Medical Ethics

Enhancing Communication Did Not Improve Care

A controlled trial to improve care for seriously ill hospitalized patients. The study to understand prognoses and preferences for outcomes and risks of treatment (SUPPORT). The SUPPORT Principal Investigators. JAMA. 1996; 274:1591-8.

Because of pressures from such groups as right-to-die advocates and from a society that increasingly demands something tangible in return for each dollar spent on medical care, physicians are being called on to deliver more accurate prognoses, especially in such high-cost settings as intensive care units. The SUPPORT group had previously identified barriers to optimal patient management and physician-patient communication. They then did an experiment to determine whether removing these barriers would improve patient outcome. The long-term goal was to improve end-of-life decision making and reduce the frequency of a mechanically supported, painful, and prolonged process of dying.

In phase I, shortcomings in communication, the frequency of aggressive treatment, and characteristics of hospital death were documented. Fewer than half of physicians knew when their patients preferred to avoid cardiopulmonary resuscitation; almost half of do-not-resuscitate orders were written within 2 days of death; almost half of patients who died had spent at least 10 days in an intensive care unit; and family members reported that their dying relative had appeared to have moderate to severe pain at least half the time.

The phase II experiment involved five U.S. teaching hospitals. In a 2-year controlled clinical trial, 4804 patients and their physicians were randomly assigned to an intervention group (n = 2652) or a control group (n = 2152). Patients had life-threatening illnesses and an overall 6-month expected mortality rate of 47%. Physicians in the intervention group received daily estimates of the following variables: likelihood of their patients' survival at 6 months, derived from a prognostic scoring system; patients' or surrogates' preferences about cardiopulmonary resuscitation; and functional disability at 2 months. Specially trained nurses met with patients, families, and physicians several times to facilitate communication, advance care planning, and pain control.

Despite the intense intervention, physician-patient communication did not improve. For example, preferences for cardiopulmonary resuscitation were discussed with 40% of patients in the intervention group and 37% of those in the control group. In addition, no differences were seen in the five targeted outcomes: 1) incidence or timing of do-not-resuscitate orders (adjusted ratio, 1.02 [95% CI, 0.90 to 1.15]); 2) physicians' knowledge of their patients' preference not to be resuscitated (adjusted ratio, 1.22 [CI, 0.99 to 1.49]); 3) number of days spent in an intensive care unit, receiving mechanical ventilation, or in a coma before death (adjusted ratio, 0.97 [CI, 0.87 to 1.07]); 4) level of reported pain (adjusted ratio, 1.15 [CI, 1.00 to 1.33]); and 5) use of hospital resources (adjusted ratio, 1.05 [CI, 0.99 to 1.12]).

The conclusion that such an intensive intervention was unsuccessful is discouraging. Several reasons may explain why the researchers found no improvement. The most obvious is that the methods may have been flawed: Nurses took the lead in promoting physician-patient communication but had no authority in direct clinical care, and physicians are not known for listening to nurses' advice. Second, most of these physicians were specialists who may have communicated differently, and perhaps less effectively, than primary care physicians would have with their "own" patients. Third, physicians in general simply do not deal well with death. Finally, physicians may not trust the prognoses provided by prognostic scoring systems.

Physicians Often Do Not Listen in Life-Threatening Situations

Asch DA, Hansen-Flaschen J, Lanken PN. Decisions to limit or continue life-sustaining treatment by critical care physicians in the United States: conflicts between physicians' practices and patients' wishes. Am J Respir Crit Care Med. 1995; 151(2 pt 1):288-92.

Asch and colleagues raise the question, Do physicians communicate poorly, or do physicians—especially those working in the intensive care unit—simply disregard what patients and surrogates say about their wishes? The legal community clearly gives patients almost complete autonomy to choose their own treatment, especially if they want to discontinue treatment. The beliefs of the medical community have been less clear.

This survey study of 879 physicians practicing in adult intensive care units sought to determine physician practices with regard to limiting life-sustaining medical treatment, particularly their decisions to continue or withhold life support without the consent or against the wishes of patients or surrogates.

In 1990, a 91-item survey was mailed to all 1970 members of the Critical Care Section of the American Thoracic Society. The response rate was 53% (n = 1050). After exclusion of physicians working in pediatrics, those not in practice, and those practicing outside the United States, 879 surveys were available for analysis.

The reasons for refusing requests for withdrawing mechanical ventilation are shown in Table 1. Almost all respondents (96%) said that they had withheld or withdrawn life-sustaining therapy, and most did so frequently. Thirty-four percent of physicians continued treatment despite patients' or surrogates' wishes that it be discontinued, and many physicians unilaterally withheld (83%) or withdrew (82%) treatment they considered futile.

Thus, it appears that physicians who care for patients in the intensive care unit do not reflexively accept requests by patients or physicians to continue or limit life-sustaining treatment. Instead, they seem to consider these requests along with other factors, such as prognosis. However, despite the controversies that surround the issue of futility, many physicians practicing in an intensive care unit incorporate some concept of futility into their bedside decision making. At both extremes, the wishes of patients and surrogates are often not honored.

Health Care Workers Varied Widely in Assigning Level of Care

Cook DJ, Guyatt GH, Jaeschke R, Reeve J, Spanier A, King D, et al. Determinants in Canadian health care workers of the decision to withdraw life support from the critically ill. Canadian Critical Care Trials Group. JAMA. 1995; 273:703-8.

Cook and colleagues surveyed a broad group of health care workers, and the results may give some clue as to why the SUPPORT study failed to show differences in practice. Cook and coworkers administered a survey to physicians, housestaff, and nurses who worked in Canadian intensive care units to examine attitudes toward the withdrawal of life support.

The investigators searched the literature to find potential determinants of the withdrawal of life support. They then conducted structured interviews with a small set of health care workers. From these interviews, they developed a case-based survey and carefully pretested it for reliability and clarity. The directors of the intensive care units at each of the 37 university-affiliated hospitals distributed the questionnaire; the overall response rate was about 76%. In all, 1361 health care workers responded. They were asked to rate the importance of 17 factors considered in the decision to withdraw life support and to rate five levels of care in terms of 12 difficult clinical scenarios.

The factors most important in the decision to withdraw life support were the patients' likelihood of surviving current episodes, presence of advance directives, cognitive function before illness developed, likelihood of long-term survival, and family directives. In choosing the level of care (one of five options, ranging from stopping all life support to aggressive intervention with mechanical ventilation and dialysis) for the patient scenarios, more than half of the respondents chose the same option in only 1 of 12 scenarios.

Health care workers perceive prognostic factors as being very important in the decision to withdraw support, but they differ widely in the level of care they would provide. In this study, the health care workers did not know of the wishes of patients and surrogates; thus, the relative importance of those wishes in decision making is unknown.

The Acute Respiratory Distress Syndrome

Many early studies of the acute respiratory distress syndrome (formerly called the "adult respiratory distress syndrome") (ARDS) focused on the mechanism of tissue injury in the lung. In the past year, however, most of the new findings about the syndrome have come from epidemiologic studies.

Sepsis Covers a Spectrum of Diseases

Rangel-Frausto MS, Pittet D, Costigan M, Hwang T, Davis CS, Wenzel RP. The natural history of the systemic inflammatory response syndrome (SIRS). A prospective study. JAMA. 1995; 273:117-23.

In discussions of the acute respiratory distress syndrome and other critical care syndromes, semantics can become confusing. Therefore, some definitions are in order. As originally described [1], the term "sepsis syndrome" initially identified an infected population of patients as being at particularly high risk for ARDS; thus, ARDS is highly associated with infection. At a recent consensus conference of chest physicians [2], new, more precise definitions were produced. One newly defined condition is the "systemic inflammatory response syndrome," which arises from a nonspecific insult and includes two or more of the following: 1) temperature greater than 38 °C or less than 36 °C; 2) heart rate greater than 90 beats/min; 3) respiratory rate greater than 20 breaths/min or PCO₂ less than 32 mm Hg; and 4) leukocyte count greater than 12.0 x 10⁹/L or less than 4.0 x 10⁹/L or the presence of more than 10% immature neutrophils.

"Sepsis" is defined as the presence of the systemic inflammatory response syndrome and a confirmed infectious process. The "sepsis syndrome" has been replaced by the term "severe sepsis," which is defined as sepsis with the presence of hypotension or signs of hypoperfusion, specifically lactic acidosis, oliguria, or altered mental status. "Septic shock" is severe sepsis that remains despite fluid resuscitation and includes signs of hypoperfusion.

Rangel-Frausto and colleagues sought to define the epidemiology of these four recently classified syndromes by describing the biological response to infection. The researchers did a cohort study of patients admitted to three intensive care units and three general wards in a tertiary teaching hospital. The enrollment period was 9 months, and patients were followed for 28 days or until hospital discharge. A total of 2527 patients with the systemic inflammatory response syndrome were recruited.

Among patients with this syndrome, 649 (26%) developed sepsis, 467 (18%) developed severe sepsis, and 110 (4%) developed septic shock. As disease severity progressed in the patients with the systemic inflammatory response syndrome, more of them developed ARDS, disseminated intravascular coagulation, acute renal failure, and refractory shock. Positive blood cultures correlated with disease severity and mortality. However, 58% of patients with sepsis had negative blood cultures.

There is a clear clinical progression from the systemic inflammatory response syndrome to sepsis to severe sepsis and septic shock. In addition, it appears that these syndromes often seem to occur in the absence of confirmed infection.

More Patients Survive the Acute Respiratory Distress Syndrome

Milberg JA, Davis DR, Steinberg KP, Hudson LD. Improved survival of patients with acute respiratory distress syndrome (ARDS): 1983-1993. JAMA. 1995; 273:306-9.

Since the recognition of ARDS in 1967, mortality rates have consistently been estimated to exceed 50%, despite developments in critical supportive care. However, no formal study had assessed true mortality rates over time. Milberg and colleagues analyzed temporal trends in rates of death related to the syndrome since 1983.

Using daily surveillance of intensive care units from 1983 to 1993, Milberg and associates identified 918 consecutive adult patients meeting the following criteria for ARDS: a ratio of PaO₂ to fraction of inspired oxygen (FIO₂) of 150 or less, diffuse parenchymal infiltrates seen on a chest radiograph, a pulmonary arterial wedge pressure of 18 mm Hg or less, and no other obvious explanations for these findings. The major causes of the syndrome were sepsis (37%) and trauma (25%). The main outcome measure was death.

Between 1983 and 1987, mortality rates ranged from 53% to 68%, with no discernible trend. Mortality rates declined slightly in 1988 and 1989 and then decreased to a low of 36% (CI, 25% to 46%) in 1993. The greatest decline was seen in young patients with sepsis.

It is difficult to generalize these findings to settings beyond this large university teaching hospital, but the rate of death from ARDS seems to be decreasing. Whether this decrease is the result of earlier diagnosis, earlier treatment, or better treatment is still unknown.

Death from Lung Injury Is More Likely with Nonpulmonary Disorders

Doyle RL, Szaflarski N, Modin GW, Wiener-Kronish JP, Matthay MA. Identification of patients with acute lung injury. Predictors of mortality. Am J Respir Crit Care Med. 1995; 152:1818-24.

An international consensus conference defined the acute respiratory distress syndrome as a syndrome in which the PaO₂:FIO₂ ratio is less than 200 and bilateral infiltrates are seen on radiography [3]. However, patients often have severe lung injury that does not meet the criteria for the syndrome. For these reasons, the international consensus conference created the term "acute lung injury," which encompasses both ARDS and other, milder forms of acute abnormalities. The diagnosis of acute lung injury requires that the PaO₂:FIO₂ ratio be less than 300 and that bilateral infiltrates be seen on a chest radiograph. Doyle and colleagues compared the outcomes in patients who had acute lung injury with outcomes in patients included in earlier studies of ARDS.

A cohort of 123 consecutive patients who met the criteria for acute lung injury was identified in the adult intensive care units of a university hospital in San Francisco. Severity of lung injury was determined by using a four-point scoring system based on severity of illness seen on chest radiograph, degree of hypoxemia, level of respiratory compliance during ventilation, and positive end-expiratory pressure during ventilation. Variables that affected mortality were identified by stepwise logistic regression.

The overall mortality rate among patients with acute lung injury was 58%. Sepsis was the disorder most frequently associated with lung injury (41% of patients). The degree of lung injury in the first 3 days did not predict survival. Death was associated with nonpulmonary organ-system dysfunction, chronic liver disease, and sepsis.

Nonpulmonary disorders are the most important determinants of outcome in patients with acute lung injury. There is little difference in clinical outcome between acute lung injury and ARDS. In addition, the mortality rate from both conditions in this study was higher than that in Rangel-Frausto and colleagues' study. This finding suggests that generalization of the results of these mortality studies may not be generalizable to other institutions.

Ventilator Management in the Acute Respiratory Distress Syndrome

It is well known that traditional methods of mechanical ventilation can produce such dramatic consequences as pneumothorax and hypotension. Recent findings showing that mechanical ventilation also causes more subtle damage to lung tissue have led to innovations in ventilation.

During mechanical ventilation, alveoli are often overdistended by high tidal volumes. Several studies have shown that conventional tidal volumes (10 to 5 mL/kg of body weight) may be detrimental to patients with ARDS. The American College of Chest Physicians has recommended that alveolar (plateau) pressures not be greater than 35 cm H₂O; in a patient with ARDS, a pressure this low can translate to a tidal volume of less than 5 mL/kg.

Standard Tidal Volumes May Damage Alveoli

Roupie E, Dambrosio M, Servillo G, Mentec H, el Atrous S, Beydon L, et al. Titration of tidal volume and induced hypercapnia in acute respiratory distress syndrome. Am J Respir Crit Care Med. 1995; 152:121-8.

In patients supported by mechanical ventilation, the pressure-volume curve has a sigmoidal shape. The upper inflection point represents the point of maximal distention of the alveoli, and the lower inflection point represents the pressure at which the alveoli collapse.

Roupie and colleagues traced the pressure-volume curves of patients receiving standard mechanical ventilation and compared the end-inspiratory plateau pressure with the upper inflection point. They then assessed the results of a ventilation protocol used to reduce the tidal volume until the plateau pressure was less than the upper inflection point.

In 25 patients with ARDS and 17 patients with acute respiratory failure who did not meet the criteria for ARDS, a pressure-volume curve was constructed by altering delivered volume between 50 mL and 1600 mL. Inflation was maintained at a pressure less than 50 cm H₂O. Each pressure-volume curve was based on an average of 16 points. When a lower inflection point was found on the curve, external positive end-expiratory pressure was adjusted so that the total was equal to the pressure at the lower inflection point (or a minimum of 5 cm H₂O).

At a tidal volume of 10 mL/kg and mean (±SD) positive end-expiratory pressure of 10 ± 3 cm H₂O, all patients with ARDS had an upper inflection point of 26 ± 6 cm H₂O (range, 18 to 40 cm H₂O). The end-inspiratory plateau pressure at these ventilator settings was 29 ± 6 cm H₂O in the patients with ARDS. In 80% of the patients, the plateau pressure exceeded the upper inflection point. In 90% of the patients with ARDS, the upper inflection point was less than 35 cm H₂O (the upper limit of plateau pressure recommended by the American College of Chest Physicians). An upper inflection point was seen in only 1 of the 17 patients who had acute lung injury that was not severe enough to meet the criteria for ARDS.

If mechanical ventilation at pressures exceeding the upper inflection point is shown to be a true harbinger of ventilator-induced lung injury, the previously recommended upper limit of 35 cm H₂O for plateau pressure should be decreased.

"Open-Lung" Ventilation Strategy Appeared To Be Superior

Amato MB, Barbas CS, Medeiros DM, Schettino G de P, Lorenzi Filho G, Kairalla RA, et al. Beneficial effects of the open lung approach with low distending pressures in acute respiratory distress syndrome. A prospective randomized study on mechanical ventilation. Am J Respir Crit Care Med. 1995; 152:1835-46.

The study by Amato and colleagues describes a new approach to mechanical ventilation—the so-called "open-lung" approach, which uses a smaller tidal volume but maintains end-expiratory pressures adequate for minimizing cyclic reopening of collapsed alveoli. Amato and colleagues determined the clinical utility of this theoretically superior technique.

Twenty-eight patients with early-stage ARDS were randomly assigned to receive either conventional ventilation or the open-lung approach. Conventional ventilation (13 patients) consisted of volume-cycled ventilation with a tidal volume of 12 mL/kg, a minimum positive end-expiratory pressure of 5 cm H₂O (with a goal of a PaO₂ > 60 mm Hg and the lowest FIO₂), and a PaCO₂ less than 38 mm Hg. In the new strategy (15 patients), tidal volume was kept to less than 6 mL/kg by using a pressure-limited mode, maintaining peak alveolar pressure at less than 40 cm H₂O, and maintaining a positive end-expiratory pressure at 2 cm H₂O above the lower inflection point (always > 16 cm H₂O).

Minute volume was lower in the open-lung group (<7 L/min) than in the conventional ventilation group (about 13 L/min) (P < 0.001) (Figure 1). The new technique was associated with a higher PaCO₂ (>50 mm Hg compared with <35 mm Hg) (P < 0.001) (Figure 2). End-inspiratory plateau pressure was about 30 cm H₂O in the group receiving the new strategy and 35 to 40 cm H₂O in the group receiving conventional ventilation. The PaO₂:FIO₂ ratio was 200 to 250 in the group that received the new strategy and about 125 in the group that received conventional ventilation (P < 0.001). The mortality rates were 33% in the new therapy group and 54% in the conventional group (P = 0.13).

View larger version (26K): [in this window] [in a new window]   Figure 1. Evolution of the minute volume required during the first week of mechanical ventilation. P values were derived from unpaired t-tests between both groups for each day. Bars represent 1 SE. Reproduced with permission from Amato MB, et al. Am J Respir Crit Care Med. 1995; 152:1835-46.

View larger version (27K): [in this window] [in a new window]   Figure 2. Arterial carbon dioxide levels obtained during the first week of mechanical ventilation. P values were derived from unpaired t-tests between both groups for each day. Bars represent 1 SE. Reproduced with permission from Amato MB, et al. Am J Respir Crit Care Med. 1995; 152:1835-46.

This new strategy of mechanical ventilation resulted in better oxygenation and a tendency toward decreased mortality in patients with ARDS. The increase in PaCO₂ (and concomitant decrease in pH) was not a problem in patients subjected to open-lung ventilation. In patients with a high risk for elevated intracranial pressure, however, this increase might not be desirable.

Since this report was published, more patients have been added to the 28 patients in this small study group; the mortality trends continue to be favorable. An international clinical trial now in progress should provide further estimates of the mortality rate associated with this strategy.

Weaning from Mechanical Ventilation

Although new mechanical ventilation strategies may prevent some lung injury, the best way to avoid damaging lung tissue is to discontinue use of mechanical ventilation as quickly as possible. Previous studies have shown that efforts to wean patients make up more than 40% of the duration of mechanical ventilation [4].

Once-Daily Breathing Trial Was Best for Weaning

Esteban A, Frutos F, Tobin MJ, Alia I, Solsona JF, Valverdu I, et al. A comparison of four methods of weaning patients from mechanical ventilation. Spanish Lung Failure Collaborative Group. N Engl J Med. 1995; 332:345-50.

For years, critical care physicians have debated the optimal method of weaning patients from ventilators. Esteban and colleagues sought to resolve the debate by comparing the results of four weaning techniques: intermittent mandatory ventilation, pressure support, intermittent trials of spontaneous breathing, and a once-daily trial of spontaneous breathing.

The study was a 14-center trial of 546 patients who had received mechanical ventilation for a mean (±SD) of 7.5 ± 6.1 days and were considered by their principal physicians to be ready for weaning. In 130 patients, respiratory distress occurred within 2 hours of spontaneous breathing; these patients were then randomly assigned to receive one of the four weaning techniques. Initial settings for intermittent mandatory ventilation and pressure support were standardized.

The median duration of weaning was 5 days for patients receiving intermittent mandatory ventilation, 4 days for those receiving pressure support, 3 days for those receiving several intermittent trials of spontaneous breathing, and 3 days for those receiving a once-daily trial of spontaneous breathing (Figure 3). After adjustment in a Cox proportional-hazards model, the rate of successful weaning was higher with a once-daily trial of spontaneous breathing than with intermittent mandatory ventilation (rate ratio, 2.83 [CI, 1.36 to 5.89]) or pressure support (rate ratio, 2.05 [CI, 1.04 to 4.04]). The rate of success did not greatly differ between patients receiving intermittent mandatory ventilation and those receiving pressure support or between patients receiving once-daily trials of spontaneous breathing and those receiving several trials.

View larger version (25K): [in this window] [in a new window]   Figure 3. Kaplan-Meier curves of the probability of successful weaning with intermittent mandatory ventilation, pressure-support ventilation, intermittent trials of spontaneous breathing, and a once-daily trial of spontaneous breathing. After adjustment for baseline characteristics in a Cox proportional-hazards model, the rate of successful weaning with a once-daily trial of spontaneous breathing was 2.83 times higher than that with intermittent mandatory ventilation (P < 0.006) and 2.05 times higher than that with pressure-support ventilation (P < 0.04). Reproduced with permission from Esteban A, et al. N Engl J Med. 1995;332:345-50.

The clinical conclusions here are twofold. First, use of mechanical ventilation was discontinued without any special weaning technique in 68% of patients. (A total of 372 of the 546 patients were extubated without difficulty, but 58 of those required reintubation within 48 hours.) For the 24% (130 of 546 patients) in whom routine weaning failed, a once-daily trial of spontaneous breathing led to extubation about three times more quickly than did intermittent mandatory ventilation and about twice as quickly as did pressure support.

Noninvasive Ventilation

Endotracheal intubation has been the standard of delivering mechanical ventilation for decades. However, the use of artificial airways is complicated by infection and tracheal injury [5]. In addition, intubation inhibits airway defenses, makes swallowing extremely difficult, and prevents speech. As a result, researchers have continued to explore noninvasive ventilation techniques.

Noninvasive Ventilation Reduced Incidence of Intubation

Kramer N, Meyer TJ, Meharg J, Cece RD, Hill NS. Randomized, prospective trial of noninvasive positive pressure ventilation in acute respiratory failure. Am J Respir Crit Care Med. 1995; 151:1799-806.

Observational studies have shown that the need for intubation among patients with acute respiratory failure may be minimized if patients are fitted with noninvasive airway devices and supported by mechanical ventilation until their own ventilatory power returns. Kramer and colleagues tested these preliminary findings to determine the effect of noninvasive positive-pressure ventilation on respiratory frequency, sense of dyspnea, the need for endotracheal intubation, and length of hospital stay.

Thirty-one patients who had acute respiratory failure but were otherwise stable were randomly assigned to receive standard therapy or standard therapy plus noninvasive positive-pressure ventilation. Exclusion criteria were respiratory arrest and need for immediate intubation, hypotension, arrhythmias, upper airway obstruction, inability to clear secretions, and inability to cooperate with the nasal mask. Noninvasive positive-pressure ventilation was delivered by using a nasal mask and ventilator with an initial pressure support setting of 8 cm H₂O.

The principal physician determined that endotracheal intubation was needed in 73% of the group that received standard treatment and in 31% of the group that received noninvasive positive-pressure ventilation (P < 0.05). In the noninvasive positive-pressure ventilation group, respiratory rates and heart rates were significantly lower within 1 hour of ventilation and arterial oxygen tension was significantly improved for the first 6 hours. At 6 hours, dyspnea was less severe in the noninvasive positive-pressure ventilation group. Duration of ventilator use, hospital stay, and mortality rates were similar in the two groups.

Noninvasive ventilation through a nasal mask reduced the need for intubation in patients with acute respiratory failure. However, the clinical outcomes in this small sample were unchanged. At present, the evidence remains insufficient to warrant the routine use of noninvasive positive-pressure ventilation.

Patients with Chronic Obstructive Pulmonary Disease Responded Well to Noninvasive Positive-Pressure Ventilation

Brochard L, Mancebo J, Wysoicki M, Lofaso F, Conti G, Rauss A, et al. Noninvasive ventilation for acute exacerbations of chronic obstructive pulmonary disease. N Engl J Med. 1995; 333:817-22.

In the study by Kramer and colleagues, patients with chronic obstructive pulmonary disease responded well to noninvasive positive-pressure ventilation; however, only 23 such patients were included in the trial. Brochard and colleagues investigated the efficacy of noninvasive positive-pressure ventilation in a larger sample of patients with an acute exacerbation of chronic obstructive pulmonary disease.

Study patients were selected from a group of 275 patients admitted to intensive care units with an acute exacerbation of chronic obstructive pulmonary disease. Eighty-five (31%) met the inclusion criteria (Table 2). Exclusion criteria primarily focused on the use of sedating drugs and the presence of severe organ-system compromise. Forty-two patients received standard therapy (which included oxygen, bronchodilators, and antibiotics), and 43 patients received standard therapy plus noninvasive positive-pressure ventilation. The latter consisted of pressure support, initially adjusted to 20 cm H₂O, that was delivered through a face mask for at least 6 hours a day.

The need for endotracheal intubation (dictated by predetermined criteria) was reduced from 74% in the standard therapy group to 26% in the noninvasive positive-pressure ventilation group (P < 0.001). The frequency of complications was reduced from 48% in the standard therapy group to 16% in the experimental group (P = 0.001). Hospital stay was reduced from a mean of 35 ± 33 days in the standard therapy group to 23 ± 17 days in the experimental group (P = 0.005). The in-hospital mortality rate was 29% in the standard therapy group and 9% in the experimental group (P = 0.02).

In carefully selected patients with an acute exacerbation of chronic obstructive pulmonary disease, noninvasive ventilation apparently can decrease the need for intubation, the frequency of complications, the duration of hospital stay, and the in-hospital mortality rate. However, this was a study of efficacy, not effectiveness, and only about a third of patients met the inclusion criteria (Table 2). Most patients with chronic obstructive pulmonary disease would not meet these criteria. Still, the evidence is mounting that noninvasive positive-pressure ventilation treatment may be a reasonable alternative to immediate intubation for some patients with chronic obstructive pulmonary disease and acute respiratory failure.

Dr. Luce: Division of Pulmonary and Critical Care Medicine, San Francisco General Hospital, 1001 Potrero Avenue, Room 5K1, San Francisco, CA 94110.

Dr. Roberts (Series Editor): York Health System Medical Group, York, PA 17403.