Simple modifications of ventilator management are producing a decrease in mortality rates in ARDS patients.

In animal models, mechanical ventilation can cause acute lung injury histologically similar to that seen in adult respiratory distress syndrome (ARDS), along with an inflammatory response. This can cause progressive respiratory failure and death. In saline-lavaged rabbits, mechanical ventilation resulted in hypoxemia, increased microvascular permeability to albumin in systemic and pulmonary vessels, hyaline membrane formation, and granulocyte infiltration of the alveolar septa.1 These changes did not occur in granulocytopenic animals. Sheep with severe lung injury induced by 27 hours of ventilation with a peak inspiratory pressure (PIP) of 50 cm H2O all subsequently died, not from respiratory failure but from progressive hypotension and renal failure.2 Recent work3 has shown that injurious mechanical ventilation causes production of proinflammatory cytokines by the lung (and their release into the systemic circulation). A recent randomized trial4 in patients with ARDS also showed that the concentrations of tumor necrosis factor-µ and interleukin 1 in lung lavage fluid and in blood fell over the 36 hours following randomization into the lung-protective ventilation group, whereas, in the control group, they showed a progressive increase. Ventilation of dogs with high tidal volume (Vt) and low positive end-expiratory pressure (PEEP), after instillation of Escherichia coli into the trachea,5 resulted in severe pulmonary edema, and five of six animals developed positive blood cultures. This did not occur with the same end-inspiratory volume but high PEEP, or with a low Vt. These findings provide a plausible mechanism through which injurious mechanical ventilation could produce the multiple organ dysfunction syndrome and death, particularly in patients with ARDS.6

Mechanisms of ventilator-induced injury
Ventilator-induced lung injury occurs in animal models when lung overdistension is produced by a high end-inspiratory lung volume. The detailed mechanisms are not understood, but stress failure of rabbit lung capillaries that were perfused at increased microvascular pressure was more frequent at high lung volumes than at normal lung volumes.7 The capillary-wall stress resulting in stress failure apparently results from a combination of transmicrovascular pressure and alveolar wall stretch.7 Recent evidence8 confirmed that high microvascular pressure can act synergistically with lung overdistension to cause ventilator-induced lung injury. Therefore, to prevent ventilator-induced lung injury, it may be important to limit pulmonary artery pressure (perhaps using inhaled nitric oxide) and wedge pressure, as well as to avoid lung overdistension. Ventilator-induced lung injury also occurs in surfactant-depleted lungs without overdistension, if ventilation occurs with a low end-expiratory lung volume, allowing collapse and reexpansion with each respiratory cycle (tidal recruitment).6 This is thought to cause high shear stresses at the junctions between collapsed and aerated lung areas and in the airway epithelium during the repetitive ripping open of closed distal airways with each inspiration. Sufficient PEEP to prevent end-expiratory collapse largely prevents injury in most animal models, provided that overdistension is avoided.6 Thus, two distinct processes appear to contribute to ventilator-induced injury, and PEEP is protective.

The Lung in ARDS
Reduced compliance in early ARDS results from a reduced amount of aerated lung9; the remaining lung is collapsed and does not contribute to ventilation (thus, the term baby lung). The specific compliance or elasticity (compliance per unit of volume of aerated lung) appears to be relatively normal.9 The use of a Vt of 10 to 15 mL/kg usually results in overdistension of the aerated lung, indicated by an increased end-inspiratory plateau pressure. The nonaerated lung is found predominantly in the dependent regions (posteriorly, in supine patients),10 and is mainly a result of compression by the weight of the overlying lung. The amount of PEEP required to prevent end-expiratory collapse is usually similar to the pressure superimposed by the weight of the overlying lung (usually about 15 to 20 cm H2O), although higher PEEP may sometimes be required to achieve maximum recruitment. The lung thus behaves like a wet sponge; the distribution of water is relatively uniform throughout, but the lower regions are compressed by those above. If sufficient PEEP is applied to recruit dependent lung regions, the nondependent regions will have a high end-expiratory volume, making the use of a low Vt even more important. In the later stages of ARDS, the pathophysiology changes, and PEEP is often less effective in achieving lung recruitment. This is also often the case in ARDS due to direct pulmonary (rather than extrapulmonary) causes.

Avoidance of Ventilator-Induced Lung Injury
Strategies designed to avoid ventilator-induced lung injury therefore aim to prevent end-expiratory collapse and tidal recruitment by using sufficient PEEP (the open lung approach), and seek to avoid end-inspiratory overdistension of the aerated lung by limiting Vt, monitored by plateau pressure. In many patients, this will result in hypercapnia. The effects of hypercapnia have been reviewed elsewhere,11 but it should be stressed that most patients tolerate hypercapnia well, particularly if it develops gradually. It is usually considered to be contraindicated, however, in patients with raised intracranial pressure, and it must be monitored carefully in those with ischemic heart disease. Oxygenation is usually well maintained during hypercapnia, and tissue oxygenation may even be improved.12 In fact, the intracellular acidosis resulting from acute hypercapnia may be protective in hypoxic tissues and may prevent cell death.13 Hypercapnic acidosis was protective in an animal model of acute lung injury,14 and recently this protection was shown to be reduced by buffering; it may also be protective in other situations of cellular stress.13 Further study is required to understand the significance of these findings for patient care.

Selection of PEEP: the Pressure-Volume Curve
The best method of selecting PEEP to minimize ventilator-induced lung injury remains the subject of speculation. In early ARDS, the thoracopulmonary pressure-volume curve usually shows lower and upper inflection points. In animal models of surfactant deficiency, the use of a PEEP greater than the lower inflection point15 largely prevents lung injury (probably by preventing tidal recruitment), and this approach was associated with reduced mortality in ARDS.16 It has been suggested that the lower inflection point represents the region during inflation over which most recruitment of previously collapsed lung areas occurs, and that if tidal ventilation occurs between the pressures of the lower and upper inflection points, end-expiratory collapse and end-inspiratory overdistension should both be avoided. A mathematical model17 recently suggested, however, that the lower inflection point had little relationship with optimum PEEP, and was determined mainly by the lowest alveolar opening pressures. Recruitment continued on the linear portion of the pressure-volume curve, and an upper inflection point could occur at quite low pressures as recruitment diminished. Figure 1 (see Oct/Nov 1999 issue, page 78) shows simulations from the mathematical model, with the lung represented as four compartments with increasing superimposed gravitational pressures, from 0 in the uppermost (nondependent) to 15 cm H2O in the lowest (dependent). The black circles show the pressure-volume curve with alveolar opening pressures of zero. The plot consists of four linear segments; as the airway pressure exceeds the superimposed pressure for each compartment, that compartment becomes aerated, thus increasing the total alveolar compliance (the sum of compliances of all aerated alveoli) and, therefore, the slope of the pressure-volume plot. This is not, however, how the lung behaves; the opening pressures of collapsed alveoli greatly exceed the pressure required to maintain inflation. The opening pressure of some lung units in injured lungs may be as high as 50 cm H2O. The black triangles in Figure 1 (see Oct/Nov 1999 issue, page 78) show the plot with an opening pressure of 12 cm H2O for all alveoli. There is no volume increase until the pressure exceeds 12 cm H2O (the opening pressure of the uppermost compartment) at point a. Then there is an immediate increase in volume as the uppermost compartment snaps open (between a and segment 1) to the volume appropriate for that pressure. With a further pressure increase, alveoli in the uppermost compartment increase their volume according to their compliance (segment 1). When the pressure exceeds the opening pressure plus the superimposed pressure of the second compartment, this compartment also snaps open, causing a further incremental volume increase (between segments 1 and 2), and then the volume increases according to the total alveolar compliance of the two uppermost compartments (segment 2). This process continues until all four compartments are aerated. It can be seen that the lower inflection point is determined mainly by the opening pressures rather than by the superimposed pressure, whereas the PEEP requirement is determined by the superimposed pressure.10 The slope of the inflation pressure-volume curve between points a and b is steeper than the total compliance of all aerated alveoli because of the volume increments as each group of previously collapsed alveoli snaps open. Figure 2 (see Oct/Nov 1999 issue, page 78) shows pressure-volume curves with multiple compartments instead of the four shown in the model. The plot with opening pressures of 10 to 20 cm H2O (open circles) is now a smooth curve; the steepest part remains steeper than with an opening pressure of zero, because of recruitment, and there is an upper inflection point flexion as recruitment ends.

These predictions have recently been confirmed in a clinical study of ARDS18 showing that the pressure-volume curve measured from zero PEEP had a higher slope than that measured from a PEEP of 10 cm H2O without disconnecting the patient from the ventilator. The volume difference between the two curves gradually decreased up to a pressure of 30 cm H2O, suggesting that, on the curve measured from zero PEEP, recruitment continued up to this pressure and was not complete on the linear portion of the plot above the lower inflection point. There is also an upper inflection point as recruitment finishes and the curves merge. Therefore, one of the concepts resulting in the idea of setting PEEP above the lower inflection point is probably incorrect. Meyer et al19 have also shown, in a clinical study of ARDS, that the lower inflection point may underestimate optimum PEEP. The concept of the open lung approach is that once lung units have been recruited (perhaps through sustained inflation or a period of ventilation with a moderately high plateau pressure) it is then usually possible to maintain inflation using much lower pressures, because the closing pressures are usually much lower than the opening pressures; the lung is then ventilated on (or somewhere below) the deflation limb of the pressure-volume curve. If the upper inflection point reflects the end of recruitment in many patients (rather than overdistension),17 then it may be unnecessary to limit the plateau pressure to the pressure at the upper inflection point.

The Tidal Pressure-Volume Plot
During mechanical ventilation with zero PEEP, the tidal pressure-volume plot (with pressure on the x-axis) is usually concave upward, showing increasing compliance during inflation due to some end-inspiratory recruitment and end-expiratory collapse (tidal recruitment). This is similar to the upward curvature at the lower inflection point on the pressure-volume curve. When PEEP is progressively increased, the shape gradually changes from concave upward to concave downward, because tidal recruitment is progressively reduced and finally eliminated. When tidal recruitment is eliminated, at moderate lung volumes the tidal pressure-volume plot may be linear, but at higher lung volumes it may be concave downward, because lung compliance decreases at high lung volumes. Many ventilators allow the pressure-time plot to be displayed, rather than the pressure-volume plot. The shape of the curves is then opposite (with pressure on the y-axis). A concave-upward pressure-time curve shows decreasing compliance during inflation (possibly associated with lung overdistension), whereas a concave-downward curve shows increasing compliance during inflation, probably associated with tidal recruitment and insufficient PEEP. Ranieri et al20 have recently shown in an excised–rat-lung model that the titration of PEEP until the tidal pressure-time plot was linear, with a constant Vt, resulted in less ventilator-induced lung injury than either a concave-upward or concave-downward pressure-time curve associated with a higher or lower level, respectively, of PEEP. Therefore, a linear or slightly concave-downward tidal pressure-volume plot may indicate sufficient PEEP to prevent tidal recruitment and so minimize ventilator-induced lung injury, but more studies are required.

Effective Compliance
Another approach to determining the PEEP requirement is to measure the mean tidal-pressure–volume slope, or effective compliance, during a PEEP trial by dividing the Vt by the difference between the static end-expiratory and end-inspiratory pressures. It was suggested many years ago that optimum PEEP should be that giving the highest effective compliance. It is necessary to use a low Vt during this procedure; otherwise, at optimum PEEP, there may be considerable end-inspiratory overdistension (and a concave-downward tidal-pressure–volume plot), resulting in a low effective compliance. Optimum PEEP was usually defined, in the past, as that giving maximum oxygen delivery, whereas now the emphasis has moved toward minimizing ventilator-induced lung injury by eliminating tidal recruitment; this may require higher PEEP levels. Unfortunately, an incremental PEEP trial can be very misleading, especially with a low Vt. Tidal recruitment increases the effective compliance (because end-expiratory collapse decreases end-expiratory volume), and maximum effective compliance may occur at quite a low PEEP level, with substantial tidal recruitment. Theoretically, as PEEP approaches its optimum level and tidal recruitment is eliminated, the end-expiratory volume may increase more than the end-inspiratory volume and pressure-volume slope can decrease. Further increases in PEEP then result in increased PIP and more end-inspiratory recruitment, and compliance can increase again. Thus, during incremental PEEP, the relationship between PEEP and effective compliance is complex, and this method is not likely to predict the PEEP requirement reliably. A modification of the mathematical model, however, suggests that measuring the effective compliance during a decremental PEEP trial should theoretically predict the PEEP required to prevent end-expiratory collapse more accurately. With this procedure, maximum recruitment is achieved at the beginning of the trial, and as PEEP is reduced, compliance should increase (because there is less alveolar overdistension) until collapse commences, reducing the number of aerated alveoli. This method may cause problems of hypotension or barotrauma if not carefully managed, and it requires clinical validation before it can be recommended.

Selection of Volume
The plateau pressure gives a good estimation of peak alveolar pressure and should indicate end-inspiratory lung distension if the specific lung compliance and chest wall compliance are normal. It has been suggested that plateau pressure should be limited to less than 35 cm H2O when possible,21 but many patients with ARDS (particularly surgical patients with abdominal distension) have reduced chest-wall compliance. In such patients, the airway and pleural pressures will be higher at any given degree of lung distension, and so a higher plateau pressure (up to 40 cm H2O, or even higher in the presence of severe abdominal distension) may be acceptable. Estimation of transpulmonary pressure using an esophageal balloon may be helpful in managing such patients. A Vt of 6 to 7 mL/kg can be used initially in ARDS, and this should be reduced further if the plateau pressure exceeds 35 to 40 cm H2O. It is probably not necessary to limit the plateau pressure to that of the upper inflection point when this occurs at low pressures.

Effect of Spontaneous Breathing
The foregoing discussion assumes that the patient has relaxed respiratory muscles. In the presence of spontaneous respiratory-muscle activity, interpretation of airway pressures is more difficult, and lung overdistension can occur even with a limited plateau pressure. During volume-controlled ventilation (VCV), inspiratory muscle activity during inspiration reduces airway and pleural pressures, but transpulmonary pressure and lung volume are unaffected. During pressure-control ventilation (PCV) and pressure-support ventilation (PSV), inspiratory-muscle activity reduces pleural pressure and increases transpulmonary pressure and peak lung volume, while airway pressure remains unchanged; again, lung overdistension may occur. If the ventilator settings are initially made during paralysis, however, then with VCV, no increase in Vt will occur with subsequent inspiratory-muscle activity (instead, the PIP and plateau pressure will decrease), whereas with PCV, Vt will increase and lung overdistension may occur while airway pressures do not change. Expiratory-muscle activity at the end of inspiration has the opposite effect (except with PSV, when expiration will be triggered). Active expiration can reduce end-expiratory lung volume and oppose the effect of PEEP. Interpretation of airway pressures during spontaneous breathing can be facilitated using an esophageal balloon to estimate transpulmonary pressure, or respiratory muscle activity may be temporarily abolished by using neuromuscular blockade or opiates.

New Ventilator Modes
There is no good evidence that any of the newer modes of ventilation are better than the conventional modes at limiting ventilator-induced lung injury (nor is there evidence that either PCV or VCV is superior). The important issue seems to be sufficient PEEP and prevention of overdistension, however this is achieved. No trials have adequately compared optimum high-frequency ventilation (using recruitment maneuvers) with optimized conventional mechanical ventilation.

Clinical Evidence of Improved Outcome
An uncontrolled trial22 has suggested that reduced mortality rates can be achieved using the approaches to mechanical ventilation described here. Two randomized trials have now shown a substantial reduction in mortality rates using such approaches. Amato et al16 showed a 28-day mortality rate of 71% in the control group and 38% in the open-lung–approach group (P<.001), and the incidence rates for clinical barotrauma were 42% and 7%, respectively (P=.02). The PEEP used during the first 36 hours was positively correlated with survival and negatively correlated with barotrauma; thus, PEEP was protective when plateau pressure was limited. More recently, the US National Heart, Lung and Blood Institute multicenter ventilation trial was terminated early because of a 25% reduction in the mortality rate in the study group.23 Three trials have not shown a mortality difference using modified ventilation strategies in ARDS, but in all three trials, the PEEP level was relatively low (7.2 to 10.8 cm H2O, compared with 16.4 cm H2O in the study group in the trial of Amato’s group) and the plateau pressure was relatively low, even in the control groups (26.8 to 31.3 cm H2O, compared with 36.8 cm H2O in Amato’s trial). Therefore, it appears that most of the control-group patients in these negative trials were not exposed to lung overdistension, and in fact, there was not a large difference between ventilation strategies in the study and control groups.

There is increasing evidence that simple modifications of ventilation management can produce large reductions in mortality rates in ARDS. Further studies are required to confirm these results and to refine this approach, but the application of these simple principles now can probably provide improved care and better outcomes for our patients.

Keith G. Hickling, MD, is a physician at Queen Elizabeth Hospital, Hong Kong.

1. Kawano T, Mori S, Cybulsky M, et al. Effect of granulocyte depletion in a ventilated surfactant-depleted lung. J Appl Physiol. 1987;62:27-33.
2. Borelli M, Kolobow T, Spatola R, et al. Severe acute respiratory failure managed with continuous positive airway pressure and partial extracorporeal carbon dioxide removal by an artificial membrane lung. Am Rev Respir Dis. 1988;138:1480-1487.
3. Tremblay L, Valenza F, Ribeiro SP, et al. Injurious ventilatory strategies increase cytokines and c-fos m-RNA expression in an isolated rat lung model. J Clin Invest. 1997;99:944-952.
4. Ranieri M, Suter P, Tortorella C, et al. Effect of mechanical ventilation on inflammatory mediators in patients with acute respiratory distress syndrome. JAMA. 1999;282:54-61.
5. Nahum A, Hoyt J, Schmitz L, et al. Effect of mechanical ventilation strategy on dissemination of intratracheally instilled Escherichia coli in dogs. Crit Care Med. 1997;25:1733-1743.
6. Slutsky AS, Tremblay LN. Multiple system organ failure. Is mechanical ventilation a contributing factor? Am J Respir Crit Care Med. 1998;157:1721-1725.
7. Fu Z, Costello M, Tsukimoto K, et al. High lung volume increases stress failure in pulmonary capillaries. J Appl Physiol. 1992;73:123-133.
8. Broccard AF, Hotchkiss JR, Kuwayama N, et al. Consequences of vascular flow on lung injury induced by mechanical ventilation. Am J Respir Crit Care Med. 1998;157:1935-1942.
9. Gattinoni L, Pesenti A, Avalli L, et al. Pressure-volume curve of total respiratory system in acute respiratory failure. Am Rev Respir Dis. 1987;136:730-736.
10. Gattinoni L, D’Andrea L, Pelosi P, et al. Regional effects and mechanism of positive end-expiratory pressure in early ARDS. JAMA. 1993;269:2122-2127.
11. Hickling K. Targets during mechanical ventilation. In: Marini JJ, Slutsky AS, eds. The Physiological Basis of Ventilatory Support. New York: Marcel Dekker; 1998:655-708.
12. Joyce C, Hickling K. Permissive hypercapnia and gas exchange in lungs with high Qs/Qt: a mathematical model. Br J Anaesth. 1996;77:678-683.
13. Laffey JG, Kavanagh BP. Carbon dioxide and the critically ill: too little of a good thing? Lancet. In press.
14. Shibata K, Cregg N, Engelberts D, et al. Hypercapnic acidosis may attenuate acute lung injury by inhibition of endogenous xanthine oxidase. Am J Respir Crit Care Med. 1998;158:1578-1584.
15. Muscedere JG, Mullen JB, Gan K, Slutsky AS. Tidal ventilation at low airway pressures can augment lung injury. Am J Respir Crit Care Med. 1994;149:1327-1334.
16. Amato MB, Barbas CS, Medeiros DM, et al. Effect of a protective-ventilation strategy on mortality in the acute respiratory distress syndrome. N Engl J Med. 1998;338:347-354.
17. Hickling KG. The pressure-volume curve is greatly modified by recruitment. A mathematical model of ARDS lungs. Am J Respir Crit Care Med. 1998;158:194-202.
18. Jonson B, Richard JC, Straus C, et al. Pressure-volume curves and compliance in acute lung injury: evidence of recruitment above the lower inflection point. Am J Respir Crit Care Med. 1999;159:1172-1178.
19. Meyer EC, Barbas CSV, Grunauer MA, et al. PEEP at Pflex cannot guarantee a fully open lung after a high-pressure recruiting maneuver in ARDS patients. Am J Respir Crit Care Med. 1998;157:A694.
20. Zhang H, Aubin M, Mullen JB, et al. Pressure-time curve during constant flow inflation predicts ventilator-induced lung injury in an isolated rat lung model. Am J Respir Crit Care Med. In press.
21. Slutsky AS. Mechanical ventilation. American College of Chest Physicians Consensus Conference. Chest. 1993;104:1833-1859.
22. Hickling KG, Henderson SJ, Jackson R. Low mortality associated with pressure limited low volume ventilation with permissive hypercapnia in severe adult respiratory distress syndrome. Intensive Care Med. 1990;16:372-377.
23. News release: NHLBI clinical trial stopped early: successful ventilator strategy found for intensive care patients on life support. Available at: Accessed October 1, 1999.