DOES INCREASING OXYGEN DELIVERY IMPROVE OUTCOME IN THE CRITICALLY ILL? NO

CONTROVERSIES IN CRITICAL CARE MEDICINE

0749-0704/96 $0.00

+ .20

DOES INCREASING OXYGEN DELIVERY IMPROVE OUTCOME IN THE CRITICALLY ILL? NO Juan J. Ronco, MD, FRCP(C), John C. Fenwick, MD, FRCP(C), and Martin G. Tweeddale, MB, FRCP(C)

The strategy of increasing oxygen (0,)delivery and consumption in critically ill patients to supranormal values previously observed among survivors of critical illness is based on three tenets: First, that critically ill patients commonly die of multiple-system organ failure and that tissue hypoxia may be partially responsible for its development; second, that such tissue hypoxia may persist in critically ill patients despite aggressive early resuscitation to endpoints of tissue perfusion traditionally considered adequate; and third, that increasing 0, delivery can reverse such tissue hypoxia. There is no doubt that prompt resuscitation with rapid reversal of shock and subsequent maintenance of adequate tissue oxygenation is a major determinant of subsequent normal organ function and survival in critical illness. We believe, however, that the rationale for increasing 0, delivery and consumption to supranormal values-beyond traditional endpoints of adequate tissue perfusion-is based on an incorrect understanding of the relationship between whole-body 0, delivery and consumption. Our belief is supported by recent studies examining the relationship between whole-body 0, delivery and consumption and by the conflicting results of randomized controlled trials that have increased 0, delivery to supranormal values. We therefore address the question of whether increasing 0, delivery to supranormal values improves outcome in the critically ill, first by

From the Division of Critical Care Medicine, Department of Medicine, Vancouver Hospital and Health Sciences Centre, University of British Columbia, Vancouver, Canada

CRITICAL CARE CLINICS VOLUME 12 NUMBER 3 * JULY 1996

645

646

RONCO et a1

reviewing the relationship between whole-body 0, delivery and consumption and then by critically reviewing the randomized controlled trials that have increased 0, delivery to supranormal values. OXYGEN DELIVERY AND CONSUMPTION IN THE CRITICALLY ILL The metabolism of 0, can be characterized by two major parameters-0, delivery and 0, consumption-that normally are tightly coupled to maintain cellular energy balance.

0, delivery (mL/kg/minute) = cardiac output X arterial 0, content 0, consumption (mL/kg/minute) = cardiac output x (arterial 0, content - venous 0, content) 0, extraction ratio = [(arterial 0, content - venous 0, content)/ arterial 0, content] x 100 In healthy anesthetized dogs, at normal or supranormal values of 0, delivery, 0, consumption remains relatively constant and independent of 0, delivery (Fig. 1).As 0, delivery gradually is reduced to very low values, a proportionate increase in tissue 0, extraction from the blood maintains 0, consumption relatively constant until a critical value of 0, delivery is reached. Below that critical value, 0, extraction no longer can increase in direct proportion to further reductions in 0, delivery. As a result, 0, consumption decreases, resulting in tissue hypoxia, with its consequent anaerobic metabolism and lactic acid production.6, This phase of the 0, delivery/consumption relationship, in which there is a direct dependence of 0, consumption on 0, delivery below the critical 0, delivery value, has been termed "supply-dependent 0, consumption." In studies of healthy dogs, 0, consumption became supply dependent below a critical 0, delivery value of 8 mL/kg/minute. At that level, the corresponding 0, extraction ratio (0,ER) was 70%.29Healthy anesthetized dogs therefore display a biphasic relationship between 0, delivery and consumption, with a phase of supply-independent 0, consumption above and a phase of supply-dependent 0, consumption below the critical 0, delivery value. In contrast to these laboratory experiments in healthy animals, which have shown supply-dependent 0, consumption only at very low values of 0, delivery, clinical studies of patients who have acute respiratory distress syndrome (ARDS) or sepsis have reported supplydependent 0, consumption at normal or even supranormal values of 0, delivery, suggesting that the critical 0, delivery value is increased in such patients (Fig. 2).,, 8* 44 It was postulated,, 8, 12, 44 that such patients are unable to increase tissue 0, extraction from the blood in response to an acute decrease in 0, delivery at a time when 0, delivery is still normal or even supranormal. As a result, 0, consumption appeared to become supply dependent at an 0,ER of 50% or less, considerably below

DOES INCREASING 0, DELIVERY IMPROVE OUTCOME IN THE CRITICALLY ILL? NO

I

647

Critical O2 Delivery

Oxygen Delivery Figure 1. The biphasic relationship between oxygen delivery and oxygen consumption in healthy animals. Above the critical oxygen delivery, oxygen consumption remains relatively constant as oxygen delivery is decreased because of a proportional increase in oxygen extraction by the tissues (supply-independent oxygen consumption phase). Below the critical oxygen-delivery value, increases in oxygen extraction are inadequate to compensate for further decreases in oxygen delivery, so that oxygen consumption becomes dependent on oxygen delivery, and lactic acid, a marker of anaerobic metabolism, accumulates in the blood (supply-dependent oxygen consumption phase). (Modified from Ronco JJ, Fenwick JC, Tweeddale MG, et al: Identification of the critical oxygen delivery for anaerobic metabolism in critically ill septic and nonseptic humans. JAMA 270:1724-1730, 1993; with permission.)

Oxygen Delivery Figure 2. The reported relationship between oxygen delivery and oxygen consumption in patients with ARDS or sepsis. Clinical studies reported supply-dependent oxygen consumption at normal or even supranormal values of oxygen delivery, suggesting that the critical oxygen-delivery value is increased in these patients. Because none of these studies definitively identified the critical oxygen-delivery value, however, the true relationship between oxygen delivery and oxygen consumption in these patients cannot be defined from these studies.

648

RONCO et a1

the 0,ER of 70% found at the critical 0, delivery value in healthy dogs. If the critical 0, delivery value, in fact, is increased and the critical 0,ER decreased in patients who have ARDS or sepsis, it would provide a rationale for increasing 0, delivery to supranormal values. In none of these studies, however, was the critical 0, delivery value identified definitively. Returning to animal studies, Nelson and coworkersz9,30, 37 hypothesized that it was sepsis, not the lung injury itself, that was responsible for the supply-dependent 0, consumption in ARDS at normal or supranormal values of 0, delivery with low values of tissue 0, extraction. In their dog studies, they demonstrated that both bacteremia and endotoxemia, but not isolated acute lung significantly increased the critical 0, delivery value by 50%-from 8 to 12 mL/kg/minute (Fig. 3). More importantly, the 0,ER at the point at which 0, consumption became supply dependent fell from 70% to 51% in bacteremic dogs. They termed this phenomenon pathologic supply-dependent O2consumption to distinguish it from the supply-dependent 0, consumption seen below the "normal" critical 0, delivery value in healthy animals. Moreover, these investigators demonstrated that the ability of tissues to extract oxygen following a septic insult varied with the tissue studied. Whereas the 0,ER at the critical 0, delivery value remained relatively unchanged in skeletal rnus~le,3~ it decreased Significantly in the suggesting that the gut was a major contributor to the 0, extraction defect and, therefore, to the pathologic supply-dependent 0, consumption seen at the level of the whole body. The work of Nelson and coworkersz9, 37 clearly provided support 307

-

Normal

Oxygen Delivery Figure 3. The biphasic relationship between oxygen delivery and oxygen consumption in healthy and septic animals. Both bacteremia and endotoxemia significantly increased the critical oxygen-delivery value in septic animals. More importantly, the oxygen-extraction ratio at the point at which oxygen consumption became supply dependent fell from 70% in healthy dogs to 51% in bacteremic dogs. This phenomenon has been termed pathologic supplydependent oxygen consumption to distinguish it from the supply-dependent oxygen consumption seen below the critical oxygen-delivery value in healthy animals.

DOES INCREASING O2DELIVERY IMPROVE OUTCOME IN THE CRITICALLY ILL? N O

649

for the strategy of increasing 0, delivery to supranormal values and has been cited widely by clinical investigators who have emphasized the clinical relevance of pathologic supply-dependent 0, consumption. Bihari and coworkers,2 for example, suggested that the finding of supplydependent 0, consumption at normal or even supranormal values of 0, delivery and low values of tissue 0, extraction represents inadequate 0, delivery, with "occult" tissue hypoxia., These investigators further correlated the presence of pathologic supply-dependent 0, consumption with an increased incidence of multiple-system organ failure and mortality. In addition, our group, as well as other investigators, reported that markers of anaerobic metabolism such as the presence of metabolic acidosis or increased concentration of plasma lactate, predicted pathologic supply-dependent 0, consumption.2, 44 Based on numerous clinical studies reporting pathologic supplydependent 0, consumption in critically ill patients, and the assumption that it represents inadequate 0, delivery with persistent tissue hypoxia, treatment strategies aimed at increasing 0, delivery to supranormal values have become widely advocated.40* Although such strategies have great clinical appeal, they are not without potential complications.20 It therefore is vital to ensure that such strategies are based on accurate interpretation of well-designed clinical studies and a proper understanding of the relationship between whole-body 0, delivery and consumption. In our opinion, this has not proved to be the case. We now believe that the clinical studies reporting pathologic supply-dependent 0, consumption (and upon which treatment strategies aimed at increasing 0, delivery to supranormal values are based) should be interpreted with caution for three reasons: First, in many of these studies, methodologic error may have occurred in the determination of 0, consumption, producing an artifactual correlation between 0, delivery and consumption. Second, even in the absence of methodologic error, spontaneous changes in 0, demand, resulting in changes in 0, delivery, may have been misinterpreted as pathologic supply-dependent 0, consumption. Third, these studies were conducted with a lack of knowledge of the critical 0, delivery value and its associated critical 0, extraction ratio in normal and critically ill humans, information vital to the correct interpretation of any studies reporting supply-dependent 0, consumption. These issues are discussed subsequently. In many clinical studies of patients with ARDS or sepsis in which pathologic supply-dependent 0, consumption has been reported, 0, delivery and consumption both were calculated values, using a common set of directly measured variables-specifically cardiac output and arterial oxygen content.,, *, 12, Oxygen delivery therefore was calculated as the product of cardiac output and arterial O2 content, whereas 0, consumption was calculated as the product of cardiac output and the difference between arterial and venous 0, content. Because cardiac output and arterial 0, content were used in both calculations, a change in the value of either automatically would influence the other. The apparent relationship between calculated 0, consumption and calculated 0, deliv-

650

RONCO et a1

ery therefore potentially could be the result of an artifactual correlation arising from mathematical coupling of the shared variables cardiac output and arterial oxygen content.', 36, 42 To avoid such an artifactual correlation arising from mathematical coupling, we:', 32, 34, 35 as well as others,lg!24* 27 studied patients who had severe ARDS or sepsis and measured 0, consumption by analysis of respiratory gases (indirect calorimetry) independently of increases in 0, delivery, which were based, as in the work of others, on thermodilution cardiac output measurements. We found that 0, consumption measured by analysis of respiratory gases remained constant and independent of increases in 0, delivery from normal to supranormal values, whether the increase in 0, delivery was produced by blood transfusion or dobutamine (Fig. 4A). In contrast, when we calculated 0, consumption as others have done, it appeared to increase with increasing 0, delivery, mimicking what others have called pathologic supply-dependent 0, consumption2,8, 12, (Fig. 4B). Furthermore, with this methodology, we found no differences in the relationship between 0, delivery and consumption between patients who had normal or increased concentrations of plasma lactate. Because calculated 0, consumption using shared variables suggested pathologic supply-dependent 0, consumption, whereas directly measured 0, consumption did not, we concluded the following: First, that clinical studies reporting pathologic supply-dependent 0, consumption were the result of methodologic error.,, 8, 12, 44 Second, that when proper methodology was used, pathologic supply-dependent 0, consumption was not present in patients with ARDS or sepsis. Third, that the finding of an increased concentration of plasma lactate, by itself, cannot be considered definitive evidence of ongoing anaerobic metabolism. Fourth, that interventions to increase 0, delivery to supranormal values in the hope of increasing systemic 0, consumption in the critically ill are not justified. A second possibility that must be considered when interpreting clinical studies reporting pathologic supply-dependent 0, consumption is that the apparent relationship between 0, delivery and consumption merely reflects normal physiology, in which increases in 0, delivery occur in response to spontaneous changes in 0, demand.45Exercise is a good example of such an occurrence. 0, consumption in a patient therefore will vary depending on his or her clinical condition. Both 0, consumption and delivery will be lower when a patient is sleeping or sedated, for example, whereas both variables increase in the awake state, and will increase even further with agitation or fever. As a result, if oxygen demand were to change during clinical studies, a relationship between 0, delivery and consumption that represents normal physiology could be misinterpreted as representing pathologic supply-dependent 0, consumption. In our studies,3l, 35 great care was taken to ensure that there was no change in 0, demand. Finally and, perhaps, most importantly, the correct interpretation of any studies reporting pathologic supply-dependent 0, consumption requires knowledge of the "normal" critical 0, delivery value and the

-

250

250

p

200

-

5F 150 cz-

-

f

gs $2 100

-

f s

-

50

4

P

200

-

150

-

100

-

50

-

d

651

-

DOES INCREASING O2 DELIVERY IMPROVE OUTCOME IN THE CRITICALLY ILL? N O

\ -

-#'p..

0

250 200 150 100 50 -

8'

0 0

B

200

400

600

800

1WO

1200

Oxygen Delivery (mUmin/mz)

Figure 4. Individual responses in measured (A) and simultaneously calculated oxygen consumption (6) to increases in oxygen delivery by dobutamine in patients who had normal (left panel) or increased (right panel) concentrations of plasma lactate. Measured oxygen consumption remained constant and independent of increases in oxygen delivery from normal to supranormal values. In contrast, simultaneously calculated oxygen consumption increased with increasing oxygen delivery, mimicking what others have called pathologic supply-dependent oxygen consumption. There were no differences in the relationship between oxygen delivery and consumption between patients who had normal or increased concentrations of plasma lactate. (Modified from Ronco JJ, Fenwick JC, Wiggs BR, et al: Oxygen consumption is independent from increases in oxygen delivery by dobutamine in patients who have sepsis and normal or increased concentrations of arterial plasma lactate. Am Rev Respir Dis 147:25-31, 1993; with permission.)

0,ER at that point. In the animal studies mentioned, in which the systemic and regional critical 0, delivery and extraction ratios were reported, the values were defined for each animal by progressively decreasing 0, delivery to very low values. Similar methodology obviously cannot be employed to define the critical 0, delivery value and the critical 0,ER in either normal or critically ill humans. Although attempts have been made to identify these critical points in anesthetized they are statistically cardiac surgical22,38 and critically ill flawed. In these studies, the critical 0, delivery was calculated by regression analysis of pooled data from multiple patients rather than by regression analysis of data obtained from each patient. Pooling of data from multiple patients ignores important differences in size and metabolic rate among patients, making the determination of the critical 0, delivery value and extraction ratio inaccurate. We recently addressed this issue by reporting data strongly suggestive that the biphasic relationship between 0, delivery and consumption

652

RONCO et a1

5 -

.-E

A

. E m

4 : x

E

Y

E 0,

m

2

0

A

12

9

6

3

B

Q

4

I

I

I

1

4

8

12

16

20

Oxygen delivery (rnl/kg/rnin)

Figure 5. The relationship between oxygen delivery and oxygen consumption (A) and between oxygen delivery and plasma-lactate concentration (6)in a septic patient who had an increased concentration of plasma lactate at baseline. Above the critical oxygen-delivery value, measured oxygen consumption and plasma-lactate concentration remained relatively constant as oxygen delivery decreased. Below the critical oxygen-delivery value, measured oxygen consumption decreased as oxygen delivery decreased further, resulting in tissue hypoxia with its consequent anaerobic metabolism and lactic acid production.

found in healthy animals may be an accurate representation of human physiology during critical illne~s.3~ We measured 0, consumption using analysis of respiratory gases in critically ill dying patients while they were experiencing decreases in O2 delivery from normal and even supranormal values to very low values during withdrawal of active treatment. To avoid methodologic error caused by pooling of data, we

DOES INCREASING O2 DELIVERY IMPROVE OUTCOME IN THE CRITICALLY ILL? N O

653

determined the critical 0, delivery value for each patient by dual-line regression analysis. These individual values then were used to calculate the mean values for the critical 0, delivery and critical 0,ER. In these patients, as 0, delivery decreased progressively, measured 0, consumption remained relatively constant and independent of 0, delivery until a critical 0, delivery value of 4 mL/kg/minute was reached (approximately 180 mL/minute/m2). That value is considerably lower than that of 330 mL/minute/m2 previously calculated using pooled group data from anesthetized cardiac surgical patients, and far lower than the generally accepted supranormal target value for 0, delivery of 600 mL/ m i n ~ t e / r nAt ~ . the ~ ~ critical 0, delivery value in each patient, the mean critical 0,ER was 60%, a value much higher than that usually seen in resuscitated critically ill patients, but similar to the 0, extraction ratio attained by athletes undergoing strenuous exercise near the anaerobic threshold.29In each patient, below the critical 0, delivery value, measured 0, consumption decreased as 0, delivery decreased further, resulting in tissue hypoxia, with its consequent anaerobic metabolism, lactic acid production, and rapid death (Fig. 5). In fact, we could calculate the critical O2 delivery value from the rise in lactate equally well as from the decrease in measured 0, consumption, and the two estimates did not differ significantly (Fig. 6 ) .Furthermore, we found no differences in the critical 0, delivery value between patients who had normal 0.5 mmol/L) or increased (mean 5.4 2.4 mmol/L) (mean 1.6

*

*

,

I

0

D4-VOzP10t

DO,-Lactate Plot

Figure 6. The critical oxygen delivery determined from the relationship between oxygen delivery and oxygen consumption (DO, - VO, plot) and from the relationship between oxygen delivery and plasma-lactate concentration (DO, - lactate plot). There was no difference in the critical oxygen-delivery value calculated from the decrease in measured oxygen consumption (4.2 2 1.3 mUkg/min) and the increase in plasma-lactate concentration (5.5 f 1.3 mUkg/min).

654

RONCO et a1

concentrations of plasma lactate at baseline, and both groups showed a progressive rise in lactate concentrations as 0, delivery decreased below the critical 0, delivery value. Finally, in contrast to the case in septic dogs, we found no differences in the critical 0, delivery value and critical 0,ER ratio between septic and nonseptic patients. From this study, we first concluded that the critical 0, delivery value in both septic and nonseptic critically ill patients was far lower than previously suspected. Second, pathologic supply-dependent 0, consumption is not a component of the pathophysiologic mechanism of human critical illness. Third, the finding of an increased concentration of plasma lactate, by itself, cannot be considered definitive evidence of ongoing anaerobic metabolism. Fourth, because any patient resuscitated to traditional endpoints of adequate tissue perfusion already is above the critical 0, delivery value and 0, consumption therefore is independent of 0, delivery, interventions to increase 0, delivery to supranormal values in critically ill patients will not increase 0, consumption further and may be inappropriate. In summary, carefully conducted studies of critically ill patients employing correct methodology have not detected pathologic supplydependent 0, consumption at normal or supranormal values of 0, delivery. Although our studies and 24, 27, 31,34, 35 do not directly refute the approach of increasing 0, delivery to supranormal values in critically ill patients to improve outcome, we believe our studies have clearly delineated the relationship between whole-body 0, delivery and consumption in critically ill humans with or without sepsis. We have demonstrated that the strategy of increasing 0, delivery to supranormal values is based on an incorrect understanding of this relationship and we therefore believe that strategies aimed at increasing 0, delivery to supranormal values are inappropriate based on whole-body measurements of 0, delivery and consumption. This, of course, presumes that whole-body values for 0, delivery and consumption, in fact, can be used to infer the adequacy of tissue oxygenation. The possibility exists that whole-body measurements of O2delivery and consumption in critically ill humans may not detect abnormalities of tissue oxygenation at the organ level. Theoretically, individual tissues may remain hypoxic in spite of adequate resuscitation and such regional hypoxia may remain undetected by whole-body measurements of 0, delivery and consumption. Evidence that this, indeed, may be the case has come from recent studies employing techniques to assess the adequacy of tissue oxygenation at the organ level, such as gastric tonometry for the determination of gastric intramucosal pH or jugular bulb catheterization for the determination of jugular venous 0, saturation?, 11, 137 17, 18, 25, 26 If tissue hypoxia were present despite whole-body measurements showing adequate 0, delivery and consumption, and if the regional hypoxia can be reversed by increasing systemic 0, delivery further, then strategies to increase 0, delivery to supranormal values might be expected, in properly selected patients, to improve outcome. Because strategies to increase 0, delivery to supranormal values have potential

DOES INCREASING 0, DELIVERY IMPROVE OUTCOME IN THE CRITICALLY ILL? N O

655

serious complications,2°however, studies reporting improved outcome employing such strategies must be scrutinized carefully before the practice is recommended for indiscriminate clinical application. In the following section we critically review the randomized controlled trials that have investigated increasing 0, delivery to supranormal values. RANDOMIZED CONTROLLED TRIALS OF INCREASING OXYGEN DELIVERY IN CRITICALLY ILL PATIENTS

We have identified 12 published randomized controlled trials that report the effects of increasing 0, delivery to supranormal values on the 3941, 43, 46, 47 14, 15, 18, morbidity and mortality of critically ill Unfortunately, the results of these trials are equally divided: Six -trials found that supranormal values of 0, delivery did not decrease morbidity or mortality in critical illness4,15, 20, 41, 43, 47; the remainder reported supranormal values of 0, delivery to be b e n e f i ~ i a l .14,~ ~ 39, , 4o Differences in the results of these trials can be explained by differences in patient characteristics and methodology, including timing of hemodynamic interventions, monitoring techniques, and treatment strategies. The six randomized controlled trials in which there was no difference in mortality each randomized a heterogeneous group of critically ill patients, after their admission to the intensive care unit (ICU) to resuscitation to supranormal 0, delivery values versus a strategy of resuscitation to traditional endpoints of adequate tissue perf~sion.~, 15, 20, 41* 43, 46, 47 In addition to finding no differences in mortality, five of the trials reported no differences in the incidence of multiple-organ failure or duration of hospital stay between the two groups.3,15,20, 41, 46, 47 In one of these trials, the study was terminated because the use of dobutamine to increase 0, delivery to supranormal values may have worsened outcome.2oBecause negative outcome studies are difficult to interpret without knowing the likelihood of type I1 error, it is unfortunate that only one of these negative trials reported a prestudy estimation of sample size and power calculation.zoA further source of confusion in the literature is that three of these trials43,46, 47 (which were negative when analyzed on an intention-to-treat basis) have been used to suggest that supranormal values of 0, delivery, in fact, do improve survival. This claim was based on the performance of a post hoc comparison of subgroups of patients (from either treatment strategy) who actually attained supranormal values of O2 delivery with those who did not. We strongly disagree with this methodology. In contrast to the heterogeneous ICU patients enrolled in the negative trials, five of the six randomized controlled trials that reported an improvement in outcome studied a far more homogeneous group, consisting mainly of high-risk surgical and multiple trauma patient^.^, 5, 14,39,40 (The sixth of these trials18will be dealt with separately.) In addition to reporting a decrease in mortality, these five trials reported decreases in the incidence of multiple-organ failure and duration of hospital stay.

656

RONCO et a1

A further and, perhaps, the most important contrast between these five trials and the negative outcome trials is the timing of patient randomization. In these five trials, surgical patients were randomized in the perioperative period, rather than after their admission to the ICU, to a strategy of increasing 0, delivery to supranormal values. Early institution of such therapy may be necessary to improve outcome because, with the passage of time, the disease process may become refractory to such treatment. Although these five trials, indeed, may support increasing 0, delivery to supranormal values in high-risk surgical and trauma patients, we believe they should be interpreted with caution, for the following reasons. In four of these trials? 14, 39, 40 differences in monitoring strategies between the patients in protocol and in control groups could have significantly influenced the choice and timing of therapeutic interventions. In one of these studies,4O for example, it was estimated that 57% of control patients actually needed a pulmonary arterial catheter but did not receive one. In the fifth of these trials: the favorable outcome in the patients randomized to supranormal values of 0, delivery may have been related to a more aggressive fluid resuscitation in the preoperative period than to the increase in 0, delivery per se. In addition, in this trial: higher rates of postoperative myocardial infarction and hemorrhage may have accounted for the increased mortality in the control group. This also is the only study to increase 0, delivery by the use of dopexamine, an inotrope and selective splanchnic vasodilator. The unique properties of this drug on the splanchnic circulation, rather than its effect on systemic O2 delivery, could account for the decrease in mortality in the therapy group, but that requires confirmation. In contrast to these five trials, the sixth randomized controlled trialIs reporting an improvement in outcome followed the same practice as the negative trials, enrolling a heterogeneous group of critically ill patients after their admission to the ICU. This trial is unique, however, in its use of gastric tonometry for the detection of a low gastric intramucosal pH 11* 17, (pHi), thought to be a marker of inadequate splanchnic perfu~ion.~, 25, 26 Critically ill patients were stratified on ICU admission to normal or low gastric pHi groups. Patients admitted to the ICU with a low pHi did not show improved survival with increasing O2 delivery, whereas increasing 0, delivery in those whose pHi initially was normal but subsequently decreased during their ICU stay did improve survival. Early recognition and treatment of splanchnic hypoxia may have been responsible for the improved outcome in this study. From these 12 trials, we conclude, first, that the indiscriminate clinical application of strategies to increase 0, delivery to supranormal values in critically ill patients after admission to the ICU cannot be supported. Second, increasing 0, delivery to supranormal values may improve outcome in high-risk surgical and trauma patients when the increase in 0, delivery is initiated prior to or during surgery. Third, pHi measurement may allow early detection and treatment of splanchnic ischemia and improve outcome in a subset of critically ill patients.

DOES INCREASING 0, DELIVERY IMPROVE OUTCOME IN THE CRITICALLY ILL? NO

657

Fourth, different strategies to increase 0, delivery to supranormal levels may not be equivalent in their clinical effects and each needs to be studied independently. CONCLUSION AND RECOMMENDATIONS

We firmly believe that critically ill patients, early in their course, whose 0, delivery is judged to be inadequate by clinical and laboratory assessment of tissue perfusion, should be resuscitated rapidly and aggressively, because prompt re-establishment and maintenance of adequate tissue oxygenation are major determinants of normal organ function and survival in critical illness. We do not advocate the indiscriminate use of strategies to attain supranormal values of O2 delivery in critically ill patients after admission to the ICU for two reasons, however. First, the rationale for increasing 0, delivery to supranormal values in appropriately resuscitated and hemodynamically stable patients in the hope of improving 0, consumption is based on an incorrect understanding of the relationship between whole-body 0, delivery and consumption. Second, the results of randomized controlled trials of increasing 0, delivery to supranormal values do not support the indiscriminate application of such strategies in critically ill patients after admission to the ICU. Although a table of hemodynamic goals may be useful to speed the delivery of care, we believe that such goals should not be used alone but in each case, should be balanced by clinical judgment of adequate tissue perfusion. We are aware that systemic parameters of tissue oxygenation (hemodynamic status, O2 transport variables, plasma lactate concentration, and end-organ function) may be insensitive markers of tissue hypoxia. While we await new and reliable tools to detect tissue hypoxia at the organ leve1,16,21 we believe that a dedicated intensivist, carefully assessing the signals obtained from the clinical examination, therapeutic interventions, and laboratory assessment remains the best option for ensuring the adequacy of tissue oxygenation in critically ill patients. References 1. Archie JP Jr: Mathematical coupling of data. A common source of error. AM Surg 193:296-303, 1981 2. Bihari D, Smithies M, Gimsom A, et al: The effect of vasodilatation with prostacyclin on oxygen delivery and uptake in critically ill patients. N Engl J Med 317397403,1987 3. Bishop MH, Shoemaker WC, Appel PL, et al: Prospective randomized trial of survivors’ values of cardiac index, oxygen delivery and oxygen consumption as resuscitation endpoints in severe trauma. J Trauma 38:780-787, 1995 4. Bone RC, Slotman G, Maunder R, et al: Randomized double-blind, multicenter study of prostaglandin E, in patients with the adult respiratory distress syndrome. Chest 96:114-119, 1989

658

RONCO et a1

5. Boyd 0, Grounds RM, Bennett E D A randomized clinical trial of the effect of deliberate perioperative increase of oxygen delivery on mortality in high-risk surgical patients. JAMA 270~2699-2707,1993 6. Cain SM: Appearance of excess lactate in anaesthetized dogs during anemic and hypoxic hypoxia. Am J Physiol 209:604-610, 1965 7. Cain SM: Oxygen delivery and uptake in dogs during anemic and hypoxic hypoxia. J Appl Physiol42228-234, 1977 8. Danek SJ, Lynch JP, Weg JG, et al: The dependence of oxygen uptake on oxygen delivery in the adult respiratory distress syndrome. Am Rev Respir Dis 122:387-395, 1980 9. Dantzker DR The gastrointestinal tract. The canary of the body? JAMA 270:12471248, 1993 10. Danztker DR, Foresman B, Gutierrez G: Oxygen supply and utilization relationships: A reevaluation. Am Rev Respir Dis 143:675-679, 1991 11. Doglio G, Pusajo J, Egurrola M, et al: Gastric intramucosal pH as a prognosis index of mortality in critically ill patients. Crit Care Med 191037-1040, 1991 12. Fenwick JC, Dodek PM, Ronco JJ, et al: Increased concentrations of plasma lactate predict pathological dependence of oxygen consumption on oxygen delivery in patients with the adult respiratory distress syndrome. J Crit Care 5531-86, 1990 13. Fenwick JC, MacDonald DB, Woodhurst WB, et al: Hyperventilation induces cerebral ischemia in patients with severe closed head injury. J Neurotrauma 12:412, 1995 14. Fleming A, Bishop M, Shoemaker W, et al: Prospective trial of supranormal values as goals of resuscitation in severe trauma. Arch Surg 1271175-1181, 1992 15. Gattinoni L, Brazzi L, Pelosi P, et al: A trial of goal-oriented hemodynamic therapy in critically ill patients. N Engl J Med 333:1025-1032, 1995 16. Gutierrez G, Brown S D Gastric tonometry: A new monitoring modality in the intensive care unit. J Intensive Care Med 10334-44, 1995 17. Gutierrez G, Bismar H, Dantzker D, et a1 Comparison of gastric intramucosal pH to measures of oxygen transport and consumption in critically ill patients, Crit Care Med 20451-457, 1992 18. Gutierrez G, Palizas F, Doglio G, et a1 Gastric intramucosal pH as a therapeutic index of tissue oxygenation in critically ill patients. Lancet 339:195-199,1992 19. Hanique G, Dugernier T, Laterre PF, et al: Significance of pathologic oxygen supply dependency in critically ill patients: Comparison between measured and calculated methods. Intensive Care Med 20:12-18, 1994 20. Hayes MA, Timmins AC, Yau EHS, et al: Elevation of systemic oxygen delivery in the treatment of critically ill patients. N Engl J Med 3301717-1722, 1994 21. Ivatury RR, Simon RJ, Havriliak D, et al: Gastric mucosal pH and oxygen delivery and oxygen consumption indices in the assessment of adequacy of resuscitation after trauma: A prospective, randomized study. J Trauma 39:128-134, 1995 22. Komatsu T, Shibutani K, Kazufumi 0, et al: Critical level of oxygen delivery after cardiopulmonary bypass. Crit Care Med 15:194-197, 1987 23. Long GR, Nelson DP, Sznajder I, et a1 Systemic oxygen delivery and consumption during acute lung injury in dogs. J Crit Care 3249-255, 1988 24. Manthous CA, Schumacker PT, Pohlman A, et al: Absence of supply dependence of oxygen consumption in patients with septic shock. J Crit Care 8:203-211,1993 25. Marik PE: Gastric intramucosal pH. A better predictor of multiorgan dysfunction syndrome and death than oxygen-derived variables in patients with sepsis. Chest 104~225-230,1993 26. Maynard N, Bihari D, Beale R, et al: Assessment of splanchnic oxygenation by gastric tonometry in patients with acute circulatory failure. JAMA 270:1203-1210, 1993 27. Mira JP, Fabre JE, Baigorri F, et al: Lack of supply dependency in patients with severe sepsis. Chest 10631524-1531, 1994 28. Mohsenifar Z , Goldbach P, Tashkin DP, et al: Relationship between oxygen consumption and oxygen delivery in adult respiratory distress syndrome. Chest 84267-271, 1983 29. Nelson DP, Beyer C, Samsel RW, et al: Pathological supply dependence of oxygen uptake during bacteraemia in dogs. J Appl Physiol63:1487-1492, 1987

DOES INCREASING O2DELIVERY IMPROVE OUTCOME IN THE CRITICALLY ILL? NO

659

30. Nelson DP, Samsel RW, Wood LDH, et al: Pathological supply dependence of systemic and intestinal 0, uptake during endotoxemia. J Appl Physiol64:2410-2419, 1988 31. Phang PT, Cunningham KF, Ronco JJ, et a1 Mathematical coupling explains dependence of oxygen consumption on oxygen delivery in ARDS. Am J Respir Crit Care Med 150:318-323, 1994 32. Ronco JJ, Phang PT: Validation of an indirect calorimeter to measure oxygen consumption in critically ill patients. J Crit Care 636-41, 1991 33. Ronco JJ, Fenwick JC, Tweeddale MG, et a 1 Identification of the critical oxygen delivery for anaerobic metabolism in critically ill septic and nonseptic humans. JAMA . 2701724-1730, 1993 34. Ronco JJ, Fenwick JC, Wiggs BR, et al: Oxygen consumption is independent of increases in oxygen delivery by dobutamine in patients who have sepsis and normal or increased concentrations of arterial plasma lactate. Am Rev Respir Dis 14725-31, 1993 35. Ronco JJ, Phang PT, Walley KR, et al: Oxygen consumption is independent of changes in oxygen delivery in severe adult respiratory distress syndrome. Am Rev Respir Dis 1431267-1273, 1991 36. Russell JA, Phang PT The oxygen delivery/consumption controversy. Approaches to management of the critically ill. Am J Respir Crit Care Med 149:533-537, 1994 37. Samsel RW, Nelson DP, Sanders WM, et al: Effect of endotoxin on systemic and skeletal muscle 0, extraction. J Appl Physiol 653377-1382, 1988 38. Shibutani K, Komatsu T, Kubal K, et al: Critical level of oxygen delivery in man. Crit Care Med 11:640-643, 1983 39. Shoemaker W, Appel PL, Kram H B Role of oxygen debt in the development of organ failure, sepsis and death in high-risk surgical patients. Chest 102208-215, 1992 40. Shoemaker W, Appel PL, Kram HB, et al: Prospective trial of supranormal values in survivors as therapeutic goals in high-risk surgical patients. Chest 94:1176-1186, 1988 41. Silverman HJ, Slotman G, Bone RC, et a1 Effects of prostaglandin E, on oxygen delivery and consumption in patients with the adult respiratory distress syndrome. Chest 98:405410, 1990 42. Stratton HH, Feustel PJ, Newel1 JC: Regression of calculated variables in the presence of shared measurement error. J Appl Physiol62:2083-2093, 1987 43. Tuchshmidt J, Fried J, Astiz M, et al: Elevation of cardiac output and oxygen delivery improves outcome in septic shock. Chest 10221&220, 1992 44. Vincent JL, Roman A, De Backer D, et al: Oxygen uptake/supply dependency. Effects of short-term dobutamine infusion. Am Rev Respir Dis 141:2-7, 1990 45. Weissman C, Kemper M The oxygen uptake-oxygen delivery relationship during ICU interventions. Chest 99:430-435, 1991 46. Yu M, Levy MM, Smith P, et al: Effect of maximizing oxygen delivery on morbidity and mortality rates in critically ill patients: A prospective randomized study. Crit Care Med 21:83&838,1993 47. Yu M, Takanishi D, Myers SA, et al: Frequency of mortality and myocardial infarction during maximizing oxygen delivery: A prospective, randomized trial. Crit Care Med 23:10251032, 1995

Address reprint requests to Juan J. Ronco, MD, FRCP(C) Critical Care Medicine 855 West 12th Avenue Vancouver, BC Canada, V5Z 1M9