|
A correction has been published: N Engl J Med 1997;337(10):710.
|
| |||||||||||||||||||||||||||||||||||||||||||||||
Background In patients with sepsis the production of arachidonic acid metabolites by cyclooxygenase increases, but the pathophysiologic role of these prostaglandins is unclear. In animal models, inhibition of cyclooxygenase by treatment with ibuprofen before the onset of sepsis reduces physiologic abnormalities and improves survival. In pilot studies of patients with sepsis, treatment with ibuprofen led to improvements in gas exchange and airway mechanics.
Methods From October 1989 to March 1995, we conducted a randomized, double-blind, placebo-controlled trial of intravenous ibuprofen (10 mg per kilogram of body weight [maximal dose, 800 mg], given every six hours for eight doses) in 455 patients who had sepsis, defined as fever, tachycardia, tachypnea, and acute failure of at least one organ system.
Results In the ibuprofen group, but not the placebo group, there were significant declines in urinary levels of prostacyclin and thromboxane, temperature, heart rate, oxygen consumption, and lactic acidosis. With ibuprofen therapy there was no increased incidence of renal dysfunction, gastrointestinal bleeding, or other adverse events. However, treatment with ibuprofen did not reduce the incidence or duration of shock or the acute respiratory distress syndrome and did not significantly improve the rate of survival at 30 days (mortality, 37 percent with ibuprofen vs. 40 percent with placebo).
Conclusions In patients with sepsis, treatment with ibuprofen reduces levels of prostacyclin and thromboxane and decreases fever, tachycardia, oxygen consumption, and lactic acidosis, but it does not prevent the development of shock or the acute respiratory distress syndrome and does not improve survival.
Ibuprofen has been shown to have effects on sepsis in humans, but because of their small samples (fewer than 30 patients), previous studies have been inadequate to assess effects on mortality.26,27,28 We sought to determine whether ibuprofen can alter rates of organ failure and mortality in patients with the sepsis syndrome, how the drug affects the increased metabolic demand in sepsis (e.g., fever, tachypnea, tachycardia, hypoxemia, and lactic acidosis), and what potential adverse effects the drug has in the sepsis syndrome.
Methods
Study Patients
Seven medical centers in the United States and Canada participated in this trial, which was approved by the institutional review board at each center. Consent was obtained from all the patients or their next of kin before enrollment. Patients were recruited from intensive care units if they had a known or suspected site of serious infection, as determined on the basis of clinical data available at the time of screening, and if they met certain criteria that represented a modification of the criteria for sepsis described by Bone et al.29 and that were similar to those defined at a consensus conference.2 There were two groups of criteria, one (group 1) involving conditions present when the patient was at rest, and the second (group 2) involving dynamic variables unrelated to coexisting disease. To be eligible for the study, a patient had to meet all the group 1 criteria, as follows: a core temperature of at least 38.3°C or less than 35.5°C, a heart rate of at least 90 beats per minute (in the absence of treatment with beta-blockers), and a respiratory rate of 20 breaths per minute or more or, if the patient was receiving mechanical ventilation, a ventilatory rate greater than 10 liters per minute. In addition, the patient had to meet at least one criterion in group 2, as follows: cardiovascular dysfunction, defined as a systolic blood pressure less than 90 mm Hg or a decrease in systolic pressure by at least 40 mm Hg for more than one hour while pulmonary-artery wedge pressures remained adequate (>12 mm Hg) or at least 500 ml of saline was infused; renal-system dysfunction, defined as a urinary output of less than 30 ml per hour or less than 0.5 ml per kilogram of body weight, for at least one hour; dysfunction related to the acute respiratory distress syndrome (ARDS), as defined by a ratio of the partial pressure of arterial oxygen (in millimeters of mercury) to the fraction of inspired oxygen (PaO2/FiO2) of less than 200 in the presence of acute, diffuse bilateral infiltrates; pulmonary-system dysfunction, defined as a PaO2 of less than 70 mm Hg while the patient was breathing room air or a PaO2/FiO2 of less than 333 if the patient was receiving supplemental oxygen (if the patient had pneumonia or chronic lung disease as the solitary cause of sepsis, he or she was also required to meet the definition of ARDS); or central nervous system dysfunction, defined as a decrease in the Glasgow coma score by at least two points.
A maximum of 24 hours was allowed for the patient to meet the entry criteria and for the study drug to be administered. Patients were excluded if they were pregnant, were less than 18 years old, had suspected hypersensitivity to cyclooxygenase inhibitors, had received a cyclooxygenase inhibitor within the preceding 12 hours (or aspirin within the preceding 24 hours), or were enrolled in another experimental study or if consent could not be obtained. We also excluded patients who had suspected brain death, advanced renal failure (serum creatinine, >4 mg per deciliter [354 µmol per liter] unless the patient's condition was stable and the patient was receiving long-term dialysis) or hepatic failure (serum bilirubin, >4 mg per deciliter [68 µmol per liter]), a core temperature of less than 32.2°C, a platelet count of less than 20,000 per cubic millimeter, a granulocyte count of less than 1000 per cubic millimeter due to causes other than sepsis, human immunodeficiency virus infection, gastrointestinal bleeding, or a life expectancy of less than six hours; those who were treated with major immunosuppressive drugs; and those whose physicians were not committed to providing aggressive life support.
Treatment Assignments
Patients were assigned to the study groups in a blinded fashion in each treatment center with the use of a permuted-block randomization algorithm. They received either ibuprofen (Motrin, Upjohn), given intravenously in a dose of 10 mg per kilogram (maximal dose, 800 mg) over a period of 30 to 60 minutes every 6 hours for eight doses or placebo (glycine-buffer vehicle) administered in the same volume and at the same times. Both the patient and the care givers were unaware of the patient's treatment assignment.
Laboratory Data
Before the first dose of the study drug, a base-line medical history was taken and a physical examination was performed to document the presumed cause, site, and time of onset of the sepsis syndrome. Blood was obtained for culture from at least two sites. Infection was classified as occurring in the lungs, peritoneum, or urinary tract or at another site or an unknown site. Chest radiographs were obtained at entry and scored by the chest radiologist to indicate the presence and severity of pulmonary edema.26
The method of calculating the Acute Physiology and Chronic Health Evaluation (APACHE II) score was modified so that the score represented a point in time — that is, it was calculated from the base-line data rather than from the worst values obtained during the first 24 hours of care in the intensive care unit. Data obtained at entry and every 4 hours thereafter for the first 44 hours and then at 72, 96, and 120 hours included the patient's temperature, mean systemic blood pressure, respiratory rate, heart rate, urinary output (as an hourly average), arterial-blood gas measurements, and requirement for antipyretic agents. Values for mean blood pressure were calculated with the following formula: (0.33 x systolic pressure) + (0.67 x diastolic pressure). When a pulmonary-artery catheter was present, the cardiac output, pulmonary-artery pressure, pulmonary-wedge pressure, and central venous pressure were measured at base line and 20 hours later. Blood lactate was measured, and the delivery and consumption of oxygen calculated, at base line and 20 hours.30
Blood samples were obtained at base line and 20, 44, 72, and 120 hours after study entry for the measurement of hemoglobin, the total leukocyte count, the platelet count, bilirubin, serum aspartate aminotransferase, lactate dehydrogenase, creatinine, blood urea nitrogen, and electrolytes. Data were recorded on the patient's requirements for blood transfusion, intensive care, and mechanical ventilation.
Measurements of Ibuprofen and Eicosanoids
Samples of urine and blood were obtained at base line and 20, 44, 72, 96, and 120 hours after study entry and were maintained at -70°C. Batch assays using stable isotope-dilution methods in conjunction with gas chromatography–mass spectrometry to detect urinary 11-dehydro-thromboxane B2 and 2,3-dinor 6-keto-prostaglandin-F1
were performed by the Vanderbilt Prostaglandin Core Laboratory.31,32 Ibuprofen levels were measured in plasma samples obtained at base line and 20 hours after study entry.33
Definitions
We used the entry criteria listed in group 2 above to define shock, ARDS, pulmonary-system dysfunction, and renal dysfunction. Shock was considered to have been reversed when the systolic blood pressure was stable and greater than 90 mm Hg for 12 hours without vasopressor support. Reversal of ARDS was considered to have occurred when the patient no longer met the criteria for ARDS on the basis of either arterial-blood gas measurement or chest radiography, and when the reversal lasted at least 24 hours.
We calculated the number of failure-free days (days without organ-system dysfunction) for the clinically important organ systems, as described above, during the 30 days immediately after study entry. For example, a patient who survived 30 days and had no renal failure (as determined on the basis of urinary output) was assigned a score of 30. If there was renal failure for 10 days and the patient died on day 15, a score of 5 was assigned. To summarize these data, we calculated the total number of failure-free days for all organs as the mean number of days the patients were free of all organ failure.
Statistical Analysis
In this study, we sought to enroll a total of 525 patients who could be evaluated. This figure was based on an expected treatment-related relative reduction in mortality of 35 percent (from 30 percent to 19.5 percent) and a power of 0.80. An independent Data and Safety Monitoring Committee examined the data on each group of 105 patients. Mortality by day 30 was the variable used in considering early termination of the trial (with the O'Brien–Fleming method used to determine sequentially adjusted P values34). After the fourth interim analysis (of 415 patients), the Data and Safety Monitoring Committee was provided with conditional estimates of the power of the study to determine potential differences in mortality between the study groups. A futility index35,36 was derived in which mortality by day 30 was treated as a dichotomous outcome variable. Using this information, we stopped the trial after the enrollment of 455 patients.
Short-time-series data were analyzed by deriving the area under the curve of the response variable for each patient, dividing this area by the time the patient spent in the study, and then using a Wilcoxon rank-sum test to compare these univariate statistics.37,38 A variety of other approaches were examined, and they yielded similar results.39,40 Two-by-two contingency tables were analyzed with a chi-square statistic with Yates' correction for continuity.38 Kaplan–Meier survival curves were compared by the log-rank test.38 Ninety-five percent confidence intervals for proportions were calculated by the method of Fleiss.41 P values of less than 0.05 were considered to indicate statistical significance.
Results
Comparison of Study Groups
The seven centers participating in the trial recruited 455 patients from a total of 3311 patients who fulfilled the study criteria between October 1989 and March 1995. Among the 2862 patients excluded from the study for whom data on 30-day outcomes were available, the mortality rate was 41 percent. The major reasons for exclusion were the need for more than 24 hours to meet the entry criteria and receive the study drug (24 percent of patients), gastrointestinal bleeding (16 percent), inability to obtain consent (12 percent), use of a cyclooxygenase-blocking drug (9 percent), and thrombocytopenia or neutropenia (9 percent). All 455 recruited patients were included in the analyses presented here.
Two hundred twenty-four patients were assigned to the ibuprofen group, and 231 patients were assigned to the placebo group. Table 1 compares the base-line characteristics of the two groups and shows that the groups were balanced with respect to a variety of measures that correlate with mortality and morbidity. The mean (±SD) interval that elapsed from the time the patient met the entry criteria to the administration of the study drug was 10.7±0.6 hours in the ibuprofen group and 11.3±0.6 hours in the placebo group (P not significant). The predominant site of infection was the lung. In each group, infections associated with positive blood cultures were considered to have been treated with appropriate antibiotics in 96 percent of cases (Table 1). Rates of organ dysfunction (organ failure) were similar in the two groups at the time of randomization, except that renal dysfunction was significantly more common in the placebo group.
|
At enrollment, the patients in both groups excreted substantial amounts of prostacyclin and thromboxane metabolites — approximately 40 and 15 times the normal amounts, respectively. After 44 hours of ibuprofen therapy, the rate of excretion of the prostacyclin metabolite was 11 percent of that in the placebo group and 4 percent of the base-line value; similar reductions were observed in the ibuprofen group with regard to excretion of the thromboxane metabolite (P<0.05). At 20 hours (2 hours after the fourth dose), ibuprofen levels were 24.4±15.0 mg per deciliter.
Physiologic Responses
The use of acetaminophen to reduce temperature was not dealt with in the study protocol. At the time of randomization, 33 percent of the ibuprofen group and 29 percent of the placebo group were receiving acetaminophen for this purpose. As temperatures decreased in the ibuprofen-treated patients, their use of acetaminophen began to diminish within eight hours. This did not occur in the placebo group. Within the first 24 hours after study-drug administration, the rate of use of acetaminophen decreased to 22 percent in the ibuprofen group, but it increased to 44 percent in the placebo group (P<0.001). The decline in the heart rate during the first eight hours of treatment was significantly greater in the ibuprofen group than in the placebo group (P<0.001). In patients receiving mechanical ventilation, ibuprofen tended to decrease minute ventilation by approximately 2 liters per minute, whereas placebo did not (P = 0.16) (Figure 1). At base line, the mean blood pressure was 80±1 mm Hg in the ibuprofen group and 79±1 mm Hg in the placebo group; at 2 hours, it was 75±1 and 79±1 mm Hg, respectively; and at 24 hours, it was 80±1 and 82±1 mm Hg (P not significant for any of these comparisons).
|
Shock was present at base line in 65 percent of the ibuprofen group and 63 percent of the placebo group. There were no treatment-related differences in the duration of shock. ARDS was present at base line in 29 percent of the ibuprofen group and 28 percent of the placebo group. There was a trend in the ibuprofen group toward more days free of pulmonary dysfunction and ARDS, as well as toward more days free of organ-system failure overall (P not significant) (Figure 2).
|
|
|
Serum creatinine and urinary output were measured serially over a five-day period to evaluate the effects of the study treatment on renal function, and no significant differences were detected between the groups. Hemodialysis was required for the first time in 6 of 224 ibuprofen-treated patients (3 percent) and 13 of 231 placebo-treated patients (6 percent, P = 0.11). Serial hemoglobin levels did not differ significantly between the two groups, nor were there any marked differences in requirements for transfusion. Gastrointestinal bleeding was reported in 9 patients in the ibuprofen group and 16 patients in the placebo group. During the 30 days of follow-up, distinct second episodes of sepsis occurred in 8.2 percent of the ibuprofen group and 11.1 percent of the placebo group.
Discussion
Ibuprofen was chosen for this trial because studies in both animals and humans in which aspirin,4,5,6 indomethacin, and ibuprofen were used as inhibitors of cyclooxygenase showed improvement in either the morbidity or the mortality associated with sepsis and because clinical experience with ibuprofen suggested that the drug was safe.7,8,9,10,11,12,13,14,15,16,17,18,19,22,26,27,28,42,43 In this study, we demonstrated that ibuprofen reduces fever, tachycardia, and lactic acid levels in patients with sepsis. However, we found no significant effect on mortality or organ failure. Evaluation of renal function and hemorrhagic events failed to demonstrate adverse effects when ibuprofen was administered over a 48-hour period early in the course of sepsis.
In animals in which systemic vasodilatation is prominent, such as primates and dogs with sepsis, ibuprofen increases blood pressure and survival.17,18,19,20,21 In a previous clinical study, we found that ibuprofen was associated with a trend toward increased systemic blood pressure, reversal of shock, and normalization of pH.26 In the current trial, we found reductions in lactic acid and oxygen consumption but no changes in oxygen delivery or blood pressure. These changes may be a result of decreased metabolic demand, as evidenced by the lower temperatures and heart rates in the ibuprofen-treated patients, or by improved matching of the delivery of oxygen to its consumption at the tissue level.
The practice of fever control in critically ill patients is controversial. In this study, 44 percent of patients with sepsis received acetaminophen for this purpose, suggesting that many attending physicians desired to limit fever. Even so, temperatures remained significantly elevated in the placebo group, but they returned to normal in the ibuprofen group. These data suggest that acetaminophen was less effective than ibuprofen in reducing temperature in this population of patients.
The analysis of days free of organ failure was performed to obtain a more sensitive measure of effects than that provided by the analysis of mortality.44 This analysis was especially important because of the confounding effect of early mortality on measures such as the duration of organ failure. No consistent effects of ibuprofen treatment were noted with regard to the number of days free of organ failure, the efficiency of oxygenation, findings on chest radiography, or the requirement for mechanical ventilation. This suggests that the effects of ibuprofen on these measures, if any, are fairly small.44
High circulating levels of the eicosanoid metabolites of prostacyclin and thromboxane have been associated with increased mortality in patients with sepsis.23,24 We found urinary concentrations of prostacyclin metabolites to be approximately 40 times higher than normal, and those of thromboxane metabolites to be approximately 15 times higher than normal. Levels of these metabolites were markedly elevated for at least 48 hours in the placebo group. These findings confirm our earlier work showing marked elevations of eicosanoids in patients with sepsis, with prompt reductions after ibuprofen treatment.
The absence of improved mortality with ibuprofen could have explanations other than the simple lack of efficacy. Perhaps racial differences or physiologic conditions such as hypothermia portend a different response to the drug, as the data in Table 2 show. Selecting the proper dose and duration of ibuprofen treatment is always a concern. Our findings show evidence of marked inhibition of cyclooxygenase, which suggests that the dose of ibuprofen was adequate. However, it is possible that there are effects other than the effects on cyclooxygenase that were not manifested at the dose selected. The treatment period of 48 hours used in this study was based on the information on safety available at the start of the trial regarding the intravenous formulation of ibuprofen. The data on cyclooxygenase metabolites presented here and in our pilot study indicate that the activity of arachidonic acid metabolites persists for at least several days after the onset of sepsis, as do the corresponding changes in vital signs associated with these mediators. It is thus possible that longer-lasting therapy would have produced different results. Finally, although the patients received the study drug quite early in this study, it is possible that treatment given even earlier, or as prophylaxis, may be necessary to produce effects on mortality and organ failure.27,45
Ibuprofen can have adverse effects, especially on the renal21,46 and gastrointestinal47 systems. Measuring renal function in terms of days of renal failure, serial creatinine levels, serial determinations of urinary output, or the need for dialysis did not reveal adverse effects, nor did the serial measurement of transfusion requirements, hemoglobin levels, or the incidence of episodes of serious gastrointestinal bleeding show differences between the study groups. The slight decrease in blood pressure (by approximately 4 mm Hg) in the first hours after the administration of ibuprofen was not statistically significant, but it was intriguing nonetheless. Further analysis indicated that this change could be attributed to the influence of patients who entered the trial with hypertension. We speculate that such patients were experiencing pain or other distress, and that ibuprofen, in relieving these underlying causes of hypertension, may have allowed the blood pressure to return toward normal. It did not lower the blood pressure of normotensive or hypotensive patients.
In summary, from this double-blind, randomized, placebo-controlled study of intravenous ibuprofen in patients with sepsis we conclude that treatment with ibuprofen is safe in such patients and markedly reduces the synthesis of prostacyclin and thromboxane, but that it has no effect on survival or the development of shock or ARDS. Treatment with ibuprofen does have clear physiologic effects on fever, tachycardia, oxygen consumption, and lactic acidosis in patients with sepsis.
Supported in part by grants (HL 43167, HL 19153, HL 07123) from the National Heart, Lung, and Blood Institute and by the Bernard Werthan, Sr., Fund for Pulmonary Research.
We are indebted to the Upjohn Company for providing intravenous ibuprofen, to the attending physicians for referring patients, to the bedside critical care nurses for their help and support, and to the patients and their families for participating in this scientific investigation.
Source Information
From the Divisions of Pulmonary and Critical Care Medicine (G.R.B., A.P.W., B.W.C., B.B.S.), Biostatistics (W.D.D.), and Biomedical Engineering and Computing (S.B.H.), Vanderbilt University Medical Center, Nashville; the Program of Critical Care Medicine, Department of Medicine, St. Paul's Hospital and University of British Columbia, Vancouver, B.C., Canada (J.A.R.); the University of Miami School of Medicine and Department of Veterans Affairs Medical Center, Miami (R.S.); the Department of Pulmonary and Critical Care Medicine, Louisiana State University Medical Center, New Orleans (W.R.S.); the Division of Pulmonary and Critical Care Medicine, Harborview Medical Center, University of Washington, Seattle (K.P.S.); the Division of Pulmonary and Critical Care Medicine, Duke University Medical Center, Durham, N.C. (W.J.F.); and the Division of Pulmonary Medicine, Methodist Hospital, Indianapolis (P.E.W.).
Address reprint requests to Dr. Bernard at Rm. T-1217, Medical Center North, Vanderbilt University School of Medicine, Nashville, TN 37232.
References
to the hemodynamic changes during baboon endotoxic shock. Adv Prostaglandin Thromboxane Leukot Res 1980;7:843-849.
in patients in septic shock. Crit Care Med 1985;13:451-453. [Medline]
The following additional institutions and investigators participated in the Ibuprofen in Sepsis Study Group: St. Paul's Hospital, Vancouver, B.C., Canada — A. Drummond; the University of Miami School of Medicine and the Department of Veterans Affairs Medical Center, Miami — M. Pena; Louisiana State University Medical Center, New Orleans — B. deBoisblanc, B. Everett, C. Glenn, and D. Tebbe; Vanderbilt University Medical Center, Nashville — L.C. Carmichael, M.J. Stroud, K. Jiang, W.D. Plummer, Jr., and F.E. Carroll; Harborview Medical Center and the University of Washington, Seattle — L.D. Hudson and D. Anardi; Duke University Medical Center, Durham, N.C. — M.J. Abernathy, L. Mallatratt, and P. Weston; Methodist Hospital, Indianapolis — K. Colvin.
Project Officers and Data and Safety Monitoring Committee: University of California, San Diego — R. Spragg; Harvard University School of Medicine, Boston — R. Demling; Cleveland Clinic Foundation, Cleveland — G. Beck; University of Minnesota, Minneapolis — F. Cerra; Division of Lung Diseases, National Heart, Lung, and Blood Institute, Bethesda, Md. — L. Jensen and C. Bosken.
Study Coordinating Center: the Center for Lung Research, Departments of Medicine, Preventive Medicine, and Biomedical Engineering, Vanderbilt University School of Medicine, Nashville.
| |||||||||||||||||||||||||||||||||||||||||||||||
Related Letters:
Ibuprofen in Patients with Sepsis
Zimmerli W., Widmer A. F., Bernard G. R., Wheeler A. P., Christman B.
Extract |
Full Text
N Engl J Med 1997;
337:710, Sep 4, 1997.
Correspondence
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | TERMS OF USE | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2009 Massachusetts Medical Society. All rights reserved. |
Previous
