Tidal Volume Reduction in Patients with Acute Lung Injury When Plateau Pressures Are Not High
http://www.100md.com
《美国呼吸和危急护理医学》
Department of Medicine, Johns Hopkins University, Baltimore, Maryland
Biostatistics Center, Massachusetts General Hospital, Boston, Massachusetts
ABSTRACT
Use of a volume- and pressure-limited mechanical ventilation strategy improves clinical outcomes of patients with acute lung injury and acute respiratory distress syndrome (ALI/ARDS). However, the extent to which tidal volumes and inspiratory airway pressures should be reduced to optimize clinical outcomes is a controversial topic. This article addresses the question, "Is there a safe upper limit to inspiratory plateau pressure in patients with ALI/ARDS" We reviewed data from animal models with and without preexisting lung injury, studies of normal human respiratory system mechanics, and the results of five clinical trials of lung-protective mechanical ventilation strategies. We also present an original analysis of data from the largest of the five clinical trials. The available data from each of these assessments do not support the commonly held view that inspiratory plateau pressures of 30 to 35 cm H2O are safe. We could not identify a safe upper limit for plateau pressures in patients with ALI/ARDS.
Key Words: acute respiratory distress syndrome acute lung injury plateau mechanical ventilation
Ventilator-induced lung injury (VILI) is reduced and clinical outcomes of patients with acute lung injury and the acute respiratory distress syndrome (ALI/ARDS) are improved if the mechanical ventilation (MV) approach uses smaller tidal volumes (VT) and lower inspiratory pressures than were used in the past (1–7). However, there is controversy regarding the extent to which VT and inspiratory airway pressures should be reduced to achieve these objectives (8–18). Some investigators have recommended that VT in patients with ALI/ARDS should be reduced to maintain inspiratory plateau pressures (Pplat) of less than 30 to 35 cm H2O (or that MV with pressure-controlled modes should limit inspiratory pressures to no more than 30–35 cm H2O) (1, 11–13). This suggests that Pplat lower than 30 to 35 cm H2O may be considered safe, and that further reductions in VT and Pplat are without benefit. Several lines of evidence have been interpreted to support this suggestion, including the results of animal models of VILI (19, 20), considerations of normal human physiology (21), and comparisons of clinical trials of lung-protective MV strategies (3, 4, 22–24). The purpose of this article is to review these lines of evidence and present an original analysis of the data from a large clinical trial to determine if there is a sound scientific basis to consider VT reduction unnecessary when Pplat levels are lower than 30 to 35 cm H2O. Some of the results of the original analysis contained in this review were previously reported in an abstract (25).
ANIMAL MODELS OF VILI
In a consensus statement regarding the use of MV, the investigators said, "Based primarily on animal data, a plateau pressure 35 cm H2O is of concern. We, therefore, recommend that when plateau pressure equals or exceeds this pressure, that tidal volume can be decreased" (1). Abundant data support this statement (6, 7, 19, 20, 26–33). However, many animal studies suggest that, under some circumstances, Pplat lower than 35 cm H2O should also be of concern.
In intact rats without other causes of ALI, perivascular edema occurred in all animals ventilated for 60 min with peak inspiratory pressures (PIP) of 30 cm H2O (26, 28) and in three of six animals ventilated with PIP of 14 cm H2O (28). In isolated rat lungs, edema occurred at PIP as low as 13 cm H2O (29). In animal models with other causes of ALI, VILI has been observed at relatively low airway pressures and VT (6, 7). In intact rats with hydrochloric acid–induced lung injury, extravascular lung water was 55% lower after 4 h of MV with Pplat of 21 cm H2O (VT, 6 ml/kg) than with Pplat of 30 cm H2O (VT, 12 ml/kg). Lung water was reduced further when Pplat levels of 16 cm H2O were used (VT, 3 ml/kg; Figure E1 in the online supplement) (6). Histologic examination of lung tissue and indices of endothelial and epithelial injury also showed less injury at lower VT and airway pressures. A similar study compared alveolar protein permeability in a rabbit model of Pseudomonas pneumonia. After 8 h, the PIP of animals ventilated with VT of 6 and 15 ml/kg increased from 17 to 21 cm H2O and from 22 to 35 cm H2O, respectively. Alveolar protein permeability was 30% lower in animals that received the lower VT and PIP (7).
Although data from animal models of VILI vary considerably, depending on species, preparation (open or closed chest), duration of MV, and presence or absence of other causes of lung injury (19), these studies suggest that the Pplat or PIP threshold for VILI in rats and rabbits is considerably lower than 30 to 35 cm H2O. However, rodent chest wall compliance is high relative to lung compliance, and VILI from overdistention probably occurs at lower airway pressures than in larger animals or humans. Therefore, direct application of the results of these small animal studies to the management of clinical ALI/ARDS is difficult. Studies in large animal models may be more informative.
In isolated dog lobes ventilated for 20 to 30 min, there was little indication of VILI (increased microvascular permeability) unless PIP exceeded 42 cm H2O (32). In open-chest dogs, VILI (increased lung lymph flow and decreased lymph-to-plasma protein ratio) occurred after 30 min of MV with PIP of 64 cm H2O but not with PIP of 22 cm H2O (33). In intact lambs, lung lymph flow and protein concentration did not change significantly after 4 h of MV with PIP of 33 and then 43 cm H2O. However, lymph flow and protein concentration increased after an additional 4 h of MV with PIP of 61 cm H2O, indicating that pulmonary vascular permeability had increased (34). These studies in dogs and lambs suggest that VILI may not occur until PIP levels are as high as 40 to 60 cm H2O. However, these experiments were relatively short in duration. Longer periods of MV could lead to VILI at lower pressures. In the studies in intact lambs (34), the 4-h period of MV at the highest PIP occurred after sequential 4-h periods of MV at PIP of 33 and 43 cm H2O in the same animals. The indications of vascular injury late in the experiments may represent the cumulative effect of MV at the lower PIP in addition to VILI caused by the final period of MV at the highest PIP. Unfortunately, few studies have considered the effects of prolonged periods (many hours to days) of MV at moderate airway pressures and lung injury. However, in one notable study in intact sheep, VILI (increased lung water and decreased surfactant function) was observed after 48 h of MV with PIP of 30 cm H2O (31). In many of these animals, VT was reduced over time to limit PIP to 30 cm H2O. This suggests that a strategy of reducing VT only when PIP levels exceed 30 cm H2O may not be sufficiently protective.
In summary, abundant evidence from animal models supports the concern for Pplat greater than 35 cm H2O. Some studies suggest that VILI does not occur after relatively brief periods of MV with PIP as high as 40 to 60 cm H2O. However, there is little evidence to support the safety of prolonged MV with Pplat as high as 30 to 35 cm H2O.
CONSIDERATIONS OF NORMAL HUMAN PHYSIOLOGY
The distending pressure of the normal relaxed human respiratory system (equivalent to Pplat) at total lung capacity is approximately 37 cm H2O (21, 35). Normal humans can inspire voluntarily to their total lung capacities without apparent adverse effects, suggesting that the mechanical forces associated with these distending pressures are safe (36). However, we were unable to find any evidence that these forces and pressures at total lung capacity are safe when applied thousands of times over periods of hours or days, with or without preexisting injury. On the other hand, experimental evidence suggests that spontaneous breathing with large VT may be injurious. Sodium salicylate injected into the cisterna magna of intact sheep caused spontaneous ventilation with large VT and rapid respiratory rates (30). In a control group, sheep received sodium salicylate injections, neuromuscular blockade, and MV with physiologic VT and respiratory rates. After approximately 12 h, the alveolar-arterial PO2 gradients were significantly higher in the animals that hyperventilated spontaneously, as were the lung-to-bodyweight ratios. These findings could represent a hydrostatic effect of spontaneous breathing with pleural pressures that are lower than those that occur during MV (37). However, there were histologic changes consistent with ALI in the lungs of the animals that breathed spontaneously. These results suggest that the mechanical forces in the lungs during spontaneous breathing with supraphysiologic VT may not be safe over periods of hours to days.
REVIEW OF CLINICAL TRIALS OF LUNG-PROTECTIVE VENTILATION STRATEGIES
Five clinical trials compared outcomes of patients with ALI/ARDS randomized to either a MV strategy with higher VT and relatively high inspiratory airway pressures or a volume- and pressure-limited MV strategy (Table 1) (3, 4, 22–24). In the two studies in which volume- and pressure-limited strategies were associated with lower mortality, mean Pplat in the higher VT study groups exceeded 32 cm H2O (3, 4). In the three studies in which volume- and pressure-limited strategies were not associated with lower mortality (nonbeneficial studies), mean Pplat levels in the higher VT groups were lower than 32 cm H2O (22–24). This suggested to some that 32 cm H2O may be a safe Pplat threshold (11–13, 38). There were modest differences in the mean VT and resulting Pplat in the different study groups between the beneficial and nonbeneficial studies, and these differences may have contributed to the variable outcomes (39). However, in each of the nonbeneficial studies (22–24), the numbers of patients in the higher VT groups whose Pplat exceeded 32 cm H2O were considerably greater than in the volume- and pressure-limited groups (Figure E2) (24). If 32 cm H2O is the critical value that separates safe from unsafe Pplat, then mortality should have trended lower in the volume- and pressure-limited study groups of these trials. The absence of such trends in these studies suggests that there may have been imbalances in the randomization groups at baseline that favored patients in the higher VT study groups (40).
VT REDUCTION AND Pplat IN PATIENTS WITH ALI/ARDS
The ARDS Network trial of MV with higher versus lower VT enrolled 861 patients with ALI/ARDS (4). Of these, Pplat levels (obtained with a 0.5-s inspiratory hold) were available on the first day after randomization (Day 1 Pplat) for 93% of patients in the higher VT group (n = 399) and 90% of patients in the lower VT group (n = 388). We used data from these patients to address the following question: "Is there a safe Pplat below which there is no beneficial effect of VT reduction" To demonstrate the relationship of mortality versus Pplat for all patients, we constructed a plot of mortality against Day 1 Pplat (Figure 1). The relationship shows decreasing mortality as Day 1 Pplat declines from high to low levels. It does not reveal a safe Pplat threshold within the range of Day 1 Pplat levels measured in patients with ALI/ARDS. On the other hand, this relationship should not be interpreted to mean that VT of patients with ALI/ARDS should be decreased below 6 ml/kg to achieve very low Pplat. Most of the data represented at the very low Pplat levels were obtained in patients with relatively high respiratory system compliances while they received VT of 6 ml/kg predicted bodyweight. VT was not decreased below 6 ml/kg in these patients. Effects on mortality of reducing VT below 6 ml/kg in patients whose Pplat levels are 30 cm H2O or lower are unknown.
To assess for independent effects of VT reduction and Pplat on mortality, we constructed a multivariable logistic regression model. For this purpose, each study group was stratified by quartile of Day 1 Pplat. With this approach, we identified groups of patients who would have had similar Pplat had they been randomized to the same VT strategy. Corresponding quartiles in each randomization group have different Pplat ranges because VT is one of the determinants of Pplat (41). For example, Day 1 Pplat levels in the lowest Pplat quartile (Quartile 1) of the higher VT group (range, 16–26 cm H2O) are higher than the Day 1 Pplat levels of Quartile 1 in the lower VT group (range, 10–20 cm H2O). Stratifying patients with similar Pplat but different VT into corresponding subsets would have been inappropriate for this purpose because it would have classified patients with more severe lung injury receiving lower VT (i.e., those with the lower respiratory system compliances) into a subset that corresponded to patients with less severe injury receiving higher VT.
In a bivariate (simple) logistic regression, the lower VT strategy was associated with a lower mortality than the higher VT strategy (p = 0.02). Bivariate analysis also demonstrated that lower Pplat quartiles were associated with reduced mortality when compared with higher Pplat quartiles (p 0.039; Table 2).
In addition to VT assignment and Pplat quartile, we included in our multivariable logistic regression model any baseline characteristic in which there was a trend toward a difference (p 0.10) between corresponding quartiles. The only such variable was Acute Physiology, Age, and Chronic Health Evaluation (APACHE) III score (42), which was higher in Quartile 1 of the higher VT group (84.5 ± 24.3 vs. 78.1 ± 24.3 [mean ± SD], p = 0.07; Table E1). In the final regression model, lower VT assignment, lower Pplat quartile, and lower APACHE score were all significant predictors of lower mortality (Table 2). Importantly, the interaction between VT assignment and Pplat quartiles was not significant (p 0.23 for all quartiles). This suggests that patients in the higher VT group would have benefited from VT reduction in each of the quartiles, including the two in which Pplat levels were already 31 cm H2O or lower (Figure 2).
There were 399 patients in the two lower Pplat quartiles of this analysis, including 205 patients with higher VT whose Day 1 Pplat levels were 31 cm H2O or lower. The reduction in mortality associated with the volume- and pressure-limited strategy in these patients was 6.9%. This difference was not statistically significant (p = 0.13). However, the original clinical trial required enrollment of as many as 1,000 subjects to detect a difference in mortality between study groups of 10% (50 and 40%). A trial designed to demonstrate a difference of 6.9% would require enrollment of up to 1,900 subjects (two-sided test, = 0.05, power = 0.90).
In a previous report, we analyzed the effects of VT reduction on mortality after dividing both study groups into quartiles according to baseline respiratory system compliance (4). Mortality rates in the highest compliance quartiles (analogous to lowest Pplat quartiles) did not favor the volume- and pressure-limited group. We have more confidence in our Pplat quartile analysis because the required data were available for a greater number of patients (787 compared with 517). Moreover, VT can affect compliance when compliance is measured during tidal ventilation (higher VT was associated with higher measured compliance [4]), and the range of baseline VT was broad (17). In contrast, the Day 1 Pplat levels were obtained when VT was tightly controlled in each study group. Therefore, the Pplat quartiles better represent disease severity than the compliance quartiles. Importantly, in both the compliance and Pplat quartile analyses, the interaction between quartiles and VT study group was not significant, suggesting that VT reduction was beneficial regardless of baseline compliance or Day 1 Pplat.
CONCLUSIONS
The ARDS Network volume- and pressure-limited strategy used a VT goal of 6 ml/kg predicted bodyweight (4). With this approach, mean Pplat was approximately 25 cm H2O. Most patients would have had Pplat below 31 cm H2O on VT that was greater than 6 ml/kg and less than or equal to 12 ml/kg. Some investigators have questioned the value of VT reduction in patients with ALI/ARDS whose Pplat levels are already below 30 to 35 cm H2O. We have been unable to find justification for this position after reviewing the results of animal models and comparing the results of clinical trials of lung-protective ventilation strategies. Our secondary analysis of the ARDS Network database suggests that there was a beneficial effect of VT reduction from 12 to 6 ml/kg predicted bodyweight, regardless of the Pplat before VT reduction. We do not advocate using VT of less than 6 ml/kg predicted bodyweight to achieve very low Pplat. Potential adverse effects of volume- and pressure-limited MV should be considered in all patients. Hypercapnia may cause elevated intracranial pressure, pulmonary hypertension, decreased myocardial contractility, decreased renal blood flow, and the release of endogenous catecholamines (43, 44). Moreover, MV with low VT and Pplat may cause more atelectasis and increase requirements for higher FIO2 and positive end-expiratory pressure (4, 45). All of these variables should be considered when titrating VT (46). However, we could not substantiate the widespread belief that VT reduction is without benefit when Pplat is already lower than 30 to 35 cm H2O.
Acknowledgments
The authors thank Dr. Henry E. Fessler for reviewing the manuscript and providing useful comments.
FOOTNOTES
Supported by contracts (N01-HR46063, N01-HR46064, P50-HL073944, F32-HL078132, and K23-HL67850) with the National Institutes of Health, National Heart, Lung, and Blood Institute ARDS Clinical Trials Network.
This article has an online supplement, which is accessible from this issue's table of contents at www.atsjournals.org
Originally Published in Press as DOI: 10.1164/rccm.200501-048CP August 4, 2005
Conflict of Interest Statement: None of the authors have a financial relationship with a commercial entity that has an interest in the subject of this manuscript.
REFERENCES
Slutsky AS. Mechanical ventilation. American College of Chest Physicians' Consensus Conference. Chest 1993;104:1833–1859.
Kallet RH, Jasmer RM, Pittet JF, Tang JF, Campbell AR, Dicker R, Hemphill C, Luce JM. Clinical implementation of the ARDS network protocol is associated with reduced hospital mortality compared with historical controls. Crit Care Med 2005;33:925–929.
Amato MB, Barbas CS, Medeiros DM, Magaldi RB, Schettino GP, Lorenzi-Filho G, Kairalla RA, Deheinzelin D, Munoz C, Oliveira R, et al. Effect of a protective-ventilation strategy on mortality in the acute respiratory distress syndrome. N Engl J Med 1998;338:347–354.
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.
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;16:372–377.
Frank JA, Gutierrez JA, Jones KD, Allen L, Dobbs L, Matthay MA. Low tidal volume reduces epithelial and endothelial injury in acid-injured rat lungs. Am J Respir Crit Care Med 2002;165:242–249.
Savel RH, Yao EC, Gropper MA. Protective effects of low tidal volume ventilation in a rabbit model of Pseudomonas aeruginosa-induced acute lung injury. Crit Care Med 2001;29:392–398.
Carmichael LC. Tidal volumes in ARDS and meta-analysis. Am J Respir Crit Care Med 2003;167:933–934.
Deans KJ, Minneci PC, Cui X, Banks SM, Natanson C, Eichacker PQ. Mechanical ventilation in ARDS: one size does not fit all. Crit Care Med 2005;33:1141–1143.
Gattinoni L, Carlesso E, Cadringher P, Valenza F, Vagginelli F, Chiumello D. Physical and biological triggers of ventilator-induced lung injury and its prevention. Eur Respir J Suppl 2003;47:15s–25s.
Eichacker PQ, Gerstenberger EP, Banks SM, Cui X, Natanson C. Meta-analysis of acute lung injury and acute respiratory distress syndrome trials testing low tidal volumes. Am J Respir Crit Care Med 2002;166: 1510–1514.
Tobin MJ. Culmination of an era in research on the acute respiratory distress syndrome. N Engl J Med 2000;342:1360–1361.
Ricard JD. Are we really reducing tidal volume—and should we Am J Respir Crit Care Med 2003;167:1297–1298.
Stewart T. Controversies around lung protective mechanical ventilation. Am J Respir Crit Care Med 2002;166:1510–1514.
Petty TL. Tidal volumes in ARDS and meta-analysis. Am J Respir Crit Care Med 2002;167:933.
Dreyfuss D, Saumon G. Evidence-based medicine or fuzzy logic: what is best for ARDS management Intensive Care Med 2002;28:230–234.
Brower RG, Matthay MA, Schoenfeld D. Meta-analysis of acutel lung injury and acute respiratory distress syndrome trials. Am J Respir Crit Care Med 2002;166:1515–1517.
Vincent JL. Evidence-based medicine in the ICU: important advances and limitations. Chest 2004;126:592–600.
Dreyfuss D, Saumon G. Ventilator-induced lung injury: lessons from experimental studies. Am J Respir Crit Care Med 1998;157:294–323.
Parker JC, Hernandez LA, Peevy KJ. Mechanisms of ventilator-induced lung injury. Crit Care Med 1993;21:131–143.
Rahn H, Otis AB, Chadwick LE, Fenn WO. The pressure-volume diagram of the thorax and lung. Am J Physiol 1946;146:161–178.
Stewart TE, Meade MO, Cook DJ, Granton JT, Hodder RV, Lapinsky SE, Mazer CD, McLean RF, Rogovein TS, Schouten BD, et al. Evaluation of a ventilation strategy to prevent barotrauma in patients at high risk for acute respiratory distress syndrome. Pressure- and Volume-limited Ventilation Strategy Group. N Engl J Med 1998;338:355–361.
Brower RG, Shanholtz CB, Fessler HE, Shade DM, White P Jr, Wiener CM, Teeter JG, Dodd-o JM, Almog Y, Piantadosi S. Prospective, randomized, controlled clinical trial comparing traditional versus reduced tidal volume ventilation in acute respiratory distress syndrome patients. Crit Care Med 1999;27:1492–1498.
Brochard L, Roudot-Thoraval F, Roupie E, Delclaux C, Chastre J, Fernandez-Mondejar E, Clementi E, Mancebo J, Factor P, Matamis D, et al. Tidal volume reduction for prevention of ventilator-induced lung injury in acute respiratory distress syndrome. The Multicenter Trail Group on Tidal Volume Reduction in ARDS. Am J Respir Crit Care Med 1998;158:1831–1838.
Brower RG, Krishnan JA, Thompson BT, Wheeler AP, Weideman H, Schoenfeld D, Bernard G. Effects of tidal volume reduction (Vt-R) in acute lung injury (ALI) patients with inspiratory plateau pressures (Pplat) < 32 cm H2O before Vt-R . Am J Respir Crit Care Med 2003;167:A616.
Cilley RE, Wang JY, Coran AG. Lung injury produced by moderate lung overinflation in rats. J Pediatr Surg 1993;28:488–493.
Dreyfuss D, Soler P, Basset G, Saumon G. High inflation pressure pulmonary edema: respective effects of high airway pressure, high tidal volume, and positive end-expiratory pressure. Am Rev Respir Dis 1988;137:1159–1164.
Webb HH, Tierney DF. Experimental pulmonary edema due to intermittent positive pressure ventilation with high inflation pressures: protection by positive end-expiratory pressure. Am Rev Respir Dis 1974;110: 556–565.
Omlor G, Niehaus GD, Maron MB. Effect of peak inspiratory pressure on the filtration coefficient in the isolated perfused rat lung. J Appl Physiol 1993;74:3068–3072.
Mascheroni D, Kolobow T, Fumagalli R, Moretti MP, Chen V, Buckhold D. Acute respiratory failure following pharmacologically induced hyperventilation: an experimental animal study. Intensive Care Med 1988;15:8–14.
Tsuno K, Prato P, Kolobow T. Acute lung injury from mechanical ventilation at moderately high airway pressures. J Appl Physiol 1990;69: 956–961.
Parker JC, Townsley MI, Rippe B, Taylor AE, Thigpen J. Increased microvascular permeability in dog lungs due to high peak airway pressures. J Appl Physiol 1984;57:1809–1816.
Parker JC, Hernandez LA, Longenecker GL, Peevy K, Johnson W. Lung edema caused by high peak inspiratory pressures in dogs: role of increased microvascular filtration pressure and permeability. Am Rev Respir Dis 1990;142:321–328.
Carlton DP, Cummings JJ, Scheerer RG, Poulain FR, Bland RD. Lung overexpansion increases pulmonary microvascular protein permeability in young lambs. J Appl Physiol 1990;69:577–583.
Taylor RW, Zimmerman JL, Dellinger RP, Straube RC, Criner GJ, Davis K Jr, Kelly KM, Smith TC, Small RJ. Low-dose inhaled nitric oxide in patients with acute lung injury: a randomized controlled trial. JAMA 2004;291:1603–1609.
Marini JJ. New options for the ventilatory management of acute lung injury. New Horiz 1993;1:489–503.
Scharf SM. Mechanical cardiopulmonary interactions in critical care. In: Dantzker DR, Scharf SM, editors. Cardiopulmonary critical care, 3rd ed. Philadelphia: WB Saunders; 1998. pp.75–91.
Tobin MJ. Update in pulmonary diseases. Ann Intern Med 2005;142: 283–288.
Brower RG. Mechanical ventilation in acute lung injury and ARDS: ridal volume reduction. Crit Care Clin 2002;18:1–13.
Amato MB, Brochard L, Stewart TE, Brower RG. Meta-analysis of tidal volume in ARDS. Am J Respir Crit Care Med 2003;168:612.
Hubmayr RD, Abel MD, Rehder K. Physiologic approach to mechanical ventilation. Crit Care Med 1990;18:103–113.
Knaus WA, Wagner DP, Draper EA, Zimmerman JE, Bergner M, Bastos PG, Sirio CA, Murphy DJ, Lotring T, Damiano A. The APACHE III prognostic system: risk prediction of hospital mortality for critically ill hospitalized adults. Chest 1991;100:1619–1636.
Feihl F, Eckert P, Brimioulle S, Jacobs O, Schaller MD, Melot C, Naeije R. Permissive hypercapnia impairs pulmonary gas exchange in the acute respiratory distress syndrome. Am J Respir Crit Care Med 2000; 162:209–215.
Feihl F, Perret C. Permissive hypercapnia: how permissive should we be Am J Respir Crit Care Med 1994;150:1722–1737.
Cheney FW Jr, Burnham SC. Effect of ventilatory pattern on oxygenation in pulmonary edema. J Appl Physiol 1971;31:909–912.
Dellinger RP. Positive clinical impact of low tidal volume strategy. Crit Care Med 2005;33:1143–1144.
Cleveland WS. Robust locally weighted regression and smoothing scatterplots. J Am Stat Assoc 1979;74:829–836.(David N. Hager, Jerry A. Krishnan, Dougl)
Biostatistics Center, Massachusetts General Hospital, Boston, Massachusetts
ABSTRACT
Use of a volume- and pressure-limited mechanical ventilation strategy improves clinical outcomes of patients with acute lung injury and acute respiratory distress syndrome (ALI/ARDS). However, the extent to which tidal volumes and inspiratory airway pressures should be reduced to optimize clinical outcomes is a controversial topic. This article addresses the question, "Is there a safe upper limit to inspiratory plateau pressure in patients with ALI/ARDS" We reviewed data from animal models with and without preexisting lung injury, studies of normal human respiratory system mechanics, and the results of five clinical trials of lung-protective mechanical ventilation strategies. We also present an original analysis of data from the largest of the five clinical trials. The available data from each of these assessments do not support the commonly held view that inspiratory plateau pressures of 30 to 35 cm H2O are safe. We could not identify a safe upper limit for plateau pressures in patients with ALI/ARDS.
Key Words: acute respiratory distress syndrome acute lung injury plateau mechanical ventilation
Ventilator-induced lung injury (VILI) is reduced and clinical outcomes of patients with acute lung injury and the acute respiratory distress syndrome (ALI/ARDS) are improved if the mechanical ventilation (MV) approach uses smaller tidal volumes (VT) and lower inspiratory pressures than were used in the past (1–7). However, there is controversy regarding the extent to which VT and inspiratory airway pressures should be reduced to achieve these objectives (8–18). Some investigators have recommended that VT in patients with ALI/ARDS should be reduced to maintain inspiratory plateau pressures (Pplat) of less than 30 to 35 cm H2O (or that MV with pressure-controlled modes should limit inspiratory pressures to no more than 30–35 cm H2O) (1, 11–13). This suggests that Pplat lower than 30 to 35 cm H2O may be considered safe, and that further reductions in VT and Pplat are without benefit. Several lines of evidence have been interpreted to support this suggestion, including the results of animal models of VILI (19, 20), considerations of normal human physiology (21), and comparisons of clinical trials of lung-protective MV strategies (3, 4, 22–24). The purpose of this article is to review these lines of evidence and present an original analysis of the data from a large clinical trial to determine if there is a sound scientific basis to consider VT reduction unnecessary when Pplat levels are lower than 30 to 35 cm H2O. Some of the results of the original analysis contained in this review were previously reported in an abstract (25).
ANIMAL MODELS OF VILI
In a consensus statement regarding the use of MV, the investigators said, "Based primarily on animal data, a plateau pressure 35 cm H2O is of concern. We, therefore, recommend that when plateau pressure equals or exceeds this pressure, that tidal volume can be decreased" (1). Abundant data support this statement (6, 7, 19, 20, 26–33). However, many animal studies suggest that, under some circumstances, Pplat lower than 35 cm H2O should also be of concern.
In intact rats without other causes of ALI, perivascular edema occurred in all animals ventilated for 60 min with peak inspiratory pressures (PIP) of 30 cm H2O (26, 28) and in three of six animals ventilated with PIP of 14 cm H2O (28). In isolated rat lungs, edema occurred at PIP as low as 13 cm H2O (29). In animal models with other causes of ALI, VILI has been observed at relatively low airway pressures and VT (6, 7). In intact rats with hydrochloric acid–induced lung injury, extravascular lung water was 55% lower after 4 h of MV with Pplat of 21 cm H2O (VT, 6 ml/kg) than with Pplat of 30 cm H2O (VT, 12 ml/kg). Lung water was reduced further when Pplat levels of 16 cm H2O were used (VT, 3 ml/kg; Figure E1 in the online supplement) (6). Histologic examination of lung tissue and indices of endothelial and epithelial injury also showed less injury at lower VT and airway pressures. A similar study compared alveolar protein permeability in a rabbit model of Pseudomonas pneumonia. After 8 h, the PIP of animals ventilated with VT of 6 and 15 ml/kg increased from 17 to 21 cm H2O and from 22 to 35 cm H2O, respectively. Alveolar protein permeability was 30% lower in animals that received the lower VT and PIP (7).
Although data from animal models of VILI vary considerably, depending on species, preparation (open or closed chest), duration of MV, and presence or absence of other causes of lung injury (19), these studies suggest that the Pplat or PIP threshold for VILI in rats and rabbits is considerably lower than 30 to 35 cm H2O. However, rodent chest wall compliance is high relative to lung compliance, and VILI from overdistention probably occurs at lower airway pressures than in larger animals or humans. Therefore, direct application of the results of these small animal studies to the management of clinical ALI/ARDS is difficult. Studies in large animal models may be more informative.
In isolated dog lobes ventilated for 20 to 30 min, there was little indication of VILI (increased microvascular permeability) unless PIP exceeded 42 cm H2O (32). In open-chest dogs, VILI (increased lung lymph flow and decreased lymph-to-plasma protein ratio) occurred after 30 min of MV with PIP of 64 cm H2O but not with PIP of 22 cm H2O (33). In intact lambs, lung lymph flow and protein concentration did not change significantly after 4 h of MV with PIP of 33 and then 43 cm H2O. However, lymph flow and protein concentration increased after an additional 4 h of MV with PIP of 61 cm H2O, indicating that pulmonary vascular permeability had increased (34). These studies in dogs and lambs suggest that VILI may not occur until PIP levels are as high as 40 to 60 cm H2O. However, these experiments were relatively short in duration. Longer periods of MV could lead to VILI at lower pressures. In the studies in intact lambs (34), the 4-h period of MV at the highest PIP occurred after sequential 4-h periods of MV at PIP of 33 and 43 cm H2O in the same animals. The indications of vascular injury late in the experiments may represent the cumulative effect of MV at the lower PIP in addition to VILI caused by the final period of MV at the highest PIP. Unfortunately, few studies have considered the effects of prolonged periods (many hours to days) of MV at moderate airway pressures and lung injury. However, in one notable study in intact sheep, VILI (increased lung water and decreased surfactant function) was observed after 48 h of MV with PIP of 30 cm H2O (31). In many of these animals, VT was reduced over time to limit PIP to 30 cm H2O. This suggests that a strategy of reducing VT only when PIP levels exceed 30 cm H2O may not be sufficiently protective.
In summary, abundant evidence from animal models supports the concern for Pplat greater than 35 cm H2O. Some studies suggest that VILI does not occur after relatively brief periods of MV with PIP as high as 40 to 60 cm H2O. However, there is little evidence to support the safety of prolonged MV with Pplat as high as 30 to 35 cm H2O.
CONSIDERATIONS OF NORMAL HUMAN PHYSIOLOGY
The distending pressure of the normal relaxed human respiratory system (equivalent to Pplat) at total lung capacity is approximately 37 cm H2O (21, 35). Normal humans can inspire voluntarily to their total lung capacities without apparent adverse effects, suggesting that the mechanical forces associated with these distending pressures are safe (36). However, we were unable to find any evidence that these forces and pressures at total lung capacity are safe when applied thousands of times over periods of hours or days, with or without preexisting injury. On the other hand, experimental evidence suggests that spontaneous breathing with large VT may be injurious. Sodium salicylate injected into the cisterna magna of intact sheep caused spontaneous ventilation with large VT and rapid respiratory rates (30). In a control group, sheep received sodium salicylate injections, neuromuscular blockade, and MV with physiologic VT and respiratory rates. After approximately 12 h, the alveolar-arterial PO2 gradients were significantly higher in the animals that hyperventilated spontaneously, as were the lung-to-bodyweight ratios. These findings could represent a hydrostatic effect of spontaneous breathing with pleural pressures that are lower than those that occur during MV (37). However, there were histologic changes consistent with ALI in the lungs of the animals that breathed spontaneously. These results suggest that the mechanical forces in the lungs during spontaneous breathing with supraphysiologic VT may not be safe over periods of hours to days.
REVIEW OF CLINICAL TRIALS OF LUNG-PROTECTIVE VENTILATION STRATEGIES
Five clinical trials compared outcomes of patients with ALI/ARDS randomized to either a MV strategy with higher VT and relatively high inspiratory airway pressures or a volume- and pressure-limited MV strategy (Table 1) (3, 4, 22–24). In the two studies in which volume- and pressure-limited strategies were associated with lower mortality, mean Pplat in the higher VT study groups exceeded 32 cm H2O (3, 4). In the three studies in which volume- and pressure-limited strategies were not associated with lower mortality (nonbeneficial studies), mean Pplat levels in the higher VT groups were lower than 32 cm H2O (22–24). This suggested to some that 32 cm H2O may be a safe Pplat threshold (11–13, 38). There were modest differences in the mean VT and resulting Pplat in the different study groups between the beneficial and nonbeneficial studies, and these differences may have contributed to the variable outcomes (39). However, in each of the nonbeneficial studies (22–24), the numbers of patients in the higher VT groups whose Pplat exceeded 32 cm H2O were considerably greater than in the volume- and pressure-limited groups (Figure E2) (24). If 32 cm H2O is the critical value that separates safe from unsafe Pplat, then mortality should have trended lower in the volume- and pressure-limited study groups of these trials. The absence of such trends in these studies suggests that there may have been imbalances in the randomization groups at baseline that favored patients in the higher VT study groups (40).
VT REDUCTION AND Pplat IN PATIENTS WITH ALI/ARDS
The ARDS Network trial of MV with higher versus lower VT enrolled 861 patients with ALI/ARDS (4). Of these, Pplat levels (obtained with a 0.5-s inspiratory hold) were available on the first day after randomization (Day 1 Pplat) for 93% of patients in the higher VT group (n = 399) and 90% of patients in the lower VT group (n = 388). We used data from these patients to address the following question: "Is there a safe Pplat below which there is no beneficial effect of VT reduction" To demonstrate the relationship of mortality versus Pplat for all patients, we constructed a plot of mortality against Day 1 Pplat (Figure 1). The relationship shows decreasing mortality as Day 1 Pplat declines from high to low levels. It does not reveal a safe Pplat threshold within the range of Day 1 Pplat levels measured in patients with ALI/ARDS. On the other hand, this relationship should not be interpreted to mean that VT of patients with ALI/ARDS should be decreased below 6 ml/kg to achieve very low Pplat. Most of the data represented at the very low Pplat levels were obtained in patients with relatively high respiratory system compliances while they received VT of 6 ml/kg predicted bodyweight. VT was not decreased below 6 ml/kg in these patients. Effects on mortality of reducing VT below 6 ml/kg in patients whose Pplat levels are 30 cm H2O or lower are unknown.
To assess for independent effects of VT reduction and Pplat on mortality, we constructed a multivariable logistic regression model. For this purpose, each study group was stratified by quartile of Day 1 Pplat. With this approach, we identified groups of patients who would have had similar Pplat had they been randomized to the same VT strategy. Corresponding quartiles in each randomization group have different Pplat ranges because VT is one of the determinants of Pplat (41). For example, Day 1 Pplat levels in the lowest Pplat quartile (Quartile 1) of the higher VT group (range, 16–26 cm H2O) are higher than the Day 1 Pplat levels of Quartile 1 in the lower VT group (range, 10–20 cm H2O). Stratifying patients with similar Pplat but different VT into corresponding subsets would have been inappropriate for this purpose because it would have classified patients with more severe lung injury receiving lower VT (i.e., those with the lower respiratory system compliances) into a subset that corresponded to patients with less severe injury receiving higher VT.
In a bivariate (simple) logistic regression, the lower VT strategy was associated with a lower mortality than the higher VT strategy (p = 0.02). Bivariate analysis also demonstrated that lower Pplat quartiles were associated with reduced mortality when compared with higher Pplat quartiles (p 0.039; Table 2).
In addition to VT assignment and Pplat quartile, we included in our multivariable logistic regression model any baseline characteristic in which there was a trend toward a difference (p 0.10) between corresponding quartiles. The only such variable was Acute Physiology, Age, and Chronic Health Evaluation (APACHE) III score (42), which was higher in Quartile 1 of the higher VT group (84.5 ± 24.3 vs. 78.1 ± 24.3 [mean ± SD], p = 0.07; Table E1). In the final regression model, lower VT assignment, lower Pplat quartile, and lower APACHE score were all significant predictors of lower mortality (Table 2). Importantly, the interaction between VT assignment and Pplat quartiles was not significant (p 0.23 for all quartiles). This suggests that patients in the higher VT group would have benefited from VT reduction in each of the quartiles, including the two in which Pplat levels were already 31 cm H2O or lower (Figure 2).
There were 399 patients in the two lower Pplat quartiles of this analysis, including 205 patients with higher VT whose Day 1 Pplat levels were 31 cm H2O or lower. The reduction in mortality associated with the volume- and pressure-limited strategy in these patients was 6.9%. This difference was not statistically significant (p = 0.13). However, the original clinical trial required enrollment of as many as 1,000 subjects to detect a difference in mortality between study groups of 10% (50 and 40%). A trial designed to demonstrate a difference of 6.9% would require enrollment of up to 1,900 subjects (two-sided test, = 0.05, power = 0.90).
In a previous report, we analyzed the effects of VT reduction on mortality after dividing both study groups into quartiles according to baseline respiratory system compliance (4). Mortality rates in the highest compliance quartiles (analogous to lowest Pplat quartiles) did not favor the volume- and pressure-limited group. We have more confidence in our Pplat quartile analysis because the required data were available for a greater number of patients (787 compared with 517). Moreover, VT can affect compliance when compliance is measured during tidal ventilation (higher VT was associated with higher measured compliance [4]), and the range of baseline VT was broad (17). In contrast, the Day 1 Pplat levels were obtained when VT was tightly controlled in each study group. Therefore, the Pplat quartiles better represent disease severity than the compliance quartiles. Importantly, in both the compliance and Pplat quartile analyses, the interaction between quartiles and VT study group was not significant, suggesting that VT reduction was beneficial regardless of baseline compliance or Day 1 Pplat.
CONCLUSIONS
The ARDS Network volume- and pressure-limited strategy used a VT goal of 6 ml/kg predicted bodyweight (4). With this approach, mean Pplat was approximately 25 cm H2O. Most patients would have had Pplat below 31 cm H2O on VT that was greater than 6 ml/kg and less than or equal to 12 ml/kg. Some investigators have questioned the value of VT reduction in patients with ALI/ARDS whose Pplat levels are already below 30 to 35 cm H2O. We have been unable to find justification for this position after reviewing the results of animal models and comparing the results of clinical trials of lung-protective ventilation strategies. Our secondary analysis of the ARDS Network database suggests that there was a beneficial effect of VT reduction from 12 to 6 ml/kg predicted bodyweight, regardless of the Pplat before VT reduction. We do not advocate using VT of less than 6 ml/kg predicted bodyweight to achieve very low Pplat. Potential adverse effects of volume- and pressure-limited MV should be considered in all patients. Hypercapnia may cause elevated intracranial pressure, pulmonary hypertension, decreased myocardial contractility, decreased renal blood flow, and the release of endogenous catecholamines (43, 44). Moreover, MV with low VT and Pplat may cause more atelectasis and increase requirements for higher FIO2 and positive end-expiratory pressure (4, 45). All of these variables should be considered when titrating VT (46). However, we could not substantiate the widespread belief that VT reduction is without benefit when Pplat is already lower than 30 to 35 cm H2O.
Acknowledgments
The authors thank Dr. Henry E. Fessler for reviewing the manuscript and providing useful comments.
FOOTNOTES
Supported by contracts (N01-HR46063, N01-HR46064, P50-HL073944, F32-HL078132, and K23-HL67850) with the National Institutes of Health, National Heart, Lung, and Blood Institute ARDS Clinical Trials Network.
This article has an online supplement, which is accessible from this issue's table of contents at www.atsjournals.org
Originally Published in Press as DOI: 10.1164/rccm.200501-048CP August 4, 2005
Conflict of Interest Statement: None of the authors have a financial relationship with a commercial entity that has an interest in the subject of this manuscript.
REFERENCES
Slutsky AS. Mechanical ventilation. American College of Chest Physicians' Consensus Conference. Chest 1993;104:1833–1859.
Kallet RH, Jasmer RM, Pittet JF, Tang JF, Campbell AR, Dicker R, Hemphill C, Luce JM. Clinical implementation of the ARDS network protocol is associated with reduced hospital mortality compared with historical controls. Crit Care Med 2005;33:925–929.
Amato MB, Barbas CS, Medeiros DM, Magaldi RB, Schettino GP, Lorenzi-Filho G, Kairalla RA, Deheinzelin D, Munoz C, Oliveira R, et al. Effect of a protective-ventilation strategy on mortality in the acute respiratory distress syndrome. N Engl J Med 1998;338:347–354.
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.
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;16:372–377.
Frank JA, Gutierrez JA, Jones KD, Allen L, Dobbs L, Matthay MA. Low tidal volume reduces epithelial and endothelial injury in acid-injured rat lungs. Am J Respir Crit Care Med 2002;165:242–249.
Savel RH, Yao EC, Gropper MA. Protective effects of low tidal volume ventilation in a rabbit model of Pseudomonas aeruginosa-induced acute lung injury. Crit Care Med 2001;29:392–398.
Carmichael LC. Tidal volumes in ARDS and meta-analysis. Am J Respir Crit Care Med 2003;167:933–934.
Deans KJ, Minneci PC, Cui X, Banks SM, Natanson C, Eichacker PQ. Mechanical ventilation in ARDS: one size does not fit all. Crit Care Med 2005;33:1141–1143.
Gattinoni L, Carlesso E, Cadringher P, Valenza F, Vagginelli F, Chiumello D. Physical and biological triggers of ventilator-induced lung injury and its prevention. Eur Respir J Suppl 2003;47:15s–25s.
Eichacker PQ, Gerstenberger EP, Banks SM, Cui X, Natanson C. Meta-analysis of acute lung injury and acute respiratory distress syndrome trials testing low tidal volumes. Am J Respir Crit Care Med 2002;166: 1510–1514.
Tobin MJ. Culmination of an era in research on the acute respiratory distress syndrome. N Engl J Med 2000;342:1360–1361.
Ricard JD. Are we really reducing tidal volume—and should we Am J Respir Crit Care Med 2003;167:1297–1298.
Stewart T. Controversies around lung protective mechanical ventilation. Am J Respir Crit Care Med 2002;166:1510–1514.
Petty TL. Tidal volumes in ARDS and meta-analysis. Am J Respir Crit Care Med 2002;167:933.
Dreyfuss D, Saumon G. Evidence-based medicine or fuzzy logic: what is best for ARDS management Intensive Care Med 2002;28:230–234.
Brower RG, Matthay MA, Schoenfeld D. Meta-analysis of acutel lung injury and acute respiratory distress syndrome trials. Am J Respir Crit Care Med 2002;166:1515–1517.
Vincent JL. Evidence-based medicine in the ICU: important advances and limitations. Chest 2004;126:592–600.
Dreyfuss D, Saumon G. Ventilator-induced lung injury: lessons from experimental studies. Am J Respir Crit Care Med 1998;157:294–323.
Parker JC, Hernandez LA, Peevy KJ. Mechanisms of ventilator-induced lung injury. Crit Care Med 1993;21:131–143.
Rahn H, Otis AB, Chadwick LE, Fenn WO. The pressure-volume diagram of the thorax and lung. Am J Physiol 1946;146:161–178.
Stewart TE, Meade MO, Cook DJ, Granton JT, Hodder RV, Lapinsky SE, Mazer CD, McLean RF, Rogovein TS, Schouten BD, et al. Evaluation of a ventilation strategy to prevent barotrauma in patients at high risk for acute respiratory distress syndrome. Pressure- and Volume-limited Ventilation Strategy Group. N Engl J Med 1998;338:355–361.
Brower RG, Shanholtz CB, Fessler HE, Shade DM, White P Jr, Wiener CM, Teeter JG, Dodd-o JM, Almog Y, Piantadosi S. Prospective, randomized, controlled clinical trial comparing traditional versus reduced tidal volume ventilation in acute respiratory distress syndrome patients. Crit Care Med 1999;27:1492–1498.
Brochard L, Roudot-Thoraval F, Roupie E, Delclaux C, Chastre J, Fernandez-Mondejar E, Clementi E, Mancebo J, Factor P, Matamis D, et al. Tidal volume reduction for prevention of ventilator-induced lung injury in acute respiratory distress syndrome. The Multicenter Trail Group on Tidal Volume Reduction in ARDS. Am J Respir Crit Care Med 1998;158:1831–1838.
Brower RG, Krishnan JA, Thompson BT, Wheeler AP, Weideman H, Schoenfeld D, Bernard G. Effects of tidal volume reduction (Vt-R) in acute lung injury (ALI) patients with inspiratory plateau pressures (Pplat) < 32 cm H2O before Vt-R . Am J Respir Crit Care Med 2003;167:A616.
Cilley RE, Wang JY, Coran AG. Lung injury produced by moderate lung overinflation in rats. J Pediatr Surg 1993;28:488–493.
Dreyfuss D, Soler P, Basset G, Saumon G. High inflation pressure pulmonary edema: respective effects of high airway pressure, high tidal volume, and positive end-expiratory pressure. Am Rev Respir Dis 1988;137:1159–1164.
Webb HH, Tierney DF. Experimental pulmonary edema due to intermittent positive pressure ventilation with high inflation pressures: protection by positive end-expiratory pressure. Am Rev Respir Dis 1974;110: 556–565.
Omlor G, Niehaus GD, Maron MB. Effect of peak inspiratory pressure on the filtration coefficient in the isolated perfused rat lung. J Appl Physiol 1993;74:3068–3072.
Mascheroni D, Kolobow T, Fumagalli R, Moretti MP, Chen V, Buckhold D. Acute respiratory failure following pharmacologically induced hyperventilation: an experimental animal study. Intensive Care Med 1988;15:8–14.
Tsuno K, Prato P, Kolobow T. Acute lung injury from mechanical ventilation at moderately high airway pressures. J Appl Physiol 1990;69: 956–961.
Parker JC, Townsley MI, Rippe B, Taylor AE, Thigpen J. Increased microvascular permeability in dog lungs due to high peak airway pressures. J Appl Physiol 1984;57:1809–1816.
Parker JC, Hernandez LA, Longenecker GL, Peevy K, Johnson W. Lung edema caused by high peak inspiratory pressures in dogs: role of increased microvascular filtration pressure and permeability. Am Rev Respir Dis 1990;142:321–328.
Carlton DP, Cummings JJ, Scheerer RG, Poulain FR, Bland RD. Lung overexpansion increases pulmonary microvascular protein permeability in young lambs. J Appl Physiol 1990;69:577–583.
Taylor RW, Zimmerman JL, Dellinger RP, Straube RC, Criner GJ, Davis K Jr, Kelly KM, Smith TC, Small RJ. Low-dose inhaled nitric oxide in patients with acute lung injury: a randomized controlled trial. JAMA 2004;291:1603–1609.
Marini JJ. New options for the ventilatory management of acute lung injury. New Horiz 1993;1:489–503.
Scharf SM. Mechanical cardiopulmonary interactions in critical care. In: Dantzker DR, Scharf SM, editors. Cardiopulmonary critical care, 3rd ed. Philadelphia: WB Saunders; 1998. pp.75–91.
Tobin MJ. Update in pulmonary diseases. Ann Intern Med 2005;142: 283–288.
Brower RG. Mechanical ventilation in acute lung injury and ARDS: ridal volume reduction. Crit Care Clin 2002;18:1–13.
Amato MB, Brochard L, Stewart TE, Brower RG. Meta-analysis of tidal volume in ARDS. Am J Respir Crit Care Med 2003;168:612.
Hubmayr RD, Abel MD, Rehder K. Physiologic approach to mechanical ventilation. Crit Care Med 1990;18:103–113.
Knaus WA, Wagner DP, Draper EA, Zimmerman JE, Bergner M, Bastos PG, Sirio CA, Murphy DJ, Lotring T, Damiano A. The APACHE III prognostic system: risk prediction of hospital mortality for critically ill hospitalized adults. Chest 1991;100:1619–1636.
Feihl F, Eckert P, Brimioulle S, Jacobs O, Schaller MD, Melot C, Naeije R. Permissive hypercapnia impairs pulmonary gas exchange in the acute respiratory distress syndrome. Am J Respir Crit Care Med 2000; 162:209–215.
Feihl F, Perret C. Permissive hypercapnia: how permissive should we be Am J Respir Crit Care Med 1994;150:1722–1737.
Cheney FW Jr, Burnham SC. Effect of ventilatory pattern on oxygenation in pulmonary edema. J Appl Physiol 1971;31:909–912.
Dellinger RP. Positive clinical impact of low tidal volume strategy. Crit Care Med 2005;33:1143–1144.
Cleveland WS. Robust locally weighted regression and smoothing scatterplots. J Am Stat Assoc 1979;74:829–836.(David N. Hager, Jerry A. Krishnan, Dougl)