CRITICAL CARE PERSPECTIVE Who Cares about Preventing Acute Respiratory Distress Syndrome? Gordon D. Rubenfeld Trauma, Emergency and Critical Care Program, Sunnybrook Health Sciences Centre, and Interdepartmental Division of Critical Care Medicine, University of Toronto, Toronto, Ontario, Canada
Abstract Acute respiratory distress syndrome (ARDS) is a common, lethal, and morbid respiratory complication primarily seen in the setting of major trauma and infection. Despite advances in mechanical ventilation for ARDS, many interventions have not been successful in reducing mortality. Recent grant announcements and ongoing clinical trials indicate an interest in preventing ARDS. This Perspective challenges some of the basic assumptions of ARDS prevention and preventive care in the intensive care unit. ARDS is an organ function surrogate outcome. Studies of surrogate outcomes in medicine have repeatedly failed to show an association with patient-centered outcomes that might include mortality, quality of life, patient satisfaction, and cost. Organ failure surrogate outcomes in critical care, including oxygenation, cardiac output, and blood pressure, have similarly failed to show a consistent association with patient-centered benefit. Trials designed to demonstrate an effect on surrogate outcomes will rarely be able to demonstrate small, but important, harms so that the net benefit of prevention can be calculated. This will leave clinicians with insufficient information to balance the unknown benefits of ARDS prevention with imprecisely estimated costs or risks of prevention. Because ARDS diagnosis relies on oxygenation and the chest radiograph that might be directly
The acute respiratory distress syndrome (ARDS) is a common, lethal, and morbid respiratory complication, primarily seen in the setting of major trauma and infection (1). Although there is evidence that outcomes are improving, primarily in clinical trial populations, mortality is still considerably worse than seen after myocardial infarction or embolic stroke (2, 3). Treatment for ARDS has evolved to include robust evidence that standardized low tidal volume ventilation and prone
influenced by the prophylactic intervention, studies must be designed to insure that the prevention is not merely cosmetic. Strategies that prevent ARDS need to be tested in trials sufficiently powered to demonstrate their patient-centered costs, benefits and harms before widespread adoption.
At a Glance Commentary Scientific Knowledge on the Subject: There are limited interventions demonstrated to effectively treat acute respiratory distress syndrome (ARDS). Recent grants and studies show that funders and scientists are interested in interventions to prevent ARDS. What This Study Adds to the Field: Prevention studies raise
important questions about whether the disease or syndrome is valuable to prevent. There is limited evidence in critical care that syndrome prevention addresses important patientcentered outcomes. This Perspective provides context for thinking about prevention studies in critical care using ARDS prevention as an example.
ventilation for patients with severe hypoxemia reduces mortality. Many feel that the evidence for diuretic therapy and noninvasive ventilation for selected subsets of patients with ARDS are sufficiently mature for routine implementation. Neuromuscular blockade and extracorporeal life support have early adopters’ support, and confirmatory clinical trials are underway. Despite these successes, there is a general feeling of disappointment from the many clinical trials targeted at
reducing mortality in ARDS (4). It may be frustration from these statistically negative treatment trials supported by evidence linking routine treatments with reduced rates of ARDS (5) and, no doubt, the adage “an ounce of prevention is worth a pound of cure,” that account for increased interest in ARDS prevention. This is supported by the recent National Institutes of Health Prevention and Early Treatment of Acute Lung Injury network (6), a series of grants from the U.S. Department of Defense that
( Received in original form August 31, 2014; accepted in final form December 4, 2014 ) Correspondence and requests for reprints should be addressed to Gordon D. Rubenfeld, M.D., M.Sc., Trauma, Emergency and Critical Care Program, Sunnybrook Health Sciences Centre, Department of Medicine, University of Toronto, Toronto, ON, M4V1E5 Canada. E-mail: [email protected] Am J Respir Crit Care Med Vol 191, Iss 3, pp 255–260, Feb 1, 2015 Copyright © 2015 by the American Thoracic Society Originally Published in Press as DOI: 10.1164/rccm.201408-1574CP on December 5, 2014 Internet address: www.atsjournals.org
Critical Care Perspective
255
CRITICAL CARE PERSPECTIVE include lung injury prevention (7), and the ongoing Lung Injury Prevention Studies (8). Researchers and funders evidently believe that ARDS is worth preventing. In this Perspective, I challenge this basic assumption and also point out some fundamental limitations of preventive critical care. I do not review the biologic rationale for potential therapeutic targets or whether there is a persuasive argument that we can prevent an inflammatory cascade when we have had limited success in aborting it. Rather, I consider the entire enterprise of prevention from a trial design and interpretation perspective. First, I argue that ARDS is a surrogate endpoint, and that surrogate endpoints are notoriously unreliable in medicine and particularly so in critical care. Second, I illustrate how surrogate endpoints are difficult to interpret, because they leave clinicians with insufficient information to balance harms and benefits from trials that are underpowered to provide these data. Finally, I discuss the unique challenges that ARDS diagnosis presents when it is used as an outcome variable, particularly when some preventive interventions might have a purely cosmetic effect on detection.
ARDS Is a Surrogate Endpoint Every outcome in a clinical trial that is not a patient-centered outcome, such as mortality or quality of life, or diseases directly in the causal pathway to a patientcentered outcome, for example, stroke or myocardial infarction, is a surrogate endpoint (9, 10). Surrogate outcomes are usually laboratory, physiologic, or imaging data that are linked to patient-centered outcomes, such that the surrogate and patient-centered outcome move together. Common surrogate outcomes include hemodynamic measures, cholesterol levels, blood pressure, and tumor size. Although surrogate outcomes may be used in phase II trials to understand mechanisms of disease and to identify potentially effective treatments, there is an important difference between their use in different types of studies (11). When surrogate outcomes are used in trials to inform clinical practice, it is because they can be measured earlier, more easily, and will detect effects with smaller sample sizes than patient-centered outcomes. It is the use of surrogate 256
outcomes in trials meant to inform clinical practice where they have the greatest potential to mislead.
Surrogate Outcomes Do Not Predict Clinically Important Responses Unfortunately, the overwhelming experience in research is that surrogate outcomes are unreliable. The classic studies showing a failure of surrogate outcomes to predict patient-centered benefit were the Cardiac Arrythmia Suppresion Trials (CAST) (12, 13). In the 1980s, it was clear that survivors of myocardial infarction with high rates of premature ventricular contractions (PVC) had an independent increased risk of mortality. It seemed biologically plausible that the PVCs triggered episodes of lethal arrhythmia, and that suppressing them would improve mortality. The CAST trial was designed to test whether treating the “disease” of frequent PVCs had any effect on patient-centered outcomes. It demonstrated that, whereas three different antiarrhythmic regimens successfully reduced PVCs, they increased mortality compared with placebo. These studies showed that clinicians could not assume that preventing the surrogate of PVCs would necessarily prevent death, despite a strong biologic rationale and considerable evidence linking PVCs with death. In years since the CAST results, surrogate endpoints have not improved their stature. Even in cardiology and oncology, where the biologic link between surrogates, such as LDL, blood pressure, tumor size, or progression-free survival and important clinical outcomes, is quite strong, these outcomes are viewed skeptically (14, 15). It is not particularly surprising then that surrogate outcomes, and specifically organ function surrogate outcomes, have fared poorly in critical care. Nitric oxide synthase inhibitor and corticosteroids improve blood pressure in septic shock, but have no effect on or worsen mortality (16, 17). Renal-dose dopamine and nesiritide increase urine output or reduce pulmonary wedge pressure, but have no effect on or worsen mortality (18–20). Large tidal volumes and inhaled nitric oxide improve oxygenation in ARDS, but have no effect on or worsen mortality, and inhaled nitric oxide may cause renal failure (21, 22). Given this experience, why should we have
any confidence that preventing ARDS will improve clinically important outcomes? Certainly most, but not all, studies show that, after adjusting for risk factor and severity of illness, patients who develop ARDS have significantly higher mortality than those who do not (23–25). In addition, one of the few studies to compare patients with ARDS with an at-risk control population found that patients with ARDS had worse health-related quality of life (26). However, the majority of patients with ARDS do not die from their ARDS, or at least from hypoxemia, but from organ failure associated with the underlying risk factor, or possibly ventilator-induced organ failure from management of their ARDS (27). Patients with acute respiratory failure who do not meet the criteria for ARDS have substantial mortality, suggesting that preventing ARDS in intubated patients may leave them at considerable risk for death from other causes (28). In a careful prospective cohort of elderly patients, mechanical ventilation did not worsen the decline in cognitive function associated with sepsis, suggesting that preventing mechanical ventilation may not eliminate these sequelae (29). When compared with matched controls without ARDS, long-term survival was determined by risk factor for ARDS and not by ARDS (30). Finally, although avoiding intensive care unit (ICU) admission is almost certainly cost saving, the marginal cost savings of avoiding intubation or preventing ARDS in patients already in the ICU may be small (31). Therefore, the evidence from observational data is mixed on the attributable mortality, morbidity, and cost from ARDS. Because interventions to prevent ARDS will have their own potential costs and risks, it is simply unknown what overall effect they might have on patient-centered outcomes.
Lessons from Other Prevention Studies in Critical Care If clinical trials to prevent ARDS are successful, we can learn how that evidence might be used by looking to the large body of evidence demonstrating effective prevention of gastrointestinal bleeding, ventilator-associated pneumonia (VAP), venous thromboembolism, and catheter related bacteremia (32–35). These surrogate outcomes are also presumed to be in the causal pathway to morbidity and death in
American Journal of Respiratory and Critical Care Medicine Volume 191 Number 3 | February 1 2015
CRITICAL CARE PERSPECTIVE the ICU. However, one of the important lessons from the ICU prevention literature is how difficult it is to make inferences about attributable effects from observational studies. Confounding, competing, and time-varying risks all make simple comparisons between those with and without one of these surrogate outcomes biased (36). For example, accurate estimates of the effect of VAP on length of stay must account for the fact that patients acquire VAP after being in the ICU for a period of time. Not surprisingly, the estimates of the associated patient benefit from prevention vary with the analytic approach, and some recent analyses find relatively low estimates of attributable effect (37, 38). The limitations of observational studies in teasing out the attributable effects of surrogate outcomes can be addressed with randomized trials that link prevention with improvement in patient-centered outcomes. However, prevention trials targeting surrogate outcomes are rarely powered to demonstrate a difference in patient-centered outcome and, even when these studies are large, rarely show an effect on mortality. For example, a clinical trial with 9,417 patients showed that silver-coated endotracheal tubes reduced VAP by nearly 36%, but had no statistically significant effect on any patient-centered outcomes (39). If a statistically significant effect on mortality is noted in an underpowered prevention trial, the mortality signal is more likely to be due to chance and less likely to be reproducible than in an appropriately powered trial (40, 41). It is understandable why prevention trials rarely find mortality effects. A recent cohort study showed that 3.5% of a mixed ICU population suffered from clinically important gastrointestinal bleeding, and that that bleeding was associated with a 30% increase in mortality (42). Under the most optimistic assumptions that prevention could eliminate all clinically important gastrointestinal bleeding and that this reduced the 30% attributable mortality to zero and that the treatment had no side effects, then prevention would reduce mortality by a 1% difference. Using the same optimistic assumptions, eliminating clinically important gastrointestinal bleeding would reduce length of stay by less than 1 day in treated patients. A typical ICU clinical trial would require 40,000 patients to demonstrate Critical Care Perspective
effects this small. ARDS prevention trials will suffer from the same sample size problems. The Lung Injury Prevention with Aspirin trial is powered to detect a 10% absolute reduction in the risk of ARDS (8). If we make the optimistic assumptions that aspirin can reduce mortality in patients from 23% if they developed ARDS to 0% if prevented and that the treatment has no side effects that could increase mortality, a mortality difference of 2.3% would be observed in this prevention trial. Depending on the mortality in the control arm, it would require a sample size of about 7,000 patients to detect this effect. Because it is unlikely that ARDS prevention reduces the risk of death to zero, realistic assumptions about the mortality benefit of ARDS prevention would require substantially larger trials to detect differences in mortality.
Surrogate Outcome Studies Yield Results That Are Difficult to Apply We might be satisfied relying on these hypothetical estimates of the benefits to patients of ARDS prevention if we knew that they outweighed the potential risks. Unfortunately, clinical trials of surrogate outcomes that are underpowered to detect small mortality benefits are also underpowered to detect small risks. These risks are usually uncovered in nonrandomized studies. For example, clinical trials of acid suppression for gastrointestinal bleeding in the ICU did not disclose the potential for delirium and nosocomial infections demonstrated subsequently (43–45). This leaves clinicians balancing the theoretic, but presumed, benefit of preventing a surrogate outcome, like gastrointestinal bleeding, against these potential harms. The relative benefit and costs to patients of these tradeoffs may be quite small (46, 47). The challenge of applying the results of surrogate outcome trials is illustrated nicely by the contradictory interpretation of two important ARDS Network trials: the Methylprednisolone for Persistent ARDS and the Fluid-Management Strategies trials (48, 49). Late steroid therapy for persistent ARDS resulted in 4.4 additional ventilatorfree days compared with placebo, whereas fluid limitation and diuretics resulted in 2.5 additional ventilator-free days. Despite
more impressive results in terms of ventilator-free days and confirmatory improvements in oxygenation and compliance, Methylprednisolone for Persistent ARDS was interpreted as a negative study with potentially harmful effects. Although neither trial showed a statistically significant difference in their primary mortality outcomes, a post hoc analysis revealed that patients treated with corticosteroids after 2 weeks of ARDS had a statistically significantly higher mortality. Safety signals for infection and reintubation were also worrisome, leading the authors to advise against the use of corticosteroids. Corticosteroid advocates might emphasize the physiologic improvement and larger effect size, and discount the safety concerns, which were not translated into a net effect on ventilator-free days, a surrogate selected presumably because it captured relevant harms as well as benefits. To complicate matters more, a subsequent study demonstrated unexpected safety concerns with the fluid-management strategy, indicating that cognitive impairment might be associated with the intervention (50). These trials demonstrate how difficult it is to apply evidence from surrogate outcomes, such as ventilator-free days. Without an important patient-centered signal, clinicians need to weigh the benefits of increased ventilator-free days against the harms of reintubation, infection, and cognitive impairment. If one of these treatments had been a novel and expensive drug therapy with additional unknown risks and high cost, the decision becomes even more complex. It is easy to imagine preventive therapies for ARDS that might increase costs, delirium, bleeding, or nosocomial infection that would put physicians in the same quandary. For example, imagine that aspirin administered to at-risk patients reduces the risk of ARDS by 10%. This would be an impressive result, meaning that only 10 patients need to be treated to prevent 1 case of ARDS. What if the excess major bleeding risk was 1%? This means that for every 10 ARDS cases prevented, we cause 1 major bleed. How many fewer cases of ARDS justify an iatrogenic major bleed? Intensivists may think that “avoiding intubation” or “avoiding ARDS” are inherently beneficial to patients in the same way that avoiding cancer, stroke, or myocardial infarction are assumed to be. However, in these cases, the effects of these 257
CRITICAL CARE PERSPECTIVE diseases are well known, and there are large datasets of clinical trials and observational cohorts to estimate net benefit not available in critical illness (51). Despite the considerable data available to them, our colleagues in primary care still face a dilemma counseling patients about the tradeoffs involved in taking aspirin as primary prophylaxis for myocardial infarction, or using chronic anticoagulation to prevent stroke, or even screening for cancer (52, 53). At least in the outpatient setting, patients can weigh the outcomes and make their own decisions. In the ICU, clinicians or expert guideline panels will have to weigh these intangible costs, risks, and benefits in making decisions for patients with a much smaller evidence base.
The Syndromic Nature of ARDS Complicates Its Use as an Outcome The challenges of ARDS diagnosis and the heterogeneity of the syndrome are often invoked as explanations for slow progression in the field. Enrolling patients without “real” ARDS, whatever that entity is, in treatment trials with mortality as an outcome simply reduces the power of the study, but does not bias it. Similarly, enrolling patients with ARDS in a prevention trial who cannot have it prevented reduces the power of the trial, but does not bias it. Standardizing the diagnostic criteria for enrolment, as was done in the recent pirfenidone trials for pulmonary fibrosis, is a recommended strategy for both types of trials to increase the expected signal (54). However, the ARDS diagnostic criteria present fundamentally different problems when used to define the endpoint for a study. Prerandomization variation in
enrolment diagnosis introduces noise; postrandomization differences in outcome diagnosis can create bias. The syndrome relies on the chest radiograph and PaO2/FIO2 ratio that have both been shown to be treatment dependent and unreliable (55, 56). In theory, a patient could develop the pathophysiology of the syndrome, as unclear as that is, and not be diagnosed with ARDS if the preventive intervention directly affects gas exchange or the radiograph. Topical antibiotic prevention strategies for VAP have generated the same concerns. Antibiotic concentrations in tracheal secretions may be more effective at preventing the diagnosis of VAP than the disease (57). Analogous examples in ARDS, including inhaled nitric oxide, lung recruitment, differential ascertainment of the oxygenation status by pulse oximetry versus arterial blood gas, and assessment of the PaO2/FIO2 ratio in nonintubated patients, with inaccurate FIO2 measurement, could all affect the diagnosis of ARDS (58–60). Of course, any of these interventions might modify the underlying mechanisms of disease, but studies that rely solely on ARDS as the primary outcome will not be able to distinguish a clinically significant effect on disease process from a purely cosmetic effect on disease diagnosis. Study design modifications can minimize the potential for bias. These would include blinding the treating clinician and blinded central adjudication of the diagnosis of ARDS. Standardized ventilator settings and protocolized screening for the diagnosis of ARDS, an option for enrolment in treatment trials, may be mandatory for outcome measurement in prevention trials, particularly if the intervention is unblinded.
References 1. Rubenfeld GD, Caldwell E, Peabody E, Weaver J, Martin DP, Neff M, Stern EJ, Hudson LD. Incidence and outcomes of acute lung injury. N Engl J Med 2005;353:1685–1693. 2. Zambon M, Vincent JL. Mortality rates for patients with acute lung injury/ ARDS have decreased over time. Chest 2008;133:1120–1127. 3. Phua J, Badia JR, Adhikari NK, Friedrich JO, Fowler RA, Singh JM, Scales DC, Stather DR, Li A, Jones A, et al. Has mortality from acute respiratory distress syndrome decreased over time? A systematic review. Am J Respir Crit Care Med 2009;179:220–227. 4. Duggal A, Ganapathy A, Mohana Ratnapalan M, Adhikari NK. Pharmacological treatments for acute respiratory distress syndrome: systematic review. Minerva Anestesiol (In press)
258
Who Cares about ARDS Prevention? ARDS is associated with significant morbidity and mortality in patients with common risk factors, such as trauma and sepsis. It is clear from ongoing trials and grant announcements that researchers care about preventing ARDS. However, it is simply unknown whether clinicians, patients, or payers should care. It is difficult to deny that, all things being equal, it would be better for a patient to avoid getting ARDS. However, all things are rarely equal. Clinicians must appreciate that risks are unlikely to be fully disclosed by prevention trials using surrogate outcomes. Therefore, there will be insufficient information to make informed decisions about the tradeoff between costs, risks, and benefits in implementing this evidence. Of course, we can do ARDS prevention when it is a beneficial side effect of treatments that we should already be implementing. Examples of these would be aspiration precautions, early resuscitation in sepsis, limiting blood products, and lung-protective ventilation (61). Positive, well designed prevention studies of ARDS can yield valuable insights into the pathophysiology of the syndrome and its heterogeneity; however, novel strategies that appear to effectively prevent ARDS in phase II studies must be tested in trials sufficiently powered to demonstrate their patient-centered costs, benefits, and harms before widespread adoption. If these trials are deemed unfeasible, due to the large sample sizes required, then one must wonder whether the benefits to patients are likely to be sufficiently important to justify the entire endeavor. n Author disclosures are available with the text of this article at www.atsjournals.org.
5. Gajic O, Dabbagh O, Park PK, Adesanya A, Chang SY, Hou P, Anderson H III, Hoth JJ, Mikkelsen ME, Gentile NT, et al.; U.S. Critical Illness and Injury Trials Group: Lung Injury Prevention Study Investigators (USCIITG-LIPS). Early identification of patients at risk of acute lung injury: evaluation of lung injury prediction score in a multicenter cohort study. Am J Respir Crit Care Med 2011;183: 462–470. 6. Clinical Centers for the NHLBI Prevention and Early Treatment of Acute Lung Injury (PETAL) Clinical Trials Network. May 15, 2013 [accessed 2014 Aug 20]. Available from: http://grants.nih.gov/grants/guide/rfafiles/RFA-HL-14-014.html 7. FY14 Peer Reviewed Medical Research Program Clinical Trial Award. 2014 [accessed 2014 Aug 20]. Available from: http://cdmrp.army.mil/ funding/pa/14prmrpcta_pa.pdf
American Journal of Respiratory and Critical Care Medicine Volume 191 Number 3 | February 1 2015
CRITICAL CARE PERSPECTIVE 8. Kor DJ, Talmor DS, Banner-Goodspeed VM, Carter RE, Hinds R, Park PK, Gajic O, Gong MN; US Critical Illness and Injury Trials Group: Lung Injury Prevention with Aspirin Study Group (USCIITG: LIPS-A). Lung Injury Prevention with Aspirin (LIPS-A): a protocol for a multicentre randomised clinical trial in medical patients at high risk of acute lung injury. BMJ Open 2012;2:e001606. 9. Prentice RL. Surrogate endpoints in clinical trials: definition and operational criteria. Stat Med 1989;8:431–440. 10. Rubenfeld GD. Surrogate measures of patient-centered outcomes in critical care. In: Angus DC, Carlet J, editors. Surviving intensive care. Berlin: Springer-Verlag; 2002. 11. Rubenfeld GD, Abraham E. When is a negative phase II trial truly negative? Am J Respir Crit Care Med 2008;178:554–555. 12. Echt DS, Liebson PR, Mitchell LB, Peters RW, Obias-Manno D, Barker AH, Arensberg D, Baker A, Friedman L, Greene HL, et al. Mortality and morbidity in patients receiving encainide, flecainide, or placebo: the Cardiac Arrhythmia Suppression Trial. N Engl J Med 1991; 324:781–788. 13. The Cardiac Arrhythmia Suppression Trial II Investigators. Effect of the antiarrhythmic agent moricizine on survival after myocardial infarction. N Engl J Med 1992;327:227–233. 14. Psaty BM, Lumley T. Surrogate end points and FDA approval: a tale of 2 lipid-altering drugs. JAMA 2008;299:1474–1476. 15. Schatzkin A, Gail M. The promise and peril of surrogate end points in cancer research. Nat Rev Cancer 2002;2:19–27. 16. Lopez ´ A, Lorente JA, Steingrub J, Bakker J, McLuckie A, Willatts S, Brockway M, Anzueto A, Holzapfel L, Breen D, et al. Multiple-center, randomized, placebo-controlled, double-blind study of the nitric oxide synthase inhibitor 546C88: effect on survival in patients with septic shock. Crit Care Med 2004;32:21–30. 17. Sprung CL, Annane D, Keh D, Moreno R, Singer M, Freivogel K, Weiss YG, Benbenishty J, Kalenka A, Forst H, et al.; CORTICUS Study Group. Hydrocortisone therapy for patients with septic shock. N Engl J Med 2008;358:111–124. 18. Friedrich JO, Adhikari N, Herridge MS, Beyene J. Meta-analysis: lowdose dopamine increases urine output but does not prevent renal dysfunction or death. Ann Intern Med 2005;142:510–524. 19. O’Connor CM, Starling RC, Hernandez AF, Armstrong PW, Dickstein K, Hasselblad V, Heizer GM, Komajda M, Massie BM, McMurray JJ, et al. Effect of nesiritide in patients with acute decompensated heart failure. N Engl J Med 2011;365:32–43. [Published erratum appears in N Engl J Med 365:773. Wilson WH corrected to Tang WHW.] 20. Sackner-Bernstein JD, Kowalski M, Fox M, Aaronson K. Short-term risk of death after treatment with nesiritide for decompensated heart failure: a pooled analysis of randomized controlled trials. JAMA 2005;293:1900–1905. 21. Adhikari NK, Burns KE, Friedrich JO, Granton JT, Cook DJ, Meade MO. Effect of nitric oxide on oxygenation and mortality in acute lung injury: systematic review and meta-analysis. BMJ 2007;334:779. 22. The Acute Respiratory Distress Syndrome Network. Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome. N Engl J Med 2000;342:1301–1308. 23. Treggiari MM, Hudson LD, Martin DP, Weiss NS, Caldwell E, Rubenfeld G. Effect of acute lung injury and acute respiratory distress syndrome on outcome in critically ill trauma patients. Crit Care Med 2004;32: 327–331. 24. Shah CV, Localio AR, Lanken PN, Kahn JM, Bellamy S, Gallop R, Finkel B, Gracias VH, Fuchs BD, Christie JD. The impact of development of acute lung injury on hospital mortality in critically ill trauma patients. Crit Care Med 2008;36:2309–2315. 25. Mikkelsen ME, Shah CV, Meyer NJ, Gaieski DF, Lyon S, Miltiades AN, Goyal M, Fuchs BD, Bellamy SL, Christie JD. The epidemiology of acute respiratory distress syndrome in patients presenting to the emergency department with severe sepsis. Shock 2013;40:375–381. 26. Davidson TA, Caldwell ES, Curtis JR, Hudson LD, Steinberg KP. Reduced quality of life in survivors of acute respiratory distress syndrome compared with critically ill control patients. JAMA 1999; 281:354–360. 27. Stapleton RD, Wang BM, Hudson LD, Rubenfeld GD, Caldwell ES, Steinberg KP. Causes and timing of death in patients with ARDS. Chest 2005;128:525–532.
Critical Care Perspective
28. Luhr OR, Antonsen K, Karlsson M, Aardal S, Thorsteinsson A, Frostell CG, Bonde J; The ARF Study Group. Incidence and mortality after acute respiratory failure and acute respiratory distress syndrome in Sweden, Denmark, and Iceland. Am J Respir Crit Care Med 1999; 159:1849–1861. 29. Iwashyna TJ, Ely EW, Smith DM, Langa KM. Long-term cognitive impairment and functional disability among survivors of severe sepsis. JAMA 2010;304:1787–1794. 30. Davidson TA, Rubenfeld GD, Caldwell ES, Hudson LD, Steinberg KP. The effect of acute respiratory distress syndrome on long-term survival. Am J Respir Crit Care Med 1999;160:1838–1842. 31. Cooke CR. Economics of mechanical ventilation and respiratory failure. Crit Care Clin 2012;28:39–55 (vi.). 32. Krag M, Perner A, Wetterslev J, Wise MP, Hylander Møller M. Stress ulcer prophylaxis versus placebo or no prophylaxis in critically ill patients: a systematic review of randomised clinical trials with metaanalysis and trial sequential analysis. Intensive Care Med 2014;40: 11–22. 33. Dodek P, Keenan S, Cook D, Heyland D, Jacka M, Hand L, Muscedere J, Foster D, Mehta N, Hall R, et al.; Canadian Critical Care Trials Group; Canadian Critical Care Society. Evidence-based clinical practice guideline for the prevention of ventilator-associated pneumonia. Ann Intern Med 2004;141:305–313. 34. Alhazzani W, Lim W, Jaeschke RZ, Murad MH, Cade J, Cook DJ. Heparin thromboprophylaxis in medical-surgical critically ill patients: a systematic review and meta-analysis of randomized trials. Crit Care Med 2013;41:2088–2098. 35. Blot K, Bergs J, Vogelaers D, Blot S, Vandijck D. Prevention of central line–associated bloodstream infections through quality improvement interventions: a systematic review and meta-analysis. Clin Infect Dis 2014;59:96–105. 36. Muñoz A, Mongilardi N, Checkley W. Multilevel competing risks in the evaluation of nosocomial infections: time to move on from proportional hazards and even from hazards altogether. Crit Care 2014;18:146. 37. Bekaert M, Timsit JF, Vansteelandt S, Depuydt P, Vesin ´ A, GarrousteOrgeas M, Decruyenaere J, Clec’h C, Azoulay E, Benoit D; Outcomerea Study Group. Attributable mortality of ventilatorassociated pneumonia: a reappraisal using causal analysis. Am J Respir Crit Care Med 2011;184:1133–1139. 38. Melsen WG, Rovers MM, Koeman M, Bonten MJ. Estimating the attributable mortality of ventilator-associated pneumonia from randomized prevention studies. Crit Care Med 2011;39:2736–2742. 39. Kollef MH, Afessa B, Anzueto A, Veremakis C, Kerr KM, Margolis BD, Craven DE, Roberts PR, Arroliga AC, Hubmayr RD, et al.; NASCENT Investigation Group. Silver-coated endotracheal tubes and incidence of ventilator-associated pneumonia: the NASCENT randomized trial. JAMA 2008;300:805–813. 40. Browner WS, Newman TB. Are all significant P values created equal? The analogy between diagnostic tests and clinical research. JAMA 1987;257:2459–2463. 41. Christley RM. Power and error: increased risk of false positive results in underpowered studies. Open Epidemiol J 2010;3:16–19. 42. Cook DJ, Griffith LE, Walter SD, Guyatt GH, Meade MO, Heyland DK, Kirby A, Tryba M; Canadian Critical Care Trials Group. The attributable mortality and length of intensive care unit stay of clinically important gastrointestinal bleeding in critically ill patients. Crit Care 2001;5:368–375. 43. Eom CS, Jeon CY, Lim JW, Cho EG, Park SM, Lee KS. Use of acidsuppressive drugs and risk of pneumonia: a systematic review and meta-analysis. CMAJ 2011;183:310–319. 44. Howell MD, Novack V, Grgurich P, Soulliard D, Novack L, Pencina M, Talmor D. Iatrogenic gastric acid suppression and the risk of nosocomial Clostridium difficile infection. Arch Intern Med 2010;170: 784–790. 45. Fujii S, Tanimukai H, Kashiwagi Y. Comparison and analysis of delirium induced by histamine h(2) receptor antagonists and proton pump inhibitors in cancer patients. Case Rep Oncol 2012;5:409–412. 46. Kahn JM, Doctor JN, Rubenfeld GD. Stress ulcer prophylaxis in mechanically ventilated patients: integrating evidence and judgment using a decision analysis. Intensive Care Med 2006;32: 1151–1158.
259
CRITICAL CARE PERSPECTIVE 47. Chiasson TC, Manns BJ, Stelfox HT. An economic evaluation of venous thromboembolism prophylaxis strategies in critically ill trauma patients at risk of bleeding. PLoS Med 2009;6:e1000098. 48. Wiedemann HP, Wheeler AP, Bernard GR, Thompson BT, Hayden D, deBoisblanc B, Connors AF Jr, Hite RD, Harabin AL; National Heart, Lung, and Blood Institute Acute Respiratory Distress Syndrome (ARDS) Clinical Trials Network. Comparison of two fluidmanagement strategies in acute lung injury. N Engl J Med 2006;354: 2564–2575. 49. Steinberg KP, Hudson LD, Goodman RB, Hough CL, Lanken PN, Hyzy R, Thompson BT, Ancukiewicz M; National Heart, Lung, and Blood Institute Acute Respiratory Distress Syndrome (ARDS) Clinical Trials Network. Efficacy and safety of corticosteroids for persistent acute respiratory distress syndrome. N Engl J Med 2006;354:1671–1684. 50. Mikkelsen ME, Christie JD, Lanken PN, Biester RC, Thompson BT, Bellamy SL, Localio AR, Demissie E, Hopkins RO, Angus DC. The adult respiratory distress syndrome cognitive outcomes study: longterm neuropsychological function in survivors of acute lung injury. Am J Respir Crit Care Med 2012;185:1307–1315. 51. US Preventive Services Task Force (USPSTF). Estimating certainty and magnitude of net benefit. 2007 [accessed 2014 Aug 28]. Available from: http://www.uspreventiveservicestaskforce.org/uspstf07/ methods/benefit.htm 52. US Preventive Services Task Force. Aspirin for the prevention of cardiovascular disease: U.S. Preventive Services Task Force recommendation statement. Ann Intern Med 2009;150:396–404. 53. Protheroe J, Fahey T, Montgomery AA, Peters TJ. The impact of patients’ preferences on the treatment of atrial fibrillation: observational study of patient based decision analysis. BMJ 2000; 320:1380–1384.
260
54. King TE Jr, Bradford WZ, Castro-Bernardini S, Fagan EA, Glaspole I, Glassberg MK, Gorina E, Hopkins PM, Kardatzke D, Lancaster L, et al.; ASCEND Study Group. A phase 3 trial of pirfenidone in patients with idiopathic pulmonary fibrosis. N Engl J Med 2014;370: 2083–2092. 55. Ferguson ND, Fan E, Camporota L, Antonelli M, Anzueto A, Beale R, Brochard L, Brower R, Esteban A, Gattinoni L, et al. The Berlin definition of ARDS: an expanded rationale, justification, and supplementary material. Intensive Care Med 2012;38:1573–1582. 56. Ranieri VM, Rubenfeld GD, Thompson BT, Ferguson ND, Caldwell E, Fan E, Camporota L, Slutsky AS; ARDS Definition Task Force. Acute respiratory distress syndrome: the Berlin Definition. JAMA 2012;307:2526–2533. 57. Schultz MJ, de Jonge E, Kesecioglu J. Selective decontamination of the digestive tract reduces mortality in critically ill patients. Crit Care 2003;7:107–110. 58. Langevin PB, Hellein V, Harms SM, Tharp WK, Cheung-Seekit C, Lampotang S. Synchronization of radiograph film exposure with the inspiratory pause: effect on the appearance of bedside chest radiographs in mechanically ventilated patients. Am J Respir Crit Care Med 1999;160:2067–2071. 59. Goldstein RS, Young J, Rebuck AS. Effect of breathing pattern on oxygen concentration received from standard face masks. Lancet 1982;2:1188–1190. 60. Rice TW, Wheeler AP, Bernard GR, Hayden DL, Schoenfeld DA, Ware LB; National Institutes of Health, National Heart, Lung, and Blood Institute ARDS Network. Comparison of the SpO2/FIO2 ratio and the PaO2/FIO2 ratio in patients with acute lung injury or ARDS. Chest 2007;132:410–417. 61. Levitt JE, Matthay MA. Clinical review: early treatment of acute lung injury—paradigm shift toward prevention and treatment prior to respiratory failure. Crit Care 2012;16:223.
American Journal of Respiratory and Critical Care Medicine Volume 191 Number 3 | February 1 2015
Abstract Acute respiratory distress syndrome (ARDS) is a common, lethal, and morbid respiratory complication primarily seen in the setting of major trauma and infection. Despite advances in mechanical ventilation for ARDS, many interventions have not been successful in reducing mortality. Recent grant announcements and ongoing clinical trials indicate an interest in preventing ARDS. This Perspective challenges some of the basic assumptions of ARDS prevention and preventive care in the intensive care unit. ARDS is an organ function surrogate outcome. Studies of surrogate outcomes in medicine have repeatedly failed to show an association with patient-centered outcomes that might include mortality, quality of life, patient satisfaction, and cost. Organ failure surrogate outcomes in critical care, including oxygenation, cardiac output, and blood pressure, have similarly failed to show a consistent association with patient-centered benefit. Trials designed to demonstrate an effect on surrogate outcomes will rarely be able to demonstrate small, but important, harms so that the net benefit of prevention can be calculated. This will leave clinicians with insufficient information to balance the unknown benefits of ARDS prevention with imprecisely estimated costs or risks of prevention. Because ARDS diagnosis relies on oxygenation and the chest radiograph that might be directly
The acute respiratory distress syndrome (ARDS) is a common, lethal, and morbid respiratory complication, primarily seen in the setting of major trauma and infection (1). Although there is evidence that outcomes are improving, primarily in clinical trial populations, mortality is still considerably worse than seen after myocardial infarction or embolic stroke (2, 3). Treatment for ARDS has evolved to include robust evidence that standardized low tidal volume ventilation and prone
influenced by the prophylactic intervention, studies must be designed to insure that the prevention is not merely cosmetic. Strategies that prevent ARDS need to be tested in trials sufficiently powered to demonstrate their patient-centered costs, benefits and harms before widespread adoption.
At a Glance Commentary Scientific Knowledge on the Subject: There are limited interventions demonstrated to effectively treat acute respiratory distress syndrome (ARDS). Recent grants and studies show that funders and scientists are interested in interventions to prevent ARDS. What This Study Adds to the Field: Prevention studies raise
important questions about whether the disease or syndrome is valuable to prevent. There is limited evidence in critical care that syndrome prevention addresses important patientcentered outcomes. This Perspective provides context for thinking about prevention studies in critical care using ARDS prevention as an example.
ventilation for patients with severe hypoxemia reduces mortality. Many feel that the evidence for diuretic therapy and noninvasive ventilation for selected subsets of patients with ARDS are sufficiently mature for routine implementation. Neuromuscular blockade and extracorporeal life support have early adopters’ support, and confirmatory clinical trials are underway. Despite these successes, there is a general feeling of disappointment from the many clinical trials targeted at
reducing mortality in ARDS (4). It may be frustration from these statistically negative treatment trials supported by evidence linking routine treatments with reduced rates of ARDS (5) and, no doubt, the adage “an ounce of prevention is worth a pound of cure,” that account for increased interest in ARDS prevention. This is supported by the recent National Institutes of Health Prevention and Early Treatment of Acute Lung Injury network (6), a series of grants from the U.S. Department of Defense that
( Received in original form August 31, 2014; accepted in final form December 4, 2014 ) Correspondence and requests for reprints should be addressed to Gordon D. Rubenfeld, M.D., M.Sc., Trauma, Emergency and Critical Care Program, Sunnybrook Health Sciences Centre, Department of Medicine, University of Toronto, Toronto, ON, M4V1E5 Canada. E-mail: [email protected] Am J Respir Crit Care Med Vol 191, Iss 3, pp 255–260, Feb 1, 2015 Copyright © 2015 by the American Thoracic Society Originally Published in Press as DOI: 10.1164/rccm.201408-1574CP on December 5, 2014 Internet address: www.atsjournals.org
Critical Care Perspective
255
CRITICAL CARE PERSPECTIVE include lung injury prevention (7), and the ongoing Lung Injury Prevention Studies (8). Researchers and funders evidently believe that ARDS is worth preventing. In this Perspective, I challenge this basic assumption and also point out some fundamental limitations of preventive critical care. I do not review the biologic rationale for potential therapeutic targets or whether there is a persuasive argument that we can prevent an inflammatory cascade when we have had limited success in aborting it. Rather, I consider the entire enterprise of prevention from a trial design and interpretation perspective. First, I argue that ARDS is a surrogate endpoint, and that surrogate endpoints are notoriously unreliable in medicine and particularly so in critical care. Second, I illustrate how surrogate endpoints are difficult to interpret, because they leave clinicians with insufficient information to balance harms and benefits from trials that are underpowered to provide these data. Finally, I discuss the unique challenges that ARDS diagnosis presents when it is used as an outcome variable, particularly when some preventive interventions might have a purely cosmetic effect on detection.
ARDS Is a Surrogate Endpoint Every outcome in a clinical trial that is not a patient-centered outcome, such as mortality or quality of life, or diseases directly in the causal pathway to a patientcentered outcome, for example, stroke or myocardial infarction, is a surrogate endpoint (9, 10). Surrogate outcomes are usually laboratory, physiologic, or imaging data that are linked to patient-centered outcomes, such that the surrogate and patient-centered outcome move together. Common surrogate outcomes include hemodynamic measures, cholesterol levels, blood pressure, and tumor size. Although surrogate outcomes may be used in phase II trials to understand mechanisms of disease and to identify potentially effective treatments, there is an important difference between their use in different types of studies (11). When surrogate outcomes are used in trials to inform clinical practice, it is because they can be measured earlier, more easily, and will detect effects with smaller sample sizes than patient-centered outcomes. It is the use of surrogate 256
outcomes in trials meant to inform clinical practice where they have the greatest potential to mislead.
Surrogate Outcomes Do Not Predict Clinically Important Responses Unfortunately, the overwhelming experience in research is that surrogate outcomes are unreliable. The classic studies showing a failure of surrogate outcomes to predict patient-centered benefit were the Cardiac Arrythmia Suppresion Trials (CAST) (12, 13). In the 1980s, it was clear that survivors of myocardial infarction with high rates of premature ventricular contractions (PVC) had an independent increased risk of mortality. It seemed biologically plausible that the PVCs triggered episodes of lethal arrhythmia, and that suppressing them would improve mortality. The CAST trial was designed to test whether treating the “disease” of frequent PVCs had any effect on patient-centered outcomes. It demonstrated that, whereas three different antiarrhythmic regimens successfully reduced PVCs, they increased mortality compared with placebo. These studies showed that clinicians could not assume that preventing the surrogate of PVCs would necessarily prevent death, despite a strong biologic rationale and considerable evidence linking PVCs with death. In years since the CAST results, surrogate endpoints have not improved their stature. Even in cardiology and oncology, where the biologic link between surrogates, such as LDL, blood pressure, tumor size, or progression-free survival and important clinical outcomes, is quite strong, these outcomes are viewed skeptically (14, 15). It is not particularly surprising then that surrogate outcomes, and specifically organ function surrogate outcomes, have fared poorly in critical care. Nitric oxide synthase inhibitor and corticosteroids improve blood pressure in septic shock, but have no effect on or worsen mortality (16, 17). Renal-dose dopamine and nesiritide increase urine output or reduce pulmonary wedge pressure, but have no effect on or worsen mortality (18–20). Large tidal volumes and inhaled nitric oxide improve oxygenation in ARDS, but have no effect on or worsen mortality, and inhaled nitric oxide may cause renal failure (21, 22). Given this experience, why should we have
any confidence that preventing ARDS will improve clinically important outcomes? Certainly most, but not all, studies show that, after adjusting for risk factor and severity of illness, patients who develop ARDS have significantly higher mortality than those who do not (23–25). In addition, one of the few studies to compare patients with ARDS with an at-risk control population found that patients with ARDS had worse health-related quality of life (26). However, the majority of patients with ARDS do not die from their ARDS, or at least from hypoxemia, but from organ failure associated with the underlying risk factor, or possibly ventilator-induced organ failure from management of their ARDS (27). Patients with acute respiratory failure who do not meet the criteria for ARDS have substantial mortality, suggesting that preventing ARDS in intubated patients may leave them at considerable risk for death from other causes (28). In a careful prospective cohort of elderly patients, mechanical ventilation did not worsen the decline in cognitive function associated with sepsis, suggesting that preventing mechanical ventilation may not eliminate these sequelae (29). When compared with matched controls without ARDS, long-term survival was determined by risk factor for ARDS and not by ARDS (30). Finally, although avoiding intensive care unit (ICU) admission is almost certainly cost saving, the marginal cost savings of avoiding intubation or preventing ARDS in patients already in the ICU may be small (31). Therefore, the evidence from observational data is mixed on the attributable mortality, morbidity, and cost from ARDS. Because interventions to prevent ARDS will have their own potential costs and risks, it is simply unknown what overall effect they might have on patient-centered outcomes.
Lessons from Other Prevention Studies in Critical Care If clinical trials to prevent ARDS are successful, we can learn how that evidence might be used by looking to the large body of evidence demonstrating effective prevention of gastrointestinal bleeding, ventilator-associated pneumonia (VAP), venous thromboembolism, and catheter related bacteremia (32–35). These surrogate outcomes are also presumed to be in the causal pathway to morbidity and death in
American Journal of Respiratory and Critical Care Medicine Volume 191 Number 3 | February 1 2015
CRITICAL CARE PERSPECTIVE the ICU. However, one of the important lessons from the ICU prevention literature is how difficult it is to make inferences about attributable effects from observational studies. Confounding, competing, and time-varying risks all make simple comparisons between those with and without one of these surrogate outcomes biased (36). For example, accurate estimates of the effect of VAP on length of stay must account for the fact that patients acquire VAP after being in the ICU for a period of time. Not surprisingly, the estimates of the associated patient benefit from prevention vary with the analytic approach, and some recent analyses find relatively low estimates of attributable effect (37, 38). The limitations of observational studies in teasing out the attributable effects of surrogate outcomes can be addressed with randomized trials that link prevention with improvement in patient-centered outcomes. However, prevention trials targeting surrogate outcomes are rarely powered to demonstrate a difference in patient-centered outcome and, even when these studies are large, rarely show an effect on mortality. For example, a clinical trial with 9,417 patients showed that silver-coated endotracheal tubes reduced VAP by nearly 36%, but had no statistically significant effect on any patient-centered outcomes (39). If a statistically significant effect on mortality is noted in an underpowered prevention trial, the mortality signal is more likely to be due to chance and less likely to be reproducible than in an appropriately powered trial (40, 41). It is understandable why prevention trials rarely find mortality effects. A recent cohort study showed that 3.5% of a mixed ICU population suffered from clinically important gastrointestinal bleeding, and that that bleeding was associated with a 30% increase in mortality (42). Under the most optimistic assumptions that prevention could eliminate all clinically important gastrointestinal bleeding and that this reduced the 30% attributable mortality to zero and that the treatment had no side effects, then prevention would reduce mortality by a 1% difference. Using the same optimistic assumptions, eliminating clinically important gastrointestinal bleeding would reduce length of stay by less than 1 day in treated patients. A typical ICU clinical trial would require 40,000 patients to demonstrate Critical Care Perspective
effects this small. ARDS prevention trials will suffer from the same sample size problems. The Lung Injury Prevention with Aspirin trial is powered to detect a 10% absolute reduction in the risk of ARDS (8). If we make the optimistic assumptions that aspirin can reduce mortality in patients from 23% if they developed ARDS to 0% if prevented and that the treatment has no side effects that could increase mortality, a mortality difference of 2.3% would be observed in this prevention trial. Depending on the mortality in the control arm, it would require a sample size of about 7,000 patients to detect this effect. Because it is unlikely that ARDS prevention reduces the risk of death to zero, realistic assumptions about the mortality benefit of ARDS prevention would require substantially larger trials to detect differences in mortality.
Surrogate Outcome Studies Yield Results That Are Difficult to Apply We might be satisfied relying on these hypothetical estimates of the benefits to patients of ARDS prevention if we knew that they outweighed the potential risks. Unfortunately, clinical trials of surrogate outcomes that are underpowered to detect small mortality benefits are also underpowered to detect small risks. These risks are usually uncovered in nonrandomized studies. For example, clinical trials of acid suppression for gastrointestinal bleeding in the ICU did not disclose the potential for delirium and nosocomial infections demonstrated subsequently (43–45). This leaves clinicians balancing the theoretic, but presumed, benefit of preventing a surrogate outcome, like gastrointestinal bleeding, against these potential harms. The relative benefit and costs to patients of these tradeoffs may be quite small (46, 47). The challenge of applying the results of surrogate outcome trials is illustrated nicely by the contradictory interpretation of two important ARDS Network trials: the Methylprednisolone for Persistent ARDS and the Fluid-Management Strategies trials (48, 49). Late steroid therapy for persistent ARDS resulted in 4.4 additional ventilatorfree days compared with placebo, whereas fluid limitation and diuretics resulted in 2.5 additional ventilator-free days. Despite
more impressive results in terms of ventilator-free days and confirmatory improvements in oxygenation and compliance, Methylprednisolone for Persistent ARDS was interpreted as a negative study with potentially harmful effects. Although neither trial showed a statistically significant difference in their primary mortality outcomes, a post hoc analysis revealed that patients treated with corticosteroids after 2 weeks of ARDS had a statistically significantly higher mortality. Safety signals for infection and reintubation were also worrisome, leading the authors to advise against the use of corticosteroids. Corticosteroid advocates might emphasize the physiologic improvement and larger effect size, and discount the safety concerns, which were not translated into a net effect on ventilator-free days, a surrogate selected presumably because it captured relevant harms as well as benefits. To complicate matters more, a subsequent study demonstrated unexpected safety concerns with the fluid-management strategy, indicating that cognitive impairment might be associated with the intervention (50). These trials demonstrate how difficult it is to apply evidence from surrogate outcomes, such as ventilator-free days. Without an important patient-centered signal, clinicians need to weigh the benefits of increased ventilator-free days against the harms of reintubation, infection, and cognitive impairment. If one of these treatments had been a novel and expensive drug therapy with additional unknown risks and high cost, the decision becomes even more complex. It is easy to imagine preventive therapies for ARDS that might increase costs, delirium, bleeding, or nosocomial infection that would put physicians in the same quandary. For example, imagine that aspirin administered to at-risk patients reduces the risk of ARDS by 10%. This would be an impressive result, meaning that only 10 patients need to be treated to prevent 1 case of ARDS. What if the excess major bleeding risk was 1%? This means that for every 10 ARDS cases prevented, we cause 1 major bleed. How many fewer cases of ARDS justify an iatrogenic major bleed? Intensivists may think that “avoiding intubation” or “avoiding ARDS” are inherently beneficial to patients in the same way that avoiding cancer, stroke, or myocardial infarction are assumed to be. However, in these cases, the effects of these 257
CRITICAL CARE PERSPECTIVE diseases are well known, and there are large datasets of clinical trials and observational cohorts to estimate net benefit not available in critical illness (51). Despite the considerable data available to them, our colleagues in primary care still face a dilemma counseling patients about the tradeoffs involved in taking aspirin as primary prophylaxis for myocardial infarction, or using chronic anticoagulation to prevent stroke, or even screening for cancer (52, 53). At least in the outpatient setting, patients can weigh the outcomes and make their own decisions. In the ICU, clinicians or expert guideline panels will have to weigh these intangible costs, risks, and benefits in making decisions for patients with a much smaller evidence base.
The Syndromic Nature of ARDS Complicates Its Use as an Outcome The challenges of ARDS diagnosis and the heterogeneity of the syndrome are often invoked as explanations for slow progression in the field. Enrolling patients without “real” ARDS, whatever that entity is, in treatment trials with mortality as an outcome simply reduces the power of the study, but does not bias it. Similarly, enrolling patients with ARDS in a prevention trial who cannot have it prevented reduces the power of the trial, but does not bias it. Standardizing the diagnostic criteria for enrolment, as was done in the recent pirfenidone trials for pulmonary fibrosis, is a recommended strategy for both types of trials to increase the expected signal (54). However, the ARDS diagnostic criteria present fundamentally different problems when used to define the endpoint for a study. Prerandomization variation in
enrolment diagnosis introduces noise; postrandomization differences in outcome diagnosis can create bias. The syndrome relies on the chest radiograph and PaO2/FIO2 ratio that have both been shown to be treatment dependent and unreliable (55, 56). In theory, a patient could develop the pathophysiology of the syndrome, as unclear as that is, and not be diagnosed with ARDS if the preventive intervention directly affects gas exchange or the radiograph. Topical antibiotic prevention strategies for VAP have generated the same concerns. Antibiotic concentrations in tracheal secretions may be more effective at preventing the diagnosis of VAP than the disease (57). Analogous examples in ARDS, including inhaled nitric oxide, lung recruitment, differential ascertainment of the oxygenation status by pulse oximetry versus arterial blood gas, and assessment of the PaO2/FIO2 ratio in nonintubated patients, with inaccurate FIO2 measurement, could all affect the diagnosis of ARDS (58–60). Of course, any of these interventions might modify the underlying mechanisms of disease, but studies that rely solely on ARDS as the primary outcome will not be able to distinguish a clinically significant effect on disease process from a purely cosmetic effect on disease diagnosis. Study design modifications can minimize the potential for bias. These would include blinding the treating clinician and blinded central adjudication of the diagnosis of ARDS. Standardized ventilator settings and protocolized screening for the diagnosis of ARDS, an option for enrolment in treatment trials, may be mandatory for outcome measurement in prevention trials, particularly if the intervention is unblinded.
References 1. Rubenfeld GD, Caldwell E, Peabody E, Weaver J, Martin DP, Neff M, Stern EJ, Hudson LD. Incidence and outcomes of acute lung injury. N Engl J Med 2005;353:1685–1693. 2. Zambon M, Vincent JL. Mortality rates for patients with acute lung injury/ ARDS have decreased over time. Chest 2008;133:1120–1127. 3. Phua J, Badia JR, Adhikari NK, Friedrich JO, Fowler RA, Singh JM, Scales DC, Stather DR, Li A, Jones A, et al. Has mortality from acute respiratory distress syndrome decreased over time? A systematic review. Am J Respir Crit Care Med 2009;179:220–227. 4. Duggal A, Ganapathy A, Mohana Ratnapalan M, Adhikari NK. Pharmacological treatments for acute respiratory distress syndrome: systematic review. Minerva Anestesiol (In press)
258
Who Cares about ARDS Prevention? ARDS is associated with significant morbidity and mortality in patients with common risk factors, such as trauma and sepsis. It is clear from ongoing trials and grant announcements that researchers care about preventing ARDS. However, it is simply unknown whether clinicians, patients, or payers should care. It is difficult to deny that, all things being equal, it would be better for a patient to avoid getting ARDS. However, all things are rarely equal. Clinicians must appreciate that risks are unlikely to be fully disclosed by prevention trials using surrogate outcomes. Therefore, there will be insufficient information to make informed decisions about the tradeoff between costs, risks, and benefits in implementing this evidence. Of course, we can do ARDS prevention when it is a beneficial side effect of treatments that we should already be implementing. Examples of these would be aspiration precautions, early resuscitation in sepsis, limiting blood products, and lung-protective ventilation (61). Positive, well designed prevention studies of ARDS can yield valuable insights into the pathophysiology of the syndrome and its heterogeneity; however, novel strategies that appear to effectively prevent ARDS in phase II studies must be tested in trials sufficiently powered to demonstrate their patient-centered costs, benefits, and harms before widespread adoption. If these trials are deemed unfeasible, due to the large sample sizes required, then one must wonder whether the benefits to patients are likely to be sufficiently important to justify the entire endeavor. n Author disclosures are available with the text of this article at www.atsjournals.org.
5. Gajic O, Dabbagh O, Park PK, Adesanya A, Chang SY, Hou P, Anderson H III, Hoth JJ, Mikkelsen ME, Gentile NT, et al.; U.S. Critical Illness and Injury Trials Group: Lung Injury Prevention Study Investigators (USCIITG-LIPS). Early identification of patients at risk of acute lung injury: evaluation of lung injury prediction score in a multicenter cohort study. Am J Respir Crit Care Med 2011;183: 462–470. 6. Clinical Centers for the NHLBI Prevention and Early Treatment of Acute Lung Injury (PETAL) Clinical Trials Network. May 15, 2013 [accessed 2014 Aug 20]. Available from: http://grants.nih.gov/grants/guide/rfafiles/RFA-HL-14-014.html 7. FY14 Peer Reviewed Medical Research Program Clinical Trial Award. 2014 [accessed 2014 Aug 20]. Available from: http://cdmrp.army.mil/ funding/pa/14prmrpcta_pa.pdf
American Journal of Respiratory and Critical Care Medicine Volume 191 Number 3 | February 1 2015
CRITICAL CARE PERSPECTIVE 8. Kor DJ, Talmor DS, Banner-Goodspeed VM, Carter RE, Hinds R, Park PK, Gajic O, Gong MN; US Critical Illness and Injury Trials Group: Lung Injury Prevention with Aspirin Study Group (USCIITG: LIPS-A). Lung Injury Prevention with Aspirin (LIPS-A): a protocol for a multicentre randomised clinical trial in medical patients at high risk of acute lung injury. BMJ Open 2012;2:e001606. 9. Prentice RL. Surrogate endpoints in clinical trials: definition and operational criteria. Stat Med 1989;8:431–440. 10. Rubenfeld GD. Surrogate measures of patient-centered outcomes in critical care. In: Angus DC, Carlet J, editors. Surviving intensive care. Berlin: Springer-Verlag; 2002. 11. Rubenfeld GD, Abraham E. When is a negative phase II trial truly negative? Am J Respir Crit Care Med 2008;178:554–555. 12. Echt DS, Liebson PR, Mitchell LB, Peters RW, Obias-Manno D, Barker AH, Arensberg D, Baker A, Friedman L, Greene HL, et al. Mortality and morbidity in patients receiving encainide, flecainide, or placebo: the Cardiac Arrhythmia Suppression Trial. N Engl J Med 1991; 324:781–788. 13. The Cardiac Arrhythmia Suppression Trial II Investigators. Effect of the antiarrhythmic agent moricizine on survival after myocardial infarction. N Engl J Med 1992;327:227–233. 14. Psaty BM, Lumley T. Surrogate end points and FDA approval: a tale of 2 lipid-altering drugs. JAMA 2008;299:1474–1476. 15. Schatzkin A, Gail M. The promise and peril of surrogate end points in cancer research. Nat Rev Cancer 2002;2:19–27. 16. Lopez ´ A, Lorente JA, Steingrub J, Bakker J, McLuckie A, Willatts S, Brockway M, Anzueto A, Holzapfel L, Breen D, et al. Multiple-center, randomized, placebo-controlled, double-blind study of the nitric oxide synthase inhibitor 546C88: effect on survival in patients with septic shock. Crit Care Med 2004;32:21–30. 17. Sprung CL, Annane D, Keh D, Moreno R, Singer M, Freivogel K, Weiss YG, Benbenishty J, Kalenka A, Forst H, et al.; CORTICUS Study Group. Hydrocortisone therapy for patients with septic shock. N Engl J Med 2008;358:111–124. 18. Friedrich JO, Adhikari N, Herridge MS, Beyene J. Meta-analysis: lowdose dopamine increases urine output but does not prevent renal dysfunction or death. Ann Intern Med 2005;142:510–524. 19. O’Connor CM, Starling RC, Hernandez AF, Armstrong PW, Dickstein K, Hasselblad V, Heizer GM, Komajda M, Massie BM, McMurray JJ, et al. Effect of nesiritide in patients with acute decompensated heart failure. N Engl J Med 2011;365:32–43. [Published erratum appears in N Engl J Med 365:773. Wilson WH corrected to Tang WHW.] 20. Sackner-Bernstein JD, Kowalski M, Fox M, Aaronson K. Short-term risk of death after treatment with nesiritide for decompensated heart failure: a pooled analysis of randomized controlled trials. JAMA 2005;293:1900–1905. 21. Adhikari NK, Burns KE, Friedrich JO, Granton JT, Cook DJ, Meade MO. Effect of nitric oxide on oxygenation and mortality in acute lung injury: systematic review and meta-analysis. BMJ 2007;334:779. 22. The Acute Respiratory Distress Syndrome Network. Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome. N Engl J Med 2000;342:1301–1308. 23. Treggiari MM, Hudson LD, Martin DP, Weiss NS, Caldwell E, Rubenfeld G. Effect of acute lung injury and acute respiratory distress syndrome on outcome in critically ill trauma patients. Crit Care Med 2004;32: 327–331. 24. Shah CV, Localio AR, Lanken PN, Kahn JM, Bellamy S, Gallop R, Finkel B, Gracias VH, Fuchs BD, Christie JD. The impact of development of acute lung injury on hospital mortality in critically ill trauma patients. Crit Care Med 2008;36:2309–2315. 25. Mikkelsen ME, Shah CV, Meyer NJ, Gaieski DF, Lyon S, Miltiades AN, Goyal M, Fuchs BD, Bellamy SL, Christie JD. The epidemiology of acute respiratory distress syndrome in patients presenting to the emergency department with severe sepsis. Shock 2013;40:375–381. 26. Davidson TA, Caldwell ES, Curtis JR, Hudson LD, Steinberg KP. Reduced quality of life in survivors of acute respiratory distress syndrome compared with critically ill control patients. JAMA 1999; 281:354–360. 27. Stapleton RD, Wang BM, Hudson LD, Rubenfeld GD, Caldwell ES, Steinberg KP. Causes and timing of death in patients with ARDS. Chest 2005;128:525–532.
Critical Care Perspective
28. Luhr OR, Antonsen K, Karlsson M, Aardal S, Thorsteinsson A, Frostell CG, Bonde J; The ARF Study Group. Incidence and mortality after acute respiratory failure and acute respiratory distress syndrome in Sweden, Denmark, and Iceland. Am J Respir Crit Care Med 1999; 159:1849–1861. 29. Iwashyna TJ, Ely EW, Smith DM, Langa KM. Long-term cognitive impairment and functional disability among survivors of severe sepsis. JAMA 2010;304:1787–1794. 30. Davidson TA, Rubenfeld GD, Caldwell ES, Hudson LD, Steinberg KP. The effect of acute respiratory distress syndrome on long-term survival. Am J Respir Crit Care Med 1999;160:1838–1842. 31. Cooke CR. Economics of mechanical ventilation and respiratory failure. Crit Care Clin 2012;28:39–55 (vi.). 32. Krag M, Perner A, Wetterslev J, Wise MP, Hylander Møller M. Stress ulcer prophylaxis versus placebo or no prophylaxis in critically ill patients: a systematic review of randomised clinical trials with metaanalysis and trial sequential analysis. Intensive Care Med 2014;40: 11–22. 33. Dodek P, Keenan S, Cook D, Heyland D, Jacka M, Hand L, Muscedere J, Foster D, Mehta N, Hall R, et al.; Canadian Critical Care Trials Group; Canadian Critical Care Society. Evidence-based clinical practice guideline for the prevention of ventilator-associated pneumonia. Ann Intern Med 2004;141:305–313. 34. Alhazzani W, Lim W, Jaeschke RZ, Murad MH, Cade J, Cook DJ. Heparin thromboprophylaxis in medical-surgical critically ill patients: a systematic review and meta-analysis of randomized trials. Crit Care Med 2013;41:2088–2098. 35. Blot K, Bergs J, Vogelaers D, Blot S, Vandijck D. Prevention of central line–associated bloodstream infections through quality improvement interventions: a systematic review and meta-analysis. Clin Infect Dis 2014;59:96–105. 36. Muñoz A, Mongilardi N, Checkley W. Multilevel competing risks in the evaluation of nosocomial infections: time to move on from proportional hazards and even from hazards altogether. Crit Care 2014;18:146. 37. Bekaert M, Timsit JF, Vansteelandt S, Depuydt P, Vesin ´ A, GarrousteOrgeas M, Decruyenaere J, Clec’h C, Azoulay E, Benoit D; Outcomerea Study Group. Attributable mortality of ventilatorassociated pneumonia: a reappraisal using causal analysis. Am J Respir Crit Care Med 2011;184:1133–1139. 38. Melsen WG, Rovers MM, Koeman M, Bonten MJ. Estimating the attributable mortality of ventilator-associated pneumonia from randomized prevention studies. Crit Care Med 2011;39:2736–2742. 39. Kollef MH, Afessa B, Anzueto A, Veremakis C, Kerr KM, Margolis BD, Craven DE, Roberts PR, Arroliga AC, Hubmayr RD, et al.; NASCENT Investigation Group. Silver-coated endotracheal tubes and incidence of ventilator-associated pneumonia: the NASCENT randomized trial. JAMA 2008;300:805–813. 40. Browner WS, Newman TB. Are all significant P values created equal? The analogy between diagnostic tests and clinical research. JAMA 1987;257:2459–2463. 41. Christley RM. Power and error: increased risk of false positive results in underpowered studies. Open Epidemiol J 2010;3:16–19. 42. Cook DJ, Griffith LE, Walter SD, Guyatt GH, Meade MO, Heyland DK, Kirby A, Tryba M; Canadian Critical Care Trials Group. The attributable mortality and length of intensive care unit stay of clinically important gastrointestinal bleeding in critically ill patients. Crit Care 2001;5:368–375. 43. Eom CS, Jeon CY, Lim JW, Cho EG, Park SM, Lee KS. Use of acidsuppressive drugs and risk of pneumonia: a systematic review and meta-analysis. CMAJ 2011;183:310–319. 44. Howell MD, Novack V, Grgurich P, Soulliard D, Novack L, Pencina M, Talmor D. Iatrogenic gastric acid suppression and the risk of nosocomial Clostridium difficile infection. Arch Intern Med 2010;170: 784–790. 45. Fujii S, Tanimukai H, Kashiwagi Y. Comparison and analysis of delirium induced by histamine h(2) receptor antagonists and proton pump inhibitors in cancer patients. Case Rep Oncol 2012;5:409–412. 46. Kahn JM, Doctor JN, Rubenfeld GD. Stress ulcer prophylaxis in mechanically ventilated patients: integrating evidence and judgment using a decision analysis. Intensive Care Med 2006;32: 1151–1158.
259
CRITICAL CARE PERSPECTIVE 47. Chiasson TC, Manns BJ, Stelfox HT. An economic evaluation of venous thromboembolism prophylaxis strategies in critically ill trauma patients at risk of bleeding. PLoS Med 2009;6:e1000098. 48. Wiedemann HP, Wheeler AP, Bernard GR, Thompson BT, Hayden D, deBoisblanc B, Connors AF Jr, Hite RD, Harabin AL; National Heart, Lung, and Blood Institute Acute Respiratory Distress Syndrome (ARDS) Clinical Trials Network. Comparison of two fluidmanagement strategies in acute lung injury. N Engl J Med 2006;354: 2564–2575. 49. Steinberg KP, Hudson LD, Goodman RB, Hough CL, Lanken PN, Hyzy R, Thompson BT, Ancukiewicz M; National Heart, Lung, and Blood Institute Acute Respiratory Distress Syndrome (ARDS) Clinical Trials Network. Efficacy and safety of corticosteroids for persistent acute respiratory distress syndrome. N Engl J Med 2006;354:1671–1684. 50. Mikkelsen ME, Christie JD, Lanken PN, Biester RC, Thompson BT, Bellamy SL, Localio AR, Demissie E, Hopkins RO, Angus DC. The adult respiratory distress syndrome cognitive outcomes study: longterm neuropsychological function in survivors of acute lung injury. Am J Respir Crit Care Med 2012;185:1307–1315. 51. US Preventive Services Task Force (USPSTF). Estimating certainty and magnitude of net benefit. 2007 [accessed 2014 Aug 28]. Available from: http://www.uspreventiveservicestaskforce.org/uspstf07/ methods/benefit.htm 52. US Preventive Services Task Force. Aspirin for the prevention of cardiovascular disease: U.S. Preventive Services Task Force recommendation statement. Ann Intern Med 2009;150:396–404. 53. Protheroe J, Fahey T, Montgomery AA, Peters TJ. The impact of patients’ preferences on the treatment of atrial fibrillation: observational study of patient based decision analysis. BMJ 2000; 320:1380–1384.
260
54. King TE Jr, Bradford WZ, Castro-Bernardini S, Fagan EA, Glaspole I, Glassberg MK, Gorina E, Hopkins PM, Kardatzke D, Lancaster L, et al.; ASCEND Study Group. A phase 3 trial of pirfenidone in patients with idiopathic pulmonary fibrosis. N Engl J Med 2014;370: 2083–2092. 55. Ferguson ND, Fan E, Camporota L, Antonelli M, Anzueto A, Beale R, Brochard L, Brower R, Esteban A, Gattinoni L, et al. The Berlin definition of ARDS: an expanded rationale, justification, and supplementary material. Intensive Care Med 2012;38:1573–1582. 56. Ranieri VM, Rubenfeld GD, Thompson BT, Ferguson ND, Caldwell E, Fan E, Camporota L, Slutsky AS; ARDS Definition Task Force. Acute respiratory distress syndrome: the Berlin Definition. JAMA 2012;307:2526–2533. 57. Schultz MJ, de Jonge E, Kesecioglu J. Selective decontamination of the digestive tract reduces mortality in critically ill patients. Crit Care 2003;7:107–110. 58. Langevin PB, Hellein V, Harms SM, Tharp WK, Cheung-Seekit C, Lampotang S. Synchronization of radiograph film exposure with the inspiratory pause: effect on the appearance of bedside chest radiographs in mechanically ventilated patients. Am J Respir Crit Care Med 1999;160:2067–2071. 59. Goldstein RS, Young J, Rebuck AS. Effect of breathing pattern on oxygen concentration received from standard face masks. Lancet 1982;2:1188–1190. 60. Rice TW, Wheeler AP, Bernard GR, Hayden DL, Schoenfeld DA, Ware LB; National Institutes of Health, National Heart, Lung, and Blood Institute ARDS Network. Comparison of the SpO2/FIO2 ratio and the PaO2/FIO2 ratio in patients with acute lung injury or ARDS. Chest 2007;132:410–417. 61. Levitt JE, Matthay MA. Clinical review: early treatment of acute lung injury—paradigm shift toward prevention and treatment prior to respiratory failure. Crit Care 2012;16:223.
American Journal of Respiratory and Critical Care Medicine Volume 191 Number 3 | February 1 2015
