance at baseline for inclusion in a univariable model. We included variables with a p value less than 0 2 in unadjusted analyses in the final multivariable model. We judged a two-sided p value less than 0 05 significant. We did all analyses with SPSS version 20 and R version Role of the funding source There was no funding source for this study. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication. Results Between January, 2014, and January, 2015, 183 ICUs in 20 countries expressed interest in participating in the PRoVENT study, and 119 ICUs in 16 countries met inclusion criteria (etable 2 and etable 3). Of 3023 individuals screened for the study, 1021 patients were included; the main reason for exclusion was that they did not receive invasive ventilation (Figure 1). 86 patients had ARDS at the start of ventilation and their data were only included

11 314 in post-hoc analyses. Baseline characteristics of patients are shown in Table 1 and etable 4. Patients were followed up for a median of 17 days (IQR 9 35). Of 935 patients without ARDS who received mechanical ventilation, 282 (30%, 95% CI 27 33) were at risk of ARDS, representing 0 14 cases per ICU bed over a 1-week period. The proportion of patients at risk of ARDS showed significant geographical variation (p = 0 015): 225 of 785 in Europe (29%, 95% CI 25 32); two of 11 in North America (18%, 9 45); 46 of 106 in South America (43%, 34 53); and nine of 33 in Australasia (27%, 11 43). Median absolute V T was 500 ml (IQR ), and V T was 7 9 ml/kg PBW ( ), with no differences between patients at risk of ARDS and those not at risk (Table 2). V T was greater than 8 0 ml/kg PBW in almost 30% of patients in both risk groups (Figure 2A). V T did not differ between patients who developed ARDS and those who did not (Table 3), or other pulmonary complications (data not shown), and V T did not differ between patients surviving and those who died (data not shown). Patients at risk of ARDS received higher PEEP compared with individuals not at risk (median 6 0 cm H 2 O [IQR ] vs 5 0 cm H 2 O [ ]; p < ; Table 2; Figure 2B). PEEP did not differ between patients who developed ARDS and those who did not (Table 3), or other pulmonary complications (data not shown), but PEEP was higher in patients who died than in those who survived (data not shown). Pressure-controlled ventilation and synchronized intermittent mandatory ventilation were the most frequently used modes of invasive ventilation, with no differences noted between risk groups (Table 2). However, compared with individuals not at risk of ARDS, patients at risk were ventilated at a higher respiratory rate, received a greater FiO 2, and had augmented peak pressure and plateau pressure (p < ; Table 2 and Figure 2C). The median driving pressure in patients at risk of ARDS was slightly higher than in those not at risk (p = 0 048; Figure 2D). Descriptions of ventilatory parameters over the 1-week study period are shown in the efigure 1. Over the first 7 days, PEEP differed between patients at risk of ARDS and those not at risk (p = 0 009).

12 315 Distributions of various ventilatory parameters versus V T are presented in Figure 3. Patients were mainly ventilated with PEEP of 5 cm H 2 O or lower, independent of their risk of ARDS (Figure 3A). Roughly half the patients received ventilation with a V T of more than 8 ml/kg PBW and a plateau pressure less than 30 cm H 2 O (Figure 3B), with no differences by risk of ARDS. A combination of low V T and high respiratory rate was seen in around a third of patients, both in patients at risk of ARDS and in those not at risk (Figure 3C). Most patients received ventilation with a driving pressure below 15 cm H 2 O, irrespective of their risk of ARDS (Figure 3D). Use of adjunctive treatments was generally low, but was higher in patients at risk of ARDS versus those not at risk (10/263 [4%] vs 7/592 [1%]; p = 0 005; etable 5). Recruitment manoeuvres were the most frequently used adjuncts. All adjunctive treatments were applied after initial diagnosis of ARDS. Total pulmonary complications, ARDS, and pneumonia developed more frequently in patients at risk of ARDS than in those not at risk (Table 4). Most patients who developed ARDS did so on the second and third days of ventilation (efigure 2). The likelihood of discontinuing mechanical ventilation and 90-day survival was decreased in patients at risk of ARDS versus those not at risk (Figure 4). The number of ventilator-free days was lower in patients at risk of ARDS versus those not at risk (p = 0 002) and the length of stay in the ICU and the hospital was longer (p < ; Table 4). Mortality in the ICU, the hospital, and at 90 days was higher in patients at risk for ARDS, with roughly a third dying compared with less than a fifth of those not at risk (p < ; Table 4). Results of the false discovery rate adjustments are shown in the etable 6; there are no differences after this adjustment. Of 1021 patients included in the study, 86 were recognised as having ARDS at the start of ventilation. V T did not differ between patients with ARDS before ventilation and those without ARDS at the start of ventilation (etable 7 and efigure 3). Individuals with ARDS, however, were ventilated with higher PEEP compared with patients without ARDS at the start of ventilation. Driving pressure was higher in patients who had ARDS at the start of ventilation compared with those without ARDS at the start of ventilation (etable 7 and

13 316 efigure 3). Moreover, an association was noted between driving pressure tertile and mortality (efigure 4). The LIPS had an AUC of (95% CI ; p = 0 014; etable 8 and efigure 5). A sensitivity analysis of the performance of the LIPS model at different cutoffs is shown in the etable 8. Specificity rose with a higher cutoff, albeit with a sharp decrease in sensitivity. At a cut-off for LIPS of 4, the positive and negative likelihood ratios for development of ARDS were 1 8 (95% CI ) and 0 5 ( ), respectively, with a sensitivity of 0 67 ( ) and specificity of 0 63 ( ). Discussion Around a third of patients undergoing invasive ventilation were at risk of ARDS. Roughly half of those without ARDS received a V T greater than 8 ml/kg PBW, and the V T did not differ between patients at risk of ARDS and those not at risk; moreover, the V T was remarkably similar to that administered to patients with ARDS at onset of ventilation. PEEP was higher in patients at risk of ARDS compared with those not at risk; however, PEEP was lower than in patients with ARDS at onset of ventilation. Pulmonary complications were frequent in patients at risk of ARDS, and associated clinical outcomes were worse. We noted considerable geographical variation in the proportion of patients at risk of ARDS, ranging from 18% to 43%. Whether this difference is a reflection of seasonal differences in the risk of ARDS, or whether it is a true difference independent of risk factors (eg, influenza), is uncertain. The disparity is more likely to be accounted for by differences in casemixes attributable to factors such as admission policies or availability of beds in the ICU. The results of PRoVENT confirm those from previous investigations, in which V T has been reported as low as 7 ml/kg PBW and as high as 10 ml/kg PBW, but with decreasing trends over recent years. 1,8,22 30 Our V T findings suggest there is little or no titration on the basis of predicted bodyweight. Indeed, the median absolute V T was typically 500 ml, with a large variance when expressed in ml/kg PBW, suggesting scant individualisation. Even though V T was lower than reported previously in patients in the ICU, 1,8,22 30 the recorded V T

14 317 could still be deemed too large in many patients, since more than half received a V T greater than 8 ml/kg PBW. V T was similar in all patients in our study, irrespective of risk of ARDS. Forty-six percent of patients in our study received PEEP greater than 5 cm H 2 O, and only 5% received PEEP greater than 10 cm H 2 O. The effect of PEEP in patients without ARDS is debatable. Randomised controlled trials up to now have been too small and have mainly assessed outcomes that could be biased. 15,16,31 Notably, in one randomized controlled trial of PEEP, a higher PEEP (8 cm H 2 O vs 0 cm H 2 O) was suggested to prevent pneumonia, but this trial was underpowered for this endpoint. 15 In PRoVENT, most patients received ventilation at a low plateau pressure and high respiratory rate. A low plateau pressure is expected in a population of patients with uninjured lungs, in whom respiratory system compliance is high, thus, resulting in low airway pressure independently from V T. The finding that patients received mainly low V T at a high respiratory rate is important, because some data suggest that use of low V T could benefit even patients without ARDS. 2,9 14,32 Several investigations have shown an association between high driving pressure and mortality in patients with ARDS. 33,34 In one study in patients undergoing intraoperative ventilation under general anaesthesia, an association was noted between driving pressure and development of postoperative pulmonary complications. 35 We did not find any differences in driving pressure between patients at risk of ARDS and those not at risk, but we did note an association between higher driving pressure and mortality. Pulmonary complications are known to have an important effect on outcome in surgical patients. 17 The effect of pulmonary complications on clinical outcome in critically ill patients in the ICU without ARDS is less well understood. Our findings suggest that development of pulmonary complications is associated with a worse outcome. The proportion of patients at risk of ARDS who finally met the definition of ARDS during follow-up (19/260 [7%]) was similar to the value in another study of patients without ARDS at onset of ventilation (9%), using the same cutoff for the LIPS. 36 Even though specificity of LIPS rose with higher cutoffs, we remained with a cutoff of 4, because sensitivity became very low with

15 318 every increase in the cutoff and because this cutoff was used in original reports on this score. 20,36 Thus, LIPS might not be the best score with which to stratify patients without ARDS at onset of ventilation. Further refinements in prediction of ARDS are needed. Also, the absence of strict criteria for diagnosis of pneumonia could lead to an incorrect diagnosis. ARDS might have been incorrectly diagnosed as pneumonia in many cases, underestimating its true prevalence. Indeed, it is difficult to diagnose pneumonia in the presence of ARDS, with a sensitivity of less than 50% using conventional clinical criteria. 37 Simultaneous to the PRoVENT study, other investigators undertook the Large observational study to UNderstand the Global impact of Severe Acute respiratory FailurE (LUNG SAFE), 38 a multicentre, prospective, observational, 4-week inception cohort study. By contrast with the PRoVENT study in patients without ARDS, the investigators of LUNG SAFE enrolled mechanically ventilated patients with ARDS and prospectively assessed the burden of ARDS, management and therapeutic approaches to the disorder, and clinical outcomes, only during the winter months in the northern and southern hemispheres. PRoVENT and LUNG SAFE together provide a unique insight into worldwide practice of ventilation in patients in the ICU with and without ARDS. It is worth noting that most patients receiving mechanical ventilation in the ICU do not have ARDS. 14 PRoVENT has several limitations. First, the willingness of participating centres to join the study could have led to selection bias towards inclusion of centres with an interest in protective ventilation. Second, any prospective observational study can interfere with daily practice, because clinicians could have been keener to use lung-protective ventilation settings. Third, the number of centres per country was not limited, which could have resulted in an over-representation of some countries. Similar to other epidemiological studies, access to patients data was restricted to local investigators and researchers, and we could not control whether all patients under mechanical ventilation in participating centres were enrolled. The findings of PRoVENT extend our knowledge of ventilation practice in patients without ARDS and highlight the epidemiological characteristics of individuals at risk for the

16 319 disorder. Furthermore, they indicate the prevalence of pulmonary complications both in people at risk of ARDS and those not at risk, and their clinical outcomes. The international character of PRoVENT makes these results representative for many countries; moreover, the study s prospective design assured completeness of data collection, and the short timeframe within which data were gathered avoided the effect of practice changes over time. As such, the data presented here could function as a basis for new hypotheses and sample size calculations for future trials of mechanical ventilation. Finally, these data enable us to better interpret the findings of previous studies and their control groups. The findings of PRoVENT suggest there is potential for improvement in the management of patients without ARDS. Further refinements for prediction of ARDS are needed. Funding Support was provided solely from institutional and/or departmental sources.

17 320 Research in context Evidence before this study We searched MEDLINE, Embase, CINAHL, and Web of Science with the terms ( mechanical ventilation ) AND ( ARDS OR acute respiratory distress syndrome ) AND ( high risk OR LIPS ), with no date or language restrictions. We excluded studies of patients not receiving mechanical ventilation and those in paediatric populations. We did not find any study specifically assessing mechanical ventilation and outcomes in patients according to their risk of acute respiratory distress syndrome (ARDS) based on the Lung Injury Prediction Score (LIPS). Findings of a study using the original LIPS database suggested that clinicians seemed to respond to ARDS with a low initial tidal volume (V T ). Initial V T, however, was not associated with development of post-intubation ARDS or other outcomes. Nevertheless, this study assessed neither the proportion of patients at risk of ARDS nor possible differences in mechanical ventilation between this group of patients and those not at risk of ARDS. Added value of this study Our study is the first to focus specifically on the proportion of patients at risk of ARDS, ventilatory management of the disorder, and clinical outcomes, including pulmonary complications and mortality. Our study was prospective, with consecutive collection of data from patients and the inclusion of several intensive care units (ICUs) from different countries and continents, increasing its generalisability. We provided detailed descriptions of ventilatory parameters, pulmonary complications, and clinical outcomes. The proportion of patients at risk of ARDS was high, and clinical outcomes were worse in this subgroup than were those of patients not at risk of ARDS. Implications of all the available evidence Most patients on ventilation in the ICU do not have ARDS; however, a considerable number of individuals are at risk of this life-threatening complication. Early implementation of protective ventilation and other strategies in patients at risk of ARDS could be associated with better outcomes. Our results add to existing knowledge about epidemiological characteristics and outcomes of patients with ARDS, as described in the LUNG SAFE study,

18 321 and could be useful in planning future studies and understanding previous findings about mechanical ventilation in patients in the ICU. Further refinements for prediction of ARDS are needed.

19 322 Table 1 Baseline patients characteristics All At risk of ARDS Not at risk of ARDS (n = 935) (n = 282) (n = 653) p value a Age (years) 65 0 ( ) 65 0 ( ) 65 0 ( ) Gender (male) 62 6 (570 / 910) 62 8 (177 / 282) 62 6 (393 / 628) Ethnic origin 1 2 (11 / 903) 0 4 (1 / 282) 1 6 (10 / 621) African 1 2 (11 / 903) / 282) 0 8 (5 / 621) Afro-Caribbean 6 3 (57 / 903) 10 3 (29 / 282) 4 5 (28 / 621) Asian 84 2 (760 / 903) 79 1 (223 / 282) 86 5 (537 / 621) White European 7 1 (64 / 903) 8 2 (23 / 282) 6 6 (41 / 621) Latin American Body-mass index (kg/m 2 ) 25 5 ( ) 26 0 ( ) 25 3 ( ) Predicted bodyweight (kg) 64 2 ( ) 66 0 ( ) 64 2 ( ) Smoker Never 33 0 (298 / 902) 28 7 (81 / 282) 35 0 (217 / 620) Previous (stopped > 3 months) 17 0 (153 / 902) 16 7 (47 / 282) 17 1 (106 / 620) Former (stopped 3 months) 3 4 (31 / 902) 4 3 (12 / 282) 3 1 (19 / 620) Current 19 3 (174 / 902) 22 7 (64 / 282) 17 7 (110 / 620) Unknown 27 3 (246 / 902) 27 7 (78 / 282) 27 1 (168 / 620) Functional status Independent 75 0 (675 / 900) 66 7 (188 / 282) 78 8 (487 / 618) Partially dependent 17 6 (158 / 900) 23 0 (65 / 282) 15 0 (93 / 618) Totally dependent 4 4 (40 / 900) 7 4 (21 / 282) 3 1 (19 / 618) Unknown 3 0 (27 / 900) 2 8 (8 / 282) 3 1 (19 / 618) Reason for ICU admission Planned surgery 34 7 (313 / 902) 6 8 (19 / 281) 47 3 (294 / 621) Emergency surgery 20 7 (187 / 902) 31 7 (89 / 281) 15 8 (98 / 621) < Clinical condition 44 6 (402 / 902) 61 6 (173 / 281) 36 9 (229 / 621) NIV before intubation 7 7 (69 / 900) 15 2 (43 / 282) 4 2 (26 / 618) < Duration (min) ( ) ( ) ( ) < Risk of death (%)* 12 7 ( ) 29 4 ( ) 12 0 ( ) < LIPS** 3 5 ( ) 6 5 ( ) 2 5 ( ) < Limitation of treatment*** 3 4 (30 / 892) 6 1 (17 / 279) 2 1 (13 / 613) Unplanned admission 53 7 (483 / 900) 74 4 (209 / 281) 44 3 (274 / 619) < Reason for intubation Cardiac arrest 8 8 (79 / 900) 10 3 (29 / 282) 8 1 (50 / 618) Anesthesia for surgery (planned) 51 9 (467 / 900) 31 2 (88 / 282) 61 4 (379 / 618) < Depressed consciousness 26 6 (239 / 900) 31 9 (90 / 282) 24 1 (149 / 618) Respiratory failure 28 4 (255 / 900) 54 3 (153 / 282) 16 6 (102 / 618) < Chronic co-morbidity Hypertension 42 6 (381 / 894) 39 5 (111 / 281) 44 0 (270 / 613) Diabetes mellitus 18 5 (166 / 896) 15 3 (43 / 281) 20 0 (123 / 615) Heart failure 17 7 (158 / 894) 18 5 (52 / 281) 17 3 (106 / 613) Chronic kidney failure 10 5 (94 / 897) 12 8 (36 / 281) 9 4 (58 / 616) Cirrhosis 3 7 (33 / 896) 3 9 (11 / 281) 3 6 (22 / 615) COPD 12 0 (107 / 888) 17 9 (50 / 281) 9 4 (57 / 608) Oxygen at home 1 7 (16 / 935) 2 8 (8 / 282) 1 2 (8 / 653) Cancer 24 4 (219 / 896) 16 0 (45 / 281) 28 3 (174 / 615) < Former 7 3 (65 / 888) 5 4 (15 / 277) 8 2 (50 / 611) Current 16 4 (146 / 888) 9 4 (26 / 277) 19 6 (120 / 611) Neuromuscular disease 2 1 (19 / 895) 1 8 (5 / 281) 3 1 (19 / 614) Immunosuppression 7 8 (70 / 895) 7 8 (22 / 281) 7 5 (46 / 612) Use of NIV at home 1 2 (11 / 892) 1 8 (5 / 280) 1 0 (6 / 612) Severity of illness, SOFA score b Total 6 0 ( ) 8 0 ( ) 5 0 ( ) < Pulmonary 2 0 ( ) 2 0 ( ) 1 0 ( ) < Hematologic 0 0 ( ) 0 0 ( ) 0 0 ( ) Liver 0 0 ( ) 0 0 ( ) 0 0 ( ) 0 026

20 323 Circulation 1 0 ( ) 2 0 ( ) 0 0 ( ) < Neurology 2 0 ( ) 3 0 ( ) 2 0 ( ) < Renal 0 0 ( ) 0 0 ( ) 0 0 ( ) < Data are % (number of patients/total number of patients) or median (IQR) APACHE II: Acute Physiology and Chronic Health Evaluation II; ARDS: acute respiratory distress syndrome; COPD: chronic obstructive pulmonary disease; ICU: intensive care unit; LIPS: Lung Injury Prediction Score; SAPS III: Simplified Acute Physiology Score III; SOFA: Sequential Organ Failure Assessment; NIV: non-invasive ventilation a p represents the comparison between risk categories for each variable * Risk of death was derived from scores on APACHE II or SAPS III ** Scores range from 0 to 35, with 0 the least severe *** Decision made to withdraw treatment because of end of life Patient can have more than one diagnosis b Scores range from 0 to 24, with 0 the least severe. If data were missing, the value was omitted and the denominator adjusted accordingly

21 324 Table 2 Characteristics of invasive ventilation All (n = 935) At risk of ARDS (n = 282) Not at risk of ARDS (n = 653) p value a Mode of ventilation SIMV 26 3 (223 / 849) 29 5 (78 / 264) 24 8 (145 / 585) Pressure-controlled 22 7 (193 / 849) 22 3 (59 / 264) 22 9 (134 / 585) BiPAP / APRV 21 8 (185 / 849) 20 5 (54 / 264) 22 4 (131 / 585) Volume-controlled 13 7 (116 / 849) 16 7 (44 / 264) 12 3 (72 / 585) Pressure support 9 4 (80 / 849) 8 0 (21 / 264) 10 1 (59 / 585) PRVC 2 7 (23 / 849) 1 5 (4 / 264) 3 2 (19 / 585) ASV 2 0 (17 / 849) 0 4 (1 / 264) 2 7 (16 / 585) VAPS 0 9 (8 / 849) 1 1 (3 / 264) 0 9 (5 / 585) NAVA 0 1 (1 / 849) 0 0 (0 / 264) 0 2 (1 / 585) Other 0 4 (3 / 849) 0 0 (0 / 264) 0 5 (3 / 585) Peak pressure (cmh 2 O) 20 0 ( ) 22 0 ( ) 19 0 ( ) < Plateau pressure (cmh 2 O) b 16 0 ( ) 17 0 ( ) 15 0 ( ) < No of patients 36 7 (343 / 935) 40 8 (115 / 282) 34 9 (228 / 653) Absolute V T (ml) 500 ( ) 500 ( ) 500 ( ) V T (ml/kg PBW) 7 9 ( ) 7 6 ( ) 7 9 ( ) Control vent mode 7 7 ( ) 7 6 ( ) 7 8 ( ) Spontaneous vent mode 8 0 ( ) 7 8 ( ) 8 0 ( ) p value* (242 / 811) 33 7 (86 / 255) 28 1 (156 / 556) (347 / 811) 38 0 (97 / 255) 45 0 (250 / 556) (161 / 811) 22 0 (56 / 255) 18 9 (105 / 556) > (61 / 811) 6 3 (16 / 255) 8 1 (45 / 556) PEEP (cmh 2 O) 5 0 ( ) 6 0 ( ) 5 0 ( ) < (450 / 830) 40 2 (104 / 259) 60 6 (346 / 571) (253 / 830) 37 5 (97 / 259) 27 3 (156 / 571) (86 / 830) 15 8 (41 / 259) 7 9 (45 / 571) < > (41 / 830) 6 6 (17 / 259) 4 2 (24 / 571) Driving pressure (cmh 2 O) 10 0 ( ) 10 0 ( ) 9 0 ( ) No of patients 36 2 (339 / 935) 40 4 (114 / 282) 34 4 (225 / 653) Respiratory rate (bpm) 15 0 ( ) 16 0 ( ) 14 0 ( ) < FiO ( ) 0 5 ( ) 0 4 ( ) < Static compliance (ml/cmh 2 O) 54 2 ( ) 52 5 ( ) 56 0 ( ) Minute ventilation (L/min) 7 4 ( ) 7 6 ( ) 7 2 ( ) PaO 2 /FiO 2 (mmhg) 261 ( ) 201 ( ) 310 ( ) < PaCO 2 (mmhg) 38 0 ( ) 42 0 ( ) 37 5 ( ) < Arterial blood ph 7 36 ( ) 7 34 ( ) 7 38 ( ) < HCO 3 (meq/l) 22 0 ( ) 22 0 ( ) 22 0 ( ) Data are % (number of patients/total number of patients) or median (IQR) APRV: airway pressure release ventilation; ARDS: acute respiratory distress syndrome; ASV: adaptive support ventilation; BiPAP: biphasic positive airway pressure; FiO 2: fraction of inspired oxygen; NAVA: neurally adjusted ventilatory assist; PaCO 2: partial pressure of carbon dioxide; PaO 2: partial pressure of oxygen; PBW: predicted bodyweight; PEEP: positive end-expiratory pressure; PRVC: pressure-regulated volume control: SIMV: synchronised intermittent mandatory ventilation; VAPS: volume-assured pressure support; V T: tidal volume; bpm: breaths per minute a p represents the comparison between risk categories for each variable b Plateau pressure values are restricted to patients in whom this value was reported and in whom an control mode was used * p represents the comparison between modes

22 325 Table 3 Ventilatory parameters in patients who developed ARDS during follow-up compared with those who did not At risk of ARDS (n = 282) Not at risk of ARDS (n = 653) Developed ARDS Did not develop ARDS Developed ARDS Did not develop ARDS p value (n = 19) (n = 263) (n = 17) (n = 636) p value V T (ml/kg PBW) 7 6 ( ) 7 7 ( ) ( ) 7 9 ( ) Plateau pressure (cmh 2 O) 19 0 ( ) 17 0 ( ) ( ) 15 0 ( ) Driving pressure (cmh 2 O) 11 0 ( ) 10 0 ( ) ( ) 9 0 ( ) PEEP (cmh 2 O) 6 0 ( ) 6 0 ( ) ( ) 5 0 ( ) FiO ( ) 0 5 ( ) ( ) 0 4 ( ) Respiratory rate (bpm) 17 0 ( ) 16 0 ( ) ( ) 14 0 ( ) Data are median (IQR) ARDS: acute respiratory distress syndrome; PBW: predicted body weight; PEEP: positive end-expiratory pressure; FiO 2: fraction of inspired oxygen; V T: tidal volume

23 326 Table 4 Outcomes of patients receiving invasive ventilation, by risk of ARDS All (n = 935) At risk of ARDS (n = 282) Not at risk of ARDS (n = 653) p value a Adjusted hazard ratio (95% CI) b p value Pulmonary Complications Total 27 2 (222 / 816) 35 4 (92 / 260) 23 4 (130 / 556) ( ) Pneumonia 10 7 (85 / 816) 14 1 (36 / 260) 9 0 (49 / 556) ( ) ARDS 4 6 (36 / 816) 7 7 (19 / 260) 3 2 (17 / 556) ( ) Mild 1 3 (10 / 816) 1 6 (4 / 260) 1 1 (6 / 556) Moderate 2 7 (21 / 816) 5 2 (13 / 260) 1 5 (8 / 556) Severe 0 6 (5 / 816) 0 8 (2 / 260) 0 6 (3 / 556) Pneumothorax 1 4 (11 / 816) 1 6 (4 / 260) 1 3 (7 / 556) ( ) Pleural effusion 9 5 (74 / 816) 13 0 (32 / 260) 7 8 (42 / 556) ( ) Atelectasis 8 5 (67 / 816) 9 3 (23 / 260) 8 1 (44 / 556) ( ) Cardiogenic pulmonary oedema 1 9 (15 / 816) 3 6 (9 / 260) 1 1 (6 / 556) ( ) New pulmonary infiltrates 2 2 (17 / 816) 3 6 (9 / 260) 1 5 (8 / 556) ( ) Extra-Pulmonary Complications Acute kidney injury 19 0 (152 / 798) 29 8 (74 / 248) 14 2 (78 / 550) < ( ) Risk 4 5 (36 / 798) 8 1 (20 / 248) 2 9 (16 / 550) Injury 4 9 (39 / 798) 7 3 (18 / 248) 3 8 (21 / 550) Failure 7 1 (57 / 798) 11 3 (28 / 248) 5 3 (29 / 550) < Loss 1 3 (10 / 798) 1 2 (3 / 248) 1 3 (7 / 550) End-stage 1 3 (10 / 798) 2 0 (5 / 248) 0 9 (5 / 550) Renal replacement therapy 4 4 (35 / 798) 6 0 (15 / 248) 3 6 (20 / 550) ( ) Infection 8 5 (68 / 798) 12 3 (31 / 248) 6 8 (37 / 550) ( ) Length of Stay All patients in the ICU (days) 4 0 ( ) 7 0 ( ) 3 0 ( ) < ( ) c Surviving patients (days) 4 0 ( ) 7 0 ( ) 3 0 ( ) < ( ) c All patients in hospital (days) 16 5 ( ) 22 0 ( ) 14 0 ( ) ( ) c Surviving patients (days) 17 0 ( ) 27 0 ( ) 14 0 ( ) < ( ) c Mechanical Ventilation Tracheostomy 6 9 (54 / 785) 11 3 (27 / 240) 5 0 (27 / 545) ( ) Duration of ventilation All patients 2 0 ( ) 2 0 ( ) 2 0 ( ) ( ) c Surviving patients 2 0 ( ) 2 0 ( ) 2 0 ( ) ( ) c 0 063

24 327 Ventilator-free days at day 28 d 25 0 ( ) 24 0 ( ) 25 0 ( ) Mortality ICU 16 8 (128 / 760) 29 1 (66 / 227) 11 6 (62 / 533) < ( ) Hospital 20 6 (160 / 775) 31 9 (74 / 232) 15 8 (86 / 543) < ( ) Day 21 1 (197 / 935) 31 2 (88 / 282) 16 7 (109 / 653) < ( ) Data are % (number of patients/total number of patients) or median (IQR) ARDS: acute respiratory distress syndrome; ICU: intensive care unit; HR: hazard ratio; CI: confidence interval; SOFA: Sequential Organ Failure Assessment a p represents the comparison between risk categories for each variable b Frailty model adjusted for body-mass index, functional status, risk of death, and SOFA total at baseline c Coefficient of a multi-level linear regression d In patients who died while receiving invasive mechanical ventilation, invasive ventilation-free days are counted as 0

25 328 Figure Legends Figure 1 Screening and enrolment Figure 2 Ventilation parameters in patients at risk of ARDS versus those not at risk Figure 3 Distribution of ventilatory parameters on the first day of ventilation in patients at risk of ARDS versus those not at risk Figure 4 Outcome of inpatients at risk of ARDS versus those not at risk

26 329 Figure 1 Screening and enrolment ARDS: acute respiratory distress syndrome; IRB: institutional review board

27 330 Figure 2 Ventilation parameters in patients at risk of ARDS versus those not at risk Cumulative frequency distribution of (A) V T, (B) PEEP, (C) plateau pressure, and (D) driving pressure. Vertical dotted line represents the cutoff for each variable and horizontal dotted lines represent the respective proportion of patients reaching each cutoff. ARDS: acute respiratory distress syndrome; PBW: predicted bodyweight; PEEP: positive end-expiratory pressure; V T: tidal volume

28 331 Figure 3 Distribution of ventilatory parameters on the first day of ventilation in patients at risk of ARDS versus those not at risk Distribution of V T against (A) PEEP, (B) plateau pressure, (C) respiratory rate, and (D) driving pressure. Dotted horizontal and vertical lines represent the respective cutoffs for each variable. ARDS: acute respiratory distress syndrome; PEEP: positive end-expiratory pressure; PBW: predicted bodyweight; V T: tidal volume

29 332 Figure 4 Outcome of inpatients at risk of ARDS versus those not at risk (A) Probability of discontinuing mechanical ventilation. (B) Probability of 90-day survival. p values for log-rank test (unadjusted) and for the frailty model (adjusted by baseline imbalance). ARDS: acute respiratory distress syndrome

30 333 References 1. Esteban A, Frutos-Vivar F, Muriel A, et al. Evolution of mortality over time in patients receiving mechanical ventilation. Am J Respir Crit Care Med 2013; 188: Goligher EC, Ferguson ND, Brochard LJ. Clinical challenges in mechanical ventilation. Lancet 2016; 387: Slutsky AS, Ranieri VM. Ventilator-induced lung injury. N Engl J Med 2013; 369: Acute Respiratory Distress Syndrome Network. Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome. N Engl J Med 2000; 342: Amato MB, Barbas CS, Medeiros DM, et al. Effect of a protective-ventilation strategy on mortality in the acute respiratory distress syndrome. N Engl J Med 1998; 338: Briel M, Meade M, Mercat A, et al. Higher vs lower positive end-expiratory pressure in patients with acute lung injury and acute respiratory distress syndrome: systematic review and meta-analysis. JAMA 2010; 303: Dellinger RP, Levy MM, Rhodes A, et al. Surviving sepsis campaign: international guidelines for management of severe sepsis and septic shock: Crit Care Med 2013; 41: ARDS Definition Task Force. Acute respiratory distress syndrome: the Berlin Definition. JAMA 2012; 307: Determann RM, Royakkers A, Wolthuis EK, et al. Ventilation with lower tidal volumes as compared with conventional tidal volumes for patients without acute lung injury: a preventive randomized controlled trial. Crit Care 2010; 14:R Serpa Neto A, Cardoso SO, Manetta JA, et al. Association between use of lungprotective ventilation with lower tidal volumes and clinical outcomes among patients without acute respiratory distress syndrome: a meta-analysis. JAMA 2012; 308: Serpa Neto A, Simonis FD, Barbas CS, et al. Lung-protective ventilation with low tidal volumes and the occurrence of pulmonary complications in patients without acute respiratory

31 334 distress syndrome: a systematic review and individual patient data analysis. Crit Care Med 2015; 43: Serpa Neto A, Simonis FD, Barbas CS, et al. Association between tidal volume size, duration of ventilation, and sedation needs in patients without acute respiratory distress syndrome: an individual patient data meta-analysis. Intensive Care Med 2014; 40: Serpa Neto A, Nagtzaam L, Schultz MJ. Ventilation with lower tidal volumes for critically ill patients without the acute respiratory distress syndrome: a systematic translational review and meta-analysis. Curr Opin Crit Care 2014; 20: Ferguson ND. Low tidal volumes for all? JAMA 2012; 308: Manzano F, Fernández-Mondéjar E, Colmenero M, et al. Positive-end expiratory pressure reduces incidence of ventilator-associated pneumonia in nonhypoxemic patients. Crit Care Med 2008; 36: Pepe PE, Hudson LD, Carrico CJ. Early application of positive end-expiratory pressure in patients at risk for the adult respiratory-distress syndrome. N Engl J Med 1984; 311: The PROVE Network Investigators for the Clinical Trial Network of the European Society of Anaesthesiology. High versus low positive end-expiratory pressure during general anaesthesia for open abdominal surgery (PROVHILO trial): a multicentre randomized controlled trial. Lancet 2014; 384: Retamal J, Bugedo G, Larsson A, Bruhn A. High PEEP levels are associated with overdistension and tidal recruitment/derecruitment in ARDS patients. Acta Anaesthesiol Scand 2015; 59: Dreyfuss D, Saumon G. Ventilator-induced lung injury: lessons from experimental studies. Am J Respir Crit Care Med 1998; 157: Gajic O, Dabbagh O, Park PK, et al. Early identification of patients at risk of acute lung injury: evaluation of lung injury prediction score in a multicenter cohort study. Am J Respir Crit Care Med 2011; 183: