This Article Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Tobin, M. J. Search for Related Content PubMed PubMed Citation Articles by Tobin, M. J.

Critical Care Medicine in AJRCCM 2002

Division of Pulmonary and Critical Care Medicine, Loyola University of Chicago Stritch School of Medicine and Hines Veterans Affairs Hospital, Hines, Illinois

Correspondence and requests for reprints should be addressed to Martin J. Tobin, M.D., Division of Pulmonary and Critical Care Medicine, Hines Veterans Affairs Hospital, Route 111N, Hines, IL 60141. E-mail: mtobin2{at}lumc.edu

	CONTENTS

TOP CONTENTS MECHANICAL VENTILATION ACUTE LUNG INJURY AND... SEPSIS AND SHOCK VENTILATOR-ASSOCIATED PNEUMONIA COMMUNITY-ACQUIRED PNEUMONIA NOSOCOMIAL INFECTIONS MONITORING INHALED NITRIC OXIDE TOXICOLOGY ETHICAL ISSUES NONPULMONARY CRITICAL CARE REFERENCES

 
Mechanical Ventilation (23)

  Conventional Approaches (1)

  Patient–Ventilator Interaction (4)

  Nonconventional Modes (6)

  Protective Ventilation (3)

  Liquid Ventilation (2)

  Ventilator-Induced Lung Injury (5)

  Patient Posture (1)

  Adjunctive Therapy (1)

Acute Lung Injury and Acute Respiratory Distress Syndrome (18)

  Epidemiology and Genetics (3)

  Diagnostic Tests (2)

  Physiologic and Radiologic Studies (2)

  Animal Models (6)

  Cellular and Molecular Mechanisms (3)

  Treatment (1)

  Outcome (1)

Sepsis and Shock (15)

  Mechanisms in Patients and Volunteers (4)

  Endotoxemia in Animals (4)

  Sepsis in Animals (3)

  Treatment of Sepsis (4)

Ventilator-Associated Pneumonia (7)

  Incidence (1)

  Diagnosis (3)

  Treatment (3)

Community-Acquired Pneumonia (2)

Nosocomial Infections (2)

Monitoring (1)

  Pressure–Volume Curves (1)

Inhaled Nitric Oxide (1)

Toxicology (1)

Ethical Issues (4)

Nonpulmonary Critical Care (18)

  Pharmacotherapy (5)

  Renal Disorders (1)

  Gastroenterological Disorders (1)

  Cardiac Disorders (3)

  Hematological Disorders (2)

  Infectious Disorders (5)

  Rheumatological Disorders (1)

	MECHANICAL VENTILATION

TOP CONTENTS MECHANICAL VENTILATION ACUTE LUNG INJURY AND... SEPSIS AND SHOCK VENTILATOR-ASSOCIATED PNEUMONIA COMMUNITY-ACQUIRED PNEUMONIA NOSOCOMIAL INFECTIONS MONITORING INHALED NITRIC OXIDE TOXICOLOGY ETHICAL ISSUES NONPULMONARY CRITICAL CARE REFERENCES

 
Conventional Approaches
In a critical care perspective, Rouby and colleagues (1) discuss the approach to selecting the right level of positive end-expiratory pressure (PEEP) in patients with the acute respiratory distress syndrome (ARDS).

Patient–Ventilator Interaction
To determine the dynamic stability of pressure support in the absence of an airway leak, Hotchkiss and coworkers (2) used mathematical models to predict stability, and they used a test lung to confirm the predicted behavior. In the setting of airway obstruction, dynamic instability—consisting of wide swings in inspiratory time and auto-PEEP—occurred with pressure support even when patient effort was constant. "Noise" within the system, arising from errors in the measurement of peak flow or flow cutoff, accentuated the unstable behavior. The unstable behavior appeared to result from variability in the level of end-expiratory alveolar pressure, which influenced the tidal volume and end-expiratory pressure of the next breath. Instability was worsened by factors that increased the respiratory time constant relative to the respiratory frequency. The authors conclude that the use of pressure support in the setting of airway obstruction results in inherently unstable behavior, with wide breath-to-breath variations in tidal volume and auto-PEEP.

The continuation of mechanical inflation into neural expiratory time is likely to cause dynamic hyperinflation unless patients prolong their neural expiratory time or recruit their expiratory muscles. To investigate this phenomenon, Younes and coworkers (3) studied 50 patients ventilated in the proportional-assist mode. Exhalation was delayed intermittently by briefly delaying the opening of the exhalation valve (average delay of 0.78 seconds). In response to a delay in the opening of the expiratory valve, 45 patients increased the duration of neural expiratory time. The increase in duration of neural expiratory time offset the delay in expiration by only 36%. Patients did not show evidence of enhanced expiratory muscle recruitment. The breaths that followed occlusions of the expiratory valve displayed an increase in dynamic hyperinflation of 100 ml. The authors conclude that the respiratory motor response to a delay in opening of the expiratory valve on a ventilator is weak and is insufficient to prevent a worsening of dynamic hyperinflation. An editorial commentary by Brochard (4) accompanies this article.

To determine whether the quality of sleep is altered by the mode of mechanical ventilation, Parthasarathy and Tobin (5) studied 11 critically ill patients. All patients achieved sleep. Sleep fragmentation, measured as the sum of arousals plus awakenings, was greater during pressure support than during assist-control ventilation: 79 versus 54 events per hour. Six of the 11 patients developed central apneas during pressure support, but not during assist-control ventilation (by virtue of the backup rate). Heart failure was more common in the 6 patients who developed apneas than in the 5 patients without apneas: 83% versus 20%. Among the patients with central apneas, adding dead space (which increased end-tidal PCO₂ by 4.3 mm Hg) decreased the sum of arousals plus awakenings from 83 to 44 events per hour. The number of central apneas was most closely related to the difference between end-tidal CO₂ (during a mixture of wakefulness and sleep) and the apnea threshold point (r = -0.83). In patients receiving pressure support, respiratory rate was 32.6% lower and end-tidal CO₂ was 11.0% higher during sleep than during wakefulness. In patients receiving assist-control ventilation, respiratory rate was 14.9% lower and end-tidal CO₂ was 4.6% higher during sleep than during wakefulness. The authors conclude that critically ill patients experience greater fragmentation of sleep during pressure support than during assist-control ventilation because of the development of central apneas, and that this effect is especially prominent in patients with heart failure.

Nonconventional Modes
In mechanically ventilated animals, deliberate variation in the size of delivered tidal volumes has been shown to improve oxygenation in some but not all studies. In guinea pigs with endotoxin-induced lung injury, Arold and coworkers (6) determined whether a beneficial effect depends on the amount of imposed variability. Tidal volumes were varied randomly by 10, 20, 40, and 60% of the average value (rate was adjusted to achieve a constant minute ventilation). Compared with conventional ventilation, an increase in the variability of tidal volume to 40% of the average value produced a decrease in lung elastance of 14% and an increase in PO₂ from 42 to 54 mm Hg; an increase in the variability of tidal volume to 60% of the average value produced a 29% decrease in elastance and no further improvement in PO₂. A 20% increase in the variability of tidal volume was not beneficial. The authors conclude that deliberate attempts to vary delivered tidal volume produce improvements in lung mechanics and oxygenation in mechanically ventilated animals with acute lung injury, and that the benefit depends on the degree of imposed variability.

In pigs with acute lung injury caused by oleic acid and ventilated with a lung-protective strategy, Boker and coworkers (7) determined whether the deliberate variation in delivered tidal volume and respiratory rate would improve pulmonary function. Breath-to-breath variation in tidal volume was induced by adding a fractal signal. Compared with conventional ventilation at a tidal volume of 6 ml per kg, variation in delivered volume produced an increase in PO₂ (173 versus 119 mm Hg) and a decrease in shunt fraction (6 versus 9%) despite a lower peak airway pressure (21 versus 24 cm H₂O). The concentration of interleukin-8 in tracheal aspirate was inversely related to the wet:dry ratios in lung tissue. The authors conclude that deliberate variation in delivered volume and rate results in improved oxygenation at a lower peak airway pressure during protective ventilation in animals with acute lung injury.

A proposed advantage of proportional assist ventilation is the synchronization of the end of mechanical inflation with the end of a patient's inspiratory effort. Du and coworkers (8) developed a computer model to investigate how well the machine is synchronized with patient effort. The introduction of a delay in the control system caused ventilator assistance to continue for as long as 0.33 seconds after the end of patient inspiratory effort under certain conditions. The delay in termination of inspiratory assistance was proportional to the time delay within the control system, the respiratory time constant, and the gain of ventilator assistance. The authors conclude that the end of inspiratory assistance during proportional assist ventilation is not synchronized with the end of patient inspiratory effort.

To assess the safety and effectiveness of high-frequency oscillatory ventilation in adult patients with ARDS, Derdak and coworkers (9) randomized 75 patients to high-frequency oscillation and 73 patients to conventional ventilation. By design, mean airway pressure was higher in the oscillation group than in the conventional group throughout the first 72 hours (about 30 cm H₂O versus less than 24 cm H₂O). In the first 16 hours, the PO₂/FI_O2 ratio was higher in the oscillation group than in the conventional group (about 175 mm Hg versus less than about 140 mm Hg); this difference was no longer present at 24 hours. Oxygenation index decreased similarly in the two groups over the first 72 hours. Mortality at 30 days tended to be lower in the oscillation group than in the conventional group: 37% versus 52%. The two groups did not differ in hemodynamic variables, oxygenation failure, ventilation failure, barotrauma, or mucus plugging. The authors conclude the high-frequency oscillatory ventilation is safe and effective in the management of adult patients with ARDS. An editorial commentary by Froese (10) accompanies this article.

A critical goal of high-frequency oscillatory ventilation is the recruitment of lung volume through an increase in mean airway pressure. To develop an objective method for optimizing airway pressure, Habib and coworkers (11) measured changes in lung volume using respiratory inductive plethysmography before and after lung lavage in five piglets. The changes in lung volume were fitted to a model that featured an exponential rise to a maximum: V_L(t, Paw) = V_L,max ^. (1 - e^-(t/)), where V_L,max was maximum lung volume as a sigmoid function of airway pressure (Paw), t was time, and was time constant. Lavage induced a decrease in effective compliance and an increase in the time constant. The relationship between effective compliance and maximum lung volume was altered by lung lavage, and the relationship exhibited a bell-shaped derivative function. The maximum of the derivative corresponded to the maximum increase in compliance. The authors conclude that changes in lung volume measured by respiratory inductive plethysmography and fitted to a nonlinear mathematical model provide an objective method for optimizing airway pressure during high-frequency oscillatory ventilation.

Protective Ventilation
In 10 patients with ARDS, de Durante and coworkers (12) determined whether the increase in respiratory rate that accompanies the use of a low tidal volume would result in intrinsic PEEP. When the patients were ventilated with a tidal volume of 12 ml per kg, respiratory rate was 14 breaths per minute, intrinsic PEEP was 1.4 cm H₂O, and total PEEP (including an external PEEP of 10.3 cm H₂O) was 11.7 cm H₂O. When the patients were ventilated with a tidal volume of 6 ml per kg, respiratory rate was 34 breaths per minute, intrinsic PEEP was 5.8 cm H₂O, and total PEEP was 16.3 cm H₂O. The authors conclude that the high respiratory rate that accompanies use of a low tidal volume in patients with ARDS produces intrinsic PEEP.

In a critical care perspective, Eichacker and colleagues (13) present a meta-analysis of trials testing low tidal volumes in patients with ARDS. An editorial commentary by Stewart (14) accompanies the article.

Liquid Ventilation
Hirschl and coworkers (15) did a pilot study of partial liquid ventilation using perflubron in patients with ARDS (65 randomized to perflubron and 25 to conventional ventilation) for a maximum of 5 days. The rate of progression to ARDS was significantly decreased in patients treated with perflubron. The number of days free from the ventilator and mortality did not differ in the two groups. On post hoc analysis, perflubron resulted in more rapid discontinuation of ventilation in the subgroup of patients younger than 55 years. Episodes of hypoxia, respiratory acidosis, and bradycardia were more frequent in the perflubron group, but these were transient. The authors conclude that partial liquid ventilation can be performed with reasonable safety in adult patients with ARDS.

Infection with respiratory syncytial virus causes inflammation of the airway mucosa associated with activation of nuclear factor-B. Compared with control mice, Haeberle and coworkers (16) found intranasal administration of perflubron six hours after infection with respiratory syncytial virus produced significant inhibition of cellular inflammation in the lungs. Expression of the chemokines, RANTES (regulated upon activation, normal T cell expressed and secreted), macrophage inflammatory protein-1, macrophage inflammatory protein-1ß, and macrophage inflammatory protein-2, were decreased in the treated mice. Perflubron markedly decreased the activation of nuclear factor-B in the lung. The authors conclude that perflubron reduces lung inflammation in mice infected by respiratory syncytial virus through inhibition of chemokine expression and activation of nuclear factor-B.

Ventilator-Induced Lung Injury
To determine the relationship between airway pressures and barotrauma, Eisner and coworkers (17) analyzed data of 718 patients with ARDS or acute lung injury enrolled in the ARDS Network Trial. The patients did not have barotrauma at baseline. During the first 4 days of the study, the cumulative incidence of barotrauma was 13%. The risk of barotrauma increased 1.67 times for each 5-cm H₂O increment in PEEP. After controlling for age, tidal volume, plateau pressure, baseline PEEP, APACHE score, and other variables, the increased risk of barotrauma with PEEP remained significant. Barotrauma was also related to the level of PEEP on the day preceding the episode of barotrauma: the risk was increased 1.38-fold for each 5-cm H₂O increment in PEEP. The authors conclude that an increase in the level of PEEP is associated with a greater likelihood of early barotrauma in patients with ARDS or acute lung injury.

To determine whether hypercapnic acidosis protects against the development of ventilator-induced lung injury, Sinclair and coworkers (18) randomized rabbits ventilated with high tidal volumes to a PCO₂ of 40 mm Hg or a PCO₂ of 80 to 100 mm Hg. Compared with the eucapnic group, the hypercapnic group had lower plateau pressures (21 versus 27 cm H₂O), lower change in PO₂ (77 versus 165 mm Hg), lower wet-to-dry ratio (6.6 versus 9.7), lower protein in bronchoalveolar fluid (656 versus 1,350 µg per ml), lower cell count (2.8 versus 6.9 x 10⁵ nucleated cells per ml), and lower injury score (0.7 versus 7.0). The authors conclude that hypercapnic acidosis protects against ventilator-induced lung injury.

Use of high tidal volumes during mechanical ventilation may disrupt epithelial and endothelial cells. When an alveolar epithelial cell is stretched, accommodation of the increase in surface area is partly achieved by the trafficking of intracellular lipids to the plasma membrane. Vlahakis and coworkers (19) determined whether lipid trafficking induced by deformation of the matrix beneath the alveolar epithelium protects again cell injury caused by deformation. Wounding of epithelial cells was dependent on the amplitude of stretch and on the rate of stretch. Depletion of cholesterol caused less lipid trafficking resulting in greater cell wounding during stretching. Cell wounding was increased by both increases and decreases in the stiffness of the cytoskeleton, and susceptibility to injury was not correlated with changes in cell stiffness. The authors conclude that dynamic remodeling processes, such as deformation-induced lipid trafficking, are more important in protecting the plasma membrane against stress failure than is the inherent strength and organization of the cytoskeleton.

One mechanism responsible for ventilator-associated lung injury is believed to be stretch injury consequent to repetitive collapse and reopening of atelectatic regions with each breath. Collapse of dependent regions with each breath should produce large oscillations in PO₂, as the magnitude of shunt varies throughout the respiratory cycle. Baumgardner and coworkers (20) placed a fast PO₂ probe in the brachiocephalic artery of six rabbits with acute lung injury caused by saline lavage. The effect of random variations of PEEP, plateau pressure (minus PEEP), and respiratory rate on oscillations of PO₂ was modeled by multiple linear regression. Oscillations in PO₂ were 3 to 22 mm Hg before lavage and increased to 5 to 439 mm Hg after lavage; the average maximum amplitude was 390 mm Hg. Both PEEP and respiratory rate produced larger changes in the amplitude of the oscillations than did plateau pressure. The authors conclude that the large oscillations in PO₂ signify repetitive collapse and recruitment of lung regions in a model of acute lung injury, and that the changes in PO₂ in response to altering respiratory rate signify that it is not possible to reliably predict dynamic behavior from measurements of the static behavior of atelectasis.

In a critical care perspective, Hubmayr (21) discusses lung injury and recruitment, providing a skeptical look at the opening and collapse story.

Patient Posture
In a clinical commentary, Messerole and colleagues (22) discuss the pragmatics of prone positioning.

Adjunctive Therapy
In 17 patients with ARDS being ventilated with a lung-protective strategy (tidal volume lower than 8 ml per kg and PEEP at 3 to 4 cm H₂O above the lower inflection point on a pressure-volume curve), Villagra and coworkers (23) studied the effect of a recruitment maneuver. The 2-minute recruitment maneuver produced a peak inspiratory pressure of 47 cm H₂O, PEEP of 30 cm H₂O, and a 65% increase in mean airway pressure. Oxygenation did not increase significantly during the recruitment maneuver or at 15 minutes after its completion. Change in oxygenation was correlated with change in respiratory system compliance (r = 0.66) and with venous admixture (r = -0.85). Eight patients were also subsequently studied late in the course of ARDS: the results were not different. The authors conclude that a recruitment maneuver does not produce short-term improvement in oxygenation in most patients with ARDS who are being ventilated with a lung-productive strategy.

	ACUTE LUNG INJURY AND ACUTE RESPIRATORY DISTRESS SYNDROME

TOP CONTENTS MECHANICAL VENTILATION ACUTE LUNG INJURY AND... SEPSIS AND SHOCK VENTILATOR-ASSOCIATED PNEUMONIA COMMUNITY-ACQUIRED PNEUMONIA NOSOCOMIAL INFECTIONS MONITORING INHALED NITRIC OXIDE TOXICOLOGY ETHICAL ISSUES NONPULMONARY CRITICAL CARE REFERENCES

 
Epidemiology and Genetics
Angiotensin converting enzyme has effects on pulmonary vascular tone, vascular permeability, epithelial survival, and fibroblast activation. Forty-seven percent of the variance in the plasma level of angiotensin-converting enzyme is accounted for by insertion/deletion (I/D) polymorphism; the D allele is associated with a higher plasma level. To determine whether the D allele is associated with the development of ARDS, Marshall and coworkers (24) studied 96 white patients with ARDS, 88 mechanically ventilated patients with a non-ARDS cause of respiratory failure, 174 patients undergoing coronary artery bypass grafting, and 1,906 individuals from the general population. Frequency of the DD genotype was higher in the patients with ARDS (0.46) than in the other three groups: coronary artery bypass group (0.25), the non-ARDS respiratory failure group (0.24), and the healthy population (0.26). In the ARDS group, the D allele was associated with increased mortality: 54.5% for DD, 27.9% for ID, and 11.1% for II. The authors conclude that polymorphism of the gene for angiotensin-converting enzyme is associated with the development of ARDS and influences patient outcome. An editorial commentary by Stuber (25) accompanies this article.

To determine the incidence and mortality of acute lung injury and ARDS, Bersten and coworkers (26) prospectively studied every admission to all 21 adult ICUs in three Australian States over a 2-month period. The first incidences of acute lung injury and ARDS were respectively 34 and 28 cases per 100,000 per annum. The respective 28-day mortalities were 32 and 34%. The authors conclude that the incidences of acute injury and ARDS were higher and the mortality rates were lower than those previously reported.

Diagnostic Tests
Idiopathic acute eosinophilic pneumonia is characterized by acute febrile respiratory failure associated with diffuse radiographic infiltrates and pulmonary eosinophilia. Philit and coworkers (27) describe a series of 22 patients with this rare clinical entity. Within the first month after the onset of symptoms, the patients presented with severe hypoxemia (Pa_O2/FI_O2 ratio, 156 mm Hg). Fourteen patients (64%) required mechanical ventilation, 12 patients met the criteria for acute lung injury, and 8 patients met the criteria for ARDS. Radiographic studies revealed diffuse bilateral infiltrates. Bronchoalveolar lavage revealed 54% eosinophils. All patients recovered: 6 recovered without glucocorticoids and 12 received glucocorticoids. No patient relapsed. The authors conclude that idiopathic acute eosinophilic pneumonia should be considered in the differential diagnosis of ARDS, finding more than 25% eosinophilia on lavage obviates the need for lung biopsy, and a response to glucocorticoids should not be used as a diagnostic criterion. An editorial commentary by du Bois (28) accompanies this article.

Physiologic and Radiologic Studies
Acute interstitial pneumonitis causes acute alveolar damage and its sudden onset resembles ARDS. Ichikado and coworkers (29) studied whether the findings on high-resolution computed tomography could predict prognosis in 10 nonsurvivors and 21 survivors of acute interstitial pneumonia. Two independent observers unaware of patient outcome graded the findings: interobserver agreement was good ( 0.75). Compared with nonsurvivors, survivors displayed less ground-glass attenuation or consolidation associated with traction bronchiolectasis or bronchiectasis, less architectural distortion, and more ground glass attenuation or consolidation without traction bronchiolectasis. Overall scores were lower in survivors than in nonsurvivors: 223 versus 324. A score of less than 245 had a positive predictive value of 80% and a negative predictive value of 90% for survival. The authors conclude that findings on computed tomography help in predicting the prognosis in acute interstitial pneumonitis irrespective of the underlying physiologic abnormality. An editorial commentary by Hansell (30) accompanies this article.

Animal Models
To determine the effects of a decrease in tidal volume on the alveolar epithelial and endothelial barriers, Frank and coworkers (31) studied a model of acid-induced lung injury in rats. Compared with a tidal volume of 12 ml per kg, tidal volumes of 6 and 3 ml per kg led to decreases in the rates of accumulated lung water by 55 and 70%, respectively. Endothelial injury, reflected by plasma levels of von Willebrand factor antigen and permeability to albumin, was reduced equally by tidal volumes of 6 and 3 ml per kg. Type I epithelial cell injury, reflected by plasma levels of RTI40, was reduced by 46% when tidal volume was lowered from 12 to 6 ml per kg, and was reduced by a further 33% with a tidal volume of 3 ml per kg. Clearance of fluid across the alveolar epithelium was faster with a tidal volume of 3 ml per kg (24% per hour) as compared with 6 ml per kg (15% per hour) or 12 ml per kg (3% per hour). The authors conclude that ventilation with a tidal volume of 6 ml per kg as compared with 12 ml per kg protects both the alveolar epithelium and lung endothelium in animals with acute lung injury and that a further reduction in tidal volume to 3 ml per kg provides additional protection to the alveolar epithelium.

Poly (ADP-ribose) polymerose-1 (PARP-1) is a nuclear enzyme that helps repair damage to DNA. To determine whether PARP-1 participates in the acute lung injury caused by intratracheal instillation of lipopolysaccharide in mice, Liaudet and coworkers (32) suppressed PARP-1 by genetic (PARP-1^-/- mice) or pharmacologic (PJ-34, a phenanthridinone compound) methods. Compared with control experiments, the suppression of PARP-1 caused downregulation of the increases in cytokines (tumor necrosis factor-, interleukin-1ß, interleukin-6), chemokines (macrophage inflammatory protein-1 and -2), and nitric oxide induced by lipopolysaccharide. PARP-1 also improved lung morphology and it decreased high permeability pulmonary edema, lipid peroxidation, and the extravasation of neutrophils in alveolar spaces. The authors conclude that activation of poly (ADP-ribose) polymerase-1 (PARP-1) plays a central role in the development of lung inflammation after intratracheal instillation of lipopolysaccharide.

Most infants born before a gestational age of 30 weeks have been exposed to chorioamnionitis and aspiration of infected amniotic fluid. To simulate this situation and determine whether susceptibility to injury is related to gestational age, Kramer and coworkers (33) studied lambs delivered at 130 or 141 days of gestation (term is 146 days). In the preterm lambs, both a low and high dose of intratracheal endotoxin (0.1 or 10 mg per kg) caused impaired gas exchange and systemic inflammation. In the near-term lambs, a high dose of intratracheal endotoxin (10 mg per kg) caused lung inflammation without a systemic effect; systemic inflammation occurred when the intratracheal endotoxin was combined with ventilation at a high tidal volume or with intravenous endotoxin. The authors conclude that intratracheal endotoxin produces systemic inflammation in preterm lambs but not in near-term lambs, unless it is combined with ventilation using high tidal volumes.

Platelet-activating factor triggers edema formation by simultaneously activating two independent pathways: one mediated by a cyclooxygenase metabolite and the other by a pathway that is inhibited by quinine. Goggel and coworkers (34) determined whether prostaglandin E₂ mediates the pulmonary edema caused by platelet-activating factor. In isolated rat lung, administration of platelet activating factor caused edema and the release of prostaglandin E₂ in the venous effluent. A neutralizing antibody to prostaglandin E₂ attenuated the increase in lung edema. Edema induced by platelet-activating factor was less in mice deficient in E-prostanoid 3-receptor than in mice deficient in E-prostanoid-1, E-prostanoid-2, or E-prostanoid-4 receptors. Edema caused by perfusion of lungs with prostaglandin E₂ was largely prevented by inhibition of voltage-gated potassium channels (ß-dendrotoxin). The edema was not prevented by inhibition of calcium-dependent potassium currents (paxilline). Inhibition of the voltage-gated potassium channels alone also attenuated the edema caused by platelet-activating factor, and the edema was completely prevented when quinine was added. The authors conclude that pulmonary edema caused by platelet-activating factor is partly mediated by release of prostaglandin E₂, activation of E-prostanoid 3-receptor, and activation of voltage-gated potassium channels.

Damage to DNA by toxic oxygen-derived species partly mediates the injury to the alveolar epithelium caused by hyperoxia and this damage is recognized and repaired by two enzymes: human 8-oxoguanine DNA glycosylase (hOgg1) and Eschericia coli foramidopyrimidine DNA glycosylase (Fpg). To determine whether these enzymes can decrease oxygen-mediated damage to DNA, Wu and coworkers (35) use a retroviral vector to transduce lung epithelial cells with the enzymes. On exposure to hyperoxia, cells expressing either of these two enzymes displayed decreased DNA damage and increased survival. Overexpression of either enzyme decreased the toxic effects of hydrogen peroxide. The authors conclude that DNA base excision repair genes can reduce the injury to lung epithelium caused by hyperoxia.

The adherence of neutrophils to endothelial cells of the pulmonary microvasculature treated with tumor necrosis factor- induces changes in the cytoskeleton of the endothelial cells that are dependent on signaling events mediated by intercellular adhesion molecule-1. In samples taken from rats, Wang and coworkers (36) determined whether events mediated by adhesion molecules differ between endothelial cells of the pulmonary artery and endothelial cells of the pulmonary microvasculature (the site of neutrophil emigration and edema formation). In rats treated with tumor necrosis factor-, neutrophil adherence induced the formation of F-actin and apparent stiffness in endothelial cells of the pulmonary microvasculature, but not in endothelial cells of the pulmonary artery. Tumor necrosis factor- caused equivalent upregulation of intercellular adhesion molecule-1 and equivalent redistribution of the adhesion molecule by cross-linking antibodies in the two cell types. Neutrophil adherence, however, induced the production of reactive oxygen species only in endothelial cells of the pulmonary microvasculature, and not in endothelial cells of the pulmonary artery. Adhesion molecule cross-linking induced phosphorylation of p38 mitogen-activated protein kinase only in endothelial cells of the pulmonary microvasculature and not in endothelial cells of the pulmonary artery. Allopurinol, a xanthine oxidase inhibitor, inhibited the increase in p38 phosphorylation in the endothelial cells of the pulmonary microvasculature. The authors conclude that the signaling events mediated by intercellular adhesion molecule-1 and induced by neutrophil adhesion cause cytoskeletal changes in endothelial cells of the pulmonary microvasculature, but not in those of the pulmonary artery.

Cellular and Molecular Mechanisms
In 17 patients enrolled in a randomized trial of methylprednisolone versus placebo, Meduri and coworkers (37) studied whether systemic inflammation associated with unresolving ARDS causes resistance to glucocorticoids. Patients receiving methylprednisolone had progressive and sustained reductions in tumor necrosis factor-, interleukin-1ß, interleukin-6, adrenocorticotrophic hormone (ACTH), and cortisol over time. Exposure of normal blood leukocytes to plasma collected during treatment with methylprednisolone resulted in progressive increases in all aspects of activity mediated by the glucocorticoid receptor-, reduction in DNA binding of nuclear factor-B, and reductions in transcription of tumor necrosis factor- and interleukin-1ß. The authors conclude that patients with resolving ARDS have increased systemic inflammation associated with resistance to endogenous glucocorticoids and that the abnormalities are lessened by the administration of methylprednisolone.

To determine whether factors present in the alveolar milieu induce cell death, Hamacher and coworkers (38) exposed endothelial cells of the lung microvasculature to supernatant obtained by bronchoalveolar lavage from four groups of patients. Death of the endothelial cells was 0% in 10 control patients, 7% in 15 patients at risk of ARDS, 16% in 21 patients with early ARDS, and 8% in 20 patients with late ARDS. Compared with the control group, the three patient groups had increased levels of tumor necrosis factor- and angiostatin. Transforming growth factor-ß1 was not increased in the patients. Neutralization of tumor necrosis factor- and angiostatin inhibited the cytotoxicity of endothelial cells in the patients with early ARDS. The authors conclude that tumor necrosis factor- and angiostatin contribute to the endothelial injury in patients with ARDS.

Plasma levels of vascular endothelial growth factor are elevated in patients with ARDS, and the level falls as patients recover. To determine the activity of vascular endothelial growth factor in epithelial lining fluid, Thickett and coworkers (39) did bronchoscopy in 40 patients with ARDS and in 28 patients who were at risk of ARDS. The levels of vascular endothelial growth factor were lower in the patients with ARDS than in the patients at risk of ARDS (1,076 versus 7,674 pg per ml). In patients who recovered from ARDS, the level of vascular endothelial growth factor increased between Day 1 and Day 4: 1,184 versus 8,856 pg per ml. In patients with ARDS, the production of vascular endothelial growth factor was 48% less in alveolar macrophages and 44% less in alveolar neutrophils than in the patients at risk of ARDS. The level of vascular endothelial growth factor was inversely correlated with lung injury score (r = -0.54). The authors conclude that vascular endothelial growth factor increases in the alveolar compartment, but decrease in the vascular compartment of patients recovering from ARDS.

Treatment
Factor and coworkers (40) determined whether acute lung injury prevents the transfer of a protective gene. They induced lung injury by exposing rats to 100% oxygen for 48 or 60 hours, and then instilled two concentrations of an adenovirus (expressing E. coli lac Z gene) in a surfactant-based vehicle into the trachea. Both doses of the adenovirus achieved transgene expression in all segments of both acutely injured lungs and healthy lungs. The authors conclude that adenovectors can efficiently transduce the alveoli of animals with acute lung injury.

Outcome
The American Thoracic Society (41) presents a report on understanding costs and cost-effectiveness in critical care from the second American Thoracic Society workshop on outcomes research.

	SEPSIS AND SHOCK

TOP CONTENTS MECHANICAL VENTILATION ACUTE LUNG INJURY AND... SEPSIS AND SHOCK VENTILATOR-ASSOCIATED PNEUMONIA COMMUNITY-ACQUIRED PNEUMONIA NOSOCOMIAL INFECTIONS MONITORING INHALED NITRIC OXIDE TOXICOLOGY ETHICAL ISSUES NONPULMONARY CRITICAL CARE REFERENCES

 
Mechanisms in Patients and Volunteers
Orthogonal polarization spectral imaging permits assessment of the human microvasculature. De Backer and coworkers (42) used this technique to study the sublingual circulation. Microvascular blood flow was equivalent in 10 healthy subjects, 16 patients before cardiac surgery, and in 10 ICU patients who were acutely ill but free of sepsis. The density of all vessels was lower in 50 patients with severe sepsis than in the volunteers: 4.5 versus 5.4 per mm. The proportion of small vessels (less than 20 µm) was less in the patients with sepsis than in the volunteers: 48% versus 90%. Nonsurvivors had fewer perfused vessels than survivors: 85% versus 90%. The alterations were reversed by topical acetylcholine. The authors conclude that alterations in microvascular blood flow are common in patients with sepsis and are related to patient survival. An editorial commentary by Ince (43) accompanies this article.

Antigens of major histocompatibility complex type II (MHC II), expressed mainly on the surface of antigen-presenting cells, are decreased in the monocytes of many critically ill patients. Depression of the MHC II antigens is believed to contribute to increased susceptibility to infection. To define the biosynthetic steps involved in the decreased expression of MHC II molecules on monocytes, Fumeaux and Pugin (44) obtained whole blood samples from two cohorts (15 volunteers and 21 patients with sepsis of varying severity, and 11 volunteers and 31 patients with septic shock). Expression of human leukocyte antigen (HLA)-DR on monocytes was inversely related to the severity of sepsis. The defect in expression of HLA-DR on monocytes resided in an intracellular sequestration of the MHC II molecules, a post-translational event. The rate of transcription of HLA-DR or of its major transcriptional inducer, Class II transactivator, was not decreased. The levels of HLA-DR protein produced by monocytes were comparable in the patients with sepsis and the healthy volunteers. Exposing monocytes from normal donors to plasma taken from patients with septic shock caused significant endocytosis of HLA-DR, resulting in decreased expression of HLA-DR on the surface of the monocytes. This effect was blocked by a monoclonal antibody to interleukin-10, but not by antagonists to transforming growth factor-ß₁, prostaglandins, or ß-adrenergic agonists. The authors conclude that HLA-DR molecules are re-endocytosed and retained within monocytes of patients with septic shock, and that the phenomenon is partially mediated by interleukin-10. An editorial commentary by Cavaillon (45) accompanies this article.

Endotoxemia in Animals
The immunosuppressive agents, cyclosporin A and tacrolimus (FK 506), are believed to act by binding to cyclophilins and FK-binding proteins (also known as immunophilins). Fauvel and coworkers (46) studied the protective effect of these two agents on the myocardial dysfunction induced by endotoxin in rats. Both cyclosporin A and tacrolimus decreased leukocyte sequestration in the heart, leukocyte activation in the mesenteric venule, and serum concentration of tumor necrosis factor-. Cyclosporin A, but not tacrolimus, reduced end-stage nuclear apoptosis and improved myocardial function, perhaps because of the unique effects of cyclosporin A on mitochondria. The authors conclude that cyclosporin A and tacrolimus decrease the sequestration and activation of leukocytes in rats exposed to endotoxin, but only cyclosporin reduces the accompanying apoptosis and myocardial dysfunction.

To determine the effect of luteolin, a flavonoid, on inflammation, Kotanidou and coworkers (47) administered lipopolysaccharide (Salmonella enteriditis) intraperitoneally to mice. At 7 days, mortality was 52% in animals pretreated with luteolin and 96% in control animals. Luteolin reduced the levels of tumor necrosis factor- in serum, it abolished leukocyte infiltration in liver and lung, and it abolished the expression of intercellular adhesion molecule-1 in the liver. The authors conclude that the flavonoid luteolin protects against lipopolysaccharide-induced toxicity in animals through a decrease in the expression of proinflammatory molecules and leukocyte infiltration of tissues.

To identify the precise sequence of events that produce endotoxin-induced lung injury, Davidson and coworkers (48) infused endotoxin (Salmonella abortus equi) intravenously into spontaneously breathing rats. The animals developed an early marked fall in arterial PO₂ and a progressive deterioration in airway resistance, tissue resistance, and lung elastance; these changes occurred despite a 1.7-fold increase in minute ventilation and 5-fold increase in the number of sighs. The changes occurred before changes in alveolar neutrophils, macrophages, albumin flux, or edema; the latter changes were increased appreciably only at 8.5 hours. The increase in elastance preceded the increase in resistance, indicating that the change in elastance arose within the lung tissue rather than reflecting a fall in lung volume. Despite a dramatic increase in the synthesis and turnover of ³H-disaturated phosphatidylcholine, the subcellular and alveolar content of surfactant protein-A, surfactant protein-B, cholesterol, disaturated phospholipids, and phospholipid classes remained normal; the increased turnover in surfactant disaturated phospholipid was attributed to the increase in the number of sighs. The authors conclude that the initial respiratory failure caused by endotoxin is the result of ventilation-perfusion mismatch and not alveolar edema per se, and that endotoxin has effects on lung elastance and resistance that are independent of surfactant composition.

Application of exogenous CC chemokine ligand 2 (JE, monocyte chemotactic protein-1) induces monocyte accumulation in airspaces, and the combination of CC chemokine ligand 2 and endotoxin (E. coli) provokes enhanced monocyte accumulation, extensive neutrophil influx, and loss of the pulmonary endothelial–epithelial barrier. Maus and coworkers (49) studied the role of the CC chemokine receptor 2 (the receptor for CC chemokine ligand 2) in alveolar leukocyte traffic. The accumulation of monocytes in response to alveolar CC chemokine ligand 2 (alone or in combination with endotoxin) was almost completely inhibited in knockout mice with disruption of the gene for CC chemokine receptor 2 and in wild-type mice treated with MC21 (a monoclonal antibody that blocks CC chemokine receptor 2). Both approaches of interfering with CC chemokine receptor 2 also markedly decreased the accumulation of alveolar neutrophils when the mice were challenged with the combination of CC chemokine ligand 2 and endotoxin. In wild-type mice, administration of an anti–Gr-1 monoclonal antibody (which selectively depletes neutrophils) or antileukinate (an inhibitor of the CXC receptor) markedly decreased the accumulation of alveolar monocyte when the mice were challenged with the combination of CC chemokine ligand 2 and endotoxin. Treating wild-type mice with MC21 (an inhibitor of CC chemokine receptor 2) or with anti–Gr-1 (a selective depletor of neutrophils) prevented the vascular leakage that normally occurs with the combination of CC chemokine receptor 2 and endotoxin. The authors conclude that CC chemokine receptor 2 plays a central role in the recruitment of alveolar monocytes in response to CC chemokine ligand 2, alone or in combination with endotoxin, and that the interdependence of monocytes and neutrophils contribute to vascular permeability.

Sepsis in Animals
Sepsis increases inducible nitric oxide synthase and it decreases constitutive nitric oxide synthase. In a rat model of sepsis, Scott and coworkers (50) determined whether the decrease in constitutive nitric oxide synthase is caused by downregulation of this enzyme or its functional inhibition secondary to the high concentrations of nitric oxide. At 6, 12, 24, and 48 hours after cecal ligation and perforation, activity of constitutive nitric oxide synthase was depressed in all tissues. Inhibition of inducible nitric oxide synthase with aminoguanidine partially restored activity of constitutive nitric oxide synthase for up to 12 hours after injury, but activity was depressed at 24 to 48 hours. Concentrations of nitric oxide synthase protein in the endothelium declined progressively over time. The authors conclude that the decrease in constitutive nitric oxide activity is caused by functional inhibition in the early stage of experimental sepsis and by downregulation of endothelial nitric oxide synthase in the later stage.

In murine sepsis caused by cecal ligation and perforation, Wang and coworkers (51) determined the contribution of inducible nitric oxide synthase derived from inflammatory cells versus from parenchymal cells to the protein leak of acute lung injury. The studies were done in four groups of mice. Compared with wild-type mice, mice lacking inducible nitric oxide synthase (iNOS^-/-) did not develop a protein leak in the pulmonary vasculature, despite equal neutrophil infiltration. When inducible nitric oxide synthase was localized to inflammatory cells of chimeric mice (achieved by transplanting bone marrow positive for inducible nitric oxide synthase into iNOS^-/- recipients; + to - chimeric mice), protein leak occurred. When inducible nitric oxide synthase was localized to parenchymal cells of chimeric mice (transplantation of bone marrow negative for inducible nitric oxide synthase into iNOS^+/+ recipients; - to + chimeric mice), protein leak did not occur. The protein leak was also prevented by pretreatment with allopurinol (xanthine oxidase inhibitor) and superoxide dismutase (a scavenger of superoxide). The authors conclude that the microvascular protein leak of sepsis-induced lung injury is dependent on inducible nitric oxide in inflammatory cells rather than in parenchymal cells, and that the dependence of the leak on both nitric oxide and superoxide suggests that peroxynitrite rather than nitric oxide per se is responsible for the leak.

Adenosine is an endogenous regulator of leukocyte–endothelial interactions and inhibits neutrophil-mediated injury to endothelial cells. To investigate the role of endogenous adenosine in the microvascular derangements of sepsis, Cohen and coworkers (52) induced sepsis in mice by cecal ligation and puncture. Compared with control mice, the septic mice displayed increased leukocyte rolling and adhesion. Treatment of the septic mice with pentostatin, an inhibitor of adenosine deaminase, caused decreases in leukocyte rolling (1.7 versus 6.0 rolling cells per minute), leukocyte adhesion (0.6 versus 2.1 adherent cells per 100 µm per minute), and albumin leakage (0.21 versus 0.4). Pentostatin decreased the levels of interleukin-6, tumor necrosis factor-, and soluble tumor necrosis factor type II receptor. Survival at 48 hours was greater in mice treated with pentostatin than in untreated mice: 56% versus 38%. The authors conclude that adenosine plays an important role in modulating both leukocyte-dependent and leukocyte-independent mechanisms of endothelial injury in sepsis.

Treatment of Sepsis
Granulocyte-macrophage colony-stimulating factor stimulates hemopoiesis and the effector functions of granulocytes and macrophages, and it is involved in pulmonary surfactant homeostasis. Presneill and coworkers (53) did a double-blind randomized trial of recombinant granulocyte-macrophage colony-stimulating factor (at a low dose over 5 days) in 18 patients with severe sepsis and respiratory dysfunction. Patients in the active treatment group developed an increase in PO₂/FI_O2 ratio, increased peripheral blood neutrophils, decreased alveolar neutrophils, increased superoxide production by blood granulocytes, and restoration or preservation of the phagocytic function of blood leukocytes and alveolar leukocytes. There was no difference in mortality, frequency of ARDS, or extrapulmonary organ dysfunction. The authors conclude that administration of granulocyte-macrophage colony-stimulating factor to patients with sepsis results in improved gas exchange without pulmonary neutrophil infiltration, despite functional activation of both circulating neutrophils and pulmonary phagocytes. An editorial commentary by Trapnell (54) accompanies this article.

Anti-inflammatory agents have shown greater benefit in animal studies than in patient studies of sepsis. Because mortality rates in the control groups were higher in the animal studies than in the patient studies (88 versus 39%), Eichacker and coworkers (55) examined whether the risk of death would explain the differences in reported efficacy of anti-inflammatory agents. Analysis of data from 95 animal experiments revealed that 70% of the variability in the benefit obtained with anti-inflammatory agents was explained by the underlying risk of death (mortality in the control group). Analysis of data from 22 trials in patients revealed that the effects of anti-inflammatory agents fell within the 95% confidence intervals found in the animal studies. The treatment effect of activated protein C did not differ from the treatment effect of other anti-inflammatory agents. In prospective studies in 1,296 rats, doses of a number of microbial agents were varied to produce a wide range of control mortality rates. The effect of anti-inflammatory agents was again found to depend on the underlying mortality rate. The authors conclude that the efficacy of anti-inflammatory agents in sepsis depends on the underlying risk of death, and that this relationship accounts for the greater demonstrable efficacy of these agents in animal studies than in patient studies. An editorial commentary by Dinarello and Abraham (56) accompanies this article.

	VENTILATOR-ASSOCIATED PNEUMONIA

TOP CONTENTS MECHANICAL VENTILATION ACUTE LUNG INJURY AND... SEPSIS AND SHOCK VENTILATOR-ASSOCIATED PNEUMONIA COMMUNITY-ACQUIRED PNEUMONIA NOSOCOMIAL INFECTIONS MONITORING INHALED NITRIC OXIDE TOXICOLOGY ETHICAL ISSUES NONPULMONARY CRITICAL CARE REFERENCES

 
Incidence
In 52 patients (aged 70 years and older) admitted from a nursing home to the ICU with presumed pneumonia, El-Solh and coworkers (57) determined the causes of failure to respond to antibiotic therapy. All patients had failed to respond to 72 hours of antibiotic therapy before admission and required mechanical ventilation. Microbial investigations (including blood culture, serology, pleural fluid, and bronchoalveolar specimens) revealed a diagnosis in 24 (46%) patients. The most common isolates were methicillin-resistant Staphylococcus aureus (33%), enteric gram-negative bacilli (24%), and Pseudomonas aeruginosa (14%). In 20 cases of definite pneumonia, invasive bronchial sampling directed a change in antibiotics in 8 patients (40%) and discontinuation of antibiotics in 2 patients (10%). The overall hospital mortality was 42%. Mortality did not differ between patients with verified and nonverified pneumonia, or between patients who did or did not have their antibiotics changed on the basis of bronchial sampling. The authors conclude that patients admitted to an ICU from a nursing home because of suspected pneumonia and who have failed prior antibiotics should receive antimicrobial therapy directed at nosocomial pathogens.

Diagnosis
To determine whether routine microbiologic cultures of specimens before the onset of ventilator-associated pneumonia might help identify causative organisms and the selection of effective initial antibiotic therapy, Hayon and coworkers (58) prospectively studied 380 consecutive episodes of suspected ventilator-associated pneumonia. A total of 125 episodes of pneumonia were confirmed by bronchoscopic samples in 91 patients. Before the episode of pneumonia, each patient had an average of 45 specimens cultured. Of the 220 microorganisms eventually deemed responsible for the pneumonia, only 73 (33%) were recovered before the onset of the pneumonia; 342 organisms not responsible for pneumonia were also cultured, making prospective identification of the true pathogens difficult. Among the 102 episodes of pneumonia for which previous cultures of respiratory secretions had been obtained (average of 8 days before), the organisms ultimately responsible for the pneumonia were recovered from 35% of these early specimens. The authors conclude that routine microbiologic cultures obtained before an episode of ventilator-associated pneumonia provides little help in the selection of effective initial antimicrobial therapy.

To assess the reproducibility of blind protected bronchoalveolar lavage in the diagnosis of ventilator-associated pneumonia, Gauvin and coworkers (59) did two blind lavages at a 2-hour interval in 30 mechanically ventilated children (age, 52 months). Bacterial growth was present in 43% of the 60 lavages. Reproducibility for the presence of bacteria on quantitative cultures was excellent: concordance, 93%, and 0.86. Concordance was 86% for the type of bacteria and 79% for the number of bacteria. Reproducibility for the presence of neutrophils containing bacteria was perfect (concordance 100%, 1.0). One child developed a pneumothorax and one child experienced a significant increase in intracranial pressure; complications were otherwise benign and transitory. The authors conclude that blind protected bronchoalveolar lavage is a feasible and reproducible test in mechanically ventilated children.

In a critical care perspective, Michaud and colleagues (60) discuss the effect of design-related bias in studies of diagnostic tests for ventilator-associated pneumonia.

Treatment
In 36 healthy piglets receiving mechanical ventilation, Goldstein and coworkers (61) compared the concentration of amikacin in lung tissue when it was delivered by an ultrasonic nebulizer versus by intravenous infusion. Forty percent of the initial dose of aerosolized amikacin reached the tracheobronchial tree. The lung concentration of amikacin was more than 10 times higher when administered by the nebulizer than when administered intravenously. Tissue concentrations were 10 times higher than the minimal inhibitory concentrations of most gram-negative organisms. The authors conclude that the tissue concentration of amikacin was more than 10 times higher after aerosolized administration than after intravenous administration in healthy piglets receiving mechanical ventilation.

In ventilated piglets with bronchopneumonia caused by intrabronchial instillation of E. coli, Goldstein and coworkers (62) compared lung deposition and bactericidal efficiency of amikacin delivered by an ultrasonic nebulizer (45 mg per kg) or intravenously (15 mg per kg). Of the amikacin in the nebulizer, 38% reached the tracheobronchial tree. The concentration of amikacin in lung tissue was 3 to 30 times greater after administration by nebulization than after administration intravenously. Nebulized amikacin achieved a 4.7-fold greater concentration in lung segments displaying mild bronchopneumonia as compared with segments displaying severe bronchopneumonia. Nebulized amikacin achieved greater bactericidal activity: 71% of lung specimens were sterile after administration by nebulization as compared with 16% of lung specimens after administration intravenously. The authors conclude that nebulized amikacin achieves greater lung deposition and greater bactericidal activity in ventilated piglets with pneumonia than does intravenous amikacin.

In a state of the art review article, Chastre and Fagon (63) discuss ventilator-associated pneumonia.

	COMMUNITY-ACQUIRED PNEUMONIA

TOP CONTENTS MECHANICAL VENTILATION ACUTE LUNG INJURY AND... SEPSIS AND SHOCK VENTILATOR-ASSOCIATED PNEUMONIA COMMUNITY-ACQUIRED PNEUMONIA NOSOCOMIAL INFECTIONS MONITORING INHALED NITRIC OXIDE TOXICOLOGY ETHICAL ISSUES NONPULMONARY CRITICAL CARE REFERENCES

 
Kaplan and coworkers (64) did an observational cohort study to better define the incidence, patterns of care, and outcome among elderly patients admitted to hospital with community-acquired pneumonia. The study sample consisted of all Medicare recipients 65 years or older admitted to hospital with a diagnostic code for community-acquired pneumonia. Among 623,718 hospital admissions (18.3 per 1,000 of that population), 10.6% died. Between 65–69 years to older than 90 years, the incidence of community-acquired pneumonia rose fivefold and the mortality doubled. Mortality was higher in men (odds ratio, 1.21). Mean length of stay in hospital was 7.6 days, and costs per admission were $6,949. Among the 22.4% of patients admitted to an ICU, length of stay was 11.3 days and costs were $14,294. Among the 7.2% of patients requiring mechanical ventilation, length of stay was 15.7 days and costs were $23,961. Overall hospital costs were $4.4 billion, and $2.1 billion was incurred by cases managed in ICUs. The authors conclude that community-acquired pneumonia is common and frequently fatal in elderly subjects, and that it frequently necessitates admission to the ICU and results in considerable expenditure.

To compare features of community-acquired pneumonia in patients admitted to an ICU or managed outside of the ICU, Angus and coworkers (65) studied a prospective cohort of patients. Compared with 1,169 patients who were not admitted to the ICU, 170 patients requiring ICU admission were more likely to be admitted from home and to have comorbid conditions. Reasons for admission to the ICU were: respiratory failure (57%) hemodynamic monitoring (32%), and shock (16%). Compared with patients who were not admitted to the ICU, the patients requiring ICU admission had longer hospital stays (23 versus 9 days), higher hospital costs ($21,144 versus $5,785), more nonpulmonary organ dysfunction, and higher hospital mortality (18 versus 5%). Among four sets of criteria for severe pneumonia, the revised criteria of the American Thoracic Society were the best discriminator for ICU admission (areas under a receiver operating characteristic curve, 0.68), although discrimination was still relatively weak. The authors conclude that many patients with community-acquired pneumonia are admitted to the ICU and that available criteria are not robust in guiding clinical care.

	NOSOCOMIAL INFECTIONS

TOP CONTENTS MECHANICAL VENTILATION ACUTE LUNG INJURY AND... SEPSIS AND SHOCK VENTILATOR-ASSOCIATED PNEUMONIA COMMUNITY-ACQUIRED PNEUMONIA NOSOCOMIAL INFECTIONS MONITORING INHALED NITRIC OXIDE TOXICOLOGY ETHICAL ISSUES NONPULMONARY CRITICAL CARE REFERENCES

 
In 546 patients (mainly trauma and surgical ICU patients), Krueger and coworkers (66) did a prospective, stratified, double-blind trial to determine whether selective decontamination of the digestive tract alters the incidence of infections, organ dysfunctions, and mortality. Patients were randomized to intravenous ciprofloxacin (two doses of 400 mg for 4 days) plus topical gentamicin and polymyxin (applied to the nostril, mouth, and stomach throughout the ICU stay) or to intravenous placebo plus topical placebo. Patients receiving prophylactic antibiotics had fewer infections (relative risk, 0.48), especially fewer episodes of pneumonia (6 versus 29 patients), other lower respiratory tract infections (39 versus 70 patients), bloodstream infections (14 versus 36 patients), and urinary tract infections (36 versus 60 patients). Patients receiving prophylactic antibiotics had fewer episodes of severe organ dysfunction (relative risk, 0.64), especially renal dysfunction (17 versus 38 patients). Overall ICU mortality did not differ between the prophylactic antibiotic group and the placebo group: 52 versus 75 deaths. For the 237 patients with a midrange APACHE score of 20 to 29, mortality was lower in the prophylactic antibiotic group than in the placebo group: 20 versus 38 deaths. Surveillance cultures did not reveal evidence for the selection of resistant microorganisms. The authors conclude that prophylactic administration of a short course of intravenous antibiotics plus topical nonabsorbable antibiotics decreases the incidence of infections and organ dysfunction in critically ill surgical and trauma patients, and that the regimen also decreases mortality in patients with an APACHE score of 20 to 29 on admission. An editorial commentary by Aarts and Marshall (67) accompanies this article.

	MONITORING

TOP CONTENTS MECHANICAL VENTILATION ACUTE LUNG INJURY AND... SEPSIS AND SHOCK VENTILATOR-ASSOCIATED PNEUMONIA COMMUNITY-ACQUIRED PNEUMONIA NOSOCOMIAL INFECTIONS MONITORING INHALED NITRIC OXIDE TOXICOLOGY ETHICAL ISSUES NONPULMONARY CRITICAL CARE REFERENCES

 
Pressure–Volume Curves
In patients with ARDS, the inflation limb of the pressure–volume curve has a low compliance up to the lower inflection point. Vieillard-Baron and coworkers (68) determined whether the initial portion of the curve is caused by an uneven distribution of time constants resulting in an initial slow compartment. Static pressure–volume curves were obtained on the first day of ventilator management in 16 patients with ARDS. Eleven patients displayed a lower inflection point. In these 11 patients, an expiratory pause of more than 6 seconds yielded an expired volume of 172 ml (taken as the slow compartment), and occlusion of the expiratory limb for 4 seconds yielded an intrinsic PEEP of 3 cm H₂O. Five patients did not display a lower inflection point. In these 5 patients, the slow compartment was 28 ml and intrinsic PEEP was negligible. The compliance of the initial portion of the pressure–volume curve was correlated with the volume of the slow compartment (taken as a fraction of theoretical tidal volume) (r = -0.85). The authors conclude that the initial portion of the pressure–volume curve in patients with ARDS who display a lower inflection point results from a slow compartment of ventilation.

	INHALED NITRIC OXIDE

TOP CONTENTS MECHANICAL VENTILATION ACUTE LUNG INJURY AND... SEPSIS AND SHOCK VENTILATOR-ASSOCIATED PNEUMONIA COMMUNITY-ACQUIRED PNEUMONIA NOSOCOMIAL INFECTIONS MONITORING INHALED NITRIC OXIDE TOXICOLOGY ETHICAL ISSUES NONPULMONARY CRITICAL CARE REFERENCES

 
An alternative to the inhalation of nitric oxide is the administration of a prodrug that releases nitric oxide. Kirov and coworkers (69) describe the synthesis of linear polyethylenimine-nitric oxide/nucleophile adduct, a water-soluble donor of nitric oxide. The donor caused potent relaxation of precontracted rat aortic rings without inducing desensitization. The donor did not suppress the viability of human pulmonary artery epithelial cells in vitro. In 16 awake sheep, an infusion of E. coli endotoxin over 8 hours induced acute lung injury. Compared with placebo, administration of the nitric oxide donor by nebulization produced blunting of the endotoxin-mediated increases in pulmonary artery pressure, microwedge pressure, and pulmonary vascular resistance index by 40 to 70%. The donor produced a 60 to 70% decrease in the accumulation of extravascular lung water, and it attenuated the increase in right ventricular stroke work index and the decreases in right ventricular ejection fraction, stroke volume, and left ventricular stroke work index. The donor decreased venous admixture and improved arterial oxygenation. The authors conclude that an aerosolized nitric oxide donor, linear polyethylenimine-nitric oxide/nucleophile addict, attenuates endotoxin-induced acute lung injury in sheep as a result of decreasing pulmonary hypertension and edema and improving gas exchange and myocardial function.

	TOXICOLOGY

TOP CONTENTS MECHANICAL VENTILATION ACUTE LUNG INJURY AND... SEPSIS AND SHOCK VENTILATOR-ASSOCIATED PNEUMONIA COMMUNITY-ACQUIRED PNEUMONIA NOSOCOMIAL INFECTIONS MONITORING INHALED NITRIC OXIDE TOXICOLOGY ETHICAL ISSUES NONPULMONARY CRITICAL CARE REFERENCES

 
In an update in nonpulmonary critical care, Chu and colleagues (70) discuss toxicological problems in critically ill patients.

	ETHICAL ISSUES

TOP CONTENTS MECHANICAL VENTILATION ACUTE LUNG INJURY AND... SEPSIS AND SHOCK VENTILATOR-ASSOCIATED PNEUMONIA COMMUNITY-ACQUIRED PNEUMONIA NOSOCOMIAL INFECTIONS MONITORING INHALED NITRIC OXIDE TOXICOLOGY ETHICAL ISSUES NONPULMONARY CRITICAL CARE REFERENCES

 
To determine whether providing more information to family members of patients in the ICU results in better comprehension and satisfaction, Azoulay and coworkers (71) did a prospective randomized controlled trial in 34 French ICUs. On their first visit to the ICU, family representatives were randomized to receiving or not receiving a family information leaflet in addition to standard information. The leaflet provided the name of the caring physician, general information on the ICU, and the names of devices used in an ICU. Providing a leaflet decreased the frequency of poor comprehension among family members from 41 to 12%. Among the subset of family representatives with good comprehension, the leaflet resulted in a 22% higher score for satisfaction. The authors conclude that providing an information leaflet to family members of ICU patients results in better comprehension and satisfaction. An editorial commentary by Hartlieb and Sibbald (72) accompanies this article.

Living wills are written documents used by patients to convey their wishes should they become extremely ill. To determine whether the words used in living wills might give rise to misunderstanding by patients, family members, and physicians, Upadya and coworkers (73) did a questionnaire study of 152 patients, 108 family members, and 70 physicians. Of 140 patients admitted to a general ward, 17 (12%) patients wanted their living wills to preclude intubation-mechanical ventilation, and 12 (8.6%) did not want resuscitation under any circumstance. Seven of 120 (6%) physicians and 4 of 108 (3.7%) family members would not intubate or perform cardiopulmonary resuscitation even if there was a chance of recovery. Of 88 patients, 29 (33%) wanted their living wills to preclude intubation-mechanical ventilation only if their condition was deemed terminal, 46 (52%) wanted their living wills to preclude intubation even if there was a 10% chance of recovery, and 13 (15%) wanted their living wills to preclude intubation even if there was a greater than 50% chance of recovery. Equivalent results were obtained for cardiopulmonary resuscitation. The authors conclude that the interpretation of living wills differs considerably among patients, family members, and physicians.

In a critical care perspective, Luce and Rubenfeld (74) discuss the possibility of reducing health care costs by limiting intensive care at the end of life.

	NONPULMONARY CRITICAL CARE

TOP CONTENTS MECHANICAL VENTILATION ACUTE LUNG INJURY AND... SEPSIS AND SHOCK VENTILATOR-ASSOCIATED PNEUMONIA COMMUNITY-ACQUIRED PNEUMONIA NOSOCOMIAL INFECTIONS MONITORING INHALED NITRIC OXIDE TOXICOLOGY ETHICAL ISSUES NONPULMONARY CRITICAL CARE REFERENCES

 
Pharmacotherapy
Because propofol has a phenolic structure similar to that of the potent antioxidant, -tocopherol, Tsuchiya and coworkers (75) studied the effect of propofol on oxidative injury of human erythrocytes. Propofol inhibited oxidative hemolysis and oxidation of cis-parinaric acid in erythrocyte membranes; the protective effects were enhanced by ascorbic acid. By increasing the membrane fluidity of erythrocytes, propofol increased their resistance to physical hemodynamic stress. The preservation of red cell counts after surgery was greater with propofol than with sevoflurane anesthesia. The authors conclude that propofol protects erythrocytes against both oxidative and physical stress.

Particulate matter found in less expensive and counterfeit medications may cause damage when administered intravenously. To investigate this issue, Lehr and coworkers (76) injected a standard preparation of cefotaxime (Claforan) and two generic preparations of cefotaxime (proven to contain particulate matter) into hamsters. In healthy hamsters, the three preparations had no effect on capillary perfusion (as studied by intravital fluorescence microscopy). When the animals had a pre-existing ischemia–reperfusion injury of muscle, the two generic preparations—but not Claforan—caused a significant decrease in capillary perfusion. Birefringent particles causing mechanical obliteration of the muscle microcirculation were seen on histology. The authors conclude that particulate contaminants in intravenous generic medications can compromise the microcirculation of previously damaged tissue. An editorial commentary by Hall (77) accompanies this article.

The Richmond Agitation-Sedation Scale is a new 10-point scale: 0 is alert and calm, +1 to +4 is the progression from being restless to combative, and -1 to -5 is the progression from being drowsy to unarousable. In a broad spectrum of critically ill patients, Sessler and coworkers (78) tested the reliability and validity of the scale. During initial assessment involving 192 patient encounters, 5 workers (2 physicians, 2 nurses, 1 pharmacist) displayed excellent inter-rater reliability (r = 0.96; = 0.73). Across subgroups of patients (medical, surgical, coronary, neurological), inter-rater reliability proved robust (r = 0.92 to 0.98; = 0.65 to 0.80). After implementing the scale in the ICU, its reliability was tested during 101 patient encounters: a nurse educator and 27 nurses trained in use of the scale displayed excellent inter-rater reliability (r = 0.96; = 0.80). The new scale showed a good correlation with the Ramsay sedation sale (r = 0.78) and with the sedation agitation scale (r = 0.78). The authors conclude that the Richmond Agitation-Sedation Scale is easy to use, easy to recall, and valid and reliable for assessing sedated and nonsedated patients across a wide variety of ICU settings.

In a clinical commentary, Kress and colleagues (79) discuss sedation and analgesia in the intensive care unit.

Renal Disorders
In an update in nonpulmonary critical care, Block and Manning (80) discuss the prevention of renal failure in the critically ill patient.

Gastroenterological Disorders
In an update in nonpulmonary critical care, Zwischenberger and colleagues (81) discuss the surgical aspects of perforation and caustic injury of the esophagus.

Cardiac Disorders
In an update in nonpulmonary critical care, Poppas and Rounds (82) discuss congestive heart failure in the critically ill patient.

In an update in nonpulmonary critical care, Shah and Lilly (83) discuss interventional therapy for coronary artery disease.

In a clinical commentary, Vieillard-Baron and colleagues (84) discuss the demonstration of acute cor pulmonale by echo–Doppler at the bedside in the intensive care unit.

Hematological Disorders
About 5% of patients undergoing hematopoietic stem cell transplantation develop diffuse alveolar hemorrhage, with a reported mortality of about 80%. Afessa and coworkers (85) describe 48 patients with diffuse alveolar hemorrhage; 52% had an autologous transplant and 67% had a peripheral stem cell transplant. The hemorrhage occurred within one month in 28 patients. Symptoms included dyspnea in 92%, fever in 67%, cough in 56%, and hemoptysis in 15%. Admission to the ICU was required in 85% of the cases, and mechanical ventilation was required in 77%. Most patients were treated with methylprednisolone. Hospital mortality was 48%. Mortality was 28% in patients with an autologous transplant versus 70% in patients with an allogenic transplant. Mortality was 32% in patients who developed hemorrhage within the first 30 days after transplantation and 70% in patients with a late hemorrhage. The authors conclude that survival in patients with diffuse alveolar hemorrhage secondary to hematopoietic stem cell transplantation is better than previously reported, and that outcome is best in patients who had received an autologous transplant and in patients with early onset of hemorrhage.

In a clinical commentary, Afessa and colleagues (86) discuss diffuse alveolar hemorrhage in recipients of hematopoietic stem cell transplants.

Infectious Disorders
To determine the factors associated with in-hospital morbidity and mortality in patients with meningitis secondary to Streptococcus pneumoniae, Auburtin and coworkers (87) studied 80 patients. In-hospital mortality was 20%; mortality was not higher in patients who were infected with strains not susceptible to penicillin G. Multivariate analysis revealed three factors independently associated with death: platelet count less than 100 G per liter (odds ratio, 32.7), alkalosis with a pH above 7.47 (odds ratio, 33.1), and mechanical ventilation (odds ratio, 48.8). After adjusting for prognostic factors, glucocorticoids reduced mortality (odds ratio, 0.07). The authors conclude that hyperventilation should be used with caution in patients with S. pneumoniae meningitis, that nonsusceptibility to penicillin G is not associated with a worse outcome, and that glucocorticoids may be beneficial in the most severely ill patients.

To determine whether the introduction of highly active antiretroviral therapy has altered the epidemiology and outcomes of intensive care among HIV-infected patients, Morris and coworkers (88) analyzed data on all HIV-infected patients admitted to their ICUs between 1996 and 1999. An average of 89 patients were admitted each year, and survival to hospital discharge was 71%. On univariate analysis, improved survival was associated with previous highly active antiretroviral therapy (odds ratio, 1.8), a non-AIDS indication for admission (odds ratio, 3.7), a lower APACHE (acute physiology and chronic health evaluation) score (odds ratio, 5.4), and higher serum albumin (odds ratio, 4.4). Decreased survival was associated with Pneumocystis carinii pneumonia (odds ratio, 0.24), mechanical ventilation (odds ratio, 0.19), and pneumothorax (odds ratio, 0.08). On mutivariale analysis, all variables except the prior use of highly active antiretroviral therapy and pneumothorax were significant independent predictors of outcome. The authors conclude that the improved survival in HIV-infected patients admitted to an intensive care unit since the introduction of highly active antiretroviral therapy has been achieved through an increased likelihood of adm