Estimating the attributable fraction of mortality from acute respiratory distress syndrome to inform enrichment in future randomised clinical trials

Rohit Saha; Tài Pham; Pratik Sinha; Manoj V Maddali; Giacomo Bellani; Eddy Fan; Charlotte Summers; Abdel Douiri; Gordon Rubenfeld; Carolyn S Calfee; John G Laffey; Daniel F McAuley; Manu Shankar-Hari

doi:10.1136/thorax-2023-220262

. Author manuscript; available in PMC: 2024 Oct 1.

Published in final edited form as: Thorax. 2023 Jul 26;78(10):990–1003. doi: 10.1136/thorax-2023-220262

Estimating the attributable fraction of mortality from acute respiratory distress syndrome to inform enrichment in future randomised clinical trials

Rohit Saha ¹, Tài Pham ^2,^3,^4,⁵, Pratik Sinha ⁶, Manoj V Maddali ⁷, Giacomo Bellani ^8,^9,^10,¹¹, Eddy Fan ³, Charlotte Summers ¹², Abdel Douiri ¹³, Gordon Rubenfeld ¹⁴, Carolyn S Calfee ^15,¹⁶, John G Laffey ^17,¹⁸, Daniel F McAuley ^19,²⁰, Manu Shankar-Hari ²¹, on behalf of the LUNG SAFE Investigators and the European Society of Intensive Care Medicine Trials Group

^1.Critical Care, King’s College Hospital NHS Foundation Trust, London, United Kingdom.

^2.Keenan Research Centre, Li Ka Shing Knowledge Institute, St Michael’s Hospital, Unity Heath Toronto, Toronto, ON, Canada.

^3.Interdepartmental Division of Critical Care Medicine, University of Toronto, Toronto, ON, Canada.

^4.Université Paris-Saclay, AP-HP, Service de médecine intensive-réanimation, Hôpital de Bicêtre, DMU CORREVE, FHU SEPSIS, Groupe de recherche clinique CARMAS, Le Kremlin-Bicêtre, France.

^5.Université Paris-Saclay, UVSQ, Université Paris-Sud, Inserm, Equipe d’Epidémiologie respiratoire intégrative, CESP, Villejuif, France.

^6.Department of Anesthesiology, Washington University in St. Louis, St. Louis, Missouri.

^7.Division of Pulmonary, Allergy, and Critical Care Medicine, Department of Medicine, Stanford University, Stanford, CA, USA.

^8.School of Medicine and Surgery, University of Milan-Bicocca, Monza, Italy.

^9.Dept of Emergency and Intensive Care, San Gerardo Hospital, Monza, Italy.

^10.Dept of Medicine, University Health Network and Mount Sinai Hospital, Toronto, ON, Canada.

^11.Interdepartmental Division of Critical Care Medicine and Institute of Health Policy, Management and Evaluation, University of Toronto, Toronto, ON, Canada

^12.Wolfson Lung Injury Unit, Heart and Lung Research Institute, University of Cambridge, Cambridge, United Kingdom.

^13.School of Population Health & Environmental Sciences, King’s College London, London, United Kingdom.

^14.Interdepartmental Division of Critical Care Medicine, University of Toronto and Program in Trauma, Emergency and Critical Care, Sunnybrook Health Sciences Center, Toronto, ON, Canada.

^15.Cardiovascular Research Institute, University of California, San Francisco; Department of Medicine, Division of Pulmonary, Critical Care, Allergy and Sleep Medicine, University of California, San Francisco.

^16.Department of Anesthesia and Perioperative Care, University of California, San Francisco.

^17.Dept of Anesthesia and Intensive Care Medicine, Galway University Hospitals, Galway, Ireland.

^18.School of Medicine, and Regenerative Medicine Institute (REMEDI) at CÚRAM Centre for Research in Medical Devices, University of Galway, Galway, Ireland.

^19.The Wellcome-Wolfson Institute For Experimental Medicine, School of Medicine, Dentistry and Biomedical Sciences, Belfast, Ireland;

^20.Royal Victoria Hospital, Belfast Health and Social Care Trust, Belfast, Ireland.

^21.Centre for Inflammation Research, University of Edinburgh, Edinburgh, United Kingdom.

Contributorship

RS, DM, and MS-H were responsible for study concept and design. RS, TP, PS, MM, GB, JL and MS-H had full access to all the data in the study and take responsibility for the integrity and accuracy of the data analyses. RS, TP, PS, MM, GB, EF, CS, AD, GR, CC, JL, DM, MS-H contributed to data interpretation. RS and MS-H undertook the statistical analyses. RS and MS-H drafted the manuscript; RS, TP, PS, MM, GB, EF, CS, AD, GR, CC, JL, DM, MS-H contributed to critical revision of the manuscript for important intellectual content. MS-H was responsible for overall supervision. Funding for the LUNG-SAFE study was obtained by GB and JL.

RS - Rohit Saha; TP - Tài Pham, PS -Pratik Sinha; MM - Manoj V. Maddali; GB - Giacomo Bellani; EF - Eddy Fan; CS - Charlotte Summers, AD - Abdel Douiri; GR - Gordon Rubenfeld; CC - Carolyn S. Calfee; JL - John G Laffey; DM - Daniel F McAuley; MS-H - Manu Shankar-Hari

^✉

Corresponding author to whom requests for reprints and correspondence should be addressed: Manu Shankar-Hari, Chair of Translational Critical Care Medicine, The Queen’s Medical Research Institute, Edinburgh BioQuarter, Centre for Inflammation Research, 47 Little France Crescent, Edinburgh, EH16 4TJ, manu.shankar-hari@ed.ac.uk

PMCID: PMC10581447 NIHMSID: NIHMS1934563 PMID: 37495364

Abstract

Background:

Efficiency of randomised clinical trials (RCTs) of acute respiratory distress syndrome (ARDS) depends on the fraction of deaths attributable to ARDS (AF_ARDS) to which interventions are targeted. Estimates of AF_ARDS in subpopulations of ARDS could improve design of ARDS trials.

Methods:

We performed a matched case-control study using the LUNG-SAFE cohort. Primary outcome was intensive care unit mortality. We used nearest neighbour propensity score matching without replacement to match ARDS to non-ARDS populations. We derived two separate AF_ARDS estimates by matching patients with ARDS to patients with non-acute hypoxaemic respiratory failure (non-AHRF) and to patients with AHRF with unilateral infiltrates only (AHRF-UL). We also estimated AF_ARDS in subgroups based on severity of hypoxaemia, number of lung quadrants involved, and hyper- versus hypo-inflammatory phenotypes. Additionally, we derived AF_AHRF estimates by matching patients with AHRF to non-AHRF controls, and AF_AHRF-UL estimates by matching patients with AHRF-UL to non-AHRF controls.

Results:

Estimated AF_ARDS was 20.9%(95%CI 10.5–31.4%) when compared to AHRF -UL controls and 38.0%(95%CI 34.4%−41.6%) compared to non-AHRF controls. Within subgroups, estimates for AF_ARDS compared to AHRF-UL controls were highest in patients with severe hypoxaemia (41.1%(95%CI 25.2–57.1%)), in those with four quadrant involvement on chest radiography (28.9%(95%CI 13.4–44.3%)), and in the hyperinflammatory sub-phenotype (26.8%(95%CI 6.9–46.7%)). Estimated AF_AHRF was 33.8%(95%CI 30.5%−37.1%) compared to non-AHRF controls. Estimated AF_AHRF-UL was 21.3%(12.8–29.7%) compared to non-AHRF controls

Conclusions:

Overall AF_ARDS mean values were between 20.9%−38.0%, with higher AF_ARDS seen with severe hypoxaemia, four quadrant involvement on chest radiography, and hyperinflammatory ARDS.

Keywords: Acute Respiratory distress syndrome, Attributable fraction, randomised clinical trials

Introduction

Acute respiratory distress syndrome (ARDS) refers to acute hypoxaemic respiratory failure (AHRF) occurring within one week of a known clinical insult, with bilateral opacities on chest radiography that are not fully explained by effusions, lobar/lung collapse, or nodules[1]. Treatments with biological plausibility[2] and strong supporting pre-clinical evidence, when tested within randomised clinical trials (RCTs)[3], often report statistically indeterminate results (i.e., uncertainty highlighted by non-significant results of two-tailed tests, rather than proof of no difference between treatments (negative)[4]). Addressing this issue remains a major clinical and methodological challenge[2].

There are several explanations for statistically indeterminate RCT results, aside from testing of ineffective treatments. First, as ARDS is a heterogeneous syndrome, RCTs may consist of participants who either benefit, have no effect, or are harmed by the tested intervention. This explanation is supported by observations that distinct sub-phenotypes of ARDS respond differentially to treatments[5–11]. Second, there are several design issues within ARDS RCTs. This explanation is supported by observations that sample size calculations overestimate control arm event rates, and the expected average treatment effect[12].

In this manuscript, we explore another explanation – variation in the excess fraction of mortality attributable to ARDS (AF_ARDS)[13, 14] in RCT participants. Patients with ARDS may die from ARDS (i.e., AF_ARDS) or with ARDS (i.e., death may be due to risk factors like comorbidities and/or other organ dysfunction during critical illness). If we explicitly link the eligibility criteria of ARDS RCTs to AF_ARDS estimates (I.e., the excess proportion of deaths from ARDS), then efficiency of ARDS RCTs would be increased from the generic predictive and prognostic enrichment alongside increase in average treatment effect. We hypothesised that AF_ARDS will vary by severity of hypoxaemia as per the Berlin ARDS definitions[1, 15], by number of quadrants affected on chest radiography[16], and by sub-phenotype. Our hypothesis was informed by the following observations: first, two small cohort studies (eTable-1) indicated that AF_ARDS ranges between 15% and 37%[17, 18]; second, in the Berlin ARDS definitions, ARDS outcomes worsened with increasing severity of hypoxaemia[1]. Third, the hyperinflammatory ARDS sub-phenotype had higher mortality and greater treatment effect within RCTs [5–11]. Recently, these ARDS sub-phenotypes were identified within the LUNG-SAFE cohort using machine learning models [19], are available as predefined categories within the LUNG-SAFE dataset, and currently there are no AF_ARDS estimates for them. Given recent proposals to include AHRF patients (including those with unilateral infiltrates[20]) within an expanded ARDS case definition[21], we also compared AF_AHRF and AF_ARDS.

Methods

Data source

Our data source was the well described LUNG-SAFE (Large observational study to Understand the Global impact of Severe Acute respiratory FailurE) dataset. We summarise key elements of the LUNG-SAFE study design and data collection in eMethods-1. National coordinators and site investigators of the LUNG-SAFE study are listed in the online supplement. AHRF was defined as concurrent presence of: (a) arterial oxygen tension: inspired fraction of oxygen (PaO₂:FiO₂) ratio ≤ 300 mmHg; (b) new pulmonary parenchymal abnormalities (either unilateral or bilateral) on chest radiography; and (c) ventilatory support with continuous positive airway pressure or expiratory positive airway pressure or positive end expiratory pressure ≥ 5 cmH₂O. The diagnosis of ARDS in LUNG-SAFE studies was made by a computer algorithm in the analysis phase of the study using the “raw” data that made up the various components of the Berlin ARDS definition[15].

Study population

Selection criteria of AHRF/ARDS cohort reported in this manuscript were described previously by Pham et al[20]. We defined four populations for our matched cohort study, after excluding patients with congestive heart failure: (1) ARDS – AHRF patients who met the Berlin ARDS criteria, (2) AHRF – patients who met criteria for AHRF (and therefore includes all ARDS patients), (3) AHRF patients with unilateral infiltrates (AHRF-UL) - met criteria for AHRF but not ARDS, and (4) Non-AHRF controls - patients receiving non-invasive or invasive ventilation who did not meet criteria for AHRF (Figure-1). The LUNG-SAFE study collected only the following variables for non-AHRF controls - age, sex, ICU length of stay and ICU mortality.

Figure-1 — AF is the proportion of individuals with the outcome of interest e.g. death that can be attributed to the exposure e.g. ARDS. For example, AF_ARDS= [(Deaths in ARDS – Deaths in non-ARDS)/Deaths in ARDS].

Comparisons used to generate overall estimates for AF_ARDS, AF_AHRF, and AF_AHRF-UL are shown in the black rectangles. Further details on each model, and the AF estimates generated are provided in the table. To generate subpopulation estimates, analysis was stratified by severity of hypoxaemia, maximum number of quadrants involved in the first 48 hours, and ARDS sub-phenotype.

ARDS: acute respiratory distress syndrome; AF: attributable fraction; AHRF: acute hypoxaemic respiratory failure; AHRF-UL: acute hypoxaemic respiratory failure with unilateral infiltrates only

ARDS sub-phenotypes have recently been assigned in the LUNG-SAFE cohort using a clinical classifier model with a limited selection of predictor variables[19] (eMethods-2). Of note, this model used optimised probability cutoffs and did not use latent class analysis to assign sub-phenotypes. Patients without ARDS do not have sub-phenotype allocation There were no patients who had unilateral infiltrates without AHRF, to act as controls for AF_AHRF range estimates, similar to the AF_ARDS range estimates we report.

Analyses framework

The primary exposure was either ARDS, or AHRF, or AHRF-UL. The primary outcome was ICU mortality, as one of the most reported outcomes in ARDS RCTs[3]. Pre-defined enrichment categories within the primary exposures were severity of hypoxaemia, number of quadrants affected on chest radiography, and sub-phenotypes. Due to the low proportion of missing data (e-Table 2a), without any discernible pattern of missingness, these data were assumed to be missing at random, and complete-case analyses were used for all models.

Estimation of AF_ARDS requires careful selection of controls and consideration of potential confounding variables. We estimated propensity scores for the exposures using logistic regression. We used nearest neighbour matching without replacement to match exposed patients to controls. With this approach the mortality in the control groups within each prespecified ARDS severity category (severity of hypoxaemia, radiology, and sub-phenotype) would also vary based on matching, enabling estimation of variation in AF_ARDS within these categories, along with AF_ARDS range.

Propensity score models and scenarios

Model-1 scenario: AF_ARDS could be reduced with treatment to mortality seen in ICU patients of similar age and sex without AHRF (one ARDS patient was matched to two non-AHRF controls).

Model-2 scenario: AF_ARDS could be reduced with treatment to mortality seen in ICU patients with AHRF after accounting for variables commonly considered as part of eligibility criteria in ARDS RCTs at the time of randomisation such as age, sex, number of comorbidities, receipt of invasive mechanical ventilation, and illness severity (one ARDS patient was matched to one AHRF-UL control).

Model-3 scenario: AF_AHRF could be reduced with treatment to mortality seen in ICU patients of similar age and sex without AHRF (one AHRF patient was matched to two non-AHRF controls)

Model-4 scenario: AF_AHRF-UL could be reduced with treatment to mortality seen in ICU patients of similar age and sex without AHRF scenario (one AHRF-UL patient was matched to two non-AHRF controls).

Additional rationale for matching methods, covariate selection and assessment of each of four different propensity score models are available in Figure-1, eMethods-3, and eFigure-1. We used four separate logistic regression models to estimate AF (AF_ARDS, AF_AHRF, AF_AHRF-UL).

Estimation of AF_ARDS range

We used model-1, and model-2 to estimate the variation in AF_ARDS by severity of hypoxaemia categories (mild (PaO₂:FiO₂ >200mmHg); moderate (PaO₂:FiO₂ 100–200mmHg), or severe (PaO₂:FIO₂ <100mmHg)); the number of quadrants involved radiographically in the first 48 hours after ICU admission (two, three, or four), and the hyperinflammatory vs hypoinflammatory ARDS subphenotypes within LUNG-SAFE dataset[19].

Estimation of AF_AHRF

We used model-3 to estimate the variation in AF_AHRF by severity of hypoxaemia categories (mild, moderate, severe), and the number of quadrants involved radiographically in the first 48 hours after ICU admission (two, three, or four).

Estimation of AF_AHRF-UL

We used model-4 to estimate the variation in AF_AHRF-UL by severity of hypoxaemia categories (mild, moderate, severe), and the number of quadrants involved radiographically in the first 48 hours after ICU admission (one, two).

Simulating AF_ARDS based sample size estimates for different enrichment subgroups

To illustrate how AF_ARDS can influence sample size estimation, we compared predicted sample size estimates from 28 published ARDS RCTs[22–49] that used mortality as primary outcome (identified in our previous systematic review[3]) to simulated sample size estimates. We simulated sample size calculations for the scenario where AF_ARDS = 100%, and for estimates of AF_ARDS from Model-1 stratified by severity of hypoxaemia (mild, moderate, or severe), maximum number of quadrants involved on chest radiography at 48 hours (two, three, or four), and (c) sub-phenotype of ARDS (hyperinflammatory or hypoinflammatory). For all simulations, RCT control event rate was fixed at 40%, alpha at 0.05, and power at 0.80.

Sensitivity analysis

We report the following sensitivity analyses; (i) unmatched analyses in all 4 models to assess how matching - and therefore, exclusion of controls - affected overall and subpopulation estimates; and (ii) used hospital mortality as outcome measure in Model-2 - instead of ICU mortality - to assess how choice of mortality timepoint affected overall and subpopulation estimates for AF_ARDS.

χ² test was used to assess linear trends in ICU mortality across subpopulations, and to assess relationship between enrichment categories. Reported p-values are two-sided and p values less than 0.05 were considered statistically significant. All analyses were performed using R version 3.4.2 (AF[50], Matchit[51], and tidyverse[52] packages).

Results

Among 12906 admissions who received ventilatory support in the ICU, we identified 3504 eligible AHRF patients; 2653 met the Berlin ARDS criteria, and 851 patients had AHRF-UL (Figure-1). Missing data are summarised in eTable-2a. Baseline characteristics and outcomes for non-AHRF patients are summarised in eTable-2b. Patients with mild hypoxaemia most often had two quadrant infiltrates, whilst patients with severe hypoxaemia had four quadrant infiltrates (eTable-3a). There was no association between severity of hypoxaemia and ARDS subphenotype (eTable-3b).

Model-1 scenario

Model-1 scenario compared 2653 patients with ARDS matched to 5306 non-AHRF controls. Patients with ARDS had higher ICU mortality compared to controls (34.8% vs 13.8%) (Table-1). Significant linear trends in mortality were seen with severity of hypoxaemia category (mild 28.6% vs 13.2%; moderate 33.2% vs 14.6%; severe 43.0% vs 13.1%; X² = 32.7; p<0.001); and with increase in number of quadrants involved (two quadrants 27.7% vs 14.4%; three quadrants 34.7% vs 14.7%; four quadrants 40.6% vs 12.7%; X² = 90.9; p<0.001). The ICU mortality was higher for hyperinflammatory (51.7% vs 12.4%) and hypoinflammatory (28.0% vs 14.4%) ARDS, compared with non-AHRF controls (X² = 413.07; p<0.001).

Table 1.

Baseline characteristics of ARDS, AHRF, AHRF-UL and corresponding propensity matched control populations to derive ARDS/AHRF attributable fraction.

	Model-1 AF_ARDS		Model-2 AF_ARDS		Model-3 AF_AHRF		Model-4 AF_AHRF-UL
	ARDS (N = 2653)	Non-AHRF (N = 5306)	ARDS (N = 851)	AHRF-UL (N = 851)	AHRF (N = 3504)	Non-AHRF (N = 7008)	AHRF-UL (N = 851)	Non-AHRF (N = 1702)
Age (years) Mean (SD)	61.5 (16.8)	61.4 (16.8)	63.0 (16.2)	63.0 (16.9)	61.8 (16.8)	62.0 (16.8)	63.0 (16.9)	63.0 (16.8)
Sex Female n (%)	1014 (38.2)	2028 (38.2)	297 (34.9)	301 (35.4)	1315 (37.5)	2664 (38.0)	301 (35.4)	599 (35.2)
BMI Mean (SD)	27.4 (8.6)	-	26.9 (7.3)	26.7 (6.8)	27.3 (8.2)	-	26.7 (6.8)	-
Number of comorbidities n (%)
0	1053 (39.7)	-	330 (38.8)	322 (37.8)	1375 (39.2)	-	322 (37.8)	-
1	937 (35.3)	-	302 (35.5)	302 (35.5)	1239 (35.4)	-	302 (35.5)	-
2 or more	663 (25.0)	-	219 (25.7)	227 (26.7)	890 (25.4)	-	227 (26.7)	-
Risk factor for ARDS n (%)
Pneumonia	1574 (59.3)	-	510 (59.9)	433 (50.9)	2007 (57.3)	-	433 (50.9)	-
Extrapulmonary sepsis	421 (15.9)	-	141 (16.6)	115 (13.5)	536 (15.3)	-	115 (13.5)	-
Aspiration	386 (14.5)	-	114 (13.4)	157 (18.4)	543 (15.5)	-	157 (18.4)	-
Pancreatitis	57 (2.1)	-	16 (1.9)	11 (1.3)	68 (1.9)	-	11 (13)	-
Pulmonary vasculitis	12 (0.5)	-	2 (0.2)	0 (0)	12 (0.3)	-	0 (0.0)	-
Trauma	108 (4.1)	-	37 (4.3)	48 (5.6)	156 (4.5)	-	48 (5.6)	-
Inhalation	69 (2.6)	-	28 (3.3)	22 (2.6)	91 (2.6)	-	22 (2.6)	-
Pulmonary contusion	82 (3.1)	-	28 (3.3)	34 (4.0)	116 (3.3)	-	34 (4.0)	-
Burns	7 (0.3)	-	3 (0.4)	2 (0.2)	9 (0.3)	-	2 (0.2)	-
Non-cardiogenic shock	201 (7.6)	-	75 (8.8)	52 (6.1)	253 (7.2)	-	52 (6.1)	-
Drowning	2 (0.1)	-	1 (01)	0 (0)	2 (0.1)	-	0 (0.0)	-
Drug overdose	47 (1.8)	-	11 (1.3)	22 (2.6)	69 (2.0)	-	22 (2.6)	-
Transfusion-related	107 (4.0)	-	34 (4.0)	27 (3.2)	134 (3.8)	-	27 (3.2)	-
Other	67 (2.5)	-	17 (2.0)	45 (5.3)	112 (3.2)	-	45 (5.3)	-
No risk factor for ARDS	229 (8.6)	-	79 (9.3)	0 (0)	229 (6.5)	-	0 (0.0)	-
PaO₂:FiO₂ ratio on day 1 Mean (SD) ^*	159 (67.4)	-	161 (67.6)	188 (63.6)	166 (67.8)	-	188 (63.6)	-
Severity of hypoxemia on day 1 n (%)
Mild (200–300)	760 (28.6)	-	249 (29.3)	377 (44.3)	1137 (32.4)	-	377 (44.3)	-
Moderate (100–200)	1263 (47.6)	-	400 (47.0)	380 (44.7)	1643 (46.9)	-	380 (44.7)	-
Severe (<100)	626 (23.6)	-	200 (23.5)	93 (10.9)	719 (20.5)	-	93 (10.9)	-
Maximum number of quadrants involved in first 48 hours n (%) ^*
1	-	-	-	598 (70.2)	482 (13.8)	-	482 (70.2)	-
2	1045 (39.4)	-	351 (41.2)	236 (27.7)	1281 (36.6)	-	236 (27.7)	-
3	615 (23.2)	-	214 (25.1)	-	624 (17.8)	-	-	-
4	993 (37.4)	-	286 (33.6)	-	1000 (28.5)	-	-	-
Subphenotype of ARDS n (%) ^*
Hypoinflammatory	1966 (74.1)	-	616 (72.4)	-	-	-	-	-
Hyperinflammatory	687 (25.9)	-	235 (27.6)	-	-	-	-	-
SOFA score day 1 Mean (SD)	9.39 (4.08)	-	9.55 (4.19)	8.69 (3.97)	9.2 (4.1)	-	8.69 (3.97)	-
Non respiratory SOFA score day 1 ^*	6.48 (3.96)	-	6.65 (4.06)	6.04 (3.87)	6.4 (4.0)	-	6.04 (3.87)	-
Ventilated on day 1 n (%)	2151 (81.1)	-	700 (82.3)	698 (82.0)	2849 (81.3)	-	698 (82.0)	-
Duration of invasive mechanical ventilation Mean (SD) ^*	12.0 (12.9)	-	12.2 (12.8)	10.6 (12.6)	11.7 (12.9)	-	10.6 (12.6)	-
ICU LOS Mean (SD)	14.2 (14.0)	6.2 (8.2)	12.7 (13.8)	6.2 (8.2)	13.8 (14.0)	6.2 (8.2)	12.7 (13.8)	6.2 (7.9)
ICU mortality n (%)	906 (34.2)	730 (13.8)	293 (34.4)	206 (24.2)	1112 (31.7)	1000 (14.3)	206 (24.2)	247 (14.5)
Hospital LOS Mean (SD) ^*	23.1 (20.8)	-	23.6 (20.7)	22.6 (20.7)	22.9 (20.8)	-	22.6 (20.7)	-
Hospital mortality n (%) ^*	1044 (39.4)	-	342 (40.2)	259 (30.4)	1303 (37.3)	-	259 (30.4)	-

Open in a new tab

AF_ARDS in Model-1 was estimated by matching 2653 ARDS patients to non-AHRF controls that were propensity score balanced on age and sex in a 1:2 ratio. AF_ARDS in Model-2 was estimated by matching 851 ARDS patients to AHRF-UL controls in a 1:1 ratio and propensity score balanced on age, sex, number of comorbidities, and ventilation status on day 1. AF_AHRF in Model-3 was estimated by matching 3504 AHRF cases to non-AHRF controls that were propensity score balanced on age and sex in a 1:2 ratio. AF_AHRF-UL in Model-4 was estimated by matching 851 AHRF-UL cases to non-AHRF controls that were propensity score balanced on age and sex in a 1:2 ratio.

Denotes variable with missing data. Further details of missing data are available in eTable-2a

Post matching standardised mean differences (SMDs) for the matching covariates in each model are shown in eTable-4 and were all < 0.05; empirical cumulative density function plots for matching covariates were also consistent with good balance (eFigure-2). Propensity score overlap following matching is shown for each model in eFigure-3.

ARDS: acute respiratory distress syndrome; AHRF: acute hypoxaemic respiratory failure; AHRF-UL: acute hypoxaemic respiratory failure with unilateral infiltrates only; Std Diff: standardised difference; SD: standard deviation; BMI: Body Mass Index; SOFA: sequential organ failure assessment; ICU: intensive care unit; LOS: length of stay; NMV: non-matched variable.

Model-2 scenario

Model-2 scenario compared 851 patients with ARDS matched to 851 AHRF-UL controls. Patients with ARDS had higher ICU mortality (34.4% vs 24.2%; p<0.001), with the higher control arm mortality compared to model-1 reflecting the differences in matching variables between the models (Table-1). Similar to model-1, significant linear trends in mortality were seen with severity of hypoxaemia (mild 28.9% vs 19.3%; moderate 32.8% vs 28.5%; severe 44.5% vs 22.0%; X² = 58.3; p < 0.001), and with increase in the number of quadrants in patients with ARDS compared with AHRF-UL controls, the absolute differences were lower compared with model-1 (mild 28.9% vs 19.3%; moderate 32.8% vs 28.5%; severe 44.5% vs 22.0%; X² = 51.5; p<0.001). The ICU mortality was higher for hyperinflammatory (50.6% vs 26.8%) and hypoinflammatory (28.2% vs 23.2%) ARDS subphenotypes, compared with matched non-AHRF controls (X² = 147.27; p<0.001).

Range of AF_ARDS and categories using model-1, and model-2

The AF_ARDS ranges between 38.0% (95% CI 34.4% - 41.6%) from model-1(Figure-2a) and 20.9% (95% CI 10.5 – 31.4%) from model-2 (Figure-2b).

AF_ARDS varies by severity of hypoxaemia

In model-1, AF_ARDS increased with worsening severity of hypoxaemia (mild = 32.1% (95 % CI 24.4 – 39.8%), moderate = 36.7% (95% CI 29.4 – 44.1%), severe = 49.7% (95% CI 43.4 – 55.9%)) (Figure-2a).

Model-2 also highlighted increase in AF_ARDS with worsening severity of hypoxaemia (mild = 25.7% (95 % CI 6.1 – 45.3%), moderate = 6.1% (95% CI −11.6 – 23.8%), and severe = 41.1% (95% CI 25.2 – 57.1%) (Figure-2b)).

AF_ARDS varies by number of quadrants involved on chest radiography

In model-1, AF_ARDS increased with number of quadrants involved on chest radiography (two = 29.4% (95 % CI 22.7 – 36.2%), three = 38.1% (95% CI 30.8 – 45.4%), four = 45.9% (95% CI 40.7 – 51.2%)) (Figure 2a).

Model-2 also highlighted increase in AF_ARDS with number of quadrants involved on chest radiography (two = 14.4% (95 % CI −3.9 – 32.6%), three = 17.9% (95% CI −4.2 – 40.0%), four = 28.9% (95% CI 13.4 – 44.3%) (Figure-2b)).

AF_ARDS was higher in hyperinflammatory subphenotype

In model-1, AF_ARDS was higher in hyperinflammatory subphenotype (hyperinflammatory= 58.7% (95% CI 53.3 – 64.1%) vs hypoinflammatory = 28.0% (95 % CI 23.1 – 33.0%)) (Figure-2a).

Model-2 also highlighted higher AF_ARDS the hyperinflammatory subphenotype ((hyperinflammatory= 28.6% (95% CI 6.9 – 46.7%) vs hypoinflammatory 17.4% (95% CI 4.5 – 30.2%) (Figure-2b).

Model-3 scenario estimating AF_AHRF

Model-3 scenario compared 3504 patients with AHRF matched to 7008 non-AHRF controls. Patients with AHRF had higher ICU mortality compared with non-AHRF controls (31.7% vs 14.1%).

The estimated AF_AHRF from model-3 was 33.8% (95% CI 30.5% - 37.1%), which is lower than the AF_ARDS from model-1 and higher than the AF_ARDS from model-2. These differences were reflected in the severity of hypoxaemia and radiography categories (Figure-3a).

**Figure-3a.** Bar graphs show the mortality difference between AHRF population compared with propensity matched non-AHRF controls. AF_AHRF estimates stratified by severity of hypoxaemia and maximum number of quadrants involved in the first 48 hours are shown in the forest plot.

**Figure-3b.** Bar graphs show the mortality difference between AHRF-UL population compared with propensity matched non-AHRF controls. AF_AHRF-UL estimates stratified by severity of hypoxemia and maximum number of quadrants involved in the first 48 hours are shown in the forest plot.

AF: attributable fraction; AHRF: acute hypoxemic respiratory failure; AHRF-UL: acute hypoxemic respiratory failure with unilateral infiltrates only; CI: confidence interval; RD: risk difference.

Model-4 scenario estimating AF_AHRF-UL

Model-4 scenario compared 851 patients with AHRF-UL matched to 1702 non-AHRF controls. Patients with AHRF-UL had higher ICU mortality (24.2% vs 14.5%). The estimate of AF_AHRF-UL was 21.3% (95% CI 12.8% - 29.7%), which was which lower than the AF_ARDS from model-1 and comparable to model-2 (Figure-3b). In patients with unilateral, two-quadrant involvement, who would be excluded from ARDS RCTs, the estimate of AF_AHRF-UL was 22.8% (95% CI 7.0% - 38.7%), which was comparable to AF_ARDS from model-2.

Sample size requirements for ARDS RCTs change with estimated AF_ARDS

As the AF_ARDS increases in a RCT population, the sample size required will decrease for pre-specified alpha, beta, and risk reduction combinations. For example, from our current work, sample size estimates were lower for severe hypoxaemia compared to mild or moderate hypoxaemia, and lower for four quadrant radiographic involvement, compared with two or three quadrant radiographic involvement, and lower for hyperinflammatory sub-phenotype, compared with hypo-inflammatory sub-phenotype (Figure-4).

Figure-4 — These curves illustrate the AF_ARDS principle. Each curve represents the sample sizes required for different AF estimates (when control event rate is fixed at 40%). We show the estimates of AF_ARDS from Model-1, stratified by (a) severity of hypoxaemia, (b) maximum number of quadrants involved on chest radiography at 48 hours, and (c) sub-phenotype of ARDS. We contrast these against the common assumption that AF_ARDS is expected to be 100%.

Dot plots represent ARDS RCTs with mortality as primary outcome identified previously in our systematic review[12]; they correspond to the actual RRR used for sample size estimation and sample size per group in these RCTs. Median (IQR) control group mortality used for sample size calculations in these RCTs was 45.0% (33.3% - 52.5%) and RRR was 29.0% (24.5% - 33.3%). Most trials aimed for 80% power and 5% alpha. The sample size per group varied between 53 to 704 patients.

RCTs above a curve will have an adequate sample size to detect the predicted RRR.

RRR; relative risk reduction, AF; attributable fraction, ARDS: acute respiratory distress syndrome.

Sensitivity analyses

Overall unmatched estimates of AF_ARDS, AF_{AHRF and} AF_AHRF-UL were consistent with overall estimates from matched analyses. In the unmatched analyses of model-1, model- 3 and model-4 - which led to an increase in number of controls – trends within enrichment categories were no longer significant. In the unmatched analysis of model-2 - which led to an increase in number of exposed patients with ARDS – trends within enrichment categories were consistent with the matched analysis (Table-2).

Table 2.

Sensitivity analysis - Overall and subpopulation estimates of AF_ARDS, AF_AHRF and AF_AHRF-UL

	Model 1: AF_ARDS (non- AHRF controls)	Model 2: AF_ARDS (AHRF-UL controls)		Model 3: AF_AHRF (non- AHRF controls)	Model 4: AF_AHRF-UL (non-AHRF controls)
Sensitivity analysis	Unmatched analysis	Unmatched analysis	Matched analysis - hospital mortality as outcome	Unmatched analysis	Unmatched analysis
Patients with ARDS/AHRF	2653	2653	851	3504	851
Controls	8407	851	851	8407	8407
Overall AF	25.2% (22.7 – 27.6%)	22.3% (14.4 – 30.1%)	17.7% (7.7 – 27.6%)	26.5% (23.9 – 29.1%)	5.9% (4.1 – 7.7%)
PaO₂/FiO₂ ratio (mmHg)
>200	8.9% (4.5 – 13.4%)	10.7% (−1.6 – 23.0%)	25.8% (7.6 – 43.9%)	10.6% (6.3 – 15.0%)	2.5% (=2.5 – 7.5%)
100–200	17.2% (13.4 – 21.1%)	23.6% (13.2 – 34.0%)	3.1% (−13.7 – 19.9%)	18.9% (15.2 – 22.7%)	3.6% (1.3 – 8.5%)
<100	14.5% (10.7 – 18.5%)	33.9% (25.9 – 41.9%)	35.0% (18.9 – 51.2%)	15.6% (11.7 – 19.5%)	1.6% (−3.5% - 6.7%)
Maximum number of quadrants involved in first 48 hours
One	-	-	-	3.9% (−1.0 – 8.8%)	3.9% (−1.0 – 8.8%)
Two	11.3% (7.0 – 15.6%)	9.7% (−2.9 – 22.4%)	12.0% (−5.3 – 29.4%)	12.8% (8.6 – 17.0%)	2.3% (−2.7 – 7.3%)
Three	10.8% (6.6 – 15.1%)	20.0% (9.7 – 30.2%)	18.3% (−1.8 – 38.5%)	10.9% (6.6% - 15.1%)	-
Four	18.8% (15.1 – 22.4%)	36.4% (28.2 – 44.6%)	22.3% (7.0 – 37.6%)	18.7% (15.1 – 22.4%)	-
Subphenotype
Hypoinflammatory	18.3% (14.4 – 22.1%)	21.3% (10.2 – 32.4%)	14.1% (1.9 – 26.3%)	-	-
Hyperinflammatory	19.4% (15.9 – 23.0%)	31.9% (22.4 – 41.5%)	25.9% (7.4 – 44.4%)	-	-

Open in a new tab

Our primary analysis used matching without replacement with ICU mortality as the primary outcome measure. We repeated the analysis without matching in all four models to examine how matching – and therefore selection of controls, affected AF estimates. The matched analysis was also conducted using hospital mortality (instead of ICU mortality) as the outcome measure in Model-2.

AF: attributable fraction; CI; confidence interval; ARDS: acute respiratory distress syndrome; AHRF: acute hypoxaemic respiratory failure; AHRF-UL: acute hypoxaemic respiratory failure with unilateral infiltrates only

Overall estimate of AF_ARDS in model-2 was lower when hospital mortality was used as the outcome measure - instead of ICU mortality; subpopulation estimates were also consistently lower, but overall trends within enrichment categories remained consistent (Table-2).

Discussion

Using the LUNG-SAFE database, we report mean estimates of AF_ARDS between 20.9% to 38.0%. We observed a dose-response increase in AF_ARDS with severity of hypoxaemia and with quadrants of radiographic involvement. AF_ARDS was higher in the hyperinflammatory compared with the hypoinflammatory subphenotype of ARDS. Our results are consistent with previous work on AF_ARDS[17, 18] and we have extended these previous works by modelling distinct clinical scenarios, including the value of incorporating patients with AHRF in the extended ARDS case definitions[21]. Of note, our AF_ARDS estimates are consistent with a very different approach using marginal structural models, reported by Torres and colleagues[17].

We focused on several a priori defined ARDS subpopulations that have the potential for enrichment in clinical trials[53]. From our previous work, higher all-cause control-arm mortality does not necessarily generate larger average treatment effects in ARDS RCTs[12], making us hypothesise that ARDS-specific enrichment subgroups may outperform generic illness severity based prognostic enrichment.

Enrichment strategy, whether prognostic or predictive, is a trade-off between population prevalence, feasibility, and expected treatment effect[54]. In the LUNG-SAFE cohort, 23.6% of patients had severe ARDS, 36.7% had four-quadrant involvement, and 36.4% were hyperinflammatory ARDS subphenotype. Severe hypoxaemia is a potentially implementable enrichment criteria for ARDS RCTs, by using the approach highlighted within the Kigali modification of the ARDS definitions[55, 56]. Further, in previous RCTs of prone positioning and extracorporeal support[30, 57, 58] enriching on severe hypoxaemia has shown promise. Another element of the AHRF-ARDS debate is the inter- and intra-observer reliability of chest radiology, and its feasibility in resource limited settings. Whilst acknowledging this debate[59], four-quadrant involvement appears to be an enrichment marker for high AF_ARDS.

Another enrichment strategy linked to precision medicine is the subphenotyping of ARDS, which would require either measuring discriminant biomarkers with near patient testing, or implementation of machine learning-derived classifier models incorporating clinically available data. Similar subphenoptypes have been reported in non-ARDS populations including COVID-19[6], AHRF and sepsis, which potentially broadens the implications of our findings[60]. For illustration, we compared the sample size estimations from 28 ARDS RCTs[22–49], that used mortality as a primary outcome, for different AF_ARDS scenarios (Figure-4), which suggests that previous ARDS RCTs may lack sensitivity under the key assumption that only AF_ARDS deaths are affected by the tested treatment.

Our findings also lead us to consider how our work informs the debate on the need for distinction between ARDS and AHRF. Specifically, the estimate of AF_AHRF-UL for patients with unilateral, two-quadrant involvement, who would be excluded currently from ARDS RCTs, was comparable to AF_ARDS from model-2. Currently ARDS is conceptualised as a subset of AHRF; exclusion of AHRF patients with similar AF and overlapping biology[61] from the overall definition has implications for future RCTs and generalisability to clinical practice. Future research should explore the impact of including these populations in ARDS/AHRF RCTs.

Our analysis has strengths and limitations. We used the LUNG-SAFE dataset - a large multinational cohort recruited from 459 ICUs that was prospectively designed to enrol and follow up patients with AHRF and which underwent systematic validation after data collection. Our assessment of enrichment categories used inclusion criteria that would be immediately applicable to inform design of ARDS RCTs. Despite the use of propensity score methods, residual confounding remains a concern, given the available characteristics of controls and because we have not accounted for differences in mortality between study site and countries. Although we have not accounted for risk factors for ARDS, type of comorbidity, potential worsening (or improvement) over time in hypoxaemia, and geographic variations in usual care/ outcomes in these analyses, eligibility criteria in ARDS RCTs seldom stipulate these covariates[3]. The risk factor of ARDS limitation is explicit, as the LUNG-SAFE cohort did not collect ARDS risk factors for non-AHRF controls. Our primary and sensitivity analyses focus on mortality; the impact on other outcomes used in ARDS RCTs (such as ventilator free days) needs to be assessed. An implicit assumption in these models is that the putative treatment for ARDS/AHRF has no effect on non-ARDS/non-AHRF patients’ mortality. This assumption would not bias AF-ARDS estimates, as the control groups in ARDS RCTs would either not receive the intervention or those who do, will be analysed as crossovers / intention to treat framework.

Conclusions

ARDS is associated with excess mortality in critically ill patients. Our results highlight generic enrichment populations based on commonly used ARDS RCT eligibility criteria such as severity of hypoxaemia, and number of quadrants involved in chest radiography. We show that hyperinflammatory ARDS sub-phenotype has higher attributable fraction.

Supplementary Material

AF-ARDS-Supplement Thorax_revised

NIHMS1934563-supplement-AF-ARDS-Supplement_Thorax_revised.pdf^{(2.2MB, pdf)}

Key messages.

What is already known on this topic: The excess mortality - or attributable fraction (AF)-due to acute respiratory distress syndrome (ARDS) has been estimated to range between 15–37%. We do not know how this varies by severity of hypoxaemia, radiographic findings, and ARDS sub-phenotype.

What this study adds: We observed a dose-response increase in AF_ARDS with severity of hypoxaemia, quadrants of radiographic involvement, and that AF_ARDS was higher in the hyperinflammatory compared with the hypoinflammatory sub-phenotype of ARDS.

How this study might affect research, practice or policy: We highlight ARDS subpopulations that can inform enrichment options in randomised clinical trials.

Funding statement

The LUNG-SAFE study was funded and supported by the European Society of Intensive Care Medicine (ESICM), Brussels, Belgium, by St Michael’s Hospital, Toronto, Canada, and by the University of Milan-Bicocca, Monza, Italy. ESICM provided support in data collection and study coordination. ESICM, St Michael’s Hospital, Toronto, and University of Milan-Bicocca had no role in the design and conduct of the study; management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; or decision to submit the current manuscript for publication.

Competing Interest

CS is funded by UKRI [MR/S035753/1 and MR/X005070/1] and the National Institute for Health and Care Research [NIHR133788]. Work in her research group is supported by GlaxoSmithKline, the Wellcome Trust and the Cambridge NIHR Biomedical Research Centre, and she has received consultancy fees from Abbvie, Sanofi and GlaxoSmithKline. CC is funded by National Institutes of Health R35-HL140026 and reports grant funding from Roche-Genentech and Quantum Leap Healthcare collaborative, in addition to the National Institutes of Health, and consulting fees from Vasomune, Gen1e Life Sciences, Cellenkos, and Janssen.

EF reports personal fees from ALung Technologies, Aerogen, Baxter, GE Healthcare, Inspira, and Vasomune outside the submitted work.

JL is funded by a Future Research Leaders Award (16-FRL-3845) from Science Foundation Ireland and reports receiving consulting fees from Baxter Inc and from Glaxosmithkline. MS-H is funded by a clinician scientist fellowship from the National Institute for Health Research [CS-2016-16-011] and reports receiving grants from the NIHR, MRC, EME, HTA, Huo Foundation, and highlights industry support for TRAITS research programme - a Chief Scientists Office, Scotland funded time critical precision medicine in adult critically ill patients (TRAITS Programme).

Footnotes

Ethics approval

All participating ICUs individually obtained local ethics committee approval and obtained either patient consent or ethics committee waiver of consent in the original LUNG SAFE study. Research ethics board approval - St. Michael’s Hospital (one of two lead sites) REB# 13–384.

Data sharing

Requests for access to the LUNG-SAFE dataset should be submitted to research@esicm.org

Disclaimer

The views expressed in this publication are those of the authors and not necessarily those of the National Health Service, the National Institute for Health and Care Research, or the Department of Health and Social Care

References

1.Force ADT, Ranieri VM, Rubenfeld GD, Thompson BT, Ferguson ND, Caldwell E, Fan E, Camporota L, Slutsky AS: Acute respiratory distress syndrome: the Berlin Definition. JAMA 2012, 307(23):2526–2533. [DOI] [PubMed] [Google Scholar]
2.Matthay MA, Zemans RL, Zimmerman GA, Arabi YM, Beitler JR, Mercat A, Herridge M, Randolph AG, Calfee CS: Acute respiratory distress syndrome. Nat Rev Dis Primers 2019, 5(1):18. [DOI] [PMC free article] [PubMed] [Google Scholar]
3.Saha R, Assouline B, Mason G, Douiri A, Summers C, Shankar-Hari M: Impact of differences in acute respiratory distress syndrome randomised controlled trial inclusion and exclusion criteria: systematic review and meta-analysis. Br J Anaesth 2021, 127(1):85–101. [DOI] [PMC free article] [PubMed] [Google Scholar]
4.Sackett DL: Superiority trials, non-inferiority trials, and prisoners of the 2-sided null hypothesis. Evid Based Med 2004, 9(2):38–39. [Google Scholar]
5.Shankar-Hari M, Santhakumaran S, Prevost AT, Ward JK, Marshall T, Bradley C, Calfee CS, Delucchi KL, Sinha P, Matthay MA et al. : Defining phenotypes and treatment effect heterogeneity to inform acute respiratory distress syndrome and sepsis trials: secondary analyses of three RCTs. 2021, 8:10. [PubMed] [Google Scholar]
6.Sinha P, Calfee CS, Cherian S, Brealey D, Cutler S, King C, Killick C, Richards O, Cheema Y, Bailey C et al. : Prevalence of phenotypes of acute respiratory distress syndrome in critically ill patients with COVID-19: a prospective observational study. The Lancet Respiratory medicine 2020, 8(12):1209–1218. [DOI] [PMC free article] [PubMed] [Google Scholar]
7.Sinha P, Delucchi KL, Thompson BT, McAuley DF, Matthay MA, Calfee CS, Network NA: Latent class analysis of ARDS subphenotypes: a secondary analysis of the statins for acutely injured lungs from sepsis (SAILS) study. Intensive Care Med 2018, 44(11):1859–1869. [DOI] [PMC free article] [PubMed] [Google Scholar]
8.Calfee CS, Delucchi KL, Sinha P, Matthay MA, Hackett J, Shankar-Hari M, McDowell C, Laffey JG, O’Kane CM, McAuley DF et al. : Acute respiratory distress syndrome subphenotypes and differential response to simvastatin: secondary analysis of a randomised controlled trial. The Lancet Respiratory medicine 2018, 6(9):691–698. [DOI] [PMC free article] [PubMed] [Google Scholar]
9.Calfee CS, Janz DR, Bernard GR, May AK, Kangelaris KN, Matthay MA, Ware LB: Distinct molecular phenotypes of direct vs indirect ARDS in single-center and multicenter studies. Chest 2015, 147(6):1539–1548. [DOI] [PMC free article] [PubMed] [Google Scholar]
10.Famous KR, Delucchi K, Ware LB, Kangelaris KN, Liu KD, Thompson BT, Calfee CS, Network A: Acute Respiratory Distress Syndrome Subphenotypes Respond Differently to Randomized Fluid Management Strategy. Am J Respir Crit Care Med 2017, 195(3):331–338. [DOI] [PMC free article] [PubMed] [Google Scholar]
11.Calfee CS, Delucchi K, Parsons PE, Thompson BT, Ware LB, Matthay MA, Network NA: Subphenotypes in acute respiratory distress syndrome: latent class analysis of data from two randomised controlled trials. The Lancet Respiratory medicine 2014, 2(8):611–620. [DOI] [PMC free article] [PubMed] [Google Scholar]
12.Saha R, Assouline B, Mason G, Douiri A, Summers C, Shankar-Hari M: The impact of sample size mis-estimations on the interpretation of acute respiratory distress syndrome trials: systematic review and meta-analysis. CHEST. [DOI] [PubMed] [Google Scholar]
13.Sinclair JC, Haynes RB: Selecting participants that raise a clinical trial’s population attributable fraction can increase the treatment effect within the trial and reduce the required sample size. J Clin Epidemiol 2011, 64(8):893–902. [DOI] [PubMed] [Google Scholar]
14.Benichou J: A review of adjusted estimators of attributable risk. Stat Methods Med Res 2001, 10(3):195–216. [DOI] [PubMed] [Google Scholar]
15.Bellani G, Laffey JG, Pham T, Fan E, Brochard L, Esteban A, Gattinoni L, van Haren F, Larsson A, McAuley DF et al. : Epidemiology, Patterns of Care, and Mortality for Patients With Acute Respiratory Distress Syndrome in Intensive Care Units in 50 Countries. JAMA 2016, 315(8):788–800. [DOI] [PubMed] [Google Scholar]
16.Warren MA, Zhao Z, Koyama T, Bastarache JA, Shaver CM, Semler MW, Rice TW, Matthay MA, Calfee CS, Ware LB: Severity scoring of lung oedema on the chest radiograph is associated with clinical outcomes in ARDS. Thorax 2018, 73(9):840–846. [DOI] [PMC free article] [PubMed] [Google Scholar]
17.Torres LK, Hoffman KL, Oromendia C, Diaz I, Harrington JS, Schenck EJ, Price DR, Gomez-Escobar L, Higuera A, Vera MP et al. : Attributable mortality of acute respiratory distress syndrome: a systematic review, meta-analysis and survival analysis using targeted minimum loss-based estimation. Thorax 2021, 76(12):1176–1185. [DOI] [PMC free article] [PubMed] [Google Scholar]
18.Auriemma CL, Zhuo H, Delucchi K, Deiss T, Liu T, Jauregui A, Ke S, Vessel K, Lippi M, Seeley E et al. : Acute respiratory distress syndrome-attributable mortality in critically ill patients with sepsis. Intensive Care Med 2020, 46(6):1222–1231. [DOI] [PMC free article] [PubMed] [Google Scholar]
19.Maddali MV, Churpek M, Pham T, Rezoagli E, Zhuo H, Zhao W, He J, Delucchi KL, Wang C, Wickersham N et al. : Validation and utility of ARDS subphenotypes identified by machine-learning models using clinical data: an observational, multicohort, retrospective analysis. The Lancet Respiratory Medicine 2022, 10(4):367–377. [DOI] [PMC free article] [PubMed] [Google Scholar]
20.Pham T, Pesenti A, Bellani G, Rubenfeld G, Fan E, Bugedo G, Lorente JA, Fernandes ADV, Van Haren F, Bruhn A et al. : Outcome of acute hypoxaemic respiratory failure: insights from the LUNG SAFE Study. Eur Respir J 2021, 57(6). [DOI] [PubMed] [Google Scholar]
21.Matthay MA, Thompson BT, Ware LB: The Berlin definition of acute respiratory distress syndrome: should patients receiving high-flow nasal oxygen be included? The Lancet Respiratory medicine 2021, 9(8):933–936. [DOI] [PMC free article] [PubMed] [Google Scholar]
22.Barrot L, Asfar P, Mauny F, Winiszewski H, Montini F, Badie J, Quenot J-P, Pili-Floury S, Bouhemad B, Louis G et al. : Liberal or Conservative Oxygen Therapy for Acute Respiratory Distress Syndrome. New England Journal of Medicine 2020, 382(11):999–1008. [DOI] [PubMed] [Google Scholar]
23.Bollen CW, van Well GT, Sherry T, Beale RJ, Shah S, Findlay G, Monchi M, Chiche JD, Weiler N, Uiterwaal CS et al. : High frequency oscillatory ventilation compared with conventional mechanical ventilation in adult respiratory distress syndrome: a randomized controlled trial [ISRCTN24242669]. Crit Care 2005, 9(4):R430–439. [DOI] [PMC free article] [PubMed] [Google Scholar]
24.Brower RG, Lanken PN, MacIntyre N, Matthay MA, Morris A, Ancukiewicz M, Schoenfeld D, Thompson BT, National Heart L, Blood Institute ACTN: Higher versus lower positive end-expiratory pressures in patients with the acute respiratory distress syndrome. N Engl J Med 2004, 351(4):327–336. [DOI] [PubMed] [Google Scholar]
25.Combes A, Hajage D, Capellier G, Demoule A, Lavoué S, Guervilly C, Da Silva D, Zafrani L, Tirot P, Veber B et al. : Extracorporeal Membrane Oxygenation for Severe Acute Respiratory Distress Syndrome. New England Journal of Medicine 2018, 378(21):1965–1975. [DOI] [PubMed] [Google Scholar]
26.Constantin J-M, Jabaudon M, Lefrant J-Y, Jaber S, Quenot J-P, Langeron O, Ferrandière M, Grelon F, Seguin P, Ichai C et al. : Personalised mechanical ventilation tailored to lung morphology versus low positive end-expiratory pressure for patients with acute respiratory distress syndrome in France (the LIVE study): a multicentre, single-blind, randomised controlled trial. The Lancet Respiratory Medicine 2019, 7(10):870–880. [DOI] [PubMed] [Google Scholar]
27.Ferguson ND, Cook DJ, Guyatt GH, Mehta S, Hand L, Austin P, Zhou Q, Matte A, Walter SD, Lamontagne F et al. : High-frequency oscillation in early acute respiratory distress syndrome. N Engl J Med 2013, 368(9):795–805. [DOI] [PubMed] [Google Scholar]
28.Fernandez R, Trenchs X, Klamburg J, Castedo J, Serrano JM, Besso G, Tirapu JP, Santos A, Mas A, Parraga M et al. : Prone positioning in acute respiratory distress syndrome: a multicenter randomized clinical trial. Intensive Care Med 2008, 34(8):1487–1491. [DOI] [PubMed] [Google Scholar]
29.Gao Smith F, Perkins GD, Gates S, Young D, McAuley DF, Tunnicliffe W, Khan Z, Lamb SE: Effect of intravenous β−2 agonist treatment on clinical outcomes in acute respiratory distress syndrome (BALTI-2): a multicentre, randomised controlled trial. Lancet (London, England) 2012, 379(9812):229–235. [DOI] [PMC free article] [PubMed] [Google Scholar]
30.Guerin C, Reignier J, Richard JC, Beuret P, Gacouin A, Boulain T, Mercier E, Badet M, Mercat A, Baudin O et al. : Prone positioning in severe acute respiratory distress syndrome. N Engl J Med 2013, 368(23):2159–2168. [DOI] [PubMed] [Google Scholar]
31.Kacmarek RM, Villar J, Sulemanji D, Montiel R, Ferrando C, Blanco J, Koh Y, Soler JA, Martínez D, Hernández M et al. : Open Lung Approach for the Acute Respiratory Distress Syndrome. Crit Care Med 2016, 44(1):32–42. [DOI] [PubMed] [Google Scholar]
32.Kesecioglu J, Beale R, Stewart TE, Findlay GP, Rouby JJ, Holzapfel L, Bruins P, Steenken EJ, Jeppesen OK, Lachmann B: Exogenous natural surfactant for treatment of acute lung injury and the acute respiratory distress syndrome. Am J Respir Crit Care Med 2009, 180(10):989–994. [DOI] [PubMed] [Google Scholar]
33.Mancebo J, Fernández R, Blanch L, Rialp G, Gordo F, Ferrer M, Rodríguez F, Garro P, Ricart P, Vallverdú I et al. : A Multicenter Trial of Prolonged Prone Ventilation in Severe Acute Respiratory Distress Syndrome. Am J Respir Crit Care Med 2006, 173(11):1233–1239. [DOI] [PubMed] [Google Scholar]
34.Meade MO, Cook DJ, Guyatt GH, Slutsky AS, Arabi YM, Cooper DJ, Davies AR, Hand LE, Zhou Q, Thabane L et al. : Ventilation strategy using low tidal volumes, recruitment maneuvers, and high positive end-expiratory pressure for acute lung injury and acute respiratory distress syndrome: a randomized controlled trial. JAMA 2008, 299(6):637–645. [DOI] [PubMed] [Google Scholar]
35.Mentzelopoulos SD, Malachias S, Zintzaras E, Kokkoris S, Zakynthinos E, Makris D, Magira E, Markaki V, Roussos C, Zakynthinos SG: Intermittent recruitment with high-frequency oscillation/tracheal gas insufflation in acute respiratory distress syndrome. Eur Respir J 2012, 39(3):635–647. [DOI] [PubMed] [Google Scholar]
36.Mercat A, Richard JC, Vielle B, Jaber S, Osman D, Diehl JL, Lefrant JY, Prat G, Richecoeur J, Nieszkowska A et al. : Positive end-expiratory pressure setting in adults with acute lung injury and acute respiratory distress syndrome: a randomized controlled trial. JAMA 2008, 299(6):646–655. [DOI] [PubMed] [Google Scholar]
37.National Heart L, and Blood Institute PETAL Clinical Trials Network, Moss M, Huang DT, Brower RG, Ferguson ND, Ginde AA, Gong MN, Grissom CK, Gundel S, Hayden D, Hite RD, Hou PC, Hough CL, Iwashyna TJ, Khan A, Liu KD, Talmor D, Thompson BT, Ulysse CA, Yealy DM, Angus DC.: Early Neuromuscular Blockade in the Acute Respiratory Distress Syndrome. New England Journal of Medicine 2019, 380(21):1997–2008. [DOI] [PMC free article] [PubMed] [Google Scholar]
38.Papazian L, Forel JM, Gacouin A, Penot-Ragon C, Perrin G, Loundou A, Jaber S, Arnal JM, Perez D, Seghboyan JM et al. : Neuromuscular blockers in early acute respiratory distress syndrome. N Engl J Med 2010, 363(12):1107–1116. [DOI] [PubMed] [Google Scholar]
39.Peek GJ, Mugford M, Tiruvoipati R, Wilson A, Allen E, Thalanany MM, Hibbert CL, Truesdale A, Clemens F, Cooper N et al. : Efficacy and economic assessment of conventional ventilatory support versus extracorporeal membrane oxygenation for severe adult respiratory failure (CESAR): a multicentre randomised controlled trial. Lancet 2009, 374(9698):1351–1363. [DOI] [PubMed] [Google Scholar]
40.Richard C, Warszawski J, Anguel N, Deye N, Combes A, Barnoud D, Boulain T, Lefort Y, Fartoukh M, Baud F et al. : Early use of the pulmonary artery catheter and outcomes in patients with shock and acute respiratory distress syndrome: a randomized controlled trial. JAMA 2003, 290(20):2713–2720. [DOI] [PubMed] [Google Scholar]
41.Spragg RG, Taut FJ, Lewis JF, Schenk P, Ruppert C, Dean N, Krell K, Karabinis A, Gunther A: Recombinant surfactant protein C-based surfactant for patients with severe direct lung injury. Am J Respir Crit Care Med 2011, 183(8):1055–1061. [DOI] [PubMed] [Google Scholar]
42.Taccone P, Pesenti A, Latini R, Polli F, Vagginelli F, Mietto C, Caspani L, Raimondi F, Bordone G, Iapichino G et al. : Prone positioning in patients with moderate and severe acute respiratory distress syndrome: a randomized controlled trial. JAMA 2009, 302(18):1977–1984. [DOI] [PubMed] [Google Scholar]
43.Tongyoo S, Permpikul C, Mongkolpun W, Vattanavanit V, Udompanturak S, Kocak M, Meduri GU: Hydrocortisone treatment in early sepsis-associated acute respiratory distress syndrome: results of a randomized controlled trial. Crit Care 2016, 20(1):329. [DOI] [PMC free article] [PubMed] [Google Scholar]
44.Truwit JD, Bernard GR, Steingrub J, Matthay MA, Liu KD, Albertson TE, Brower RG, Shanholtz C, Rock P, Douglas IS et al. : Rosuvastatin for sepsis-associated acute respiratory distress syndrome. N Engl J Med 2014, 370(23):2191–2200. [DOI] [PMC free article] [PubMed] [Google Scholar]
45.Villar J, Ferrando C, Martínez D, Ambrós A, Muñoz T, Soler JA, Aguilar G, Alba F, González-Higueras E, Conesa LA et al. : Dexamethasone treatment for the acute respiratory distress syndrome: a multicentre, randomised controlled trial. The Lancet Respiratory Medicine 2020, 8(3):267–276. [DOI] [PubMed] [Google Scholar]
46.Wheeler AP, Bernard GR, Thompson BT, Schoenfeld D, Wiedemann HP, deBoisblanc B, Connors AF Jr., Hite RD, Harabin AL, National Heart L et al. : Pulmonary-artery versus central venous catheter to guide treatment of acute lung injury. N Engl J Med 2006, 354(21):2213–2224. [DOI] [PubMed] [Google Scholar]
47.Wiedemann HP, Wheeler AP, Bernard GR, Thompson BT, Hayden D, deBoisblanc B, Connors AF Jr., Hite RD, Harabin AL, National Heart L et al. : Comparison of two fluid-management strategies in acute lung injury. N Engl J Med 2006, 354(24):2564–2575. [DOI] [PubMed] [Google Scholar]
48.Willson DF, Truwit JD, Conaway MR, Traul CS, Egan EE: The Adult Calfactant in Acute Respiratory Distress Syndrome Trial. Chest 2015, 148(2):356–364. [DOI] [PubMed] [Google Scholar]
49.Young D, Lamb SE, Shah S, MacKenzie I, Tunnicliffe W, Lall R, Rowan K, Cuthbertson BH, Group OS: High-frequency oscillation for acute respiratory distress syndrome. N Engl J Med 2013, 368(9):806–813. [DOI] [PubMed] [Google Scholar]
50.Dahlqwist E, Magnusson PKE, Pawitan Y, Sjolander A: On the relationship between the heritability and the attributable fraction. Hum Genet 2019, 138(4):425–435. [DOI] [PMC free article] [PubMed] [Google Scholar]
51.Ho DE, Imai K, King G, Stuart EA: MatchIt: Nonparametric Preprocessing for Parametric Causal Inference. Journal of Statistical Software 2011, 42(8):1 – 28. [Google Scholar]
52.Wickham H, Averick M, Bryan J, Chang W, McGowan L, François R, Grolemund G, Hayes A, Henry L, Hester J et al. : Welcome to the Tidyverse. Journal of Open Source Software 2019, 4(43). [Google Scholar]
53.Ware LB, Matthay MA, Mebazaa A: Designing an ARDS trial for 2020 and beyond: focus on enrichment strategies. Intensive Care Med 2020, 46(12):2153–2156. [DOI] [PMC free article] [PubMed] [Google Scholar]
54.Shankar-Hari M, Rubenfeld GD: Population enrichment for critical care trials: phenotypes and differential outcomes. Curr Opin Crit Care 2019, 25(5):489–497. [DOI] [PubMed] [Google Scholar]
55.Riviello ED, Kiviri W, Twagirumugabe T, Mueller A, Banner-Goodspeed VM, Officer L, Novack V, Mutumwinka M, Talmor DS, Fowler RA: Hospital Incidence and Outcomes of the Acute Respiratory Distress Syndrome Using the Kigali Modification of the Berlin Definition. Am J Respir Crit Care Med 2016, 193(1):52–59. [DOI] [PubMed] [Google Scholar]
56.Vercesi V, Pisani L, van Tongeren PSI, Lagrand WK, Leopold SJ, Huson MMA, Henwood PC, Walden A, Smit M, Riviello ED et al. : External confirmation and exploration of the Kigali modification for diagnosing moderate or severe ARDS. Intensive Care Med 2018, 44(4):523–524. [DOI] [PubMed] [Google Scholar]
57.Goligher EC, Tomlinson G, Hajage D, Wijeysundera DN, Fan E, Juni P, Brodie D, Slutsky AS, Combes A: Extracorporeal Membrane Oxygenation for Severe Acute Respiratory Distress Syndrome and Posterior Probability of Mortality Benefit in a Post Hoc Bayesian Analysis of a Randomized Clinical Trial. JAMA 2018, 320(21):2251–2259. [DOI] [PubMed] [Google Scholar]
58.Combes A, Hajage D, Capellier G, Demoule A, Lavoue S, Guervilly C, Da Silva D, Zafrani L, Tirot P, Veber B et al. : Extracorporeal Membrane Oxygenation for Severe Acute Respiratory Distress Syndrome. N Engl J Med 2018, 378(21):1965–1975. [DOI] [PubMed] [Google Scholar]
59.Rubenfeld GD, Caldwell E, Granton J, Hudson LD, Matthay MA: Interobserver variability in applying a radiographic definition for ARDS. Chest 1999, 116(5):1347–1353. [DOI] [PubMed] [Google Scholar]
60.Heijnen NFL, Hagens LA, Smit MR, Cremer OL, Ong DSY, van der Poll T, van Vught LA, Scicluna BP, Schnabel RM, van der Horst ICC et al. : Biological Subphenotypes of Acute Respiratory Distress Syndrome Show Prognostic Enrichment in Mechanically Ventilated Patients without Acute Respiratory Distress Syndrome. Am J Respir Crit Care Med 2021, 203(12):1503–1511. [DOI] [PubMed] [Google Scholar]
61.Garcia-de-Acilu M, Marin-Corral J, Vazquez A, Ruano L, Magret M, Ferrer R, Masclans JR, Roca O: Hypoxemic Patients With Bilateral Infiltrates Treated With High-Flow Nasal Cannula Present a Similar Pattern of Biomarkers of Inflammation and Injury to Acute Respiratory Distress Syndrome Patients. Crit Care Med 2017, 45(11):1845–1853. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

AF-ARDS-Supplement Thorax_revised

NIHMS1934563-supplement-AF-ARDS-Supplement_Thorax_revised.pdf^{(2.2MB, pdf)}

[R1] 1.Force ADT, Ranieri VM, Rubenfeld GD, Thompson BT, Ferguson ND, Caldwell E, Fan E, Camporota L, Slutsky AS: Acute respiratory distress syndrome: the Berlin Definition. JAMA 2012, 307(23):2526–2533. [DOI] [PubMed] [Google Scholar]

[R2] 2.Matthay MA, Zemans RL, Zimmerman GA, Arabi YM, Beitler JR, Mercat A, Herridge M, Randolph AG, Calfee CS: Acute respiratory distress syndrome. Nat Rev Dis Primers 2019, 5(1):18. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R3] 3.Saha R, Assouline B, Mason G, Douiri A, Summers C, Shankar-Hari M: Impact of differences in acute respiratory distress syndrome randomised controlled trial inclusion and exclusion criteria: systematic review and meta-analysis. Br J Anaesth 2021, 127(1):85–101. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R4] 4.Sackett DL: Superiority trials, non-inferiority trials, and prisoners of the 2-sided null hypothesis. Evid Based Med 2004, 9(2):38–39. [Google Scholar]

[R5] 5.Shankar-Hari M, Santhakumaran S, Prevost AT, Ward JK, Marshall T, Bradley C, Calfee CS, Delucchi KL, Sinha P, Matthay MA et al. : Defining phenotypes and treatment effect heterogeneity to inform acute respiratory distress syndrome and sepsis trials: secondary analyses of three RCTs. 2021, 8:10. [PubMed] [Google Scholar]

[R6] 6.Sinha P, Calfee CS, Cherian S, Brealey D, Cutler S, King C, Killick C, Richards O, Cheema Y, Bailey C et al. : Prevalence of phenotypes of acute respiratory distress syndrome in critically ill patients with COVID-19: a prospective observational study. The Lancet Respiratory medicine 2020, 8(12):1209–1218. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R7] 7.Sinha P, Delucchi KL, Thompson BT, McAuley DF, Matthay MA, Calfee CS, Network NA: Latent class analysis of ARDS subphenotypes: a secondary analysis of the statins for acutely injured lungs from sepsis (SAILS) study. Intensive Care Med 2018, 44(11):1859–1869. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R8] 8.Calfee CS, Delucchi KL, Sinha P, Matthay MA, Hackett J, Shankar-Hari M, McDowell C, Laffey JG, O’Kane CM, McAuley DF et al. : Acute respiratory distress syndrome subphenotypes and differential response to simvastatin: secondary analysis of a randomised controlled trial. The Lancet Respiratory medicine 2018, 6(9):691–698. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R9] 9.Calfee CS, Janz DR, Bernard GR, May AK, Kangelaris KN, Matthay MA, Ware LB: Distinct molecular phenotypes of direct vs indirect ARDS in single-center and multicenter studies. Chest 2015, 147(6):1539–1548. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R10] 10.Famous KR, Delucchi K, Ware LB, Kangelaris KN, Liu KD, Thompson BT, Calfee CS, Network A: Acute Respiratory Distress Syndrome Subphenotypes Respond Differently to Randomized Fluid Management Strategy. Am J Respir Crit Care Med 2017, 195(3):331–338. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R11] 11.Calfee CS, Delucchi K, Parsons PE, Thompson BT, Ware LB, Matthay MA, Network NA: Subphenotypes in acute respiratory distress syndrome: latent class analysis of data from two randomised controlled trials. The Lancet Respiratory medicine 2014, 2(8):611–620. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R12] 12.Saha R, Assouline B, Mason G, Douiri A, Summers C, Shankar-Hari M: The impact of sample size mis-estimations on the interpretation of acute respiratory distress syndrome trials: systematic review and meta-analysis. CHEST. [DOI] [PubMed] [Google Scholar]

[R13] 13.Sinclair JC, Haynes RB: Selecting participants that raise a clinical trial’s population attributable fraction can increase the treatment effect within the trial and reduce the required sample size. J Clin Epidemiol 2011, 64(8):893–902. [DOI] [PubMed] [Google Scholar]

[R14] 14.Benichou J: A review of adjusted estimators of attributable risk. Stat Methods Med Res 2001, 10(3):195–216. [DOI] [PubMed] [Google Scholar]

[R15] 15.Bellani G, Laffey JG, Pham T, Fan E, Brochard L, Esteban A, Gattinoni L, van Haren F, Larsson A, McAuley DF et al. : Epidemiology, Patterns of Care, and Mortality for Patients With Acute Respiratory Distress Syndrome in Intensive Care Units in 50 Countries. JAMA 2016, 315(8):788–800. [DOI] [PubMed] [Google Scholar]

[R16] 16.Warren MA, Zhao Z, Koyama T, Bastarache JA, Shaver CM, Semler MW, Rice TW, Matthay MA, Calfee CS, Ware LB: Severity scoring of lung oedema on the chest radiograph is associated with clinical outcomes in ARDS. Thorax 2018, 73(9):840–846. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R17] 17.Torres LK, Hoffman KL, Oromendia C, Diaz I, Harrington JS, Schenck EJ, Price DR, Gomez-Escobar L, Higuera A, Vera MP et al. : Attributable mortality of acute respiratory distress syndrome: a systematic review, meta-analysis and survival analysis using targeted minimum loss-based estimation. Thorax 2021, 76(12):1176–1185. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R18] 18.Auriemma CL, Zhuo H, Delucchi K, Deiss T, Liu T, Jauregui A, Ke S, Vessel K, Lippi M, Seeley E et al. : Acute respiratory distress syndrome-attributable mortality in critically ill patients with sepsis. Intensive Care Med 2020, 46(6):1222–1231. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R19] 19.Maddali MV, Churpek M, Pham T, Rezoagli E, Zhuo H, Zhao W, He J, Delucchi KL, Wang C, Wickersham N et al. : Validation and utility of ARDS subphenotypes identified by machine-learning models using clinical data: an observational, multicohort, retrospective analysis. The Lancet Respiratory Medicine 2022, 10(4):367–377. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R20] 20.Pham T, Pesenti A, Bellani G, Rubenfeld G, Fan E, Bugedo G, Lorente JA, Fernandes ADV, Van Haren F, Bruhn A et al. : Outcome of acute hypoxaemic respiratory failure: insights from the LUNG SAFE Study. Eur Respir J 2021, 57(6). [DOI] [PubMed] [Google Scholar]

[R21] 21.Matthay MA, Thompson BT, Ware LB: The Berlin definition of acute respiratory distress syndrome: should patients receiving high-flow nasal oxygen be included? The Lancet Respiratory medicine 2021, 9(8):933–936. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R22] 22.Barrot L, Asfar P, Mauny F, Winiszewski H, Montini F, Badie J, Quenot J-P, Pili-Floury S, Bouhemad B, Louis G et al. : Liberal or Conservative Oxygen Therapy for Acute Respiratory Distress Syndrome. New England Journal of Medicine 2020, 382(11):999–1008. [DOI] [PubMed] [Google Scholar]

[R23] 23.Bollen CW, van Well GT, Sherry T, Beale RJ, Shah S, Findlay G, Monchi M, Chiche JD, Weiler N, Uiterwaal CS et al. : High frequency oscillatory ventilation compared with conventional mechanical ventilation in adult respiratory distress syndrome: a randomized controlled trial [ISRCTN24242669]. Crit Care 2005, 9(4):R430–439. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R24] 24.Brower RG, Lanken PN, MacIntyre N, Matthay MA, Morris A, Ancukiewicz M, Schoenfeld D, Thompson BT, National Heart L, Blood Institute ACTN: Higher versus lower positive end-expiratory pressures in patients with the acute respiratory distress syndrome. N Engl J Med 2004, 351(4):327–336. [DOI] [PubMed] [Google Scholar]

[R25] 25.Combes A, Hajage D, Capellier G, Demoule A, Lavoué S, Guervilly C, Da Silva D, Zafrani L, Tirot P, Veber B et al. : Extracorporeal Membrane Oxygenation for Severe Acute Respiratory Distress Syndrome. New England Journal of Medicine 2018, 378(21):1965–1975. [DOI] [PubMed] [Google Scholar]

[R26] 26.Constantin J-M, Jabaudon M, Lefrant J-Y, Jaber S, Quenot J-P, Langeron O, Ferrandière M, Grelon F, Seguin P, Ichai C et al. : Personalised mechanical ventilation tailored to lung morphology versus low positive end-expiratory pressure for patients with acute respiratory distress syndrome in France (the LIVE study): a multicentre, single-blind, randomised controlled trial. The Lancet Respiratory Medicine 2019, 7(10):870–880. [DOI] [PubMed] [Google Scholar]

[R27] 27.Ferguson ND, Cook DJ, Guyatt GH, Mehta S, Hand L, Austin P, Zhou Q, Matte A, Walter SD, Lamontagne F et al. : High-frequency oscillation in early acute respiratory distress syndrome. N Engl J Med 2013, 368(9):795–805. [DOI] [PubMed] [Google Scholar]

[R28] 28.Fernandez R, Trenchs X, Klamburg J, Castedo J, Serrano JM, Besso G, Tirapu JP, Santos A, Mas A, Parraga M et al. : Prone positioning in acute respiratory distress syndrome: a multicenter randomized clinical trial. Intensive Care Med 2008, 34(8):1487–1491. [DOI] [PubMed] [Google Scholar]

[R29] 29.Gao Smith F, Perkins GD, Gates S, Young D, McAuley DF, Tunnicliffe W, Khan Z, Lamb SE: Effect of intravenous β−2 agonist treatment on clinical outcomes in acute respiratory distress syndrome (BALTI-2): a multicentre, randomised controlled trial. Lancet (London, England) 2012, 379(9812):229–235. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R30] 30.Guerin C, Reignier J, Richard JC, Beuret P, Gacouin A, Boulain T, Mercier E, Badet M, Mercat A, Baudin O et al. : Prone positioning in severe acute respiratory distress syndrome. N Engl J Med 2013, 368(23):2159–2168. [DOI] [PubMed] [Google Scholar]

[R31] 31.Kacmarek RM, Villar J, Sulemanji D, Montiel R, Ferrando C, Blanco J, Koh Y, Soler JA, Martínez D, Hernández M et al. : Open Lung Approach for the Acute Respiratory Distress Syndrome. Crit Care Med 2016, 44(1):32–42. [DOI] [PubMed] [Google Scholar]

[R32] 32.Kesecioglu J, Beale R, Stewart TE, Findlay GP, Rouby JJ, Holzapfel L, Bruins P, Steenken EJ, Jeppesen OK, Lachmann B: Exogenous natural surfactant for treatment of acute lung injury and the acute respiratory distress syndrome. Am J Respir Crit Care Med 2009, 180(10):989–994. [DOI] [PubMed] [Google Scholar]

[R33] 33.Mancebo J, Fernández R, Blanch L, Rialp G, Gordo F, Ferrer M, Rodríguez F, Garro P, Ricart P, Vallverdú I et al. : A Multicenter Trial of Prolonged Prone Ventilation in Severe Acute Respiratory Distress Syndrome. Am J Respir Crit Care Med 2006, 173(11):1233–1239. [DOI] [PubMed] [Google Scholar]

[R34] 34.Meade MO, Cook DJ, Guyatt GH, Slutsky AS, Arabi YM, Cooper DJ, Davies AR, Hand LE, Zhou Q, Thabane L et al. : Ventilation strategy using low tidal volumes, recruitment maneuvers, and high positive end-expiratory pressure for acute lung injury and acute respiratory distress syndrome: a randomized controlled trial. JAMA 2008, 299(6):637–645. [DOI] [PubMed] [Google Scholar]

[R35] 35.Mentzelopoulos SD, Malachias S, Zintzaras E, Kokkoris S, Zakynthinos E, Makris D, Magira E, Markaki V, Roussos C, Zakynthinos SG: Intermittent recruitment with high-frequency oscillation/tracheal gas insufflation in acute respiratory distress syndrome. Eur Respir J 2012, 39(3):635–647. [DOI] [PubMed] [Google Scholar]

[R36] 36.Mercat A, Richard JC, Vielle B, Jaber S, Osman D, Diehl JL, Lefrant JY, Prat G, Richecoeur J, Nieszkowska A et al. : Positive end-expiratory pressure setting in adults with acute lung injury and acute respiratory distress syndrome: a randomized controlled trial. JAMA 2008, 299(6):646–655. [DOI] [PubMed] [Google Scholar]

[R37] 37.National Heart L, and Blood Institute PETAL Clinical Trials Network, Moss M, Huang DT, Brower RG, Ferguson ND, Ginde AA, Gong MN, Grissom CK, Gundel S, Hayden D, Hite RD, Hou PC, Hough CL, Iwashyna TJ, Khan A, Liu KD, Talmor D, Thompson BT, Ulysse CA, Yealy DM, Angus DC.: Early Neuromuscular Blockade in the Acute Respiratory Distress Syndrome. New England Journal of Medicine 2019, 380(21):1997–2008. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R38] 38.Papazian L, Forel JM, Gacouin A, Penot-Ragon C, Perrin G, Loundou A, Jaber S, Arnal JM, Perez D, Seghboyan JM et al. : Neuromuscular blockers in early acute respiratory distress syndrome. N Engl J Med 2010, 363(12):1107–1116. [DOI] [PubMed] [Google Scholar]

[R39] 39.Peek GJ, Mugford M, Tiruvoipati R, Wilson A, Allen E, Thalanany MM, Hibbert CL, Truesdale A, Clemens F, Cooper N et al. : Efficacy and economic assessment of conventional ventilatory support versus extracorporeal membrane oxygenation for severe adult respiratory failure (CESAR): a multicentre randomised controlled trial. Lancet 2009, 374(9698):1351–1363. [DOI] [PubMed] [Google Scholar]

[R40] 40.Richard C, Warszawski J, Anguel N, Deye N, Combes A, Barnoud D, Boulain T, Lefort Y, Fartoukh M, Baud F et al. : Early use of the pulmonary artery catheter and outcomes in patients with shock and acute respiratory distress syndrome: a randomized controlled trial. JAMA 2003, 290(20):2713–2720. [DOI] [PubMed] [Google Scholar]

[R41] 41.Spragg RG, Taut FJ, Lewis JF, Schenk P, Ruppert C, Dean N, Krell K, Karabinis A, Gunther A: Recombinant surfactant protein C-based surfactant for patients with severe direct lung injury. Am J Respir Crit Care Med 2011, 183(8):1055–1061. [DOI] [PubMed] [Google Scholar]

[R42] 42.Taccone P, Pesenti A, Latini R, Polli F, Vagginelli F, Mietto C, Caspani L, Raimondi F, Bordone G, Iapichino G et al. : Prone positioning in patients with moderate and severe acute respiratory distress syndrome: a randomized controlled trial. JAMA 2009, 302(18):1977–1984. [DOI] [PubMed] [Google Scholar]

[R43] 43.Tongyoo S, Permpikul C, Mongkolpun W, Vattanavanit V, Udompanturak S, Kocak M, Meduri GU: Hydrocortisone treatment in early sepsis-associated acute respiratory distress syndrome: results of a randomized controlled trial. Crit Care 2016, 20(1):329. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R44] 44.Truwit JD, Bernard GR, Steingrub J, Matthay MA, Liu KD, Albertson TE, Brower RG, Shanholtz C, Rock P, Douglas IS et al. : Rosuvastatin for sepsis-associated acute respiratory distress syndrome. N Engl J Med 2014, 370(23):2191–2200. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R45] 45.Villar J, Ferrando C, Martínez D, Ambrós A, Muñoz T, Soler JA, Aguilar G, Alba F, González-Higueras E, Conesa LA et al. : Dexamethasone treatment for the acute respiratory distress syndrome: a multicentre, randomised controlled trial. The Lancet Respiratory Medicine 2020, 8(3):267–276. [DOI] [PubMed] [Google Scholar]

[R46] 46.Wheeler AP, Bernard GR, Thompson BT, Schoenfeld D, Wiedemann HP, deBoisblanc B, Connors AF Jr., Hite RD, Harabin AL, National Heart L et al. : Pulmonary-artery versus central venous catheter to guide treatment of acute lung injury. N Engl J Med 2006, 354(21):2213–2224. [DOI] [PubMed] [Google Scholar]

[R47] 47.Wiedemann HP, Wheeler AP, Bernard GR, Thompson BT, Hayden D, deBoisblanc B, Connors AF Jr., Hite RD, Harabin AL, National Heart L et al. : Comparison of two fluid-management strategies in acute lung injury. N Engl J Med 2006, 354(24):2564–2575. [DOI] [PubMed] [Google Scholar]

[R48] 48.Willson DF, Truwit JD, Conaway MR, Traul CS, Egan EE: The Adult Calfactant in Acute Respiratory Distress Syndrome Trial. Chest 2015, 148(2):356–364. [DOI] [PubMed] [Google Scholar]

[R49] 49.Young D, Lamb SE, Shah S, MacKenzie I, Tunnicliffe W, Lall R, Rowan K, Cuthbertson BH, Group OS: High-frequency oscillation for acute respiratory distress syndrome. N Engl J Med 2013, 368(9):806–813. [DOI] [PubMed] [Google Scholar]

[R50] 50.Dahlqwist E, Magnusson PKE, Pawitan Y, Sjolander A: On the relationship between the heritability and the attributable fraction. Hum Genet 2019, 138(4):425–435. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R51] 51.Ho DE, Imai K, King G, Stuart EA: MatchIt: Nonparametric Preprocessing for Parametric Causal Inference. Journal of Statistical Software 2011, 42(8):1 – 28. [Google Scholar]

[R52] 52.Wickham H, Averick M, Bryan J, Chang W, McGowan L, François R, Grolemund G, Hayes A, Henry L, Hester J et al. : Welcome to the Tidyverse. Journal of Open Source Software 2019, 4(43). [Google Scholar]

[R53] 53.Ware LB, Matthay MA, Mebazaa A: Designing an ARDS trial for 2020 and beyond: focus on enrichment strategies. Intensive Care Med 2020, 46(12):2153–2156. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R54] 54.Shankar-Hari M, Rubenfeld GD: Population enrichment for critical care trials: phenotypes and differential outcomes. Curr Opin Crit Care 2019, 25(5):489–497. [DOI] [PubMed] [Google Scholar]

[R55] 55.Riviello ED, Kiviri W, Twagirumugabe T, Mueller A, Banner-Goodspeed VM, Officer L, Novack V, Mutumwinka M, Talmor DS, Fowler RA: Hospital Incidence and Outcomes of the Acute Respiratory Distress Syndrome Using the Kigali Modification of the Berlin Definition. Am J Respir Crit Care Med 2016, 193(1):52–59. [DOI] [PubMed] [Google Scholar]

[R56] 56.Vercesi V, Pisani L, van Tongeren PSI, Lagrand WK, Leopold SJ, Huson MMA, Henwood PC, Walden A, Smit M, Riviello ED et al. : External confirmation and exploration of the Kigali modification for diagnosing moderate or severe ARDS. Intensive Care Med 2018, 44(4):523–524. [DOI] [PubMed] [Google Scholar]

[R57] 57.Goligher EC, Tomlinson G, Hajage D, Wijeysundera DN, Fan E, Juni P, Brodie D, Slutsky AS, Combes A: Extracorporeal Membrane Oxygenation for Severe Acute Respiratory Distress Syndrome and Posterior Probability of Mortality Benefit in a Post Hoc Bayesian Analysis of a Randomized Clinical Trial. JAMA 2018, 320(21):2251–2259. [DOI] [PubMed] [Google Scholar]

[R58] 58.Combes A, Hajage D, Capellier G, Demoule A, Lavoue S, Guervilly C, Da Silva D, Zafrani L, Tirot P, Veber B et al. : Extracorporeal Membrane Oxygenation for Severe Acute Respiratory Distress Syndrome. N Engl J Med 2018, 378(21):1965–1975. [DOI] [PubMed] [Google Scholar]

[R59] 59.Rubenfeld GD, Caldwell E, Granton J, Hudson LD, Matthay MA: Interobserver variability in applying a radiographic definition for ARDS. Chest 1999, 116(5):1347–1353. [DOI] [PubMed] [Google Scholar]

[R60] 60.Heijnen NFL, Hagens LA, Smit MR, Cremer OL, Ong DSY, van der Poll T, van Vught LA, Scicluna BP, Schnabel RM, van der Horst ICC et al. : Biological Subphenotypes of Acute Respiratory Distress Syndrome Show Prognostic Enrichment in Mechanically Ventilated Patients without Acute Respiratory Distress Syndrome. Am J Respir Crit Care Med 2021, 203(12):1503–1511. [DOI] [PubMed] [Google Scholar]

[R61] 61.Garcia-de-Acilu M, Marin-Corral J, Vazquez A, Ruano L, Magret M, Ferrer R, Masclans JR, Roca O: Hypoxemic Patients With Bilateral Infiltrates Treated With High-Flow Nasal Cannula Present a Similar Pattern of Biomarkers of Inflammation and Injury to Acute Respiratory Distress Syndrome Patients. Crit Care Med 2017, 45(11):1845–1853. [DOI] [PubMed] [Google Scholar]

PERMALINK

Estimating the attributable fraction of mortality from acute respiratory distress syndrome to inform enrichment in future randomised clinical trials

Rohit Saha

Tài Pham

Pratik Sinha

Manoj V Maddali

Giacomo Bellani

Eddy Fan

Charlotte Summers

Abdel Douiri

Gordon Rubenfeld

Carolyn S Calfee

John G Laffey

Daniel F McAuley

Manu Shankar-Hari

Abstract

Background:

Methods:

Results:

Conclusions:

Introduction

Methods

Data source

Study population

Figure-1. Flowchart of patients screened and included in the models used to generate overall and subpopulation estimates of AFARDS, AFAHRF and AFAHRF-UL.

Analyses framework

Propensity score models and scenarios

Estimation of AFARDS range

Estimation of AFAHRF

Estimation of AFAHRF-UL

Simulating AFARDS based sample size estimates for different enrichment subgroups

Sensitivity analysis

Results

Model-1 scenario

Table 1.

Model-2 scenario

Range of AFARDS and categories using model-1, and model-2

Figure-2. Overall and subpopulation estimates of AFARDS.

AFARDS varies by severity of hypoxaemia

AFARDS varies by number of quadrants involved on chest radiography

AFARDS was higher in hyperinflammatory subphenotype

Model-3 scenario estimating AFAHRF

Figure-3. Overall and subpopulation estimates of AFAHRF and AFAHRF-UL.

Model-4 scenario estimating AFAHRF-UL

Sample size requirements for ARDS RCTs change with estimated AFARDS

Figure-4. Illustrative examples of sample size calculations for different AFARDS scenarios.

Sensitivity analyses

Table 2.

Discussion

Conclusions

Supplementary Material

Key messages.

Funding statement

Competing Interest

Footnotes

References

Associated Data

Supplementary Materials

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases

Figure-1. Flowchart of patients screened and included in the models used to generate overall and subpopulation estimates of AF_ARDS, AF_AHRF and AF_AHRF-UL.

Estimation of AF_ARDS range

Estimation of AF_AHRF

Estimation of AF_AHRF-UL

Simulating AF_ARDS based sample size estimates for different enrichment subgroups

Range of AF_ARDS and categories using model-1, and model-2

Figure-2. Overall and subpopulation estimates of AF_ARDS.

AF_ARDS varies by severity of hypoxaemia

AF_ARDS varies by number of quadrants involved on chest radiography

AF_ARDS was higher in hyperinflammatory subphenotype

Model-3 scenario estimating AF_AHRF

Figure-3. Overall and subpopulation estimates of AF_AHRF and AF_AHRF-UL.

Model-4 scenario estimating AF_AHRF-UL

Sample size requirements for ARDS RCTs change with estimated AF_ARDS

Figure-4. Illustrative examples of sample size calculations for different AF_ARDS scenarios.