Limitations of clinical trials in acute lung injury and acute respiratory distress syndrome John J. Marini

Transcription

1 Limitations of clinical trials in acute lung injury and acute respiratory distress syndrome John J. Marini Purpose of review To review the challenges and limitations of randomized clinical trials in acute respiratory distress syndrome, with special emphasis on those pertaining to ventilatory management. Recent findings Superbly executed randomized trials of ventilatory strategy have garnered deserved attention from the critical care community and yet have illustrated the limitations of our current approach to clinical research in this area. Inexact definitions, incomplete mechanistic understanding of complex pathophysiology, inappropriate outcome variables, diverse therapeutic environments, lengthy data acquisition time and ethical constraints on trial design limit the applicability of randomized control trial methodology to acute respiratory distress syndrome and acute lung injury. As yet, clinical practice does not seem to have been greatly impacted by the implications of completed randomized controlled trials per se. Recent issues, both ethical and interpretive, regarding control group participants have raised troubling and theoretically important issues that are yet to be fully resolved. Summary Without tighter definitions of the condition under treatment, more specific targets for interventions to act upon, stratification that recognizes key interactive elements, and cointerventions based on better mechanistic understanding, randomized controlled trials of new drugs, ventilatory strategy, and other management approaches in acute respiratory distress syndrome are likely to remain a blunt instrument for investigation. As valuable as they are for calling important therapeutic principles to attention and for helping to suggest general guidelines for care, the limitations of randomized controlled trials for treating the individual with acute respiratory distress syndrome must be acknowledged. Keywords acute respiratory distress syndrome, clinical trials, ethics, ventilatory management Curr Opin Crit Care 12: ß 2006 Lippincott Williams & Wilkins. University of Minnesota, Minneapolis/St Paul, Minnesota, USA Correspondence to John J. Marini MD, Professor of Medicine, Regions Hospital, Pulmonary/Critical Care, MS-11203B, 640 Jackson St., St Paul, MN 55101, USA Tel: ; fax: ; john.j.marini@healthpartners.com Current Opinion in Critical Care 2006, 12:25 31 Abbreviations ALI acute lung injury ARDS acute respiratory distress syndrome ICU intensive care unit PEEP positive end-expiratory pressure RCT randomized clinical trial ß 2006 Lippincott Williams & Wilkins Introduction Undeniable recent progress has been made toward understanding and managing the set of clinical problems we recognize as acute lung injury (ALI) and acute respiratory distress syndrome (ARDS). After decades without progress, mortality rates have apparently fallen over only a few years to much lower levels than before [1,2]. Although some might argue that this improvement reflects a classification artifact or beneficial reduction in the incidence or nature of the underlying problems, a better case can be made for improved management, especially in the area of ventilatory support. The caregiver s first responsibility is to do no harm; yet, overwhelming evidence from the laboratory setting documents the potential for adverse patterns of ventilation to retard healing or to further damage the lung [3 ]. Moreover, emerging clinical data demonstrate that adherence to less aggressive ventilatory management translates into survival benefit [4 6]. Better understanding of mechanical stress, strain, and mechano-biologic signaling [7,8], of nonventilatory cofactors of injury [9], and of the inflammatory process itself promises to open avenues of intervention that may further reduce the morbidity of this still devastating clinical disease. Understandably, there is a strong desire to test directly the relevance of laboratory principles in a clinical environment before they are introduced into practice, and already adopted standards should be subjected to scrutiny when new evidence emerges. For many, the randomized and controlled clinical trial (RCT), particularly one whose endpoint is a relevant clinical outcome, provides the most convincing evidence of benefit from intervention or lack thereof. In fact, the results of large outcome-oriented clinical trials positive or negative often serve as the basis for management guidelines recommended by opinion leaders and professional organizations [10 12]. Unfortunately, RCT methodology, while quite appropriate for characterizing populations 25

2 26 Respiratory system and therefore for demonstrating general principles, is not always well suited to setting prescriptions for individual case management. Despite good intentions, strenuous effort and admirable execution, even our best efforts have sparked a firestorm of controversy [13]. Recent experience, reviewed herein, highlights the problems and limitations of RCTs for guiding clinicians who confront ALI/ARDS. Problems of randomized controlled trials in acute respiratory distress syndrome and acute lung injury The concepts of standardizing practice and eliminating variability to encourage better quality control were borrowed from those of industry, where they have been in place for the better part of a century [14]. In producing an industrial product, it makes undeniable sense to find out what works and what does not, as the fabricator starts with uniform raw materials, strives for a uniform final product, and seeks to understand what facilitates and what disrupts the inanimate manufacturing procedure. With these conditions met, improving the perfected process and eliminating variability then become transcendent goals that, when accomplished, have durable import. Economic incentives are perfectly aligned with such behaviors. Although emulating this successful strategy in medicine may be laudable (if a touch naïve), none of these valued manufacturing tenets hold up well in intensive care unit (ICU) practice, specifically with respect to the biologically diverse problem set we label collectively as ALI/ARDS. In a complex environment lacking precise definitions, even an excellent RCT can only provide a general proof of principle or a set of unanticipated observations that help inform clinical practice or generate new questions for investigation. If lacking a rigorous design shaped by knowledge of physiologic mechanism or flawless execution and interpretation, these resource-costly, time consuming, and high-impact studies have the potential to prove confusing or frankly misleading. The sobering experience of the past decade illustrates that, even at present, clinical trials in ARDS/ALI are beset by several major and perhaps insurmountable challenges. Definitional uncertainty Traditional statistical methods work well when the independent and dependent variables can be identified unequivocally. As understood from the first years of ICU experience with this entity, however, ALI/ARDS is a syndrome that arises in diverse circumstances not a disease with a well characterized physiologic or biologic signature [15 17]. In studies of oncology and cardiovascular therapeutics, successful clinical trials with valid implications for the individual patient can be conducted when the nature and severity of disease expression are certain. Clearly, this is not the case for ARDS; reliance on the broadly inclusive American European consensus committee (AECC) definitions [18] has led to the inclusion of patients who exhibit an extraordinary range of respiratory mechanics and severity of illness. The AECC criteria stress acute onset of illness, radiographic criteria, impaired oxygenation, and clinical absence of cardiogenic edema any one of which can be influenced by management decisions or interpretive bias. Even if loosely defined, ARDS/ALI is not very common [19]. This obligates clinical trialists to engage in multicenter cooperative studies for timely completion. Even then, years may elapse before data collection ends. Enrollment may further be hampered by competition among different RCTs for scarce patients. (Two studies of prone positioning in ARDS that offered no monetary incentive suffered from slow enrollment, encouraging their premature closure [20,21].) The liberal AECC inclusion criteria certainly allow for faster enrollment in clinical trials and for more timely study completion. Disturbingly for studies of mechanical ventilation aimed at lung stretch, however, no components relate to respiratory mechanics, positive end-expiratory pressure (PEEP), ventilator settings, comorbidities, pace of illness or responsiveness to therapy [18]. Furthermore, although minute ventilation and dead space are likely to correlate both with severity of lung injury and risk for death [22], no trial undertaken thus far has stratified on those variables. Therefore, patients who qualify by oxygenation criteria at low levels of PEEP might be excluded if PEEP were raised. Patients who respond quickly to diuretics or steroids may only transiently carry the ARDS label, whichever of the tested therapies they receive. (An enlightened approach that recognizes these factors was undertaken by a Spanish cooperative group (ARIES) that required the oxygenation criterion to be met before and after a 24 h waiting period while receiving moderate PEEP [6].) Conversely, patients with underlying comorbidities who are refractory to treatment and therefore may be destined to die of their illness independently of ventilator strategy are also so classified. With regard to the hardest definitional component, multiquadrant bilateral pulmonary infiltrates, qualified experts many of them coinvestigators on multicenter trials do not always agree [23,24]. No reliable blood or body fluid bio-signature for identifying or grading ALI has yet been identified. In the absence of such a troponin for the lung clinical judgment still holds sway as to what is ARDS and what is not. Complexity of pathological processes When mortality is used as an endpoint, it is subject to each variable that affects that outcome. Strong observational evidence already at hand indicates that numerous

3 Limitations of clinical trials Marini 27 factors influence the fate of patients with ARDS. Some relate to etiology, some to comorbid factors, and some to stage and severity of the disorder itself [25,26]. Patients sustaining ARDS in a trauma setting, for example, are generally considered to have a better prognosis than others [27 ]. Management style and decisions that have little to do with ventilation per se exert a powerful impact; withdrawal of care is among the most common reasons for death within and outside the ICU environment, even when ultimate survivability is possible [28]. As attractive as the unequivocal binary death or survival endpoint might appear from some perspectives, it is one that has many potential contributors. To many, it seems more reasonable to select the criterion variable of the RCT so as to reflect the likely contribution of the tested intervention. It is quite likely that some interventions with potentially powerful influences on outcome are conditionally important that is, important only under certain conditions. In experimental ventilation studies there appear to be numerous examples. Breathing frequency is important only if tissue stresses are excessive [29,30]; tidal volume is important only if airway pressure rises into a dangerous range [31]; PEEP is important only when plateau pressure is high enough; heating may have protective or damaging effects on ventilator-induced lung injury (VILI) expression, depending on the timing of its application [32,33 ]. The list of these conditional influences continues to grow. Clinical data suggest that prone positioning may influence mortality only in the most critically ill, those exposed to high tidal volumes, and those who experience recruitment in the process [21,34]. To lump all potentially pathological and therapeutic conditions together without controlling for them (a practical necessity for trial completion) may well mask any but the very strongest effect. To categorically discard treatments that are potentially useful for some patients based on group mean data is particularly hazardous. Awareness of such codependencies may help design a trial capable of uncovering therapeutic differences in a relatively small number of patients [4,6] (see following section). An important lesson learned from the detailed imaging studies of recent years is that the lung and chest wall form a respiratory system whose mechanical complexity and regional variations seriously compromise attempts to apply a single airway pressure or flow that obtains a uniform result within the same lung [35]. Similarly, variation in chest wall stiffness among patients strongly influences the relationships of airway pressure to tidal volume and precludes bold statements about what values of either define their safety limits for all patients. The very complexity of the pathological processes we confront in this arena cripples the statistical models on which success is judged. How does one confidently power a study whose population is diverse and whose true variance used in the power calculation is only guessed at? Even Emil Fisher, the originator of linear model statistical methods, would have had difficulty if his marbles were neither black nor white but shades of gray. Diversity of therapeutic environments Generally speaking, many important details of clinical management are addressed in study protocols but many more are not and cannot be. This compromise is a matter of practicality, difficulty of reaching consensus, and deference to the integrative judgment of the clinician at the bedside. It is difficult enough to adhere to a detailed protocol controlling even a few salient variables (e.g. the PEEP and FiO 2 adjustment table of ARDSnet [5]). When such protocols are followed, they often represent a departure from everyday practice, contributing to the difficulty of translating study results to implications for care ( internal versus external validity) [36]. Yet, not to control variables potentially important to outcome (some of which are not yet known) lessens the chance of discovering a positive result. Tradeoffs among potentially important therapeutic options are inevitable (e.g. PEEP versus FiO 2, wet versus dry approach to fluid management), and for a given patient might often be made differently by various physicians, depending on skill, experience and priorities. Staying with that same example, one physician might point to the experimental evidence that high levels of pulmonary microvascular pressure increase the likelihood of VILI [37,38], while another might consider preservation of urinary function or adequate cardiac priming as transcendent considerations. Which approach is considered appropriate by the same clinician may even change over time, depending on his or her evaluation of the broader clinical picture as the problem evolves. Eroded strength of the difference signal Mortality rate in ARDS fell progressively through the early and middle portion of the 1990s well in advance of the results of any positive ventilator trial [39]. In part, this trend may reflect closer adherence of the clinical practitioner to principles of lung protection emerging from the laboratories and uncontrolled clinical studies of the preceding years [40,41]. Mortality in ARDS may now be less than half what it was 15 years ago [39,42]. Whatever the reason, the dropping mortality rate made randomized comparisons of new treatments to the control strategy less likely to show a mortality difference. Furthermore, as already noted, mortality is a variable that is influenced by many factors, further diluting signal strength. Finally, reasonable but unfortunate choices of tested ventilatory variables or their magnitude (tidal volume rather than plateau pressure [43 45], magnitude of recruiting pressure [46], plateau pressure limits [47 ]) resulted in

4 28 Respiratory system considerable overlap between tested cohorts with regard to the underlying ventilation variable most likely to drive mortality (high plateau pressure with insufficient PEEP). Trial design and execution One strong message that has emerged from discussions surrounding published RCTs of mechanical ventilation in ARDS is that protocols that do not have a strong physiologic rationale hold limited value for practice, no matter how well intentioned or executed. To be applicable to the individual patient being treated, carefully collected data that simulate clinical practice must be examined for patients indistinguishable from that patient in question on all important characteristics. Yet, all multicenter trials of which I am aware have mixed together patients of varying severity and respiratory mechanical characteristics (Fig. 1). Several such trials have attempted to answer (tacitly or explicitly) such broad and difficult questions as Is it better to use a smaller or larger tidal volume? Should we employ high or low PEEP? Is prone positioning or a recruiting maneuver helpful? As already noted, the valid answers to such questions are virtually always conditioned by other key variables that are not subject to protocol control. Moreover, the dose of the tested intervention and its timing of application and withdrawal may be very important to intervention efficacy and trial outcome. Generally speaking, only two levels of the tested variable are tested against each other. What should this imply for all other combinations not tested? Even for an apparently dichotomous variable such as prone positioning, the dosing issue isn t that easy when in the course should proning be performed? For how Figure 1 Wide scatter of tidal volume and plateau pressure in patients before trial entry in the ARDSnet randomized controlled trial illustrates their mechanical diversity Source: Reproduced with permission from [31]. many hours per day should proning be applied? When should prone positioning be stopped? Who are the patients most likely to benefit? What exactly are we expecting proning to do recruit or oxygenate? The latter is simple enough to measure, but what about recruiting efficacy? (Should either be expected to strongly influence mortality, the selected outcome variable?) Along the same line, do recruiting maneuvers have a place in ARDS management? Not all recruiting maneuvers are equally effective or tolerated [48,49], and not all patients are recruitable [50]. Even among those who potentially are, the effect of a given maneuver depends on its magnitude and the starting conditions of the patient. Patients already well recruited are not likely to benefit much; those whose lungs open only at very high pressures may be in desperate need of a maneuver [51]. The two published studies of a high PEEP/low tidal volume that observed the open lung /minimized tissue strain concepts of lung protection developed in the laboratory prioritized recruitment and were able to demonstrate differences in the expected direction with relatively few enrolled patients [4,6]. Ethical issues All physicians are agreed that no matter how interesting or important the question, no patient who enters a clinical trial should be placed at undue risk for harm. An extension of this bedrock principle is that no study patient should be intentionally deprived of clearly useful treatment (e.g. withholding antibiotics in the control arm of an infection study). Such considerations rightly limit the scope of questions that may be tested in an RCT. When comparisons are made between strategies that widely separate the tested variable (e.g. very high and very low tidal volume), the signal is amplified and the power of the study increases. Yet, neither arm may reflect prevailing practice, and in recent years cogent arguments have been advanced that conducting an RCT that involves control and test arms at the extremes of accepted practice does not guarantee that patients in either limb stand to benefit over routine management [52,53]. Indeed, participation may deprive both of their best chance to receive optimal care. As demonstrated in the large sample of patients screened for the low versus high tidal volume multicenter trial (ARMA) conducted by the ARDS Network [5] and from an independent sampling of physicians from 361 ICUs in 20 countries about the same period [54 ], there is a wide variation of tidal volumes in use for patients who carry a diagnosis of ARDS, with a median value that approximates 8.5 ml/kg of measured body weight. The ARMA tested values (6 and 12 ml/kg of ideal body weight) lie near opposite tails of this bell-shaped distribution. Interestingly, the international sampling could find no obvious independent contribution of plateau pressure to mortality

5 Limitations of clinical trials Marini 29 Figure 2 Comparison of mortality in the patients qualifying for but excluded from study entry (vertical bars) and the outcomes observed within the ARMA trial itself Source: Reproduced with permission from [52 ]. Mortality 50 rate (%) ml/kg tidal volume Noneligible participants 6 ml/kg tidal volume N = Incorrectly excluded Enrolled in other trials last 30 days Patient refused enrollment Unable to consent / consent not obtained Physician refused enrollment Inclusion criteria > 36 h over a rather wide range extending from 15 to 30 cmh 2 O [54 ], suggesting that the injury risk attributed to a rising plateau is less a continuous function than a threshold-like variable. In similar fashion, the patients excluded from participation in the ARMA trial (for whatever reason) who were exposed to a similarly wide spectrum of tidal volumes and pressures as were the random international sample experienced a mortality rate as low as that of the 6 ml/kg group [52 ] (Fig. 2). Unfortunately the mortality of the ARMA control group appears to have been impressively higher than this, although still less than that expected from historical experience. Thus, ARMA may not have shown the inherent value of lower tidal volumes, but the risk of high inflation pressures. Making the assumption that the lower tidal volume used in the ARMA tidal volume trial was the most appropriate one to use for all patients, and respecting the raw correlation observed between plateau pressure and mortality observed in that study [31] imposed serious ethical constraints on subsequent trials that tested PEEP (ALVEOLI) [46] and recruiting maneuvers [46] as single independent variables. In effect, the conditional importance of PEEP and recruiting maneuvers that is strongly implied by laboratory data could not be effectively tested due to the valid ethical constraints imposed by its ARMA predecessor. Thus, in these examples the sequence of RCTs may have imposed appropriate ethical limits and doomed the PEEP trial to a negative outcome, independent of high PEEP s merit in other settings. That higher PEEP on the first day is important when high plateau pressures are in use has been strongly suggested by the analysis of Barbas et al. [55]. Planning and conducting RCTs in this field is a very difficult business indeed. Protracted data acquisition time Even if AECC defined, completing an ARDS clinical trial requires relatively long data acquisition time. When the design, analysis and publication lag times are accounted for, years may elapse before the study result is known. Without a strong incentive to finish financial or other studies may fall short of their intended enrollment number [20,21]. Strictly speaking, the relevance of the published conclusions bears only on the data sample at the time of its collection, and medical practice changes spontaneously at a rapid rate. Reproducibility of observation is essential to good science. Yet, given these delays, RCTs are not amenable to the test of replication. For a variety of reasons very few RCTs addressing the same question are conducted contemporaneously, and virtually no confirmatory trials are repeated using exactly the same protocol conducted against a background of unchanging practice. Perhaps it is no surprise that in the field of mechanical ventilation of ARDS, there has yet to be a replicated RCT of any of the published trials. Temporary clinical relevance As the see-sawing estrogen in post-menopausal women debate illustrates [56], contradictory results sometimes follow the publication of interesting clinical research [57,58]. Usually this occurs because the subsequent studies alter the definitional criteria or trial protocol,

6 30 Respiratory system but sometimes quite parallel designs draw different conclusions. Highly cited clinical research trials are no exception; a recent analysis found that a disturbing proportion of highly cited clinical research studies were contradictory of or had substantially smaller effects than the original practice-changing work [58]. We need look no further than the tidal volume story to see an example in the field of ARDS clinical research; tidal volume was concluded to be inconsequential in at least three well conducted trials [43 45] before apparently being contradicted by ARMA [5]. In fact, all four trials may be compatible if the relevant conditions and results of those studies are closely examined [59]. Failure to influence practice To directly apply the median result of a protocol conducted on a loosely defined population to the care of the individual patient invites problems, for reasons already outlined. It should be remembered that few trials can be truly blinded, and fewer still have unbiased funding. Whatever one s position regarding the rightful place of RCTs, it seems undeniable that their effect on practice has been much less than anticipated or desired by the trial designers [60,61]. The potential reasons for this disconnect are many and are still being investigated. Simple explanations are that physicians are not paying attention or that they prefer to practice in their ingrained pattern despite better evidence. To some extent, both criticisms are undoubtedly true. Yet, failure to adopt RCT-derived recommendations might also be rational if the study reported was conducted in patients who do not resemble those that the clinician encounters, if the correspondence between the internally valid study and the external practice environment is not a close one, or if the one size fits all interpretation of the study clashes with the practitioner s own experience [52 ]. Conclusion In the end, randomized clinical trials of therapy in ARDS should not be conducted with the intention of determining what is best for the individual but rather to demonstrate the potential applicability of principles gathered from other laboratory and clinical evidence. Without tighter definitions of the condition under treatment, stratification on key interactive elements of pathology and cointerventions, better mechanistic understanding and more specific targets for interventions to act upon, RCTs of drugs, ventilatory strategy, and other management approaches are likely to remain a blunt scientific instrument and an occasionally misleading undertaking that carries disproportionate impact. As valuable as they might be for calling important therapeutic principles to attention and for helping to develop general guidelines for care, the limitations of RCTs for treating the individual with ARDS will continue to generate controversy. References and recommended reading Papers of particular interest, published within the annual period of review, have been highlighted as: of special interest of outstanding interest Additional references related to this topic can also be found in the Current World Literature section in this issue (p. 69). 1 Ashbaugh DB, Bigelow DB, Petty TL. Acute respiratory distress in adults. Lancet 1967; 2: Steinberg KP, Hudson LD. Acute lung injury and acute respiratory distress syndrome: the clinical syndrome. Clin Chest Med 2000; 21: Plotz FB, Slutsky AS, van Vught AJ, Heijnen CJ. Ventilator-induced lung injury and multiple system organ failure: a critical review of facts and hypotheses. Intensive Care Med 2004; 30: A superb summary of the experimental database that documents the potential for ventilator-induced injury and fuels concern toward its clinical consequences. 4 Amato MBP, Barbas CSV, Medeiros DM, et al. Effect of a protective-ventilation strategy on mortality in the acute respiratory distress syndrome. N Engl J Med 1998; 338: The ARDS 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: Villar J, Kacmarek RM, Perez-Mendez L, Aguirre-Jaime. The ARIES Network. A high PEEP-low tidal volumeventilatory strategy improves outcome in persistent ARDS. Crit Care Med (in press). 7 Pugin J. Molecular mechanisms of lung cell activation induced by cyclic stretch. Crit Care Med 2003; 31 (4 Suppl):S200 S Dos Santos CC, Slutsky AS. Invited review: mechanisms of ventilator-induced lung injury: a perspective. J Appl Physiol 2000; 89: Ferrer LE, Marini JJ. Modulators of ventilator-induced lung injury. In: Vincent J- L, editor Yearbook of intensive care and emergency medicine. Berlin: Springer; pp Meade MO, Herridge MS. An evidence-based approach to acute respiratory distress syndrome. Respir Care 2001; 46: Sevransky JE, Levy MM, Marini JJ. Mechanical ventilation in sepsis-induced acute lung injury/acute respiratory distress syndrome: an evidence-based review. Crit Care Med 2004; 32 (11 Suppl):S548 S Dellinger RP, Carlet JM, Masur H, et al. Surviving Sepsis Campaign Management Guidelines Committee. Surviving sepsis campaign guidelines for management of severe sepsis and septic shock. Crit Care Med 2004; 32: Stewart TE. Controversies around lung protective ventilation. Am J Respir Crit Care Med 2002; 166: Walton M. The deming management method. New York: Penguin Putnam; Schuster DP. Identifying patients with ARDS: time for a different approach. Intensive Care Med 1997; 23: Brochard L, Brun-Buisson C. Clinical trials in acute respiratory distress syndrome: what is ARDS? Crit Care Med 1999; 27: Petty TL. The adult respiratory distress syndrome (confessions of a lumper ). Am Rev Respir Dis 1975; 111: Bernard GR, Artigas A, Brigham KL. The American-European Consensus Conference on ARDS: definitions, mechanisms, relevant outcomes, and clinical trial coordination. Am J Respir Crit Care Med 1994; 149: Goss CH, Brower RG, Hudson LD, et al. Incidence of acute lung injury in the United States. Crit Care Med 2003; 31: Mancebo J, Rialp G, Fernández R, et al. Prone vs supine position in ARDS patients: results of a randomized multicenter trial [abstract]. Am J Respir Crit Care Med 2003; 168 (4 Pt II):167A. 21 Gattinoni L, Tognoni G, Pesenti A, et al. Effect of prone positioning on the survival of patients with acute respiratory failure. N Engl J Med 2001; 345: Nuckton TJ, Alonso JA, Kallet RH, et al. Pulmonary dead-space fraction as a risk factor for death in the acute respiratory distress syndrome. N Engl J Med 2002; 346: Rubenfeld GD, Caldwell E, Granton J, et al. Interobserver variability in applying a radiographic definition for ARDS. Chest 1999; 116: Meade MO, Cook RJ, Guyatt GH, et al. Interobserver variation in interpreting chest radiographs for the diagnosis of acute respiratory distress syndrome. Am J Respir Crit Care Med 2000; 161:85 90.

7 Limitations of clinical trials Marini Villar J, Perez-Mendez L, Kacmarek RM. Current definitions of acute lung injury and the acute respiratory distress syndrome do not reflect their true severity and outcome. Intensive Care Med 1999; 25: Gattinoni L, Pelosi P, Suter PM, et al. Acute respiratory distress syndrome caused by pulmonary and extrapulmonary disease different syndromes? Am J Respir Crit Care Med 1998; 158: Dicker RA, Morabito DJ, Pittet JF, et al. Acute respiratory distress syndrome criteria in trauma patients: why the definitions do not work. J Trauma 2004; 57: ; discussion Patients with ARDS developing from different pathological roots do not closely resemble each other in terms of clinical behavior or response to ventilation therapy. 28 Fuhrman BP, Abraham E, Dellinger RP. Futility of randomized, controlled ARDS trials: a new approach is needed. Crit Care Med 1999; 27: Hotchkiss JR, Blanch LL, Murias G, et al. Effects of decreased respiratory frequency on ventilator induced lung injury. Am J Resp Crit Care Med 2000; 161: Rich PB, Reickert CA, Sawada S, et al. Effect of rate and inspiratory flow on ventilator-induced lung injury. J Trauma 2000; 49: Brower RG, Matthay M, Schoenfeld D. Meta-analysis of acute lung injury and acute respiratory distress syndrome trials. Am J Respir Crit Care Med 2002; 166: Ribeiro SP, Rhee K, Tremblay L, et al. Heat stress attenuates ventilatorinduced lung dysfunction in an ex vivo rat lung model. Am J Respir Crit Care Med 2001; 163: Suzuki S, Hotchkiss JR, Takahashi T, et al. Effect of core body temperature on ventilator-induced lung injury. Crit Care Med 2004; 32: With respect to high body temperature and VILI, timing may be everything, as shown in a comparison of this experimental study and the one above [32]. 34 Gattinoni L, Vagginelli F, Carlesso E, et al. Prone-Supine Study Group. Decrease in PaCO2 with prone position is predictive of improved outcome in acute respiratory distress syndrome. Crit Care Med 2003; 31: Rouby JJ, Puybasset L, Nieszkowska A, Lu Q. Acute respiratory distress syndrome: lessons from computed tomography of the whole lung. Crit Care Med 2003; 31 (4 Suppl):S285 S Cabana MD, Rand CS, Powe NR, et al. Why don t physicians follow clinical practice guidelines? A framework for improvement. JAMA 1999; 282: Hotchkiss JR, Simonson DA, Marek DJ, et al. Pulmonary microvascular fracture in a patient with acute respiratory distress syndrome. Crit Care Med 2002; 30: Hotchkiss JR, Blanch LL, Naviera A, et al. Relative roles of vascular and airspace pressures in ventilator induced lung injury. Crit Care Med 2001; 29: Milberg JA, Davis DR, Steinberg KP, Hudson LD. Improved survival of patients with acute respiratory distress syndrome (ARDS): JAMA 1995; 273: Dreyfuss D, Saumon G. Ventilator-induced lung injury: lessons from experimental studies. Am J Respir Crit Care Med 1998; 157: Hickling KG, Henderson SJ, Jackson R. Low mortality associated with low volume pressure limited ventilation with permissive hypercapnia in severe adult respiratory distress syndrome. Intensive Care Med 1990; 16: Steltzer H, Krafft P. Prognosis of ARDS patients: light at the end of the tunnel? Intensive Care Med 1997; 23: Stewart TE, Meade MO, Cook DJ, et al. Evaluation of a ventilation strategy to prevent barotrauma in patients at high risk for acute respiratory distress syndrome. N Engl J Med 1998; 338: Brochard L, Roudot-Thoraval F, Roupie E, et al. Tidal volume reduction for prevention of ventilator-induced lung injury in the acute respiratory distress syndrome. Am J Respir Crit Care Med 1998; 158: Brower RG, Shanholtz CB, Fessler HE, et al. Prospective, randomized, controlled clinical trial comparing traditional versus reduced tidal volume ventilation in acute respiratory distress syndrome. Crit Care Med 1999; 27: ARDS Clinical Trials Network, National Heart, Lung and Blood Institute, National Institutes of Health. Effects of recruitment maneuvers in patients with acute lung injury and acute respiratory distress syndrome ventilated with high positive end expiratory pressure. Crit Care Med 2003; 31: ARDS Clinical Trials Network. Higher versus lower positive end-expiratory pressures in patients with the acute respiratory distress syndrome. N Engl J Med 2004; 351: Nicely executed study with a result that squares with those expected from laboratory data higher PEEP is not necessarily a helpful strategy if plateau pressure is constrained. 48 Lim S-C, Adams AB, Simonson DA, et al. Intercomparison of recruitment maneuvers in three models of acute lung injury. Crit Care Med 2004; 32: Efficacy of recruiting maneuvers may depend on the type of maneuver and the causative stimulus for lung injury. 49 Lim C-M, Jung H, Koh Y, et al. Effect of alveolar recruitment maneuver in early acute respiratory distress syndrome according to antiderecruitment strategy, etiological category of diffuse lung injury, and body position of the patient. Crit Care Med 2003; 31: Grasso S, Mascia L, Del Turco M, et al. Effects of recruiting maneuvers in patients with acute respiratory distress syndrome ventilated with protective ventilatory strategy. Anesthesiology 2002; 96: Zigelman C, Briskin A, Hersch M. Extremely high-pressure lung recruitment maneuver may be life saving in the most severe cases of acute lung injury/ acute respiratory distress syndrome. Crit Care Med 2004; 32: Deans KJ, Minneci PC, Cui X, et al. Mechanical ventilation in ARDS: one size does not fit all. Crit Care Med 2005; 33: An interesting analysis directed toward translating the uncontested results of the well executed ARDSnet tidal volume trial into valid messages relevant to clinical practice. Notably, patients who qualified for but were excluded from the ARMA trial experienced mortality rates indistinguishable from the low tidal volume (6 ml/kg) group. 53 Eichacker PQ, Gerstenberger EP, Banks SM, et al. Meta-analysis of acute lung injury and acute respiratory distress syndrome trials with low tidal volumes. Am J Respir Crit Care Med 2002; 166: Ferguson ND, Frutos-Vivar F, Esteban A, et al. Mechanical Ventilation International Study Group. Airway pressures, tidal volumes, and mortality in patients with acute respiratory distress syndrome. Crit Care Med 2005; 33: As several surveys have shown before, this interesting paper documents the wide range of tidal volumes employed in the treatment of this diverse condition. No independent effect of end-inspiratory plateau pressure on mortality was evident for any pressure less than 30 cmh 2 O or for tidal volumes spanning the wide range in current use. 55 Barbas CSV, Medeiros DM, Magaldi RB, et al. High PEEP levels improved survival in ARDS patients [abstract]. Am J Respir Crit Care Med 2002; 165:A Rossouw JE, Anderson GL, Prentice RL, et al. Writing Group for the Women s Health Initiative Investigators. Risks and benefits of estrogen plus progestin in healthy postmenopausal women: principal results from the Women s Health Initiative randomized controlled trial. JAMA 2002; 288: Ioannidis JP. Contradicted and initially stronger effects in highly cited clinical research. JAMA 2005; 294: A disturbing scholarly analysis that indicates that the results of even large RCTs may not hold up as new data come to light. 58 Poynard T, Munteanu M, Ratziu V, et al. Truth survival in clinical research: an evidence-based requiem? Ann Intern Med 2002; 136: Amato M, Brochard L, Stewart T, Brower R. Metaanalysis of tidal volume in ARDS. Am J Respir Crit Care Med 2003; 168: Villar J, Perez-Mendez L, Aguirre-Jaime A, Kacmarek RM. Why are physicians so skeptical about positive randomized controlled clinical trials in critical care medicine? Intensive Care Med 2005; 31: Are practicing physicians inappropriate in their reticence to be early adopters and accept the results of RCTs?. 61 Rubenfeld GD, Cooper C, Carter G, et al. Barriers to providing lung-protective ventilation to patients with acute lung injury. Crit Care Med 2004; 32: