Driving pressure and acute respiratory distress syndrome in critically ill patients

ORIGINAL ARTICLE

Driving pressure and acute respiratory distress syndrome

in critically ill patients

R

AIKO

BLONDONNET,

1,2

E

LODIE

JOUBERT,

1

T

HOMAS

GODET,

1

P

AULINE

BERTHELIN,

1

T

HIBAUT

PRANAL,

1

L

AURENCE

ROSZYK,

2,3

R

USSELL

CHABANNE,

1

N

ATHANAEL

EISENMANN,

4

A

LEXANDRE

LAUTRETTE,

5

C

ORINNE

BELVILLE,

2

S

OPHIE

CAYOT,

1

T

HIERRY

GILLART,

1

B

ERTRAND

SOUWEINE,

5

D

AMIEN

BOUVIER,

2,3

L

OIC

BLANCHON,

3

V

INCENT

SAPIN,

2,3

B

RUNO

PEREIRA,

6

J

EAN

-M

ICHEL

CONSTANTIN

1,2

AND M

ATTHIEU

JABAUDON

1,2

1

Department of Perioperative Medicine, CHU Clermont-Ferrand, Clermont-Ferrand, France;2GReD, CNRS UMR 6293, INSERM U1103, Université Clermont Auvergne, Clermont-Ferrand, France;3Department of Medical Biochemistry and Molecular Biology, CHU Clermont-Ferrand, Clermont-Ferrand, France;4Intensive Care Unit, Jean Perrin Comprehensive Cancer Center,

Clermont-Ferrand, France;5Medical Intensive Care Unit, CHU Clermont-Ferrand, Clermont-Ferrand, France;6Department of Clinical Research and Innovation (DRCI), CHU Clermont-Ferrand, Clermont-Ferrand, France

ABSTRACT

Background and objective: Elevated driving pressure

(

ΔP) may be associated with increased risk of acute

respiratory

distress

syndrome

(ARDS)

in

patients

admitted via the emergency department and with

post-operative pulmonary complications in surgical patients.

This study investigated the association of higher

ΔP

with the onset of ARDS in a high-risk, intensive care

unit (ICU) population.

Methods: This is a secondary analysis of a prospective

multicentre observational study. Data for this ancillary

study were obtained from intubated adult patients with

at least one ARDS risk factor upon ICU admission

enrolled in a previous multicentre observational study.

Patients were followed up for the development of ARDS

within 7 days (primary outcome). Univariate and

multi-variate analyses tested the association between

ΔP

(measured at ICU admission (baseline) or 24 h later

(day 1)) and the development of ARDS.

Results: A total of 221 patients were included in this

study, among whom 34 (15%) developed ARDS within

7 days. These patients had higher baseline

ΔP than

those

who

did

not

(mean

 SD: 12.5  3.1 vs

9.8

 3.4 cm H

2

O, respectively,

P = 0.0001). The

associ-ation between baseline

ΔP and the risk of developing

ARDS was robust to adjustment for baseline tidal

vol-ume, positive-end expiratory pressure, illness severity,

serum lactate and sepsis, pneumonia, severe trauma

and shock as primary ARDS risk factors (odds ratio:

1.20; 95% CI: 1.03

–1.41; P = 0.02). The same results

were found with day 1

ΔP.

Conclusion: Among at-risk ICU patients, higher

ΔP

may

identify

those

who

are

more

likely

to

develop ARDS.

Clinical trial registration:

NCT02070536 at ClinicalTrials.gov

Key words:

acute respiratory distress syndrome, driving pres-sure, intensive care unit, mechanical ventilation, risk prediction.

Abbreviations:

ΔP, driving pressure; ARDS, acute respiratory distress syndrome; AUROC, area under an ROC curve; CRS, respiratory-system compliance; FiO2, fraction of inspired oxygen; ICU, intensive care unit; IQR, interquartile range; PaO2, partial pressure of arterial oxygen; PBW, predicted body weight; PEEP, positive-end expiratory pressure; Pplat, inspiratory plateau pressure; RAGE, receptor for advanced glycation end-products; ROC, receiver-operating characteristic; RR, respiratory rate; SAPS, Simpli_{fied Acute Physiology Score; VILI,} ventilator-induced lung injury; VT, tidal volume.

INTRODUCTION

Acute respiratory distress syndrome (ARDS) is an

under-recognized and undertreated syndrome

1,2

_with

high mortality and limited effective therapy.

3–6

One

major challenge in targeting prevention and early

treat-ment of ARDS is the inability to accurately predict

which patients will develop the syndrome.

7

Tidal volume (V

T

), as adjusted to predicted body

weight (PBW), often serves as a surrogate for the risk of

developing ventilator-induced lung injury (VILI) in

Correspondence: Matthieu Jabaudon, Department of Perioperative Medicine, CHU Clermont-Ferrand, Université Clermont Auvergne, CNRS UMR 6293, INSERM U1103, GReD, 1 Place Lucie Aubrac, 63003 Clermont-Ferrand Cedex 1, France. Email: mjabaudon@chu-clermontferrand.fr

Received 26 March 2018; invited to revise 26 June 2018; revised 4 July 2018; accepted 9 August 2018 (Associate Editor: Yuanlin Song; Senior Editor: Lutz Beckert).

S U M M A R Y A T A G L A N C E

The driving pressure at admission to the intensive

care unit in intubated patients under controlled

ventilation identi

fies patients with clinical risk

factor(s), who develop acute respiratory distress

syndrome within 7 days.

mechanically ventilated patients.

3,8

_{However, because}

ARDS is characterized by a decrease in

respiratory-system compliance (C

RS

), that is a decrease in the

func-tional size of the lung, Amato et al. recently reported

from retrospective secondary analyses of clinical ARDS

trials that scaling V

T

to C

RS

was a better predictor of

outcomes than V

T

, positive end-expiratory pressure

(PEEP) and other risk covariates.

9

_{The driving pressure}

(

ΔP = V

T

/C

RS

) can be easily estimated at the patient

bedside as the difference between inspiratory plateau

pressure (Pplat) and PEEP.

Recent studies suggest a role of

ΔP in the risk of

developing VILI in ARDS patients under extracorporeal

membrane oxygenation

10

_{or with brain injury.}

11

_In

addition,

ΔP and Pplat were associated with mortality

and development of ARDS in patients who presented

to the emergency department requiring mechanical

ventilation.

12

_{A recent meta-analysis of patient data}

from randomized controlled trials of lung-protective

ventilation during general anaesthesia for surgery

dem-onstrated that high intraoperative

ΔP was associated

with more post-operative pulmonary complications.

13

Because most patients with ARDS are identi

fied

within 7 days of recognition of the underlying risk

factor(s),

1,14

_{we postulated that}

ΔP could be a risk

pre-dictor for ARDS development in patients without ARDS

but with at least one ARDS risk factor upon admission

to the intensive care unit (ICU). To assess this

associa-tion, we did a secondary analysis of individual patient

data that were prospectively obtained during a large

multicentre observational study.

15

METHODS

Participants

This is a secondary analysis of a prospective

multicen-tre observational study. Data used in this ancillary

study were prospectively obtained from patients

previ-ously enrolled in a large multicentre observational

study of the predictive values of RAGE (the receptor for

advanced glycation end-products) isoforms and gene

variants for the onset of ARDS (PrediRAGE study).

15

Between June 2014 and January 2015, 500 critically ill

adult patients in whom at least one ARDS risk factor

was identi

fied

1

_{were enrolled. Patients were excluded if}

they were admitted for an isolated neurological or

neu-rosurgical diagnosis without any signi

ficant medical

co-morbidities.

16

_{After ventilatory settings were unchanged}

for at least 30 min,

ΔP was measured as the difference

between static Pplat (measured at end inspiration

dur-ing an inspiratory hold manoeuvre of 3 s) and the level

of PEEP. Patients who were not intubated, who

received pressure support ventilation, a PEEP <5 cm

H

2

O or had respiratory rates (RR) that were higher than

the ventilator settings (suggesting the presence of

venti-latory efforts)

9

_{and those for whom}

_{ΔP was not}

accessi-ble were excluded from this secondary analysis. The

first assessment of ΔP defined the ‘baseline’ timepoint;

by study design, it was performed as soon as possible

after ICU admission and study enrolment. When

avail-able,

ΔP was measured both at baseline and 24 h later

(day 1).

Our

Institutional

Review

Board

approved

the

research protocols for both the primary and ancillary

studies (Comité de Protection des Personnes Sud Est VI,

approval number AU1073). All participants, or their

next of kin, provided written consent to participate.

There was no deviation from the approved protocols.

Primary outcome and additional variables

The primary outcome was the difference in baseline

ΔP between patients who developed ARDS within

7 days after study enrolment and those who did not.

The primary outcome follow-up time was chosen a

priori because ARDS criteria must be met by de

finition

within 1 week of a known clinical insult or new or

worsening respiratory symptoms.

1

_{ARDS was de}

_fined

by physicians caring for the patients, based on criteria

from the Berlin de

finition.

1

_{Chest radiographs and}

arte-rial blood gases were routinely performed at least daily,

as per local protocols, and more frequently if prompted

by clinical symptoms (e.g. new or worsening

respira-tory symptoms). Physicians caring for the patients

assessed the presence of ARDS criteria at least daily,

and this procedure was conducted uniformly for all

enrolled patients among participating ICU. Patients

who met criteria for ARDS at initial assessment, or

within the subsequent 24 h, were excluded from

analy-sis to ensure removal of ARDS that was present at

baseline.

Secondary outcomes included differences in

ΔP on

day 1 between patients who developed ARDS within

7 days and those who did not, and differences in

base-line or day 1

ΔP between patients who developed

ARDS within 30 days and those who did not. There

were no missing data regarding the primary outcome

for patients included in this secondary analysis.

Statistical analysis

Categorical data are expressed as numbers and

per-centages, and quantitative data as mean

 SD or

median and interquartile range (IQR). To compare

baseline characteristics between groups (patients who

developed ARDS and those who did not), Student

’s

t-test or Mann

–Whitney test were considered for

quan-titative parameters according to t-test assumptions

(normality assessed using Shapiro

–Wilk test and

homo-scedasticity by Fisher

–Snedecor). Categorical data were

compared among groups using chi-square or Fisher

’s

exact test.

17

Multivariate generalized linear mixed models

(logis-tic for binary endpoint) were performed to study

ΔP as

a risk predictor of ARDS development, considering

adjustment for potential confounding factors

deter-mined according to univariate analysis (

α ≤0.15) and

clinical relevance

18

_{: V}

T

, PEEP, RR, the ratio of partial

pressure of arterial oxygen to fraction of inspired

oxy-gen (PaO

2

/FiO

2

), baseline severity of illness (Simpli

fied

Acute Physiology Score (SAPS) II),

19,20

_{serum lactate,}

sepsis, pneumonia, severe trauma and shock as

pri-mary ARDS risk factors. Results were expressed as odds

ratios (OR) and 95% CI. The multicollinearity between

factors was studied, particularly with

ΔP. Univariate

correlations

between

quantitative

outcomes

were

assessed using Pearson and Spearman correlation

coef-ficients according to statistical distribution.

Discrimination of

ΔP was tested by calculating the

area under a receiver-operating characteristic curve

(AUROC). We calculated sensitivity and speci

ficity at

the cut-off point that minimized the distance to perfect

sensitivity and speci

ficity (coordinates (0,1) on the

graph), and multiple cut-off values of

ΔP were tested

for speci

ficity and sensitivity. Several indexes reported

in the literature were explored: Youden and Liu.

A two-sided P-value <0.05 was considered

statisti-cally signi

ficant. Statistical analysis was performed with

Stata

software

(v14,

StataCorp,

College

Station,

TX, USA).

RESULTS

Baseline characteristics

The

flow chart of the ancillary study is presented in

Figure 1. The baseline characteristics of the study

sam-ple are described in Table 1. Of the 221 patients

included, 34 (15%) developed ARDS by day 7 (median

time (IQR) to ARDS onset: 3 (2

–5) days). There was no

signi

ficant difference in baseline characteristics,

coex-isting conditions, primary admission diagnosis, ARDS

risk factors, SAPS II or the need for vasopressor support

between patients who went on to develop ARDS and

those who did not. Patients who developed ARDS had

longer durations of mechanical ventilation and of ICU

Figure 1 Flow diagram of the ancillary study. Personnel shortage was another identified reason for non-enrolment of eligible subjects and some subjects were missed without a given explanation. ARDS, acute respiratory distress syndrome; ICU, intensive care unit.

stay than those who did not. Other parameters and

concurrent treatments are reported in Table 2. Patients

who developed ARDS had higher RR (P = 0.01) and

serum lactate (P = 0.0004) than those who did not. The

main respiratory parameters, indices of lung injury

severity and clinical outcomes of patients who

devel-oped ARDS at day 7 are presented in Table 3.

Predictive value of baseline driving pressure

for subsequent ARDS

Baseline

ΔP was significantly higher among patients

who developed ARDS at day 7 compared with those

who did not (12.5

 3.1 vs 9.8  3.4 cm H

2

O,

respec-tively, P = 0.0001) (Fig. 2, Table 2).

Unadjusted

analyses

examined

the

relationship

between baseline features and development of ARDS at

day 7 in our cohort; in these analyses, baseline features

such as higher baseline

ΔP (P < 0.001), V

T

(P < 0.001),

PEEP (P < 0.001), Pplat (P < 0.001), RR (P < 0.001) and

PaO

2

/FiO

2

(P < 0.001) were signi

ficantly associated with

higher risk of developing subsequent ARDS (Table 4).

When measured on day 1 (n = 87),

ΔP was also

higher among patients who developed ARDS at day

7 compared with those who did not (12.6

 4.1 vs

9.5

 3.8 cm H

2

O, respectively, P = 0.005).

Multivariate adjustment of predictor models

Variables that were signi

ficant in univariate analyses

(but not already included in the SAPS II, e.g. age), as

Table 1 Main baseline characteristics and clinical outcomes

Whole cohort (n = 221) No ARDS (n = 187) Develop ARDS (n = 34) P-value Age (years) 67 13 66 15 70 16 0.2 Male sex 148 (67) 123 (66) 25 (74) 0.4 BMI (kg/m2_{) 26.7}  5.3 26.4_{ 5.1} 28.3_{ 5.8} 0.06

Primary admission diagnosis

Cardiac 9 (4) 7 (4) 2 (6) 0.6 Respiratory 140 (63) 118 (63) 22 (66) 0.6 Gastrointestinal 43 (20) 34 (18) 9 (26) 0.3 Infectious 66 (30) 54 (29) 12 (35) 0.5 Neurological 17 (8) 13 (7) 4 (12) 0.5 Major surgery 67 (30) 54 (29) 13 (38) 0.3 Other 8 (4) 6 (3) 2 (6) 0.6

Coexisting chronic conditions

Atherosclerosis 55 (25) 45 (24) 10 (29) 0.5 Diabetes 32 (15) 28 (15) 4 (12) 0.8 Hypertension 94 (43) 77 (41) 15 (44) 0.9 Dyslipidaemia 51 (23) 41 (22) 10 (29) 0.4 Current smoking 62 (28) 50 (27) 12 (35) 0.4 Asthma 8 (4) 6 (3) 2 (6) 0.6 COPD 24 (11) 17 (9) 7 (21) 0.06

Chronic renal failure requiring dialysis 6 (3) 6 (3) 0 (0) 0.6

Liver cirrhosis 7 (3) 6 (3) 1 (3) 1

Cancer 27 (12) 22 (12) 5 (15) 0.8

Primary ARDS risk factor at admission

Shock 78 (35) 69 (37) 9 (25) 1 Sepsis 25 (11) 19 (10) 6 (18) 0.2 Pneumonia 12 (5) 9 (5) 3 (9) 0.4 Aspiration 7 (3) 6 (3) 1 (3) 1 Severe trauma 35 (16) 30 (16) 5 (15) 1 Pancreatitis 5 (2) 4 (2) 1 (3) 1 Drug overdose 12 (5) 11 (6) 1 (3) 0.7 High-risk surgery 99 (45) 82 (44) 17 (50) 0.6

Lung Injury Prediction Score 5.1 2.7 4.9 2.6 5.9 3.2 0.2

Simplified Acute Physiology Score II 52 17 52 17 52 18 0.9

Vasopressor use at admission 79 (36) 65 (35) 14 (41) 0.6

Duration of mechanical ventilation (days) 4 (1_–13) 3 (1_–11) 13 (5_–19) 0.0001

Duration of ICU stay (days) 9 (5–21) 8 (4–20) 20 (7–28) 0.001

30-Day mortality 45 (20) 36 (19) 9 (26) 0.3

Data are presented as mean SD, median (interquartile range) or n (%). P-values were calculated for comparisons between patients who developed ARDS at day 7 and those who did not. Analysis was performed using Wilcoxon rank-sum, chi-square test or Fisher exact test as appropriate. Percentages may not exactly total 100% because of rounding. The BMI is the weight in kilograms divided by the square of the height in metres.

well as other non-signi

ficant but clinically relevant

vari-ables, were used to compute OR for ARDS

develop-ment at day 7 using multivariate models taking into

account centre effect as random variable in mixed

model. Pplat was not included in multivariate analysis

because of its multicollinearity with PEEP (Spearman

’s

Table 2 Respiratory, clinical and biological parameters at baseline

Whole cohort (n = 221) No ARDS (n = 187) Develop ARDS (n = 34) P-value Respiratory parameters VT(mL/kg of PBW) 7.6 1.3 7.7 1.2 7.4 1.4 0.4 PEEP (cm H2O) 6.8 2.1 6.6 2.0 8.0 2.4 0.2 Pplat (cm H2O) 17.1 3.9 16.4 3.6 20.5 3.5 0.9 ΔP (cm H2O) 10.2 3.5 9.8 3.4 12.5 3.1 0.0001

Respiratory rate (per min) 18 5 18 4 20 6 0.01

PaO2/FiO2(mm Hg) 269 88 280 84 204 81 0.9

Number of abnormal quadrants on chest radiograph 1.5 0.8 1.4 0.8 1.7 0.9 0.4

FiO2(%) 55 20 54 19 63 20 0.7

PaO2(mm Hg) 147 80 152 81 114 63 0.1

PaCO2(mm Hg) 39 7 39 7 42 8 0.3

Clinical parameters

Mean arterial pressure (mm Hg) 81 12 81 12 79 11 0.5

Body temperature (C) 36.5 1.2 36.4 1.25 36.9 1.2 0.8

Biological parameters

Serum creatinine (μmol/L) 100 67 98 68 110 60 0.4

Serum lactate (mmol/L) 2.8 3.1 2.7 2.8 3.5 4.6 0.0004

Arterial pH 7.34 0.09 7.34 0.09 7.31 0.10 0.4

Serum bicarbonate (mmol/L) 21_{ 5} 21_{ 4} 21_{ 5} 0.1

Leukocytes (G/L) 13.2 8.1 13.2 8.2 13.2 7.9 0.9

Platelets (G/L) 192 105 196 103 166 116 0.4

Concurrent treatment

Corticosteroids 10 (5) 8 (4) 2 (6) 0.7

Neuromuscular blocking agents 3 (1) 3 (2) 0 (0) 0.5

Data are presented as mean_{ SD or n (%). P-values were calculated for comparisons between patients who developed ARDS at} day 7 and those who did not. Analysis was performed using Wilcoxon rank-sum, chi-square test or Fisher exact test as appropriate. Percentages may not exactly total 100% because of rounding.

ΔP, driving pressure; ARDS, acute respiratory distress syndrome; FiO2, fraction of inspired oxygen; PaCO2, partial pressure of arte-rial carbon dioxide; PaO2, partial pressure of arterial oxygen; PBW, predicted body weight; PEEP, positive end-expiratory pressure; Pplat, inspiratory plateau pressure; VT, tidal volume.

Table 3 Main clinical outcomes of patients who developed ARDS at day 7 (n = 34), with respiratory parameters and indices of lung injury severity at ARDS onset

Time from inclusion to ARDS onset (days) 3.3_{ 1.9}

VT(mL/kg of PBW) 6.8 11 PEEP (cm H2O) 8.4 2.4 Pplat (cm H2O) 24.3 3.9 ΔP (cm H2O) 15.8 2.9 PaO2/FiO2(mm Hg) 126 35 Cause of ARDS

Pulmonary cause (pneumonia, aspiration) 31 (97)

Extrapulmonary cause 1 (3)

Sepsis 16 (53)

Clinical outcomes at day 30

Survival 25 (76)

Duration of ICU stay (days) 20 (7_–28) Mechanical ventilation duration (days) 13 (5–19) Data are presented as mean_{ SD, median (interquartile} range) or n (%). Percentages may not exactly total 100% because of rounding.

ΔP, driving pressure; ARDS, acute respiratory distress syn-drome; FiO2, fraction of inspired oxygen; ICU, intensive care unit; PaO2, partial pressure of arterial oxygen; PBW, predicted body weight; PEEP, positive end-expiratory pressure; Pplat, inspiratory plateau pressure; VT, tidal volume.

Figure 2 Baseline driving pressure in patients who went on to develop acute respiratory distress syndrome (ARDS) at day 7 (n = 34) and those who did not (n = 187). Data are presented as mean SD (P = 0.0001).

rho = 0.44, P < 0.0001) and

ΔP (Spearman’s rho = 0.83,

P < 0.0001).

After multivariate adjustment, higher

ΔP at baseline

remained associated with an increased risk of

develop-ing ARDS at day 7 (OR for each one-point increment:

1.20; 95% CI: 1.03

–1.41, P = 0.02; n = 124) even after

adjustment (Table 4). Using the same model, higher

ΔP on day 1 was associated with an increased risk of

developing ARDS at day 7 (OR for each one-point

increment: 1.39; 95% CI: 1.05

–1.84, P = 0.02, n = 71).

Discrimination of clinical and biomarker

predictive models

An ROC curve of baseline

ΔP in differentiating between

the presence and absence of ARDS at day 7 was

con-structed (Fig. 3). The AUROC was 0.72 (95% CI:

0.60

–0.83, P = 0.0007) for a cut-off value of 11.5 cm

H

2

O, with a sensitivity of 72%, a speci

ficity of 73%, a

positive predictive value of 37% (95% CI: 23

–52%) and

a negative predictive value of 92% (95% CI: 85

–97%). A

cut-off value of baseline

ΔP >16.5 cm H

2

O was

predic-tive of subsequent ARDS development with a sensitivity

of 24% and a speci

ficity of 90%, whereas a cut-off value

of baseline

ΔP <7.5 cm H

2

O was predictive of not

developing ARDS with a sensitivity of 93% and a

speci-ficity of 23%. When ΔP was measured on day 1, the

AUROC was 0.73 (95% CI: 0.57

–0.90, P = 0.005) for a

cut-off value of 11.5 cm H

2

O, with a sensitivity of 60%

and a speci

ficity of 79%.

Secondary outcomes

No signi

ficant correlation was observed between

base-line

ΔP and time from inclusion to ARDS onset, PaO

2

/

FiO

2

upon ARDS onset, lowest PaO

2

/FiO

2

during ARDS

and respiratory parameters upon ARDS onset such as

V

T

, PEEP or Pplat in patients who developed ARDS at

day

7

(Spearman

’s rho = 0.13 (P = 0.5), −0.14

(P = 0.5),

−0.26 (P = 0.2), 0.12 (P = 0.7), −0.17

(P = 0.4) and

−0.38 (P = 0.3), respectively). There was

no difference in baseline

ΔP with regards to 30-day

mortality (P = 0.7), the duration of mechanical

ventila-tion or the length of stay in the ICU (P = 0.9). Thirteen

patients (6%) from our cohort had later onset of ARDS

and developed the syndrome between day 7 and day

Table 4 OR for developing ARDS at day 7 after univariate and multivariate analyses in critically ill patients at risk of developing ARDS (n = 124)

Variable

Univariate analysis Multivariate analysis

OR 95% CI P OR 95% CI P

BaselineΔP† 1.23 1.10–1.37 <0.001 1.20 1.03–1.41 0.02

Baseline PaO2/FiO2† 0.99 0.98–0.99 <0.001 0.99 0.98–0.99 0.007 Baseline Simpli_{fied Acute Physiology Score II}† 1.00 0.98_–1.03 0.8 0.97 0.95_–1.01 0.2

Baseline VT† 0.77 0.53–1.12 0.2 0.81 0.52–1.25 0.3

Baseline PEEP† 1.47 1.19–1.83 <0.001 1.21 0.92–1.58 0.2

Baseline RR† 1.13 1.03–1.24 <0.001 0.98 0.85–1.12 0.8

Primary ARDS risk factor

Sepsis 2.23 0.75–6.63 0.1 0.72 0.13–4.13 0.7 Pneumonia 2.11 0.51–8.67 0.3 2.51 0.32–19.79 0.4 Shock 0.95 0.44–2.05 0.9 1.80 0.51–6.33 0.4 Severe trauma 0.90 0.31_–2.55 0.8 2.77 0.57_–13.53 0.2 Age† 1.01 0.99–1.04 0.4 — — — Baseline Pplat† 1.32 1.18–1.48 <0.001 — — — Serum lactate† 1.07 0.96–1.19 0.2 — — —

Variables that were significant in univariate analyses (but not already included in Simplified Acute Physiology Score II, e.g. age), as well as other non-significant but clinically relevant variables, were used to compute OR for ARDS development at day 7 using multivar-iate logistic regression models taking into account centre effect. Pplat was not included in multivarmultivar-iate analysis because of its multicol-linearity with PEEP and_ΔP.

†_{OR and 95% CI are for each one-point increment.}

ΔP, driving pressure; ARDS, acute respiratory distress syndrome; FiO2, fraction of inspired oxygen; PaO2, partial pressure of arterial oxygen; PEEP, positive end-expiratory pressure; Pplat, inspiratory plateau pressure; RR, respiratory rate; VT, tidal volume.

Figure 3 Receiver-operating characteristic (ROC) curve of base-line driving pressure (_{ΔP) in differentiating between the} develop-ment and the absence of developdevelop-ment of acute respiratory distress syndrome (ARDS) at day 7. The area under the ROC curve was 0.72 (95% CI: 0.60_{–0.83, P = 0.0007) for a cut-off value} of 11.5 cm H2O, with a sensitivity of 72% and a specificity of 73%.

30 after inclusion. Baseline

ΔP was similar between

these patients and those who did not develop ARDS

after day 7 (10.7

 3.7 vs 10.1  3.5 cm H

2

O,

respec-tively, P = 0.6).

Similar results were found with

ΔP as measured on

day 1.

DISCUSSION

In patients admitted to the ICU with identi

fied risk

fac-tors of ARDS, both baseline and day 1

ΔP were

signifi-cantly higher among patients who developed ARDS

within 7 days compared with those who did not.

Higher

ΔP was associated with increased risk of

devel-oping ARDS even after adjustment for severity of

ill-ness, respiratory parameters and ARDS risk factors.

Mortality of ARDS patients is high,

2

_{and current}

research initiatives include primary prevention.

21–23

A key

challenge is to accurately identify patients in whom ARDS

is likely to develop and who would bene

fit most

preven-tive measures.

23

_{Clinical prediction scores can identify}

patients with a known clinical risk factor, or new or

wors-ening respiratory symptoms, who are more likely to

develop ARDS.

21

_{However, there is still an urgent need to}

widely implement measures such as lung-protective

mechanical ventilation, aggressive resuscitation,

reduc-tion of transfusion and prevenreduc-tion of common

complica-tions.

24

_{The identi}

_{fication of clinical or biological}

variables may, therefore, be important to assess

preven-tive strategies, early disease detection and treatment for

ARDS in patients who are most likely to bene

fit.

2,25

This study is the

first to report that higher ΔP is

associated with an increased risk of developing ARDS

in ICU patients with ARDS risk factor(s),

1

_{which is in}

line with previous reports on the importance of

ΔP in

the development, severity and outcome of lung

injury.

9,13,26–28

Because it is de

fined as the amount of

cyclic parenchymal deformation imposed on ventilated

and preserved lung units,

9

_{ΔP may serve as a reliable}

surrogate marker of cyclic lung strain that is most

accessible at the bedside.

29

_{Lung-protective ventilation}

strategies maintain alveolar aeration, prevent

overex-pansion of the lung and limit

ΔP, and thereby are

thought to reduce VILI.

8

_Because

_{ΔP is the tidal}

increase in static transrespiratory pressure, it is

propor-tional to V

T

, with respiratory-system elastance (the

inverse of compliance) being the constant of

propor-tionality and a re

flect of the severity and extent of lung

injury.

The association between higher

ΔP and the risk of

ARDS development may be attributable to the fact that

the variables that de

fine ΔP (V

T

and C

RS

) are

them-selves highly associated with lung injury.

30–33

However,

our

findings should be considered

hypothesis-generating rather than de

finitive because the ‘baby

lung

’ concept (in which some portion of the lung in

patients with ARDS is collapsed or

flooded and does

not participate in gas exchange) may not be applicable

to most critically ill patients without ARDS.

34

_Limiting

ΔP may therefore be used to scale the delivered breath

to the size of the lung that is available to participate in

gas exchange, which may be more biologically relevant

than scaling to PBW.

8,9,13,26

Our study has some limitations. First, this study lacks

a validation cohort and no sample size was speci

fically

calculated for this analysis. However, although future

prospective studies remain necessary to validate our

findings, the effect size of baseline ΔP to predict ARDS

development was 0.8 in this study and estimated

statis-tical power was 0.99 with a two-sided type I error of

0.05. In addition, the question of whether manipulating

ΔP in critically ill patients at risk for ARDS could

pre-vent ARDS development remains unanswered. Second,

the selection of potential confounders was limited to

clinical data collected by the original study. Third, our

analysis does not account for baseline chest wall

ela-stance, although the cyclic gradient of pressures across

the lung (that may generate parenchymal injury during

ventilation) might be lower in patients with increased

chest wall elastance.

33

_{Fourth, our conclusions on}

_ΔP

are only valid for ventilation in which patients are not

making respiratory efforts, as it is dif

ficult to interpret

ΔP in actively breathing patients. Finally, our findings

may be true in patients with high risk of ARDS only, as

assessed by high lung injury prediction scores in our

cohort, but not in all critically ill patients

16

_{; therefore,}

better identifying clinical risk factors remains of crucial

importance in predicting risk for ARDS

21

_{and elevated}

ΔP might only provide some additional value in this

speci

fic population.

This study also has several strengths. First, this is the

first report, in a large multicentre cohort of 221 selected,

at-risk patients with clinical ARDS risk factor(s),

1

_{of a}

role for

ΔP in the risk of developing ARDS within

7 days. Importantly, the same signi

ficant associations

with ARDS development were found when

ΔP was

measured at ICU admission or 24 h later. Second, our

cohort was primarily designed for early collection of

data in patients admitted to the ICU, allowing us to

include many at-risk patients before ARDS

develop-ment. This is a rather dif

ficult population to study,

although it is also the population for whom prediction

could be the most important and likely to provide the

most bene

fit.

23

_{Third, there were only few exclusion}

criteria in this study, allowing for the recruitment of a

broad range of critically ill patients with various ARDS

risk factors, especially extrapulmonary risk factors,

thus possibly reinforcing the generalizability of our

findings.

In conclusion, this study is the

first to suggest that

an elevated

ΔP is associated with an increased risk of

ARDS development in patients at risk for the syndrome,

independent from V

T

, PEEP and other baseline risk

covariates. However, whether manipulation of

ΔP

might contribute to prevent ARDS in at-risk patients

remains unknown.

Data availability statement

The data sets analysed during the current study are available from the corresponding author on reasonable request.

Acknowledgements

The authors wish to thank the nurses and staff from participating ICU, and the technicians and staff from the Department of Medi-cal Biochemistry and Molecular Biology, CHU Clermont-Ferrand,

and from the Université Clermont Auvergne, Clermont-Ferrand, France. This work was supported by grants from the Auvergne Regional Council (‘Programme Nouveau Chercheur’ 2013), the French Agence Nationale de la Recherche and Direction Générale de l_{’Offre de Soins (‘Programme de Recherche} Translationnelle en Santé’ ANR-13-PRTS-0010) and from CHU Clermont-Ferrand (‘Appel d’Offre Interne 2013’). The funders had no in_{fluence in the study design, conduct and analysis or in the} preparation of this article.

REFERENCES

1 ARDS Definition Task Force, Ranieri VM, Rubenfeld GD, Thompson BT, Ferguson ND, Caldwell E, Fan E, Camporota L, Slutsky AS. Acute respiratory distress syndrome: the Berlin Defini-tion. JAMA 2012; 307: 2526–33.

2 Bellani G, Laffey JG, Pham T, Fan E, Brochard L, Esteban A, Gattinoni L, van Haren F, Larsson A, McAuley DF et al.; LUNG SAFE Investigators; ESICM Trials Group. Epidemiology, patterns of care, and mortality for patients with acute respiratory distress syn-drome in intensive care units in 50 countries. JAMA 2016; 315: 788–800.

3 Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory dis-tress syndrome. The Acute Respiratory Disdis-tress Syndrome Net-work. N. Engl. J. Med. 2000; 342: 1301–8.

4 National Heart, Lung, and Blood Institute Acute Respiratory Dis-tress Syndrome (ARDS) Clinical Trials Network, Wiedemann HP, Wheeler AP, Bernard GR, Thompson BT, Hayden D, deBoisblanc B, Connors AF Jr, Hite RD, Harabin AL. Comparison of twofluid-management strategies in acute lung injury. N. Engl. J. Med. 2006; 354: 2564–75.

5 Papazian L, Forel J-M, Gacouin A, Penot-Ragon C, Perrin G, Loundou A, Jaber S, Arnal J-M, Perez D, Seghboyan J-M et al.; ACURASYS Study Investigators. Neuromuscular blockers in early acute respiratory distress syndrome. N. Engl. J. Med. 2010; 363: 1107–16.

6 Guérin C, Reignier J, Richard J-C, Beuret P, Gacouin A, Boulain T, Mercier E, Badet M, Mercat A, Baudin O et al.; PROSEVA Study Group. Prone positioning in severe acute respiratory distress syn-drome. N. Engl. J. Med. 2013; 368: 2159–68.

7 Matthay MA, Zimmerman GA, Esmon C, Bhattacharya J, Coller B, Doerschuk CM, Floros J, Gimbrone MA Jr, Hoffman E, Hubmayr RD et al. Future research directions in acute lung injury: summary of a National Heart, Lung, and Blood Institute Working Group. Am. J. Respir. Crit. Care Med. 2003; 167: 1027–35. 8 Slutsky AS, Ranieri VM. Ventilator-induced lung injury. N. Engl.

J. Med. 2013; 369: 2126–36.

9 Amato MBP, Meade MO, Slutsky AS, Brochard L, Costa ELV, Schoenfeld DA, Stewart TE, Briel M, Talmor D, Mercat A et al. Driving pressure and survival in the acute respiratory distress syn-drome. N. Engl. J. Med. 2015; 372: 747–55.

10 Serpa Neto A, Schmidt M, Azevedo LCP, Bein T, Brochard L, Beutel G, Combes A, Costa ELV, Hodgson C, Lindskov C et al. Associations between ventilator settings during extracorporeal membrane oxygenation for refractory hypoxemia and outcome in patients with acute respiratory distress syndrome: a pooled indi-vidual patient data analysis: mechanical ventilation during ECMO. Intensive Care Med. 2016; 42: 1672–84.

11 Tejerina E, Pelosi P, Muriel A, Peñuelas O, Sutherasan Y, Frutos-Vivar F, Nin N, Davies AR, Rios F, Violi DA et al.; for VENTILA Group. Association between ventilatory settings and development of acute respiratory distress syndrome in mechanically ventilated patients due to brain injury. J. Crit. Care 2017; 38: 341–5. 12 Fuller BM, Page D, Stephens RJ, Roberts BW, Drewry AM,

Ablordeppey E, Mohr NM, Kollef MH. Pulmonary mechanics and mortality in mechanically ventilated patients without acute respira-tory distress syndrome: a cohort study. Shock 2018; 49: 311–6.

13 Neto AS, Hemmes SNT, Barbas CSV, Beiderlinden M, Fernandez-Bustamante A, Futier E, Gajic O, El-Tahan MR, Ghamdi AAA, Günay E et al. Association between driving pressure and develop-ment of postoperative pulmonary complications in patients under-going mechanical ventilation for general anaesthesia: a meta-analysis of individual patient data. Lancet Respir. Med. 2016; 4: 272–80.

14 Hudson LD, Milberg JA, Anardi D, Maunder RJ. Clinical risks for development of the acute respiratory distress syndrome. Am. J. Respir. Crit. Care Med. 1995; 151: 293–301.

15 Jabaudon M, Berthelin P, Pranal T, Roszyk L, Godet T, Faure J-S, Chabanne R, Eisenmann N, Lautrette A, Belville C et al. Receptor for advanced glycation end-products and ARDS prediction: a mul-ticentre observational study. Sci. Rep. 2018; 8: 2603.

16 Agrawal A, Matthay MA, Kangelaris KN, Stein J, Chu JC, Imp BM, Cortez A, Abbott J, Liu KD, Calfee CS. Plasma angiopoietin-2 pre-dicts the onset of acute lung injury in critically ill patients. Am. J. Respir. Crit. Care Med. 2013; 187: 736–42.

17 Zhang Z. Univariate description and bivariate statistical inference: thefirst step delving into data. Ann. Transl. Med. 2016; 4: 91. 18 Zhang Z. Model building strategy for logistic regression: purposeful

selection. Ann. Transl. Med. 2016; 4: 111.

19 Le Gall JR, Lemeshow S, Saulnier F. A new Simplified Acute Physi-ology Score (SAPS II) based on a European/North American multi-center study. JAMA 1993; 270: 2957–63.

20 Le Gall JR, Neumann A, Hemery F, Bleriot JP, Fulgencio JP, Garrigues B, Gouzes C, Lepage E, Moine P, Villers D. Mortality prediction using SAPS II: an update for French intensive care units. Crit. Care 2005; 9: R645–52.

21 Gajic O, Dabbagh O, Park PK, Adesanya A, Chang SY, Hou P, Anderson H 3rd, Hoth JJ, Mikkelsen ME, Gentile NT et al.; U.S. Critical Illness and Injury Trials Group: Lung Injury Prevention Study Investigators (USCIITG-LIPS). 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: 462–70.

22 Levitt JE, Calfee CS, Goldstein BA, Vojnik R, Matthay MA. Early acute lung injury: criteria for identifying lung injury prior to the need for positive pressure ventilation*. Crit. Care Med. 2013; 41: 1929–37.

23 Beitler JR, Schoenfeld DA, Thompson BT. Preventing ARDS: pro-gress, promise, and pitfalls. Chest 2014; 146: 1102–13.

24 Li G, Malinchoc M, Cartin-Ceba R, Venkata CV, Kor DJ, Peters SG, Hubmayr RD, Gajic O. Eight-year trend of acute respiratory dis-tress syndrome: a population-based study in Olmsted County, Minnesota. Am. J. Respir. Crit. Care Med. 2011; 183: 59–66. 25 Laffey JG, Talmor D. Predicting the development of acute

respira-tory distress syndrome: searching for the“troponin of ARDS”. Am. J. Respir. Crit. Care Med. 2013; 187: 671–2.

26 Samary CS, Santos RS, Santos CL, Felix NS, Bentes M, Barboza T, Capelozzi VL, Morales MM, Garcia CSNB, Souza SAL et al. Biologi-cal impact of transpulmonary driving pressure in experimental acute respiratory distress syndrome. Anesthesiology 2015; 123: 423–33.

27 Costa ELV, Slutsky AS, Amato MBP. Driving pressure as a key ven-tilation variable. N. Engl. J. Med. 2015; 372: 2072.

28 Loring SH, Malhotra A. Driving pressure and respiratory mechan-ics in ARDS. N. Engl. J. Med. 2015; 372: 776–7.

29 Protti A, Andreis DT, Monti M, Santini A, Sparacino CC, Langer T, Votta E, Gatti S, Lombardi L, Leopardi O et al. Lung stress and strain during mechanical ventilation: any difference between stat-ics and dynamstat-ics? Crit. Care Med. 2013; 41: 1046–55.

30 Serpa Neto A, Cardoso SO, Manetta JA, Pereira VGM, Espósito DC, Pasqualucci M de OP, Damasceno MCT, Schultz MJ. Association between use of lung-protective ventilation with lower tidal volumes and clinical outcomes among patients without acute respiratory distress syndrome: a meta-analysis. JAMA 2012; 308: 1651–9.

31 Neto AS, Simonis FD, Barbas CSV, Biehl M, Determann RM, Elmer J, Friedman G, Gajic O, Goldstein JN, Linko R et al.;

PROtective Ventilation Network Investigators. Lung-protective ventilation with low tidal volumes and the occur-rence of pulmonary complications in patients without acute respiratory distress syndrome: a systematic review and indi-vidual patient data analysis. Crit. Care Med. 2015; 43: 2155–63.

32 Webb HH, Tierney DF. Experimental pulmonary edema due to intermittent positive pressure ventilation with high inflation pres-sures. Protection by positive end-expiratory pressure. Am. Rev. Respir. Dis. 1974; 110: 556–65.

33 Dreyfuss D, Soler P, Basset G, Saumon G. High inflation pressure pulmonary edema. Respective effects of high airway pressure, high

tidal volume, and positive end-expiratory pressure. Am. Rev. Respir. Dis. 1988; 137: 1159–64.

34 Gattinoni L, Pesenti A. The concept of“baby lung”. Intensive Care Med. 2005; 31: 776–84.

Supplementary Information

Additional supplementary information can be accessed via the html version of this article at the publisher_{’s website.}

Visual Abstract

Higher

ΔP may identify those more

likely to develop ARDS among at-risk ICU patients.