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 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 2003

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 CRITICAL CARE REFERENCES

 
Critical Care (115)

  Mechanical Ventilation (42)

    Conventional Approaches (2)

    Patient-Ventilator Interaction (6)

    Non-Conventional Modes (2)

    Protective Ventilation (2)

    Liquid Ventilation (1)

    Ventilator-Induced Lung Injury (15)

    Ventilator-Induced Diaphragmatic Injury (2)

    Weaning (3)

    Patient Posture (1)

    Non-Invasive Ventilation (3)

    Adjunctive Therapy (5)

  Acute Lung Injury and Acute Respiratory Distress Syndrome (16)

    Epidemiology and Genetics (2)

    Animal Models (4)

    Cellular and Molecular Mechanisms (3)

    Fluid Biology (4)

    Treatment (2)

    Review Articles (1)

  Sepsis and Shock (19)

    Epidemiology and Genetics (5)

    Mechanisms in Patients and Volunteers (2)

    Endotoxemia in Animals (3)

    Sepsis in Animals (3)

    Treatment of Sepsis (3)

    Nonseptic Causes of Shock (3)

  Ventilator-Associated Pneumonia (4)

  Severe Acute Respiratory Syndrome (SARS) (5)

  Nosocomial Infections (3)

  Massive Pulmonary Hemorrhage (1)

  Chronic Obstructive Pulmonary Disease (2)

  Monitoring (4)

    General (2)

    Hemodynamic (2)

  Nitric Oxide (6)

  Ethical Issues (3)

  Nonpulmonary Critical Care (10)

    Pharmacotherapy (1)

    Trauma (2)

    Gastroenterological Disorders (1)

    Cardiac Disorders (1)

    Hematological Disorders (2)

    Infectious Disorders (2)

    Neurologic and Neuromuscular Disorders (1)

	CRITICAL CARE

TOP CONTENTS CRITICAL CARE REFERENCES

 
Mechanical Ventilation
Conventional approaches.
To determine the level of understanding of mechanical ventilation among internal medicine residents, Cox and coworkers (1) administered a 19-question examination to 259 residents at 31 residency programs. The average score on the test was 74% correct (range, 37 to 100%). Most residents correctly identified tension pneumothorax (86% correct) and clinical findings suggestive of severe hypotension secondary to auto positive end-expiratory pressure (intrinsic PEEP) (93% correct). High rates of incorrect answers were found for the setting of tidal volume in patients with the acute respiratory distress syndrome (ARDS) (48% incorrect), identifying whether a patient was ready for a weaning trial (38% incorrect), and recognizing an indication for noninvasive ventilation (27% incorrect). Higher scores were associated with closed-unit versus open-unit organization (76 versus 71% correct), resident perception of greater versus lesser knowledge (79 versus 71% correct), and graduation from a U.S. versus a foreign medical school (75 versus 69% correct). Only 46% of residents were satisfied with their training in mechanical ventilation. The authors conclude that residents in internal medicine programs may not be gaining knowledge essential for providing care to patients requiring mechanical ventilation. An editorial commentary by Dunn (2) accompanies this article.

Patient-ventilator interaction.
Sharshar and coworkers (3) designed a ventilator that adjusts airway pressure in proportion to the generation of transdiaphragmatic pressure during inspiration. The ventilator was evaluated in seven healthy subjects during normocapnia and with increases in carbon dioxide tension (PCO₂) of 3, 6, and 9 mm Hg. Each inflation cycle was triggered by either a preset increase in transdiaphragmatic pressure or preset inspiratory flow threshold, whichever occurred first (it was pressure in six subjects and flow in one subject). Untimely triggering or intense (or prolonged) triggering efforts were not seen at any level of hypercapnia. During servoventilator assistance, airway pressure increased in proportion to variation in transdiaphragmatic pressure. Both with and without the servoventilator, hypercapnia produced increases in tidal volume, respiratory rate, pressure–time product, and intrinsic PEEP. Pressure–time product was lower with the servoventilator than without it. The authors conclude that a servoventilator driven by transdiaphragmatic pressure delivers an increase in airway pressure in proportion to subject effort, and is effective in decreasing subject effort during normocapnia and hypercapnia without inducing untimely or excessive triggering attempts. An editorial commentary by Sinderby (4) accompanies this article.

When mechanical ventilation is delivered under normocapnic conditions, its cessation is accompanied by central apneas. To determine whether increases in tidal volume or increases in respiratory rate are responsible for the change in respiratory rhythm, Rice and coworkers (5) increased tidal volume (to 135–220% of eucapnic volume) and respiratory rate (1 or 3 breaths above eucapnic rate) in seven healthy subjects during NREM sleep. During controlled ventilation, an increase in tidal volume (by 65% or more of eucapnic volume) plus an increase in rate (of 1 breath per minute) eliminated transdiaphragmatic pressure. Cessation of controlled ventilation was accompanied by prolongation of expiratory time (by two to four times the control value); the increases in expiratory time were proportional to the increases in tidal volume during controlled ventilation but were independent of further increases in ventilator rate. During and after assist-control ventilation, increases in tidal volume (to 135 to 220% of eucapnic volume) were accompanied by decreases in transdiaphragmatic pressure in proportion to the increase in tidal volume; each ventilator inflation was actively triggered and the increase in expiratory time on cessation of mechanical ventilation was less than 20% of that seen after cessation of controlled ventilation. The authors conclude that both increases in ventilator rate and delivered volume during mechanical ventilation cause inhibition of respiratory motor output via nonchemical mechanisms (neuromechanical effects of repeated and augmented lung inflation), and that the resetting of respiratory rhythm is greater with additional increase in ventilator rate than with increase in delivered volume alone.

The ventilatory response to hypoxia in the presence of hypocapnia is controversial. Corne and coworkers (6) used volume-cycled ventilation to measure the ventilatory response to hypoxia during eucapnia and hypocapnia. The response of respiratory muscle pressure to hypoxia (expressed as cm H₂O per percentage change in oxygen saturation) was 0.53 at eucapnia, 0.26 at end-tidal PCO₂ of 6 mm Hg below eucapnia, and 0.003 at PCO₂ of 12 mm Hg below eucapnia. Similar reductions were evident for respiratory rate (expressed as breaths per minute per percentage change in oxygen saturation): 0.17 at eucapnia, 0.11 at PCO₂ of 6 mm Hg below eucapnia, and 0.01 at PCO₂ of 12 mm Hg below eucapnia. The responses of respiratory muscle pressure and respiratory rate at a PCO₂ of 12 mm Hg below eucapnia were not significantly different from zero. The authors conclude that regardless of the strength of the hypoxic ventilatory response at eucapnia, the responses are lost when PCO₂ is reduced (in a stable manner) by 6 to 12 mm Hg below eucapnia.

To determine the prevalence of noise in the intensive care unit (ICU) and its effect on sleep, Gabor and coworkers (7) did polysomnography on 7 mechanically ventilated patients and 6 healthy subjects over 24 hours in the ICU. Time-synchronized environmental monitoring was also performed. The average level of noise was 51 dB in the open ICU and 43 dB in an isolated single room (noise is 70 dB in a busy office, and 40 dB in a bedroom). In the patients, increases in sound occurred 37 times per hour of sleep and were responsible for 21% of all arousals and awakenings. Patient-care activities occurred 8 times per hour and were responsible for 7% of all arousals and awakenings. Compared with an open ICU, healthy subjects slept longer when placed in a single ICU room (9.5 versus 8.2 hours); sleep architecture was not altered. The authors conclude that noise and patient-care activities account for 28% of arousals and awakenings in critically ill patients receiving mechanical ventilation.

Arterial pulse pressure is known to increase during mechanical inflation and decrease during exhalation secondary to cyclic changes in left-ventricular stroke output. To determine whether the changes during inflation result from improvement in left-ventricular systolic function, Vieillard-Baron and coworkers (8) did transesophageal echocardiography, Doppler studies, and arterial pressure measurements in 31 septic patients. Lung inflation produced a 16.6% increase in left-ventricular stroke volume, which was directly related to a 13.1% increase in left-ventricular diastolic volume and a 12.2% increase in left-ventricular ejection fraction. The authors conclude that the improvement in left-ventricular function during mechanical inflation is primarily the result of increased left-ventricular filling.

Nonconventional modes.
To determine whether the mode of ventilation alters lung strain, Edibam and coworkers (9) studied 18 patients with acute lung injury. Thirty minutes of ventilation with three modes—volume-control, pressure-control, and pressure-controlled inverse-ratio—had equivalent effects on static mechanics, oxygenation, and hemodynamics. Nonlinear behavior, as a fraction of total elastance, was higher with pressure-control (36%) than with volume-control (25%) and lowest with pressure-controlled inverse-ratio (16%). Computed tomography revealed about 25% more overinflated units in the left subcarinal region with pressure-controlled inverse-ratio ventilation than with either of the two other modes. The authors conclude that lung strain in patients with acute lung injury is greater with pressure-controlled inverse-ratio ventilation than with volume-control ventilation, and least with pressure-control ventilation.

In a critical care perspective, Bollen and colleagues (10) present a meta-analysis on use of high-frequency and conventional ventilation.

Protective ventilation.
To determine whether clinicians changed the setting of tidal volume between 1994 and 2001, Weinert and coworkers (11) analyzed data on 398 patients with acute lung injury. In the first 5 years, tidal volume was 9.5 ml per kg (measured body weight). Tidal volume decreased between mid to late 1998; it was 8.9 ml per kg for the last 2 years of the study (the change in tidal volume occurred before initial release of results from the ARDS Network trial in 1999). A tidal volume of 6 ml per kg or less was used in 0.9% of patients. Between the first and third day of mechanical ventilation, the setting of tidal volume was decreased by only 33 ml. The authors conclude that clinicians started ventilating acute lung injury patients with lower tidal volumes in 1998 as compared with 1994, but the decrease in tidal volume was modest. An editorial commentary by Ricard (12) accompanies this article.

Liquid ventilation.
During partial liquid ventilation in the setting of ventilator-induced lung injury in rats, Ricard and coworkers (13) studied the influence of varying the dose of perfluorocarbon from 0 to 20 ml per kg. A high dose of perfluorocarbon (20 ml per kg) aggravated the lung capillary leak induced by ventilation at a tidal volume of 33 ml per kg. The leak was decreased partially by PEEP of 5 cm H₂O. A low dose of perfluorocarbon (10 ml per kg) decreased the amount of capillary leak induced by ventilation without perfluorocarbon or with a high dose of perfluorocarbon. PEEP had no effect on the capillary leak observed at the low dose of perfluorocarbon. The high dose of perfluorocarbon was accompanied by a 68% increase in functional residual capacity (FRC); this gas trapping was reduced by PEEP of 5 cm H₂O. The authors conclude that the effect of perfluorocarbon on the vulnerability to ventilator-induced lung injury is complex: low doses of perfluorocarbon protect against lung capillary leak whereas high doses aggravate the leak (and the latter can be reduced by the addition of PEEP).

Ventilator-induced lung injury.
To determine whether ventilator-induced lung injury is accompanied by stress failure of the plasma membrane, Gajic and coworkers (14) subjected perfused rat lungs to one of four ventilator strategies. The lungs were perfused with a membrane-impermeable marker, propidium iodide, and the number of marker-positive cells per alveolus (assessed by confocal microscopy) was taken as a measure of stress failure of the plasma membrane. Ventilation with a tidal volume of 40 ml per kg plus zero PEEP caused more plasma membrane wounds than did ventilation with a tidal volume of 6 ml per kg plus PEEP of 3 cm H₂O: 0.35 versus 0.04 marker-positive cells per alveolus. The number of wounded cells was correlated with gain in lung weight, change in dynamic compliance, and histologic injury score. In a second set of experiments, the propidium iodide marker was perfused either during or after ventilation with a high tidal volume for 30 minutes. Injured cells decreased by about 66% in the period immediately after ventilation. The authors conclude that ventilator-induced lung injury is accompanied by plasma membrane stress failure, and that discontinuation of ventilation results in wound healing (restoration of integrity of the plasma membrane). An editorial commentary by McNeil (15) accompanies this article.

In pigs with surfactant deactivation induced by Tween instillation, Halter and coworkers (16) determined whether recruited alveoli would again collapse or become unstable if PEEP was not applied after the recruitment maneuver. The number and stability of alveoli was measured by in vivo microscopy. A recruitment maneuver consisting of peak pressure of 45 cm H₂O plus PEEP of 35 cm H₂O for 1 minute opened a significant number of alveoli, which remained stable during the maneuver. After the recruitment maneuver, pigs ventilated with PEEP of 10 cm H₂O showed improved oxygenation and alveolar stability; but pigs ventilated with PEEP of 5 cm H₂O showed significant alveolar instability. The authors conclude that collapsed alveoli opened by a recruitment maneuver collapse again if ventilation is not accompanied by an adequate level of PEEP. An editorial commentary Kavanagh (17) accompanies this article.

Copland and coworkers (18) used gene arrays to assess global changes in gene expression after exposing in vivo adult rat lungs to 30 minutes of ventilation at high tidal volume (25 ml per kg). Thirty minutes of high-volume ventilation did not induce discernable change in lung histology or lung mechanics; obvious injury occurred when ventilation was continued for 90 minutes. Thirty minutes of high-volume ventilation, however, did cause significant upregulation of 10 genes and suppression of 12 genes. Among the upregulated genes were transcription factors, stress proteins, and inflammatory mediators. The downregulated genes were exemplified by metabolic regulatory genes. Temporal studies revealed that Egr-1 and c-Jun were increased early and before heat shock protein 70 and interleukin-1ß. All four of the genes were upregulated primarily in the bronchiolar airway epithelium. Ninety minutes of high-volume ventilation caused an increase in intracellular interleukin-1ß protein. The authors conclude that specific patterns of gene activation and suppression precede the lung injury caused by high-volume ventilation. An editorial commentary by Jacobson and Garcia (19) accompanies this article.

Hypercapnic acidosis protects against lung injury caused by a direct pulmonary insult, but it is not known if it protects against injury originating in a nonpulmonary source. To investigate this issue, Laffey and coworkers (20) used a splanchnic ischemia-reperfusion injury to produce lung injury in rats. Compared with control conditions, hypercapnia (resulting from the addition of CO₂ to inspired gas) attenuated protein leakage, improved oxygenation, and preserved lung mechanics. Hypercapnia had both therapeutic and prophylactic actions. Protection was dose-dependent, although benefit at inspired CO₂ concentrations above 5% was small. Protection occurred despite pulmonary artery pressure being higher than the pressure during normocapnia. Hypercapnia did not alter the injury to the bowel caused by ischemia and reperfusion. The authors conclude that hypercapnia protects against lung injury in rats caused by bowel injury resulting from ischemia and reperfusion.

To determine the effect of ventilation with and without a recruitment strategy on cardiopulmonary function, Duggan and coworkers (21) studied rats receiving conventional ventilation. Mortality was lower in rats ventilated with recruitment maneuvers plus PEEP than in rats ventilated without recruitment maneuvers: 59 versus 100%. Ventilation without recruitment maneuvers also resulted in greater hypoxemia, lung stiffness, pulmonary hypertension, microvascular leak, and lactic acidosis. Rats ventilated without recruitment maneuvers exhibited right-ventricular dysfunction on echocardiography and endothelial disruption on electron microscopy. The authors conclude that atelectasis arising from lack of recruitment during mechanical ventilation is accompanied by microvascular leak and right-ventricular dysfunction, which may contribute to systemic abnormalities and increased mortality.

To determine whether ventilator-induced lung injury causes a systemic microvascular leak that is dependent on nitric oxide synthase expression, Choi and coworkers (22) ventilated rats for 2 hours with tidal volumes of 7 or 20 ml per kg. The larger tidal volume induced a significant microvascular leak in both the lungs and the kidneys, and also increased the serum concentration of vascular endothelial growth factor. The larger tidal volume increased the expression of endothelial nitric oxide synthase in lung and kidney tissue, but not inducible nitric oxide synthase. An inhibitor of nitric oxide synthase, N-nitro-L-arginine methyl ester, attenuated the microvascular leak in the lungs and kidneys. The authors conclude that ventilator-induced lung injury results in a systemic microvascular leak that is mediated by endothelial nitric oxide synthase.

Mechanical ventilation at high tidal volume downregulates the function of alveolar sodium, potassium–ATPase, and impairs clearance of lung liquid. Adir and coworkers (23) investigated whether the overexpression of sodium, potassium–ATPase in alveolar epithelium increases liquid clearance in rats. Seven days before inducing lung injury with a tidal volume of 40 ml per kg, the rats were infected with a replication-incompetent adenovirus to overexpress sodium, potassium-ATPase ß₁ subunit gene. Compared with sham or animals infected with a null gene, the animals infected with the ß₁ subunit gene exhibited greater lung liquid clearance, greater sodium, potassium-ATPase activity, and more abundant protein. The authors conclude that overexpression of the ß₁ subunit gene for sodium, potassium-ATPase in alveolar epithelium increases activity of sodium, potassium-ATPase and lung liquid clearance in rats with ventilator-induced lung injury. An editorial commentary by Saumon (24) accompanies this article.

Mechanical distortion of blood vessels is known to activate endothelial cells. To determine whether airway distension with the application of PEEP promotes leukocyte recruitment, Lim and Wagner (25) used intravital microscopy in rat tracheas. Normal mechanical ventilation produced no change in leukocyte rolling velocity and the number of adherent cells over 2 hours. Ventilation with PEEP of 8 cm H₂O for 1 hour caused a decrease in leukocyte rolling velocity and an increase in adhesion. PEEP did not alter leukocyte recruitment in the mesenteric circulation. Application of PEEP distal to the site of measurement in the airway did not induce leukocyte recruitment. Pretreatment with endothelin receptor and selectin inhibitors blocks the effect of PEEP on leukocyte recruitment. The authors conclude that airway distension induced by PEEP leads to inflammatory leukocyte trafficking in the airways. An editorial commentary by Uhlig (26) accompanies this article.

In a critical care perspective, Dreyfuss and colleagues (27) discuss the physiological and clinical relevance of lung-borne cytokines during ventilator-induced lung injury.

Tierney (28) recalls early studies of ventilator-induced lung injury.

Ventilator-induced diaphragmatic injury.
In rats, Racz and coworkers (29) studied the effects of 24 hours of mechanical ventilation on muscle protein and transcription factors at the levels of both messenger RNA and protein. The diaphragm of ventilated rats exhibited decreases in messenger RNA of MyoD, myosin heavy chain-2a and -2b, and sarcoplasmic/endoplasmic reticulum calcium–ATPase-1a; messenger RNA of myogenin was increased. Protein expression of MyoD and myogenin followed the changes in messenger RNA; isoforms of myosin heavy chains did not change. Parallel experiments on the gastrocnemius revealed that immobilization induced deconditioning; passive shortening induced no additional effects. The two muscles exhibited a similar pattern of change with the exception of protein expression of MyoD, which increased in the gastrocnemius and decreased in the diaphragm (messenger RNA decreased in both muscles). The authors conclude that controlled mechanical ventilation alters the levels of messenger RNA for muscle heavy chain isoforms and sarcoplasmic/endoplasmic reticulum calcium–ATPase isoforms within 24 hours.

In a state of the art review article, Laghi and Tobin (30) discuss disorders of the respiratory muscles.

Weaning.
Laghi and coworkers (31) asked, "Do patients who fail a trial of weaning from mechanical ventilation develop diaphragmatic fatigue?". Transdiaphragmatic pressure was recorded during twitch stimulation of the phrenic nerves before and 30 minutes after completing a spontaneous breathing trial (lasting up to 60 minutes) in 11 weaning-failure and 8 weaning-success patients. Twitch transdiaphragmatic pressure was 8.9 cm H₂O before the trials and 9.4 cm H₂O after the trials in the weaning-failure patients; the corresponding values in the weaning-success patients were 10.3 and 11.2 cm H₂O. The failure of twitch pressure to decrease in the weaning-failure patients appears to have arisen because physicians interrupted the trial based on clinical manifestations of respiratory distress before patients had sufficient time to develop contractile fatigue. Weaning-failure patients displayed increased recruitment of rib cage and expiratory muscles, and considerable diaphragmatic weakness. The authors conclude that patients failing a trial of weaning from mechanical ventilation do not develop low-frequency fatigue of the diaphragm, although many patients display severe diaphragmatic weakness.

Ferrer and coworkers (32) randomized 43 patients who failed trials of spontaneous breathing on three consecutive days to extubation plus noninvasive ventilation or to continued intubation plus daily spontaneous breathing trials. The trial was stopped early because the end-point was reached after studying half the planned study population. Compared with conventional weaning, patients weaned with noninvasive ventilation had a shorter duration of invasive ventilation (9.5 versus 20.1 days), shorter stay in ICU (14.1 versus 25.0 days), shorter hospital stay (27.8 versus 40.8 days), less need for tracheotomy (1.5 versus 13.6%), lower incidence of nosocomial pneumonia (5.2 versus 13.6%), lower incidence of septic shock (2.1 versus 9.4%), greater ICU survival (19.9 versus 13.6%), and greater 90-day survival (about 75% versus 40%). Conventional weaning was independently associated with decreased ICU survival (odds ratio, 6.6) and decreased 90-day survival (odds ratio, 3.5). The authors conclude that use of noninvasive ventilation in patients who fail three consecutive days of conventional weaning results in shorter duration of mechanical ventilation, fewer complication, less need for tracheotomy, and increased ICU and 90-day survival. An editorial commentary by Navalesi and coworkers (33) accompanies this article.

Patient posture.
Because the prone position increases the homogeneity of lung inflation, Pelosi and coworkers (34) determined whether a sigh causes greater recruitment in the prone position than in the supine position. In 10 patients with ARDS, three consecutive sighs (increase in plateau pressure to 45 cm H₂O) increased PO₂ more in the prone position than in the supine position (37 versus 16 mm Hg), and the sighs also increased end-expiratory lung volume more in the prone position than in the supine position (277 versus 68 ml). Improvement was sustained for one hour after discontinuation of the sighs. The increase in PO₂ with sighs was correlated with the change in end-expiratory lung volume (r = 0.82). The authors conclude that the recruitment effect of sighs is greater in the prone position than in the supine position.

Noninvasive ventilation.
To determine whether noninvasive ventilation is of benefit for treating cardiogenic pulmonary edema in the emergency department, Nava and coworkers (35) did a multicenter, randomized trial in 130 patients in acute respiratory failure secondary to cardiogenic pulmonary edema. Compared with medical therapy and supplemental oxygen, noninvasive ventilation with pressure support achieved more rapid improvements in dyspnea, respiratory rate, and Pa_O2/FI_O2 ratio. The rate of intubation, duration of hospital stay, and hospital mortality were equivalent for the two groups. In the subgroup of patients with hypercapnia, noninvasive ventilation achieved more rapid improvement in PCO₂ than did standard therapy, and also decreased the rate of intubation: 6 versus 29%. Adverse effects, including myocardial infarction, were evenly distributed in the two groups. The authors conclude that use of noninvasive ventilation for the treatment of acute respiratory failure secondary to cardiogenic pulmonary edema in the emergency department achieves more rapid improvements in dyspnea, respiratory rate, and arterial oxygenation, without affecting intubation rate or mortality, although intubation rate is decreased in the subgroup of patients with hypercapnia. An editorial commentary by Bersten (36) accompanies this article.

To determine whether use of noninvasive ventilation is beneficial for the treatment of acute hypoxemic respiratory failure, Ferrer and coworkers (37) did a randomized trial in 105 patients with severe hypoxemia (PO₂ less than 60 mm Hg or oxygen saturation less than 90% while breathing an oxygen concentration of 50%). Compared with oxygen therapy, noninvasive ventilation decreased the need for intubation (13.3% versus 28.5%), incidence of septic shock (6.1% versus 17.3%), and ICU mortality (9.2% versus 21.4%); 90-day survival was also increased. Noninvasive ventilation achieved a greater improvement in PO₂ and respiratory rate over time. On multivariate analysis, noninvasive ventilation was independently associated with decreased risk of intubation and decreased 90-day mortality. The authors conclude that use of noninvasive ventilation in patients with acute hypoxemic respiratory failure decreased the rate of intubation and septic shock and improved survival as compared with high-concentration oxygen-therapy.

Adjunctive therapy.
In nine patients with neuromuscular disease receiving mechanical ventilation through a cuffless tracheostomy, Prigent and coworkers (38) compared the effect of assist-control ventilation and bilevel positive-pressure ventilation (pressure support and PEEP) on speech production. At rest, breathing pattern was equivalent with the two modes. Speech was accompanied by a 20% increase in inspiratory time during bilevel ventilation but not during assist-control ventilation. The volume released by the ventilator during speech was greater with bilevel ventilation than during assist-control ventilation: 172 versus 26 ml. The maximum duration of speech (as a fraction of a respiratory cycle) was greater with bilevel ventilation than with assist-control ventilation: 74 versus 39%. Speech continued into expiration with bilevel ventilation and three patients were able to speak continuously during several respiratory cycles. Comfort during speech was greater with bilevel ventilation. The authors conclude that the duration of speech during a respiratory cycle was greater during bilevel positive-pressure ventilation (pressure support and PEEP) than during assist-control ventilation. An editorial commentary by Hoit and Banzett (39) accompanies this article.

In nine patients with acute lung injury, Maggiore and coworkers (40) studied the fall in end-expiratory lung volume (derecruitment) during endotracheal suctioning. Suctioning performed after disconnection from the ventilator produced a decrease in lung volume of 1,466 ml. The decrease in lung volume during suctioning after disconnection from the ventilator resulted almost equally from simple disconnection (728 ml) and application of negative pressure (737 ml). The decrease in lung volume was less when suctioning was performed via a catheter introduced through the swivel adaptor of the catheter mount (733 ml) or with use of a closed suctioning system (531 ml). The decrease in lung volume during suctioning was no longer statistically significant when pressure support produced peak inspiratory pressures of 40 cm H₂O during suctioning through the swivel (168 ml) or a closed system (284 ml). Oxygenation paralleled the changes in lung volumes. The authors conclude that endotracheal suctioning decreases end-expiratory lung volume in patients with acute lung injury, and that the decreases can be minimized by avoiding disconnection from the ventilator and through the use of a closed-suction system.

During bench application of modern ventilators, Miller and coworkers (41) investigated the key variables affecting aerosol delivery and also assessed the relationship between bench predictions and in vivo end points in patients. With the bench studies, inhaled mass (percentage of nebulizer charge) ranged between 5.7% and 37.4%, and breath-activated nebulization and humidity were found to be the most important determinants of aerosol delivery. In patients, levels of deposited antibiotics in sputum of patients ranged from 1.1 to 19.6 µg per ml per mg, and the level correlated with predictions from the bench model. The authors conclude that bench models of aerosol delivery during mechanical ventilation can predict the delivery of aerosols in intubated patients. An editorial commentary by Dhand (42) accompanies this article.

Acute Lung Injury and Acute Respiratory Distress Syndrome
Epidemiology and genetics.
To assess pulmonary function and quality of life at 1 year after surviving ARDS, Orme and coworkers (43) studied 66 patients who had been entered in a trial of high versus low tidal volumes. Pulmonary function at 1 year was not affected by the setting of tidal volume. About 80% of survivors had reduced diffusing capacity, 20% had airflow obstruction, and 20% had chest restriction. Scores for depression and anxiety were within the normal range. Measures of quality of life were decreased: decreased physical functioning, physical ability to maintain roles, body pain, general health, and vitality. Abnormalities in pulmonary function were correlated with decreases in health-related quality of life for domains reflecting physical function. The authors conclude that survivors of ARDS exhibit abnormal pulmonary function at 1 year, which is related to decreased quality of life.

To determine whether ARDS complicating bacteremic sepsis has an independent effect on mortality, Eggimann and coworkers (44) analyzed data on 4,530 admissions to an ICU. There were 196 cases of bacteremic sepsis and 31 (16%) of these patients developed ARDS. On Cox proportional hazards regression, ARDS was associated with increased mortality (unadjusted hazard ratio, 1.8). After adjusting for comorbid factors present before the onset of sepsis, ARDS was still associated with increased mortality (adjusted hazard ratio, 2.2). After adjusting for nonpulmonary organ dysfunction and microbiologic factors, ARDS was not associated with increased mortality (adjusted hazard ratio, 0.6). The authors conclude that ARDS complicating bacteremic sepsis is not independently associated with short-term mortality once the severity of illness and nonpulmonary organ dysfunction have been taken into account.

Animal models.
It is unknown which of the three isoforms of inducible nitric oxide synthase contributes to the acute lung injury that results with burns and smoke inhalation. In control group of sheep, Enkhbaatar and coworkers (45) found that third-degree burns involving 40% of the body surface plus insufflation of cotton smoke (48 breaths of less than 40°C) induced multiple features of ARDS. The abnormalities were accompanied by large increases in tracheal blood flow and increased plasma levels of nitrate/nitrite. A second group of sheep received a constant infusion of a highly selective inhibitor of inducible nitric oxide synthase (BBS-2), commencing 1 hour after injury. This inhibitor caused attenuation of all aspects of the pulmonary pathology, and lessened the increase in tracheal blood flow and plasma levels of nitrate/nitrite. The authors conclude that inducible nitric oxide synthase is a key mediator of the ARDS that results from burn and smoke inhalation injury in sheep.

Rocco and coworkers (46) studied the effect of low-dose glucocorticoids on the lung injury caused by intraperitoneal administration of paraquat in rats. Both 10 and 25 mg/kg of paraquat induced increases in lung resistive, viscoelastic, and static elastance pressures on the first day after exposure. Hysterestivity was increased only with a paraquat dose of 25 mg/kg. Administration of methylprednisolone at either 1 or 6 hours produced attenuation of mechanical changes after 25 mg/kg of paraquat, but had no effect on changes induced by 10 mg/kg of paraquat. Collagen and elastic fibers increased in a dose-dependent fashion. Methylprednisolone prevented the increase in collagen and avoided elastogenesis. The authors conclude that methylprednisolone prevents the abnormal respiratory mechanics associated with mild lung injury caused by paraquat and minimizes the changes in tissue impedance and the extracellular matrix associated with severe lung injury caused by paraquat.

The hydrophobic surfactant protein C binds to lipopolysaccharide (which is found in airborne particles) and one of its cellular receptors, CD14. Augusto and coworkers (47) investigated the influence of surfactant protein C on the responses of immunocompetent cells to lipopolysaccharide. When associated with vesicles of dipalmitoylphosphatidylcholine, surfactant protein C inhibited the mitogenic effect of lipopolysaccharide to a macrophage cell line (RAW 264.7). Under similar conditions, surfactant protein C inhibited the mitogenic effect of lipopolysaccharide on mouse splenocytes, inhibited the lipopolysaccharide-induced production of tumor necrosis factor- by peritoneal and alveolar macrophages, and of nitric oxide by the macrophage cell line. In contrast, surfactant protein C did not affect production of tumor necrosis factor- induced by a lipopeptide or production of nitric oxide induced by picolinic acid. The lipopolysaccharide-binding capacity of surfactant protein C was resistant to peroxynitrite (a known mediator of acute lung injury formed by the reaction of nitric oxide with superoxide anions). The authors conclude that surfactant protein C resists degradation under inflammatory conditions and traps lipopolysaccharide, preventing it from inducing the production of noxious mediators in alveolar cells.

Many genes elicited by bacteria in the lungs are regulated by B sites in DNA, which bind nuclear factor-B proteins. Mizgerd and coworkers (48) determined whether the p50 subunit of nuclear factor-B limits the expression of B-associated genes and prevents excessive inflammation and injury. During pneumonia caused by Escherichia coli, gene-targeted deficiency of p50 caused increased mortality, but it did not alter bacterial clearance from mouse lungs. Deficiency of p50 caused increased expression of proinflammatory cytokines. The dysregulation produced aggravation of inflammation (increased neutrophil recruitment), pulmonary edema, worse gas exchange, and leak of protein and bacteria from the lungs with consequent bacteremia and multiple organ failure. The authors conclude that the endogenous p50 subunit of nuclear factor-B limits the inflammatory injury that occurs during pneumonia.

Cellular and molecular mechanisms.
To determine whether insulin-like growth factor-I contributes to the fibroproliferative phase of ARDS, Krein and coworkers (49) studied lung biopsies from eight patients who displayed organizing diffuse alveolar damage. Immunochemical studies revealed enhanced staining for insulin-like growth factor-I and its receptor, collagen I, collagen III, smooth muscle actin, CD68, and proliferating cell nuclear antigen. Staining for insulin-like growth factor-I and its receptor was prominent in alveolar and interstitial macrophages and in a variety of mesenchymal cells. Staining for insulin-like growth factor-I was correlated with CD68-positive cells, enhanced collagen I, collagen III, and proliferating cell nuclear antigen immunoreactivity, suggesting that the growth factor plays a role in the deposition of protein in the extracellular matrix and the proliferation of cells in the fibroproliferative phase of ARDS. The authors conclude the insulin-like growth factor-I is increased in the fibroproliferative phase of ARDS.

In a state of the art review article, Kinnula and Crapo (50) discuss superoxide dismutase in lung disease.

In a state of the art review article, de Perrot (51) discusses ischemia-reperfusion injury that results after lung transplantation.

Fluid biology.
Tumor necrosis factor- activates sodium channels in Type II alveolar epithelial cells, and this effect could have favorable consequences for fluid clearance from the lung. In mouse and rat lung models, Elia and coworkers (52) investigated the relative contribution of tumor necrosis factor- receptor-dependent and receptor-independent activities to the capacity for fluid resorption. In an in situ mouse lung model, fluid resorption secondary to tumor necrosis factor- was functional in mice that were genetically deficient in the receptor for tumor necrosis factor-, indicating that the receptor-independent effect is important. In an ex vivo rat lung model, a tumor necrosis factor- tip peptide (a peptide that mimics the lectin-like domain of the cytokine, and activates sodium transport by a receptor-independent mechanism), in contrast with tumor necrosis factor-, produced progressive improvement in lung mechanics after alveolar flooding (and a reduction in lung water). The authors conclude that receptor-independent activities of murine tumor necrosis factor- predominate in the resorption of alveolar edema in rats and mice. An editorial commentary by Berthiaume (53) accompanies this article.

Pulmonary edema caused by scorpion venom is attributed by increased pulmonary vascular permeability. To determine effect of scorpion venom on clearance of alveolar fluid, Comellas and coworkers (54) injected rats intraperitoneally with venom (from Tityus serrulatus). Wet-to-dry weight ratio of the lung increased and clearance of lung edema decreased by about 60%. Protein abundance of the ₁- and ß₁-subunits of sodium-potassium-ATPase decreased at the basolateral membranes of type II cells of the alveolar epithelium. The authors conclude that scorpion venom decreases clearance of alveolar fluid probably secondary to downregulation of sodium, potassium-ATPase in the alveolar epithelium.

To determine whether inhibiting the production of nitric oxide by the lung influences the clearance of alveolar fluid, Tsubochi and coworkers (55) instilled endotoxin into the trachea of adult rats. Levels of nitric oxide in the lung reached a maximum at 6 hours after instillation of endotoxin, and the production was accompanied by increases in cyclic guanosine monophosphate. Clearance of alveolar fluid decreased at 6 hours and then increased at 24 hours; these changes were related to the function of sodium channels sensitive to amiloride. Administration of gadolinium chloride and aminoguanidine decreased the levels of nitric oxide and cyclic guanosine monophosphate, and inhibited the changes in alveolar fluid clearance. Inducible nitric oxide synthase was abundantly expressed in the cytoplasm of alveolar macrophages. The authors conclude that endotoxin causes the production of nitric oxide by alveolar macrophages and leads to changes in alveolar fluid clearance.

Treatment.
In 40 patients with ARDS, Spragg and coworkers (56) did a Phase I/Phase II randomized trial of a recombinant surfactant protein C-based surfactant (venticute). Patients received one or two doses of exogenous surfactant on four occasions over 24 hours. Treatment was well tolerated. No benefit was noted. At 48 hours, bronchoalveolar lavage reflected exogenous surfactant components, did not show improvement in lowering of surface tension, and had lower level of interleukin-6 than did the control group. Exogenous surfactant was not detected in lavage fluid at 120 hours. The authors conclude that a recombinant protein C-based surfactant can be safely administered to patients with acute lung injury.

Richard and coworkers (57) investigated the reliability of positron emission tomographic imaging in detecting the expression of genes delivered to the lungs by viral vectors. The study was conducted in normal rats and an enhanced mutant herpes simplex virus-1 thymidine kinase was used as the reporter gene. Rats were studied 3 days after the intratracheal administration of a replication-incompetent adenovirus containing a fusion gene for the mutant kinase and green fluorescent protein. The rats were injected with an imaging substance for the viral kinase, 9-(4-[¹⁸F]-fluoro-3-hydroxymethylbutyl)guanine; images were obtained 1 hour later. Measurements derived from imaging were linearly correlated with measures of thymidine kinase activity and green fluorescent protein levels in tissue (r² = 0.96). Imaging detected expression of thymidine kinase in 15 of 16 rats, even at low viral doses. Imaging for the viral kinase was correlated with in vitro assays for both kinase activity (r² = 0.48) and fluorescent protein (r² = 0.46). The authors conclude that positron emission tomographic imaging is a sensitive and quantitative method for noninvasive detection of pulmonary reporter gene expression.

Review article.
In a summary report from a National Heart, Blood, and Lung Institute Workshop, Matthay and colleagues (58) discuss future research directions in acute lung injury.

Sepsis and Shock
Epidemiology and genetics.
To determine the clinical significance of the criteria for the systemic inflammatory response syndrome (SIRS), Alberti and coworker (59) analyzed risk factors for hospital mortality in 3,608 ICU patients in the European Sepsis Study. Hospital mortality was variable: about 25% in patients with uncomplicated infection or sepsis, 40% in patients with severe sepsis, and 60% in patients with septic shock. Outcome was equivalent for patients who had infection and met SIRS criteria (sepsis) and patients who had infection and did not meet SIRS criteria (infection only). When patients were stratified according to severity of sepsis, the number of fulfilled SIRS criteria had no influence on mortality. On Cox regression, factors associated with outcome included comorbid conditions, severity of acute illness and acute organ dysfunction, shock, nosocomial infection, infection caused by aerobic gram-negative bacilli, enterobacteria, Staphylococcus aureus, and infection from a digestive source or unknown source. The authors conclude that characterization of patients with infection on the basis of SIRS criteria has no prognostic implication, whereas categorization on the basis of organ dysfunction or shock has prognostic significance.

To characterize the epidemiology of septic shock, Annane and coworkers (60) analyzed data on 100,554 admissions to 22 ICUs in Paris. The overall frequency of septic shock was 8.2 per 100 admissions. Frequency increased from 7% of admissions in 1993 to 9.7% of admissions in 2000. Both the rate of pulmonary infection and septic shock caused by multiresistant bacteria increased over time. Crude mortality was 62.1% in 1993 and 55.9% in 2000. Compared with patients admitted to the ICU without sepsis, the increased risk of death secondary to septic shock was 25.7; the matched odds ratio of death was 3.9. The authors conclude that the frequency of septic shock is increasing and is involving more multiresistant strains, and that mortality is decreasing but remains higher than in nonseptic critically ill patients.

To characterize the epidemiology of sepsis in children, Watson and coworkers (61) analyzed data on 1,586,253 hospitalizations in children, 19 years of age or younger. These were 42,364 cases of severe sepsis per year nationally in children (0.56 cases per 1,000 population per year). Incidence was highest in infants (5.2 per 1,000), lower in 10- to 14-year-old children (0.2 per 1,000), and 15% higher in boys than in girls. Half the cases had an underlying disease, and 23% were low-birth-weight newborns. The most common infections were respiratory (37%) and primary bacteremia (25%). Hospital mortality was 10.3%, mean length of stay was 31 days, and cost was $40,600. Annual costs were estimated at $1.97 billion nationally. The authors conclude that there are more than 42,000 cases of sepsis with 4,400 associated deaths in the United States each year.

Schaaf and coworkers (62) asked, "Are patients who are genetically predisposed to increased production of interleukin-10 (as determined by interleukin-10–1082 polymorphism) at increased risk of severe pneumococcal infection leading to septic shock?". Polymorphisms were studied by the polymerase chain reaction in 69 patients with pneumococcal disease (61 with community-acquired pneumonia, 5 with meningitis, and 3 with pneumonia and meningitis) and 50 control subjects. No significant genotype differences were seen between the patients and the control subjects. Of 69 patients with pneumococcal disease, 13 developed septic shock. Interleukin-10 allele G–homozygous patients had the highest risk for septic shock (odds ratio, 6.1). Release of interleukin-10 from whole blood was 46% greater in patients who were homozygous for the interleukin-10 allele G than in nonhomozygotes. No association was found between tumor necrosis factor genotypes and sepsis severity. The authors conclude that patients who are homozygous for interleukin-10–1082 allele G (associated with increased production of interleukin-10) are at increased risk of developing septic shock from pneumococcal infection.

In the proximal coding region of the gene for the lipopolysaccharide binding protein, a T G single nucleotide polymorphism at nucleotide 292 has been reported as a risk for sepsis after trauma. Barber and O'Keefe (63) used multiple analytical methods to characterize this coding region. The single nucleotide polymorphism at 292 was not found, but an adjacent nucleotide (291) was found to be polymorphic. Among 151 trauma patients, 25% developed severe sepsis and 13% died. Severe sepsis occurred in 13 of 50 C-allele carriers and 24 of 101 TT homozygotes. Genotype was not associated with sepsis or death. The authors conclude that a single nucleotide polymorphism exists at position 291 in the proximal coding region of the gene for lipopolysaccharide binding protein, and that this polymorphism is not associated with complicated sepsis after trauma.

Mechanisms in patients and volunteers.
The transcriptional regulatory factor, nuclear factor-B, modulates many of the proinflammatory mediators implicated in acute lung injury. In 30 patients with acute lung injury, Yang and coworkers (64) examined the relationship between clinical outcome and activation of nuclear factor-B in peripheral neutrophils. Nuclear translocation of nuclear factor-B depended on activation of p38 mitogen-activated protein and Akt kinases. Diminished activation of nuclear factor-B or Akt, but not p38 mitogen-activated protein, in the early period after intubation was associated with a shorter duration of mechanical ventilation and improved survival. The authors conclude that alterations in neutrophil activation patterns in patients with early acute lung injury, particularly involving the ability to accumulate nuclear factor-B in the nucleus after relevant stimuli, influences subsequent clinical course.

In a clinical commentary, Vieillard-Baron and colleagues (65) discuss the use of echocardiography in assessing hemodynamic instability in sepsis.

Endotoxemia in animals.
ß-Lapachone, a 1,2-naphthoquinone, is a novel chemotherapeutic agent, which is capable of suppressing expression and function of nitric oxide synthase in rat alveolar macrophages. Tzeng and coworkers (66) investigated the molecular mechanisms for the actions of ß-lapachone on the response of rat alveolar macrophages to lipopolysaccharide. Lipopolysaccharide increased the production of nitrite and expression of inducible nitric oxide synthase in macrophages; these actions were inhibited by coincubation with ß-lapachone. Lipopolysaccharide induced the production of tumor necrosis factor-; this action was inhibited by ß-lapachone. Lipopolysaccharide induced phosphorylation of protein tyrosine and binding activity of nuclear factor-B in macrophages; these actions were significantly inhibited by ß-lapachone. In in vivo studies, ß-lapachone protected against the development of lung edema, protein expression of inducible nitric oxide synthase, activation of nuclear factor-B, plasma nitrite, serum tumor necrosis factor-, and mortality induced by lipopolysaccharide. The authors conclude that ß-lapachone suppresses the induction of inducible nitric oxide synthase and production of tumor necrosis factor- mediated by inhibiting the phosphorylation of tyrosine kinase and activation of nuclear factor-B that is caused by lipopolysaccharide.

In a rat model of acute lung injury induced by lipopolysaccharide, Agorreta and coworkers (67) investigated the effect of hypoxia on expression of nitric oxide synthase-2 (the endothelial, inducible isoform of the enzyme). Exposure to hypoxia alone had no effect on the expression of nitric oxide synthase-2 in rat lungs. Intraperitoneal injection of lipopolysaccharide increased the level of nitric oxide synthase-2 in the lungs, which was further enhanced by concomitant exposure to hypoxia. Resident lung cells showed no changes in the expression of nitric oxide synthase-2 under any conditions. In in vitro experiments, lung epithelial and endothelial cell lines showed no detectable expression of nitric oxide synthase-2 with any treatment. In a macrophage cell line, expression of nitric oxide synthase-2 in response to lipopolysaccharide was not affected by concomitant exposure to hypoxia. The authors conclude that the increase in nitric oxide synthase-2 in acute lung injury caused by lipopolysaccharide results from recruitment of leukocytes that produce the enzyme, and that concomitant hypoxia increases leukocyte recruitment and further enhances the expression of nitric oxide synthase-2.

Because mitochondria are the principal organelles that consume oxygen and generate reactive oxygen species, Suliman and coworkers (68) determined whether cell activation by lipopolysaccharide would cause damage to mitochondria through activation of mitochondrial DNA. A single injection of lipopolysaccharide caused a decrease in copy number of mitochondrial DNA in the liver of rats. Lipopolysaccharide caused an oxidant-dependent 3.8-kb mitochondrial DNA deletion in the region encoding NADH dehydrogenase subunits 1 and 2 and cytochrome c oxidase subunit I (which correlated with depletion of mitochondrial glutathione). Expression of mitochondrial messenger RNA and transcription of mitochondrial RNA were suppressed; expression of messenger RNA was increased for selected nuclear-encoded mitochondrial proteins. Resolution of the damage to mitochondrial DNA was mediated by importing mitochondrial transcription factor A protein (a central regulator of the copy number of mitochondrial DNA), accompanied by binding of mitochondrial protein extract to the mitochondrial transcription factor A DNA–binding site. The authors conclude that hepatic injury induced by lipopolysaccharide is associated with a specific oxidative mitochondrial DNA deletion, which inhibits mitochondrial transcription and is restored by activation of mechanisms that lead to biogenesis.

Sepsis in animals.
Activation of coagulation by tissue factor is an early event in the pathogenesis of acute lung injury and organ failure that results from sepsis. To determine whether blockade of tissue factor would attenuate these injuries by preventing fibrin deposition and inflammation, Carraway and coworkers (69) administered heat-killed E. coli intravenously to 12 baboons, followed 12 hours later by an infusion of live E. coli. Six animals received site-inactivated Factor VIIa (a competitive inhibitor of tissue factor) at 2 hours after the live bacteria, and six animals received vehicle. Animals treated by site-inactivated Factor VIIa displayed less severe lung injury, preserved gas exchange, better lung compliance, better histology scores, lower lung wet-to-dry weight ratios, higher urinary output, attenuation of metabolic acidosis, and protection of renal tubular architecture. Treated animals exhibited attenuation of coagulopathy and lower levels of plasma interleukin-6, interleukin-8, and soluble tumor necrosis factor receptor-1 than did the control animals. The authors conclude that blockade of coagulation attenuates acute lung and renal injury in established gram-negative sepsis accompanied by antiinfammatory effects.

ß2 microglobulin knockout mice (ß2M^-/-) lack CD8⁺ T and natural killer T cells. Sherwood and coworkers (70) determined whether such mice have increased mortality when sepsis is induced by cecal ligation and puncture. Although the ß2 microglobulin knockout mice survived for longer than did wild-type mice, all eventually died. When ß2 microglobulin knockout mice were treated with anti-asialoGM1, to deplete natural killer cells, long-term survival exceeded 70%. Compared with wild-type mice, ß2 microglobulin knockout mice treated with anti-asialoGM1 did not exhibit hypothermia or metabolic acidosis after induction of sepsis and they produced less proinflammatory cytokines. Septic ß2 microglobulin knockout mice treated with anti-asialoGM1 experienced an increase in mortality with adoptive transfer of CD8⁺ T and natural killer cells. CD8 knockout mice treated with anti-asialoGM1, which are deficient in CD8⁺ T cells and natural killer cells, had a greater than 40% survival rate after inducing sepsis. Treating wild-type mice with antibodies to CD8 and asialoGM1 increased survival. The authors conclude that increased survival of ß2-microglobulin knockout mice after cecal ligation and perforation is largely consequent to depletion of CD8⁺ T and natural killer cells.

In a sheep model of septic shock secondary to peritonitis, Sun and coworkers (71) compared the effects of low-dose vasopressin, with or without norepinephrine, on hemodynamics, histology, and survival. Mean arterial pressure was maintained with all treatments. Blood flow in the superior mesenteric artery was lower in sheep treated with vasopressin plus norepinephrine than in sheep treated with vasopressin alone. Compared with sheep treated with Ringer's lactate (control) or norepinephrine alone, sheep treated with vasopressin alone or vasopressin plus norepinephrine experienced less of an increase in blood lactate and ileal PCO₂-gap. Sheep treated with vasopressin had a higher urinary output than the other groups. Survival time was 17 hours in the control group, and was higher with norepinephrine alone (20 hours), vasopressin alone (30 hours), and vasopressin plus norepinephrine (30 hours). Tissue injury was less severe in sheep receiving vasopressin alone or vasopressin plus epinephrine. The authors conclude that low-dose vasopressin (alone or in combination with norepinephrine) improves mesenteric blood flow, limits lactic acidosis and tissue injury, and increases survival in sheep with septic shock secondary to peritonitis.

Treatment of sepsis.
In 40 patients with septic shock, Keh and coworkers (72) studied the effects of hydrocortisone on the balance between proinflammation and antiinflammation. Patients were randomized to receive hydrocortisone (100 mg loading dose followed by 10 mg per hour) for the first 3 days followed by placebo for 3 days, or vice versa. Hydrocortisone induced increases in arterial pressure and systemic vascular resistance, and decreases in heart rate, cardiac index, and norepinephrine requirement. Hydrocortisone inhibited the formation of nitric oxide (reflected by a reduction in the ratio of plasma nitrite to nitrate), which was correlated with the reduction in vasopressor support. Hydrocortisone caused attenuation of the inflammatory response (interleukin-6 and interleukin-8), endothelial activation (soluble E-selectin), neutrophil activation (expression of CD11b and CD64), and the antiiflammatory response (soluble tumor necrosis factor receptors I and II and interleukin-10). In peripheral blood monocytes, expression of human leukocyte antigen-DR was slightly depressed, and phagocytosis and interleukin-12 (a monocyte-activating cytokine) were increased. The authors conclude that low-dose hydrocortisone rapidly induces hemodynamic stabilization in patients with septic shock, probably secondary to decreased formation of nitric oxide, and increases the antiinfammatory rather than the immunosuppressive response to stress. An editorial commentary by Bornstein and Briegel (73) accompanies this article

Conventional diet (and lipid emulsions) predominates in n-6 fatty acids (including arachidonic acid), whereas fish-based lipids yield n-3 fatty acids (rich in eicosapentaenoic acid and docosahexaenoic acid). To determine the effect of n-3 fatty acids on fat profiles and monocyte cytokine production, Mayer and coworkers (74) entered 21 patients with sepsis into a 5-day randomized trial. Before the trial, plasma-free fatty acid concentrations were greatly increased in the septic patients, and n-6 fatty acids greatly exceeded n-3 fatty acids. The high n-6/n-3 ratio was maintained during conventional lipid infusions. Within 2 days of starting the n-3 fatty acid infusion, free n-3 fatty acids increased, the n-3/n-6 ratio was reversed, and n-3 fatty acids were rapidly incorporated into the membranes of mononuclear leukocytes. Generation of proinflammatory cytokines by mononuclear leukocytes was markedly amplified during n-6 lipid infusion and suppressed during n-3 lipid infusion. After termination of n-3 lipid infusion, the concentration of free n-3 fatty acid and cytokine synthesis by mononuclear leukocytes decreased. The authors conclude that use of n-3 fatty acid infusions may achieve a favorable effect on inflammation and immunologic function in patients with sepsis in addition to nutrition.

Nonseptic causes of shock.
To assess the proinflammatory response, antiinflammatory response, site of mediator production, and route of cytokine diffusion in acute pancreatitis, Dugernier and coworkers (75) obtained simultaneous samples of ascites, thoracic lymph, and blood at the onset of end-organ dysfunction and for the subsequent 6 days in 60 patients. Tumor necrosis factor- and interleukin-1ß were found in less than 15% of blood and lymph samples. Secondary proinflammatory and antiinflammatory cytokines were elevated early and throughout the sampling period in all compartments. Cytokine levels decreased from ascites to lymph to blood, suggesting a splanchnic origin. Prolonged diversion of ascites and lymph did not alter cytokine gradients, suggesting transfer of mediators via the splanchnic blood circulation. The authors conclude that the magnitude of the inflammatory response in patients with acute pancreatitis is proportional to pancreatic damage and end-organ dysfunction, that inflammatory mediators are primarily released from the splanchnic area, mediators gain access to the systemic compartment mainly by the portal and suprahepatic circulations, the proinflammatory response is dominant in the pancreas and splanchnic areas, and the antiinflammatory response is dominant in the thoracic duct and systemic circulation. An editorial commentary by Raraty and Neoptolemos (76) accompanies this article.

To determine whether critical oxygen delivery, the point at which oxygen consumption becomes limited by oxygen delivery, is higher when hypoxia is caused by stagnant flow versus anemia, Morita and coworkers (77) induced sepsis in rats by cecal legation and perforation. Rats were randomized to anemic hypoxia (induced by isovolumic hemodilution) or stagnant hypoxia (induced by stepwise inflation of a balloon in the right atrium). The level for critical oxygen delivery did not differ between anemic hypoxia and stagnant hypoxia in the septic animals. The critical hemoglobin concentration for anemic hypoxia was equivalent for septic animals and control animals, indicating that tolerance to acute anemia is not altered by sepsis. The authors conclude that critical oxygen delivery did not differ between anemic and stagnant hypoxia in either septic or control rats.

Ventilator-Associated Pneumonia
To investigate trade-offs in deciding optimal strategies for the management of ventilator-associated pneumonia, Ost and coworkers (78) performed a decision-analysis model involving 16 management strategies. Initial coverage with three antibiotics was superior to expectant management or use of one or two antibiotics: survival of 54% versus 66%; cost of $41,483 versus $55,447 per survivor. Testing with nonbronchoscopic mini-bronchoalveolar lavage decreased costs ($39,967 versus $41,483) and antibiotic use (39 versus 63 antibiotic days per survivor), but did not improve survival. The authors conclude that use of mini-bronchoalveolar lavage and three antibiotics is the optimal strategy for minimizing cost, minimizing use of antibiotics, and maximizing survival in patients with ventilator-associated pneumonia. An editorial commentary by Amin and Hebert (79) accompanies this article.

The clinical pulmonary infection score has been reported to have a sensitivity and specificity of greater than 90% for diagnosis of pneumonia. In 79 episodes of suspected ventilator-associated pneumonia, Fartoukh and coworkers (80) tested the reliability of the score against culture of bronchoalveolar lavage fluid as a reference standard. A physician's estimate of the probability of pneumonia had a sensitivity of 50% and a specificity of 58%. The clinical pulmonary infection score was not statistically higher in 40 episodes of confirmed pneumonia than in 39 episodes not confirmed: mean score of 6.5 versus 5.9. The score had a sensitivity of 60% and a specificity of 59%. Addition of gram-stain results of directed or blinded protected telescoping catheter to the score increased sensitivity to 78% and specificity to 56%. Addition of gram-stain results of bronchoalveolar lavage to the score achieved a sensitivity of 85% and specificity of 49%. The authors conclude that the clinical pulmonary infection score has low diagnostic accuracy and the addition of gram-stain results from a protected telescoping catheter or bronchoalveolar lavage may improve clinical decision-making.

To define the microbiology of severe aspiration pneumonia in institutionalized elderly patients, El-Solh and coworkers (81) prospectively studied 95 patients with severe aspiration pneumonia older than 65 years of age admitted to their ICU from a long-term care facility. Quantitative bronchial samples were obtained in 95 patients and 67 pathogens were identified. Organisms were gram-negative enteric bacilli (49%), anaerobic bacilli (16%), and S. aureus (12%); Prevotella and Fusobacterium species were the most common anaerobes. Aerobic gram-negative bacilli were recovered in conjunction with 55% of anaerobic isolates. Dental plaque did not differ between the aerobic and anaerobic groups. Functional status was the only determinant of anaerobic bacteria. Seven patients with anaerobic isolates initially received inadequate antimicrobial therapy, yet 6 had an effective clinical response. Crude mortality was 33% for the aerobic group and 36% for the anaerobic group. Multivariate analysis revealed hypoalbuminemia and the burden of comorbid diseases as independent risk factors for poor outcome. The authors conclude that anaerobes represent a significant proportion of oral flora in institutionalized elderly patients, but their role in causing aspiration pneumonia has been overemphasized.

Severe Acute Respiratory Syndrome
To assess the relative efficacy of high-dose pulse glucocorticoids versus nonpulse glucocorticoids in the management of patients with severe acute respiratory syndrome (SARS), Ho and coworkers (82) retrospectively analyzed data on 72 patients with probable SARS. Pulse therapy (methylprednisolone of at least 500 mg daily) was used in 17 patients, and nonpulse therapy (methylprednisolone less than 500 mg daily) was initially used in 55 patients. After 21 days, mortality was equivalent in the two groups, as was cumulative dose of glucocorticoids, rate of ICU admission and mechanical ventilation, and hematologic and biochemical variables. Patients receiving pulse glucocorticoids had better radiographic outcomes, lower oxygen requirements, and less likelihood of requiring rescue therapy with pulse glucocorticoids than did the patients treated with nonpulse glucocorticoids. The authors conclude that patients with SARS achieve greater benefit with the use of high-dose pulse glucocorticoid therapy as compared with patients treated with lower dosages of glucocorticoids. An editorial commentary by Bernard (83) accompanies this article.

In a clinical commentary, Tsang and Lam (84) discuss the experience of managing patients with SARS at Hong Kong University.

In a pulmonary perspective, Shortridge (85) discusses the role of zoonotic incursions from southern China in the development of SARS and influenza.

In an occasional essay, Zhong and Zeng (86) discusses strategies for fighting SARS.

Nosocomial Infections
Nitric oxide produced within the paranasal sinuses defends against bacteria and modulates mucociliary clearance. To determine whether decreased formation of nitric oxide might contribute to impaired host defense, Deja and coworkers (87) measured levels of nitric oxide directly within maxillary sinuses of 11 patients with sepsis and radiologic maxillary sinusitis and 11 control patients free of airway and systemic inflammation. Patients with radiologic maxillary sinusitis had lower levels of nitric oxide in the maxillary sinus than did the control subjects: 31 versus 2,554 ppb. On immunohistochemical and in situ hybridization studies, expression of inducible nitric oxide synthase in patients with sinusitis was one-tenth of that in the control group. In the control group, the major site of nitric oxide production was the cilia and microvilli of the maxillary sinus epithelium. The authors conclude that patients with radiologic maxillary sinusitis and sepsis have markedly decreased production of nitric oxide within the sinuses resulting in impaired local host defense. An editorial commentary by Rouby (88) accompanies this article.

Destruction of the extracellular matrix is common in patients with hospital-acquired pneumonia. Hartog and coworkers (89) did mini-bronchoalveolar lavages in 30 patients with hospital-acquired pneumonia and 16 control subjects, and assessed activity of matrix metalloproteinases. Compared with the control subjects, the patients had 10-fold increases in metalloproteinase-8 and -9, whereas tissue inhibitor of metalloproteinase-1 was not increased. The active form of metalloproteinase-9 was found in 80% of the patients with pneumonia, but in none of the control subjects; neutrophils were the main source of the enzyme. Basal release of metalloproteinase was higher from pulmonary neutrophils than from blood neutrophils; pulmonary neutrophils were maximally activated. Concentrations of metalloproteinase were five times higher in patients with positive cultures than in patients with negative cultures. The levels of metalloproteinase were related to clinical severity. The authors conclude the neutrophil-derived metalloproteinases are increased in hospital-acquired pneumonia and are associated with positive cultures and clinical severity.

Massive Pulmonary Hemorrhage
Massive hemoptysis in patients with severe leptospirosis is resistant to conventional therapy and commonly fatal. In six patients with leptospirosis complicated by respiratory failure, shock, and multiple organ failure, Pea and colleagues (90) used desmopressin (1-deamino-8-D-arginine vasopressin [DDAVP]) to treat massive pulmonary hemorrhage. Desmopressin caused rapid cessation of hemoptysis in every case, and five patients eventually recovered. The authors conclude that desmopressin is a promising agent for treating massive pulmonary hemorrhage in patients with leptospirosis.

Chronic Obstructive Pulmonary Disease
To assess the role of neutrophilia, and neutrophil chemokine and receptor gene expression in severe exacerbations of chronic obstructive pulmonary disease (COPD), Qiu and coworkers (91) did endobronchial biopsies in 15 patients with COPD who were intubated for an acute exacerbation (FEV₁, 36% of predicted), 7 patients with stable COPD (FEV₁, 51% of predicted), and 15 control subjects intubated for a nonrespiratory surgical procedure (FEV₁, 98% of predicted). Compared with the patients with stable COPD, the patients with an acute exacerbation of COPD had a 97-fold increase in neutrophilia; gene expression for epithelial-derived neutrophil attractant-78 (CXCL5) was increased 6-fold, interleukin-8 (CXCL8) was increased 6-fold, CXCR1 was increased 3-fold, and CXCR2 was increased 7-fold. In patients with an exacerbation of COPD, the number of neutrophils was correlated with cells positive for CXCR2 messenger RNA (r = 0.79), but not with cells positive for CXCR1 messenger RNA. Neutrophil number was not correlated with the duration of intubation or viral infection. The authors conclude that a severe exacerbation of COPD causes increased neutrophil recruitment that is associated with upregulation of the neutrophil-selective cytokines, CXCL5 (epithelial-derived neutrophil attractant-78) and CXCL8 (interleukin-8), and with CXCR1 and CXCR2, the relevant receptors on which these neutrophil chemoattractant ligands act. An editorial commentary by Saetta and coworkers (92) accompanies this article.

Monitoring
General.
In a critical care perspective, Bates and Young (93) discuss the use of fuzzy logic for decision-making in the ICU.

Milic-Emili (94) recalls early studies on measurement of esophageal pressure.

Hemodynamic.
An increase in pulmonary alveolar pressure above pulmonary venous pressure is thought to interrupt the continuous fluid column between the pulmonary artery and the left atrium; if so, pulmonary artery occlusion pressure will not reflect left atrial pressure. To evaluate this line of thinking, Albert and Lamm (95) obtained measurements of left-atrial pressure when alveolar pressure exceeded both pulmonary arterial pressure and pulmonary venous pressure. Pulmonary arterial pressure and left-atrial pressure were set at 5 cm H₂O above alveolar pressure; alveolar pressure was varied from 0 to 25 cm H₂O; and left atrial pressure was decreased in decrements of 2 to 4 cm H₂O. At all alveolar pressures greater than 0 cm H₂O, pressure in the pulmonary artery accurately reflected left-atrial pressure, even when both were exceeded by alveolar pressure. The authors conclude that a pulmonary artery catheter can accurately estimate left-atrial pressure under Zone 1 conditions (when alveolar pressure exceeds both pulmonary artery pressure and left-atrial pressures).

Use of femoral-vein catheters is associated with a high risk of thrombosis despite prophylactic therapy. To assess the cost-effectiveness of routine lower extremity Doppler ultrasound screening in patients with femoral vein catheters, Cox and coworkers (96) developed a decision model. The analysis was based on a hypothetical cohort of 60-year-old medical patients being treated for acute respiratory failure. The costs of the ultrasound strategy were $8,688 per quality-adjusted life-year gained, $5,305 per pulmonary embolism averted, and $99,286 per death from pulmonary embolism averted. By varying in-hospital mortality, prevalence of deep-vein thrombosis, and ultrasound accuracy, the best-case cost was $1,170 and the worst-case costs was $35,342 per quality-adjusted life-year gained. Allowing for variation in variables of uncertain value, the median cost was $12,793 per quality-adjusted life-year gained. The authors conclude that routine ultrasound screening of patients with femoral vein catheters may improve outcomes at acceptable costs.

Nitric Oxide
To determine whether the individual response to inhaled nitric oxide changes over a prolonged course of therapy, Gerlach and coworkers (97) did a prospective, randomized study in 40 patients with ARDS. Before treatment with inhaled nitric oxide, a dose–response curve revealed maximal improvement in Pa_O2/FI_O2 ratio at 10 ppm. After 4 days of continuous treatment (10 ppm), the dose–response curve to nitric oxide was shifted to the left with a peak response at 1 ppm. At higher doses (10 and 100 ppm), oxygenation deteriorated and several patients displayed no response to nitric oxide. This effect was not seen in the untreated control group. The authors conclude that patients with ARDS display enhanced sensitivity to nitric oxide within a few days of treatment, placing patients at risk of overdosing and worsening oxygenation.

Inhalation of nitric oxide has been advocated as a method to prevent ischemia–reperfusion injury after lung transplantation. Meade and coworkers (98) did a randomized controlled trial of inhaled nitric oxide (22 ppm) versus placebo in 83 patients, initiated 10 minutes after reperfusion. PO₂/FI_O2 was equivalent in the nitric oxide and placebo groups, 361 versus 351, on admission to the ICU. Severe hypoxemia, PO₂/FI_O2 less than 150, taken as an index of severe reperfusion injury, was present in 14.6% of the nitric oxide group and 9.5% of the control group. The nitric oxide and placebo groups had equivalent times to first trial of spontaneous breathing (medians of 25 versus 27 hours), successful extubation (32 versus 34 hours), ICU discharge (3 days for both), and hospital discharge (27 versus 29 days). Five patients in the nitric oxide and six in the placebo group died in the hospital. The authors conclude that inhaled nitric oxide shortly after reperfusion does not alter clinical outcome in patients undergoing lung transplantation. An editorial commentary by Glanville (99) accompanies this article.

To measure fractions of exhaled nitric oxide in intubated, mechanically ventilated patients, Tornberg and coworkers (100) developed a method for obtaining multiple single-breath measurements at preset expiratory flows. A suction ejection system connected to a restriction valve was used to obtain multiple single-breath exhalations. Intubation produced a 50% decrease in the fraction of exhaled nitric oxide and a 36% decrease in the airway wall transfer rate of nitric oxide. Neither the fraction of nitric oxide originating in alveoli or in airway wall epithelium was affected by intubation. The peak concentration of nitric oxide after 20 seconds of apnea was similar to the value of nitric oxide calculated as originating in the airway wall epithelium. The authors conclude that the vacuum aspiration method for multiple single-breath measurements in mechanically ventilated patients enables the calculation of alveolar and bronchial fractions of nitric oxide.

In dogs with cardiac failure induced by procainamide, Natori and coworkers (101) studied the cardiovascular effects of inhaled nitric oxide. Inhaled nitric oxide did not alter baseline left-ventricular function in control dogs or in dogs with heart failure. Inhaled nitric oxide decreased pulmonary vascular resistance in dogs with heart failure. Inhaled nitric oxide caused unexpected increases in left-ventricular end-diastolic pressure in the presence of both vasodilating agents (acetylcholine and nitroglycerin) and vasoconstricting agents (norepinephrine and angiotensin-II). End-diastolic left-ventricular dimension and wall stress were increased by inhaled nitric oxide, whereas end-systolic variables were not affected. The authors conclude that inhaled nitric oxide reduces pulmonary vascular resistance in dogs with heart failure but it also increases left-ventricular preload and end-diastolic wall stress.

To determine whether endogenous endothelium-dependent vasoactive substances cause release of nitric oxide that can be detected in exhaled air, Malmstrom and coworkers (102) administered intravenously various pharmacologic compounds in anesthetized pigs and humans. Administration of acetylcholine, bradykinin, substance P, endothelin-1, and nitroglycerine produced dose-dependent increases of exhaled nitric oxide in pigs. Each compound caused a highly individual and reproducible release pattern. Angiotensin-converting enzyme inhibition enhanced the release of nitric oxide induced by bradykinin. Endothelin receptor antagonism reduced the response of exhaled nitric oxide to endothelin-1. Atropine abolished the response to acetylcholine. Inhibition of nitric oxide synthase abolished the basal levels of exhaled nitric oxide, and also decreased the exhaled nitric oxide response to all compounds except nitroglycerine. In six human subjects, acetylcholine evoked a dose-dependent increased in exhaled nitric oxide. The authors conclude that release of nitric oxide in response to endogenous vasoactive compounds can be measured online in exhaled air.

Ethical Issues
To evaluate the perceptions of caregivers involved in the withholding and withdrawing of life support in ICUs, Ferrand and coworkers (103) sent open-ended questionnaires to staff in 133 French ICUs. Completed responses were received from 3,156 nurses and 521 physicians (response rate, 42%). Decision-making processes were perceived as satisfactory by 73% of physicians and 33% of nurses. More than 73% of physicians believed that decision-making should be collaborative; yet only 27% of nurses and 50% of physicians believed that nurses were actually involved. Fear of litigation modified the information given by physicians to nurses, competent patients, and families. The authors conclude that nurses are commonly not involved and frequently dissatisfied with decisions about withholding and withdrawing of life support, and that fear of litigation curtails the information discussed. An editorial commentary by Morris (104) accompanies this article.

In an editorial, Tobin (105) discusses the role of a journal in a scientific controversy.

Nonpulmonary Critical Care
Pharmacotherapy.
To determine whether daily interruption of sedative infusions in mechanically ventilated patients induces long-term psychological abnormalities, Kress and coworkers (106) studied 13 patients who had experienced daily interruption of sedative therapy and 17 patients who had been awakened only at the discretion of the ICU team. At 6 months (or longer) after hospital discharge, patients subjected to daily interruption of sedatives had a lower total Impacts of Events score (a measure of signs of posttraumatic stress disorder) (11.2 versus 27.3), a trend toward a lower incidence of posttraumatic stress disorder (0% versus 32%) and a trend toward a better total Psychosocial Adjustment to Illness score (46.8 versus 54.3). The authors conclude that daily interruption of sedative infusion in mechanically ventilated patients decreases signs of posttraumatic stress disorder and tends to decrease the incidence of posttraumatic stress disorder.

Trauma.
Major trauma is associated with a decreased capacity of leukocytes to produce inflammatory cytokines. To identify pathways that remain functional in monocytes, Adib-Conquy and coworkers (107) studied 48 patients with major trauma. Hyporeactivity of leukocytes was restricted to gram-negative bacteria, E. coli endotoxin, and unmethylated bacterial DNA. In contrast, Leptospira interrogans endotoxin–induced production of tumor necrosis factor was equivalent to that in healthy subjects. Production of tumor necrosis factor and interleukin-6 in response to gram-positive bacteria was not altered. Expression of toll-like receptor 2 on the monocytes of patients was not reduced, whereas expression of toll-like receptor 4 was reduced. Downregulation of toll-like receptor 4 did not fully explain the hyporeactivity to gram-negative bacteria. Release of antiinflammatory cytokines in response to stimulation with microbial products was greater in the patients than in the control subjects. The increased production of interleukin-10 involved p38 mitogen-activated protein kinase, Sp-1 transcription factor, heterotrimeric Gi protein, and phosphatidylinositol-3'-kinase. The authors conclude that leukocyte deactivation in patients with trauma is not a generalized phenomenon but depends on the stimulus and the signaling pathway.

To better define the pathology of blast injury to the lung, Tsokos and coworkers (108) studied autopsy specimens from eight people who died from close-range detonation of chemical explosives (four homicides, three suicides, and one accident). Histologic, immunohistochemical, and scanning electron microscopy revealed diffuse alveolar overdistension, circumscribed interstitial hemorrhages showing a cuff-like pattern around pulmonary vessels, venous air embolism, bone marrow embolism, and pulmonary fat embolism. Pulmonary fat embolism in this setting was independent of ventilation and was sufficient to produce ARDS in survivors. The authors conclude that blast injury to the lung secondary to close-range chemical detonation produces a relatively uniform picture of alveolar over-distension, circumscribed hemorrhage, venous embolism, and pulmonary fat embolism.

Gastroenterological disorders.
In an update in nonpulmonary critical care, Marrero and colleagues (109) discuss advances in critical care hepatology.

Cardiac disorders.
Arterial hypertension, a common finding in patients with sleep apnea, is the main risk factor for aortic dissection. To investigate the relationship between aortic dissection and obstructive sleep apnea syndrome, Sampol and coworkers (110) studied 19 consecutive patients with thoracic aortic dissection and 19 matched patients with hypertension. An apnea–hypopnea index of more than 5 events per hour was observed in 13 patients (68%) from each group. The average apnea–hypopnea index was greater in the aortic dissection group than in the control group: 28 versus 11 events per hour. An apnea–hypopnea index of more than 30 events per hour was noted in 7 of the patients with aortic dissection versus 1 of the patients with hypertension. The authors conclude that patients with thoracic aortic dissection have a high prevalence of obstructive sleep apnea syndrome.

Hematological disorders.
Azoulay and coworkers (111) report 20 patients with respiratory impairment as the presenting manifestation of undiagnosed acute monocytic leukemia. All of the patients had myeloid leukemia of the AML5 subtype. The median leukocyte count was 98,250 per mm³ and all but one patient had circulating monocytic cells. Patients presented with a respiratory rate of 33 breaths per minute, PO₂ (room air) of 45 mm Hg, and variable changes on chest X-ray. Alveolar hemorrhage was the main finding on bronchoalveolar lavage. Respiratory function deteriorated in every patient with a few hours of starting chemotherapy. Ten of 15 patients requiring mechanical ventilation died. The authors conclude that acute pulmonary infiltration resulting in respiratory failure can be the presenting manifestation of acute monocytic leukemia, and that patients require intensive management in anticipation of deterioration after commencing chemotherapy.

More than 40% of patients who develop acute chest syndrome secondary to sickle cell disease have fat droplets in their alveolar macrophages, which are suggestive of pulmonary fat embolism. To determine the reliability of induced sputum for diagnosing fat embolism, Lechapt and coworkers (112) did two studies. In 20 patients with acute chest syndrome, the number of Oil Red O–stained macrophages in induced sputum was correlated with the number in bronchoalveolar lavage fluid (Spearman's coefficient, 0.66). In a second group of 60 patients with episodes of acute chest syndrome, sputum induction was successful in 47 patients. A diagnosis of pulmonary fat embolism (based on 5% of macrophages staining with Oil Red O) was made in 29 of the 47 patients with acute chest syndrome (61.7%) and none of 9 patients who did not have acute chest syndrome. Compared with patients who had acute chest syndrome but did not have pulmonary fat embolism, the patients with fat embolism were more likely to have additional extrathoracic pain (76 versus 50%), neurologic symptoms (7 versus 0%), abnormal transaminases (28 versus 17%), and a lower differential platelet count (-49 versus 85). The authors conclude that staining of induced sputum is a useful test for diagnosing pulmonary fat embolism in patients who have acute chest syndrome secondary to sickle cell disease.

Infectious disorders.
To describe the clinical spectrum and factors that determine mortality in patients with imported malaria, Bruneel and coworkers (113) studied 188 patients admitted to an ICU in Paris with severe and/or complicated imported malaria. The mean age was 38 years, 51% of the patients were nonimmune whites, 94% acquired Plasmodium falciparum in sub-Saharan Africa, and 96% had taken inadequate antimalarial chemoprophylaxis. Mortality was 10% in 93 patients with severe malaria and zero in 95 patients with less severe malaria. The main factors associated with mortality were Simplified Acute Physiology Score (SAPS), shock, acidosis, coma, pulmonary edema, and coagulation disorders. Bacterial coinfection was relatively common and a possible contributor to mortality. The authors conclude that severe imported malaria carries a mortality of 10% even when treated in a highly experienced ICU, and that the presence of any degree of neurologic, acid–base, circulatory, or pulmonary failure should prompt admission to the ICU. An editorial commentary by White (114) accompanies this article.

Neurologic and neuromuscular disorders.
In an update in nonpulmonary critical care, Deem and colleagues (115) discuss acquired neuromuscular disorders in critically ill patients.

	FOOTNOTES

 
Supported by a Merit Review grant from the Veterans Affairs Research Service.

Conflict of Interest Statement