Effect of Aspirin on Development of ARDS in At-Risk Patients Presenting to the Emergency Department: The LIPS-A Randomized Clinical Trial

Importance  Management of acute respiratory distress syndrome (ARDS) remains largely supportive. Whether early intervention can prevent development of ARDS remains unclear.

Objective  To evaluate the efficacy and safety of early aspirin administration for the prevention of ARDS.

Design, Setting, and Participants  A multicenter, double-blind, placebo-controlled, randomized clinical trial conducted at 16 US academic hospitals. Between January 2, 2012, and November 17, 2014, 7673 patients at risk for ARDS (Lung Injury Prediction Score ≥4) in the emergency department were screened and 400 were randomized. Ten patients were excluded, leaving 390 in the final modified intention-to-treat analysis cohort.

Interventions  Administration of aspirin, 325-mg loading dose followed by 81 mg/d (n = 195) or placebo (n = 195) within 24 hours of emergency department presentation and continued to hospital day 7, discharge, or death.

Main Outcomes and Measures  The primary outcome was the development of ARDS by study day 7. Secondary measures included ventilator-free days, hospital and intensive care unit length of stay, 28-day and 1-year survival, and change in serum biomarkers associated with ARDS. A final α level of .0737 (α = .10 overall) was required for statistical significance of the primary outcome.

Results  Among 390 analyzed patients (median age, 57 years; 187 [48%] women), the median (IQR) hospital length of stay was 6 3-10) days. Administration of aspirin, compared with placebo, did not significantly reduce the incidence of ARDS at 7 days (10.3% vs 8.7%, respectively; odds ratio, 1.24 [92.6% CI, 0.67 to 2.31], P = .53). No significant differences were seen in secondary outcomes: ventilator-free to day 28, mean (SD), 24.9 (7.4) days vs 25.2 (7.0) days (mean [90% CI] difference, −0.26 [−1.46 to 0.94] days; P = .72); ICU length of stay, mean (SD), 5.2 (7.0) days vs 5.4 (7.0) days (mean [90% CI] difference, −0.16 [−1.75 to 1.43] days; P = .87); hospital length of stay, mean (SD), 8.8 (10.3) days vs 9.0 (9.9) days (mean [90% CI] difference, −0.27 [−1.96 to 1.42] days; P = .79); or 28-day survival, 90% vs 90% (hazard ratio [90% CI], 1.03 [0.60 to 1.79]; P = .92) or 1-year survival, 73% vs 75% (hazard ratio [90% CI], 1.06 [0.75 to 1.50]; P = .79). Bleeding-related adverse events were infrequent in both groups (aspirin vs placebo, 5.6% vs 2.6%; odds ratio [90% CI], 2.27 [0.92 to 5.61]; P = .13).

Conclusions and Relevance  Among at-risk patients presenting to the ED, the use of aspirin compared with placebo did not reduce the risk of ARDS at 7 days. The findings of this phase 2b trial do not support continuation to a larger phase 3 trial.

Trial Registration  clinicaltrials.gov Identifier: NCT01504867

Quiz Ref IDAcute respiratory distress syndrome (ARDS) remains a life-threatening critical care syndrome^1-3 characterized by alveolar-capillary membrane injury and hypoxemic respiratory failure. The median time to onset of ARDS is 2 days after hospital presentation.⁴ The period between hospital presentation and development of ARDS presents a brief window of opportunity for ARDS prevention. Therefore, a major inherent challenge for ARDS prevention trials is early and accurate identification and treatment of patients at risk.

Quiz Ref IDMechanistically, ARDS has been viewed as an inflammatory condition. Recently, additional pathways have been described with accumulating evidence suggesting an important role for platelets in both the onset^5-7 and resolution^8-10 of lung injury. Observational studies have suggested a potential preventive role for antiplatelet therapy in patients at high risk for ARDS.^11-14 To further understand the role of aspirin for the prevention of ARDS, a randomized clinical trial was performed aiming to test the efficacy and safety of aspirin for the prevention of ARDS among at-risk patients: the Lung Injury Prevention Study with Aspirin (LIPS-A).

This was a multicenter, double-blind, placebo-controlled, parallel-group, phase 2b, randomized clinical trial. The full study design and study procedures are published elsewhere¹⁵ and are included in Supplement 1. The study was approved by the institutional review boards of all participating locations prior to the initiation of study-related activities. Written informed consent was obtained from the patient, next of kin, or the legal representative of the patient for those unable to provide consent due to their medical condition(s). The patient or surrogate was informed about the right to withdraw from the study at any point. Patients who were unable to provide consent prior to randomization due to their medical condition(s) were informed accordingly if they regained consciousness.

Patients aged 18 years or older admitted to the hospital through the emergency department with elevated risk for developing ARDS based on a calculated lung injury prediction score (LIPS ≥4)⁴ were considered for inclusion in the trial. The LIPS threshold of 4 was previously identified as the cut point that optimized both sensitivity and specificity of the predictive model. The aim of this trial was to better define the role of aspirin as a potential ARDS prevention intervention (as opposed to a therapy for established ARDS). Therefore, patients with prevalent ARDS at the time of screening were excluded. Patients presenting to the emergency department who were already receiving antiplatelet therapies were also excluded. Considerations related to this exclusion included the ethical implications of discontinuing antiplatelet therapies in patients for whom they had previously been prescribed as well as the potential confounding effects of preadmission antiplatelet therapies (and their potential for extended impact on platelet function even after discontinuation) with the primary outcome of interest. Additional exclusion criteria are described in Supplement 1. Shortly after trial onset, the data and safety monitoring board removed prevalent chronic kidney disease or acute kidney injury as exclusion criteria. Randomization was required to be completed within 12 hours of presentation to the participating hospital. Information on race, stipulated by the study funding agency, was collected from participants via self-report.

Eligible participants were centrally randomized in a 1:1 ratio to the aspirin or placebo treatment group using Medidata Balance. Dynamic minimization with a second guess probability of 0.2 was used to randomly allocate treatment assignments while stratifying by center.¹⁶ The study participant, clinical team, and all members of the study team were blinded to treatment allocation.

The first dose of study drug was administered within 24 hours after presentation to the hospital. For patients randomized to the intervention group, a 325-mg loading dose of aspirin was administered on day 1, followed by 81 mg of aspirin once daily up to day 7, hospital discharge, or death, whichever occurred first. The placebo group received an identical-appearing capsule filled with lactose powder. The dose of aspirin selected for this trial was influenced by existing literature noting low-dose aspirin at 81 mg/d was effective in elevating plasma levels of anti-inflammatory lipoxins and inhibiting platelet thromboxane activity with only a slight increase in effect at higher doses.^17,18 A larger loading dose of aspirin (325 mg) was selected in an effort to mitigate potential risks related to insufficient dosing of study medication.

Important co-interventions, including mechanical ventilation, aspiration precautions, infection control, and fluid and transfusion management, were standardized across sites using the web-based tool Checklist for Lung Injury Prevention (CLIP)¹⁹ (see full protocol in Supplement 1).

The primary outcome was the development of ARDS, as defined by Berlin criteria (modified to require invasive mechanical ventilation),²⁰ within 7 days of hospital admission. The inclusion of invasive mechanical ventilation as a requirement for adjudicating ARDS was pursued to mitigate concerns related to the more subjective nature of implementing noninvasive ventilator support. In addition, restriction of mechanical ventilator support to include only invasive mechanical ventilation provided a greater degree of consistency with prior ARDS clinical trials. To assess for the primary outcome, study participants were screened daily for receipt of mechanical ventilation and determination of the partial pressure of arterial oxygen (Pao₂) or oxygen saturation to fraction of inspired oxygen ratio (Spo₂:Fio₂). If the study participant’s Spo₂:Fio₂ ratio was consistently below 315,²¹ hypoxemia was confirmed with measurement of arterial blood gas. Chest radiographs for all intubated patients with a Pao₂:Fio₂ ratio of 300 or less were independently reviewed by both site investigator and a member of the trial’s executive committee (D.T.) for bilateral infiltrates consistent with ARDS. Disagreements were resolved by 3 additional investigators blinded to the initial ARDS adjudication (D.J.K., O.G., M.N.G.). Study participants who died or were discharged from the hospital before day 7 without meeting criteria for ARDS were adjudicated as not having ARDS. Secondary outcome assessments included ventilator-free days to hospital day 28, intensive care unit (ICU) and hospital lengths of stay, and 28-day and 1-year mortality.

Plasma samples were obtained at baseline (after randomization, before administration of study drug), on study day 1 (approximately 24 hours after the first dose), and on study day 4 for enzyme-linked immunosorbent assays (in duplicate) of 9 plasma biomarkers previously found to be associated with the development of ARDS. These include surfactant protein D (SP-D), a marker of lung epithelial injury (BioVendor Inc); angiopoietin 2 (Ang-2), a marker and mediator of endothelial injury (R&D Systems); interleukins IL-1β, IL-2, IL-4, IL-6, IL-8, and IL-10; and tumor necrosis factor α (TNF-α), markers of inflammation (Meso Scale Diagnostics).

This study was designed as a phase 2b clinical trial using an a priori α= .10 and planned interim analysis, which was conducted with an information fraction of 62.5% (n = 250 participants). This shifted the final α level from a planned .0889 to .0737 using the prespecified O’Brien-Fleming–like α spending function.^22,23 Therefore, for the primary end point to be statistically significant, the 2-sided P value would need to be <.0737. For secondary end points, P <.10 was considered statistically significant. No adjustment for multiple testing was applied to reported P values, and these analyses should be interpreted as exploratory.

We estimated the sample size of 197 per group based on the following assumptions: (1) an ARDS development rate of 18%, (2) a minimum clinically relevant effect of 10 percentage points, and (3) a final 2-sided α of .0889, adjusted for a planned interim analysis at 50% information fraction. The effect size of 10% was chosen by the study’s executive team at the time of protocol creation based on the impression that this represents a clinically relevant between-group difference in ARDS event rates. Conservative sample size estimates were based on ARDS event rates of 25% and 15% to decrease the relative risk and increase the binomial proportion variance. Target randomization was set at 200 participants per group (400 total) to allow for attrition.

Conditional logistic regression was used to test the primary hypothesis that early aspirin administration would decrease the rate of ARDS. Clinical site was included in the model as a stratification variable. This analysis was supplemented by Cochran-Mantel-Haenszel stratified analysis with odds ratios computed for each site. We used the Breslow-Day test to test for differences in aspirin effect by site. Secondary binary end points were tested using unconditional tests and time-to-event analyses were conducted using Kaplan-Meier estimator. Continuous measures were tested between groups using Wilcoxon rank sum and 2-sample t tests. Numerical summaries of these variables are presented as median (quartile 1-quartile 3) and mean (SD) unless otherwise specified. The protocol-specified full intention-to-treat (ITT) analysis set was modified to account for withdrawal of consent or ineligibility based on inclusion or exclusion criteria. We conducted analyses on the resultant full analysis set (ie, a modified ITT analysis set). Plasma biomarkers were analyzed separately using mixed models to test for the fixed effects of day of treatment, treatment assignment, and the treatment by time. Participants were modeled with a random intercept. Biomarker concentrations were log transformed prior to analysis, and for concentrations below the assay lower limit of detection (LLD), a numeric value was imputed as 0.5 × assay LLD.

In addition to statistical criteria for significance, the study included a priori “go-no-go” definitions for recommending continuation to phase 3 study (see section 10.3.5 in the protocol in Supplement 1). Briefly, continuation to phase 3 would occur with a positive primary outcome finding along with an acceptable safety profile. An acceptable safety profile was defined as a serious adverse event profile for aspirin that was not statistically worse than placebo (95% CI for the relative risk of any serious adverse event covers the null value of relative risk = 1.0). The “no-go decision” was defined as early termination by the data and safety monitoring board for safety or unfavorable risk/benefit ratio. An indeterminate case in which there was a non–statistically significant effect but this effect was in a clinically meaningful direction was also defined. Final statistical analyses were conducted using the base SAS System version 9.4 and SAS/STAT version 14.1.

Between January 2, 2012, and November 17, 2014, 7673 patients were screened at 16 medical centers from across the United States (Figure; eTable 1 in Supplement 2). The most common reasons for exclusion included antiplatelet therapy at the time of presentation (n = 3052), inability to obtain informed consent within the prespecified 12-hour window (n = 1299), and suspicion for active bleeding at time of initial evaluation (n = 999). After excluding 7273 screen failures, 400 patients were randomized. Ten randomized patients were excluded from subsequent analyses (aspirin, n = 7; placebo, n = 3; 6 for withdrawal of consent and 4 for ineligibility discovered after randomization), leaving 195 patients in each group of the modified ITT analysis set.

Baseline demographics and clinical characteristics, according to treatment allocation, are presented in Table 1. Randomization procedures were effective at equalizing distributions of baseline variables. The median (interquartile range [IQR]) time from hospital presentation to randomization was 7.3 (5.1-10.2) hours. The median age was 57 (45-68) years and 52% (203/390) were male. Baseline lung injury prediction scores (LIPS) were not significantly different between groups, with a median (IQR) LIPS of 6.0 (5.0-7.5) in the aspirin group and 5.5 (4.5-7.0) in the placebo group. Major risk factors for ARDS were similarly distributed in both treatment groups.

Of 2049 potential study drug administration episodes, 1742 (85%) were provided as per protocol. Of the 195 patients randomized to receive aspirin, 185 (95%) received at least 1 dose of study medication. In the placebo group, 189 (96.9%) received at least 1 dose of study drug. The median (IQR) number of doses was not statistically different (4 [2-7] for aspirin and 5 [3-7] for placebo; P = .19). The median time from randomization to first study drug administration was 12.7 (7.9-21.2) hours in the aspirin group and 12.4 (8.8-19.3) hours in the placebo group (P = .86). Reasons for study participants not receiving at least 1 of their assigned study medications are provided in the Figure.

In the modified ITT analysis set, 37 patients (9.5%) developed ARDS within 7 days of randomization: 20 patients (10.3%) in the aspirin group compared with 17 patients in the placebo group (8.7%), for a site-adjusted odds ratio (92.6% CI) of 1.24 (0.67-2.31); P = .53. The distribution of ARDS by enrolling site is shown in the online data supplement (eFigure 1 in Supplement 2). The Breslow-Day test for homogeneity did not suggest there was significant variation in the odds ratio by site (P = .19).

There was no signal for a beneficial effect of aspirin across secondary efficacy outcome measures (Table 2). A total of 18 patients (9.2%) in the aspirin group and 18 patients (9.2%) in the placebo group died by 28 days (28-day survival, 90% vs 90%, hazard ratio [90% CI], 1.03 [0.60-1.79]; log-rank P = .92). Longer-term mortality over the year of follow-up showed the same pattern as shorter-duration mortality analyses (1-year survival, 73% vs 75%; hazard ratio [90% CI], 1.06 [0.75-1.50]; log-rank P = .79; eFigure 2 in Supplement 2). No statistically significant differences were noted in the additional secondary clinical outcomes including need for mechanical ventilation, ventilator-free days at day 28, ICU length of stay, or hospital length of stay. Patients in the aspirin group were more likely to be admitted to the ICU (59.0% vs 50.3%; odds ratio [90% CI], 1.42 [1.02-1.99]; P = .08). In the biomarker analysis, IL-2 was higher in the aspirin group vs placebo on day 1 (P = .08) with a time × treatment interaction effect (P = .08) that met the prespecified level of significance (P < .10). However, there were no other between-group differences or time × treatment interaction effects for any of the other biomarker levels analyzed at baseline, day 1, or day 4 (P > .10) (eFigure 3 in Supplement 2).

No statistically significant differences were found in measures of safety (Table 3; eTable 2 in Supplement 2). Study-reported adverse events were observed in 7.7% (30/390) of the participants. Of these, bleeding-related adverse events were reported in 11 of 195 patients (5.6%) assigned to the aspirin group and 5 of 195 patients (2.6%) assigned to the placebo group (odds ratio [90% CI], 2.27 [0.92-5.61]; P = .13). Moderate or severe bleeding-related adverse events were infrequent in both groups [aspirin (n = 8) vs placebo (n = 4), 4.1% vs 2.1%, odds ratio [90% CI], 2.04 [0.74-5.67]; P = .24). Development of worsening renal function as estimated by the modified RIFLE criteria (Risk, Injury, Failure, Loss, End-stage Renal Disease²⁴) did not differ statistically by treatment assignment. Detailed numerical summaries of changes in renal function are included in Table 3.

Quiz Ref IDWe report the results of a multicenter, randomized, double-blind, placebo-controlled phase 2b trial evaluating early aspirin administration for prevention of ARDS. In at-risk patients, initiating aspirin therapy within 24 hours of presentation to the emergency department was safe when compared with placebo. However, early aspirin therapy did not decrease the primary outcome of ARDS development or improve any of the secondary outcomes. As such, the results of this phase 2b trial did not meet the prespecified criteria to recommend pursuing a larger, phase 3 study.

Despite substantial improvements over the past 2 decades,^1,25 mortality in severe ARDS remains as high as 40% to 50%.^3,20 The search for further reductions in mortality has shifted attention from targeted therapeutics administered after the development of critical illness to earlier interventions designed to ameliorate or even prevent organ failure.^26,27 The majority of patients and clinicians are willing to consider an acceptably safe, early intervention before the development of disease in order to prevent later, more severe consequences. In this paradigm, timely risk stratification is essential to maximize administration to patients who have the highest probability of benefit and to minimize the risk to patients who will never go on to develop the disease.

The time course of ARDS development following presentation to the emergency department is rapid, with most cases developing within 1 to 2 days of initial hospitalization.^4,28 To effectively test a true prevention strategy, we based patient identification and risk stratification on early determination of LIPS of 4 or higher in the emergency department or ICU coupled with a 12-hour trial eligibility window, well in advance of the usual time frame for enrolling participants in ARDS treatment trials.

Early aspirin administration for ARDS prevention was tested based on the body of existing experimental data demonstrating alterations in platelet function during the development of ARDS.²⁹ Platelet activation, aggregation, and sequestration, as well as modulation of anti-inflammatory lipid mediators, including leukotrienes, thromboxane, and prostaglandins, have all been implicated as important mediators of ARDS progression and severity.^5,6,8,9,29 Aspirin directly modifies these mechanistic pathways, making it a plausible preventive and therapeutic measure in this setting.^5,30,31 To date, clinical studies have suggested conflicting benefits of aspirin or other nonsteroidal anti-inflammatory drugs in the setting of both sepsis and lung injury.^11-14,32 In this first randomized clinical trial evaluating aspirin for the prevention of ARDS, no effects on the primary or secondary clinical outcomes were noted. Increased IL-2 concentrations after treatment with aspirin were present on day 1, consistent with prior work suggesting increased IL-2 production following aspirin administration.³³ These results may signify a biological effect of aspirin. However, type I error cannot be excluded and any potential biological effect of aspirin was not associated with a difference in clinical outcomes nor alterations in the other 8 plasma biomarkers of inflammation, endothelial injury, and lung epithelial injury.

Quiz Ref IDIn trials of early aspirin administration for acute coronary syndrome and stroke, the excess risk of major extracranial bleeding has been estimated as a proportional increase of approximately half of the baseline absolute risk of bleeding, a risk offset by the much greater absolute benefits of treatment.³⁴ We observed similar risks for moderate or severe bleeding events; however, the results of the investigation failed to associate aspirin administration with any benefits of reduced rate of ARDS, intensity of hospital use, or mortality.

In addition to testing the specific hypotheses of this trial, we gleaned potentially useful additional information during this investigation. The study demonstrated that the conduct of a large multicenter ARDS prevention trial, though challenging, was feasible. Unlike most ARDS treatment trials in which the primary screening and enrollment activity occurs in the ICU, the primary screening environment for this investigation was the emergency department. Furthermore, the window for randomization was limited (12 hours from presentation). These constraints required meaningful modifications to the traditional screening and recruitment strategies used in prior ARDS treatment trials. Computer-assisted high-throughput screening algorithms greatly facilitated study activities at participating centers where they could be implemented. Modifications in study coordinator screening times (extending screening activities through the evenings to better capture the presentation times of the target population) also facilitated the identification of study participants.

Strengths of this investigation include the robust study design, secondary review of ARDS outcomes by an independent review panel, implementation of the LIPS risk prediction score to enrich the study population, and use of the CLIP¹⁹ to assist with standardizing important co-interventions that may have otherwise confounded the interactions of interest.

Quiz Ref IDThere were also several limitations that deserve discussion. First and foremost was the lower than expected rate of ARDS development (9.5% vs 18% expected). Although the reasons for this unexpectedly low rate of ARDS remain unclear, considerations include suboptimal performance of the LIP score, biased enrollment as a result of the informed consent process (less severely ill patients being more likely to provide consent), effectiveness of adherence with CLIP elements in reducing the risk of hospital-acquired ARDS, or temporal reductions in the rate of ARDS from the time of study planning to actual study conduct. In addition to the lower than expected rate of ARDS, low rates of mechanical ventilation, acute kidney injury, and mortality suggest that the enrolled study population may have had a more modest overall severity of illness than what was anticipated at study onset. As a result, the external validity of our findings in a cohort of critically ill patients with greater severity of illness remains unclear. Still, the results of this trial appear robust and consistent between the clinical and biomarker outcome measures.

Second, despite the intention of early identification of patients at risk for ARDS to facilitate the testing of aspirin as a prevention intervention, a large number of potential study participants (n = 1152) had bilateral infiltrates consistent with ARDS at the time of screening. This limitation highlights the challenge of pursuing ARDS prevention strategies even when targeting screening efforts to environments such as the emergency department. Also noted is the larger than expected number of patients who were excluded for prevalent antiplatelet therapy or who were thought to be at high risk for bleeding. In addition to potentially biasing the study population toward a less severely ill cohort, thereby contributing to the lower than expected rate of ARDS, these exclusions may further limit the generalizability of the study findings.

Third, the time from randomization to first drug administration was longer than anticipated at study onset. The goal was to encourage informed consent, randomization, and first study medication administration to be as accelerated as ethically possible. Delays were occasionally experienced due to the need for legally authorized representative consent as well as the lack of a 24-hour research pharmacy at many of the participating institutions.

Fourth, it is also possible the aspirin dose chosen for this study was too low. Although prior studies informed the dosing regimen used in this trial,^17,18 larger doses of aspirin or extended duration of administration may have resulted in different outcomes. Additional mechanistic studies on the effect of aspirin on thromboxane and platelet-neutrophil function are in progress to better address this question.

Among at-risk patients presenting to the emergency department, the use of aspirin compared with placebo did not reduce the risk of ARDS at 7 days. The findings of this phase 2b trial do not support continuation to a larger phase 3 trial.

Corresponding Author: Daryl J. Kor, MD, Mayo Clinic College of Medicine, Mayo Clinic, 200 First St SW, Rochester, MN 55905 (kor.daryl@mayo.edu).

Correction: This article was corrected on September 13, 2016, to clarify exclusion criteria in the flow diagram and the Results section.

Published Online: May 15, 2016. doi:10.1001/jama.2016.6330

Author Contributions: Drs Kor and Carter had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: All authors.

Acquisition, analysis, or interpretation of data: Kor, Carter, Park, Festic, Banner-Goodspeed, Hinds, Talmor, Ware, Gong.

Drafting of the manuscript: Kor, Carter, Park, Festic, Banner-Goodspeed, Hinds, Talmor, Gong.

Critical revision of the manuscript for important intellectual content: Kor, Carter, Park, Festic, Banner-Goodspeed, Talmor, Gajic, Ware, Gong.

Statistical analysis: Kor, Carter, Talmor.

Obtained funding: Kor, Hinds, Talmor, Gong.

Administrative, technical, or material support: Kor, Banner-Goodspeed, Hinds, Talmor,

Study supervision: Kor, Banner-Goodspeed, Hinds, Talmor, Gajic, Gong.

Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr Kor reports grants from the National Heart, Lung, and Blood Institute (NHLBI) during the conduct of the study as well as grants from NHLBI, personal fees from NHLBI, and personal fees from UptoDate outside the submitted work. Dr Park reports grants from NHLBI and nonfinancial support from the National Center for Advancing Translational Sciences during the conduct of the study. Mr Hinds reports grants from NHLBI during the conduct of the study. Dr Talmor reports grants from NHLBI during the conduct of the study. Dr Gajic reports grants from NHLBI during the conduct of the study. Dr Ware reports grants from the National Institutes of Health during the conduct of the study as well as consulting fees from Abbott, GlaxoSmithKline, and Hemocue, and grants from Boehringer Ingleheim and Global Blood Therapeutics. Dr Gong reports grants from NHLBI during the conduct of the study as well as grants from the Centers for Medicare & Medicaid Services, the Food and Drug Administration, and the National Institute on Aging with nonfinancial support from La Jolla Pharmaceutical outside the submitted work.

NIH NHLBI Lung Injury Prevention Study with Aspirin Network of Investigators: Beth Israel Deaconess Medical Center (D. Talmor*, V.M. Banner-Goodspeed, T.L. Henson, A.L. Mueller, V.A. Nielsen, L.V. Officer, H. Yuan), Bridgeport Hospital (D.A. Kaufman*), Brigham and Women’s Hospital (P.C. Hou*, R.E. Abdulnour, I.P. Aisiku, R.C. Dwyer, G. Frendl, R.E. Gish, E. Goralnick, T.M. Kuczmarski, B.D. Levy, S. Parmar, J.S. Rempell, R.R. Seethala, M.Q. Wilson), Duke University Medical Center (I.J. Welsby*, W.G. Drake, J. Davies), Massachusetts General Hospital (E. Bajwa*, K. Briat, K. Cosgrove, C. Holland), Mayo Clinic (E. Festic*, O. Gajic*, D.J. Kor*, R. Hinds, V. Bansal, R.K. Lingenini, T.M. Gunderson), Montefiore Medical Center (M.N. Gong*, G. Soto, S.J. Hsieh, A. Hope, M. Martinez, J. Salcedo, J. Lora), Stanford University (J.E. Levitt*, A. Asuni, R. Vojnik), Temple University (N. Marchetti*, N.T. Gentile*, K. Dehnkamp, V. Kalugdan), University of Florida (M.C. Elie-Turenne*, H. Alnuaiman, M. Plourde), University of Illinois College of Medicine at Chicago (R. Sadikot*), University of Louisville Medical Center (O. Akca*, R. Cavallazzi, M. Platt), University of Michigan (P. Park*, K.J. Brierley, J. Cherry-Bukowiec, L.M. Napolitano, K. Raghavendran, J. Younger), University of Washington- Harborview Medical Center (T.R. Watkins *, C.L. Hough*, S.A. Gundel, L. Hogl, A. Minhas, E. Tran), Wake Forest University School of Medicine (J.J. Hoth*, C. Wells, B.K. Yoza). Clinical Coordinating Center: D.S. Talmor*, V.M. Banner-Goodspeed. Data Coordinating Center: O. Gajic, D.J. Kor, R.E. Carter, R. Hinds, R.K. Lingenini, T.M. Gunderson. Biospecimen and Knowledge Translation Coordinating Center: M.N. Gong*, G. Soto. Data and Safety Monitoring Board: G. Martin (Chair), Y. Arabi, S. Carson, N. Ferguson, J. Mandrekar. Independent Medical Monitors: P.F. Clardy, M.D. Howell. National Heart, Lung, and Blood Institute: A. Harabin, P. Ghofrani.

* denotes Principal Investigator.

Funding/Support: This study is supported by NIH grants U01-HL108712-01, KL2 RR024151, K23HL112855, UL1TR000433 and the Mayo Clinic Critical Care Research Committee.

Role of the Funder/Sponsor: Neither the National Heart, Lung, and Blood Institute nor the Mayo Clinic Critical Care Research Committee had any role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.