Serial Blood Lactate Levels Can Predict the Development of Multiple Organ Failure

Following Septic Shock Jan Bakker, MD, Philippe Gris, MD, Michel Coffernils, MD, Robert J. Kahn, MD,

Jean-Louis Vincent, MD, PhD, Brussels, Belgium

BACKGROUND: Despite successful initial resusci- tation, septic shock frequently evolves into mul- tiple system organ failure (MSOF) and death. Since blood lactate levels can reflect the degree of cellular derangements, we examined the rela- tion between serial blood lactate levels and the development of MSOF, or mortality, in patients with septic shock.

PATIENTS AND METHODS: In 87 patients with a first episode of septic shock, we measured initial lac- tate (at onset of septic shock), final lactate (be- fore recovery or death), “lactime” (time during which blood lactate was a2.0 mmol/L), and the area under the curve (AUC) for abnormal values (above 2.0 mmol/L). These measurements were correlated with survival and organ failure and scored for four systems (ie, respiratory, renal, hepatic, and coagulation), adding to a maximal score of 8.

RESULTS: Thirty-three (38%) patients survived. of the 54 (62%) nonsurvivors, the 13 patients who died during the first 24 hours of septic shock had higher initial blood lactate levels than those who died later (mean * standard deviation 9.6 f 5.3 mmol/L versus 5.6 f 3.7 mmol/L, P ~0.05). The 74 patients who survived the first 24 hours

of shock were studied in more detail. On pres- entation, survivors had a significantly higher mean arterial pressure (76 + 12 mm Hg versus 63 f 20 mm Hg, P qO.001) and arterial pH (7.40 f 0.07 versus 7.37 * 0.09, P ~0.05) than nonsur- vivors. Although the differences in initial blood lactate levels between survivors and nonsur- vivors did not reach statistical significance (4.7 f 2.5 mmol/L versus 5.6 + 3.7 mmol/L), only the survivors had a significant decrease during the first 24 hours of septic shock. The survivors had a significantly lower lactime and AUC than the nonsurvivors. The duration of lactic acidosis

From the Department of Intensive Care, Erasme University Hospital, Free University of Brussels, Brussels, Belgium.

Requests for reprints should be addressed to Jean-Louis Vincent, MD, PhD, Department of Intensive Care, Erasme University Hospital, Route de Lennik 808, 1070 Brussels, Belgium.

Manuscript submitted February 7, 1995 and accepted April 10, 1995.

was the best predictor of survival (multiple re- gression analysis, R* = 0.266, P <O.OOl), fol- lowed by age, heart rate, and mean arterial pres- sure. Patients with lower organ failure scores had lower initial blood lactate, lactime, and AUC. The duration of lactic acidosis was the only sig- nificant predictor of organ failure.

CONCLUSIONS: In patients with septic shock, se- rial determinations of blood lactate levels are good predictors of the development of MSOF and death. In this respect, the duration of lactic acidosis is more important than the initial lac- tate value. Although a number of factors may contribute to hyperlactatemia, these observa- tions are compatible with a direct role of pro- longed tissue hypoxia in the development of complications following septic shock. Am J Surg. 1996;171:221-226.

S eptic shock remains an important cause of morbidity and mortality in critically ill patients.’ Despite progress in the immediate resuscitation from shock,

death sometimes occurs after a prolonged complicated course leading to multiple system organ failure (MSOF). The role of tissue hypoxia in the development of MSOF is supported by several arguments. First, it is well established that oxygen availability is essential to cell function and that prolonged hypoxia leads to organ damage. Second, severe sepsis in pa- tients has been associated with the presence of a pathologic dependency of oxygen uptake on oxygen delivery (DO,), suggesting the persistance of tissue hypoxia.‘,’ Third, several recent studies have suggested that the maintenance of supra- normal DO2 values may reduce organ failure and mortality in critically ill patients.‘,’ Fourth, the persistance of gut is- chemia has been implicated in the development of MSOF.6s’ Fifth, the immune response is altered by the presence of hy- poxia or lactic acidosis. sa9 The mediators of sepsis, in turn, may worsen tissue hypoxia by increasing cellular oxygen de- mand, thereby altering the oxygen extraction and reducing myocardial contractility. lc Thus, an interrelation between sepsis and hypoxia may take place.

Finally, blood lactate levels, that are supposed to reflect the magnitude of the anaerobic metabolism related to cellular hy- poxia, have been well correlated to survival from various forms of acute circulatory failure.‘,“-‘4 Nevertheless, the in- terpretation of elevated blood lactate levels in sepsis is lim- ited by several important factors. I5 First blood lactate levels reflect not only the production hut also the elimination of the molecule, and the lactate clearance can be sometimes

THE AMERICAN JOURNAL OF SURGERY* VOLUME 171 FEBRUARY 1996 221

;LOOD LACTATE IN SEPTIC SHOCWBAKKER ET AL 1 ,

TABLE I

System/Variable

Respiratory

Organ Failure Score

No Yes

Mechanical ventilation >24 hours PaO,/FiO, 1250 mm Hg”

Renal

0 1 0 1

BUN >40 mg/dL and/or creatinine >180 FmoVL BUN >80 mg/dL and/or creatinine >350 pmol/L+

Hepatic

0 0

1 1

Bilirubin >34 pmoVL’ (without hemolysis) 0 1 AST or ALT >80 U/L 0 1

Coagulation

Platelets ~100 X 10$/L PT ~60% or aPn >80 set’

Score

0 1

0 1 0 8

*Or 50% change from baseline if inif~ally abnormal. :Or doubling from baseiine if !nitia//y abnormai. BUN = blood urea nitrogen, AST = aspartate aminotransferase, ALT = ala- nine aminotransferase, PT = prothrombrn time, aPR = activated part/a/ ihromboplastin time

TABLE II

Sources of Infection and Type of Microorganisms Isolated in All Patients

Bacterial Source of Gram Gram Viral &

Infection Negative Positive Mixed Fungal Unknown Total

Lung 20 8 7 1 1 37 Abdomen 16 2 2 3 4 27

Urogenital 9 1 0 0 0 10

system Skin 1 2 1 1 0 5 CNS 0 2 0 0 0 2

Other 1 5 0 0 0 6 Total 47 20 10 5 5 87

CNS = cenirai nervous system.

protracted, especially in the presence of liver impairment. Second, sepsis can increase lactate production by increasing the glycolysis of the cells. Third, increased lactate production in sepsis may result from other cellular alterations than hy- poxia, in particular, the inhibition of pyruvate metabolism.lh Finally, lactate levels reflect only global changes but give no indication about the regional distribution of blood flow.

One could speculate that the evaluation of serial blood lac- tate levels may have a better prognostic value than a single measurement, as it may take into account the duration of the tissue injury. By repeating lactate measurements every 20 minutes in a series of patients with shock, Vincent et al’* showed that the survivors had an early reduction in blood lactate levels. The short duration of the study, however, did not allow the investigators to correlate the severity of lactic acidosis to morbidity. In the present study, we examined the relation between the time course of blood lactate measure- ment and morbidity and mortality following septic shock.

PATIENTS AND METHODS We studied 87 consecutive adult patients who developed

a first and unique episode of septic shock. Circulatory shock

-----3 INITIAL +8h +16h +24h FINAL

Time I Figure 1. Time course of blood lactate levels for the 33 sur-

vivors and the 41 nonsurvivors. Initial values were taken at on- set of shock, and the final values at time of recovery or before death (mean f standard error of the mean). ‘P ~0.05; “P<O.Ol

(survivors versus nonsurvivors); #P ~0.05 versus initial blood

lactate level.

was defined by the development of arterial hypotension (sys- tolic arterial pressure ~90 mm Hg requiring the administra- tion of vasopressors) and hyperlactatemla (blood lactate level >2.0 mmol/L) in the presence nf signs of altered tissue perfusion such as oliguria (urine output ~20 mL/h) or altered mental status. Each patient also h,ld sepsis as defined by the presence of fever (temperature above 38. S’C), an abnormal white blood cell count (above 13.0 X lo”/L or below 4.0 X 109/L), and either positive blood cultures or a documented source of infection.

Treatment of shock followed a standard protocol. The un- derlying infection was treated with antibiotics and surgery when needed. Hypoxemia was treated with oxygen therapy and generally with mechanical ventilation. Each patient was monitored with an arterial catheter and a pulmonary artery catheter. Imtlal treatment of shock included the adminis- tration of a 4% colloid solution (Stable Solution of Plasma Proteins, Red Cross of Belgium) until the cardiac output reached a plateau or the pulmonary artery balloon-occluded pressure (PAOP) reached 16 mm Hg. As hypotension per- sisted, dopamine was started and the dose increased until the systolic arterial pressure reached at least 90 mm Hg. If this end point was not obtained with ;I dose of 20 pg/kg per minute of dopamine, norepinephrine was added at titrated doses until hypotension was controlled. No low-dose dopamine was used to increase urine output. Dobutamine was aJded in the presence of an inappropriate cardiac out- put (~3 L/minim’), an increase in PAOP above 18 mm Hg during the fluid challenge, or a persistant lactic acidosis.”

A total of 326 measurements were obtained during the sep- tic shock episode. Initial measurements were obtained within 4 hours after the onset of septic shock. Final measurements were obtained after recovery of shock (when the systolic ar- terial pressure w;1s above 90 mm Hg without administration of vasopressors) or before death.

The organ failure (OF) score was calculated based on the functioning of four systems and totalled a possible 8 points (Table I). Patients with a history of respiratory, renal, or bil- iary tract disease were scored according to their baseline

222 THE AMERICAN JOURNAL OF SURGERYa VOLUME 171 FEBRUARY 1996

1 BLOOD LACTATE IN SEPTIC SHOCWBAKKER ET AL

P=NS

P <O.OOl

P <O.OOl

Figure 2. Initial blood lactate level, time during which blood lac tate exceeded 2.0 mmol/L (lactime), and area under the curvy (AUC) for the 33 survivors c7 and the 41 nonsurvivors BE (mean f standard error of the mean). NS = not significant.

PaOz/FiOz ratio, blood urea nitrogen, and creatinine, biliru bin, aspartate aminotransferase, and alanine aminotrans ferase levels, respectively. No patient with chronic renal fail ure requiring dialysis or advanced hepatic failure was enterer into the study. Bilirubin levels were scored in the absence of hemolysis, biliary obstruction, or resorption of large hematomas. Coagulation parameters were scored in the ah sence of anticoagulant medication.

Each patient was monitored with a pulmonary artery cathete (Baxter Healthcare, Irvine, California) and an arteria (femoral) catheter. Intravascular pressures were determined a

TABLE III

Initial Data in Survivors and Nonsurvivors

Survivors Nonsurvivors

(n = 33) (n=41)

Heart rate (bpm) 96r 15 112 * 31

Mean arterial pressure (mm Hg) 76k12 63 + 20

Cardiac index (L/min/m*) 3.8 + 1.3 3.5 It 1.2

SVRI (dynes/sec/cm-5/m2) 1,582 + 681 1,624 + 1,108

Wa 7.40 f 0.07 7.37 f 0.09+ Lactate (mmol/L) 4.7 + 2.5 5.6 + 3.7

DO, (mL/min/m’) 553 + 207 493 k 185 VO, (mL/min/m*) 169 + 53 164 + 59

T CO 001 between survivors and nonsurwors. +P 405 between surwors and nonsurwors SW = system/c vascular resisfance index. pHa = arterial pH; DOi = oxy- gen delwery, VO‘? = oxygen consumpbon

TABLE IV

Discriminants of Survival by Multiple Regression Analysis

of All Measurements in the 74 Patients Surviving the First 24 Hours of Septic Shock

R2 at End

of Stepwise P Value of the Variable Regression Analysis R2 Change

Lactime (h) 0.266 0.001 Age (Y) 0.384 0.001

Heat-l rate (bpm) 0.440 0.02 Mean arterial pressure (mm Hg) 0.485 0.04

end-expiration. Cardiac output was determined by the ther- modilution technique (computer 9520A or REF-1, Baxter Healthcare) using 3 to 5 bolus injections of cold (< 10°C) D5W (CO-set system, Baxter Healthcare). Immediately after each cardiac output determination, arterial and mixed venous blood samples were drawn for determination of blood gases (ABU or ABL 300, Radiometer, Denmark) and hemoglobin satura- tions (Hemoximeter OSM-3, Radiometer). Arterial blood lac- tate concentrations were determined enzymatically (auto-an- alyzer 705, Hitachi, Tokyo, Japan). A value higher than 2.0 mmol/L was considered abnormal.

Cardiac index, DOz, and oxygen consumption (VO,) were calculated by standard formulas.’ “Lactime” was defined as the time during which the blood lactate levels remained above 2.0 mmol/L. The area under the curve (AUC) for ab- normal values (above 2.0 mmol/L) of lactate levels was cal- culated using the initial and final hlood lactate levels, and assuming a linear regression, according to the equation:

AUC = 0.5 X lactime X (initial lactate + final lactate) - 2 X lactime.

Results were analyzed by analysis of variance for repeated measurements, Newman-Keul’s multiple range tests, and Student’s t-test when appropriate. All data are presented as mean + standard deviation unless stated otherwise. P co.05 was accepted as statistically significant.

RESULTS The age of the 87 patients (64 male, 23 female) ranged

from 24 to 86 years (mean 55 + 13). The most common sources of infection were the lungs and the abdomen. Gram-

THE AMERICAN JOURNAL OF SURGERYO VOLUME 171 FEBRUARY 1996 223

LOOD LACTATE IN SEPTIC SHOCK/BAKKER ET AL

TABLE V

Initial and Final Data According to Organ Failure Score in the 74 Patients Who Survived the First 24 Hours of Septic Shock

Organ Failure Score l-2 Organ Failure Score 3-4 Organ Failure Score 5-8 Organ Failure Score 7-8

(n = 10) (n = 23) (n = 28) (n = 13)

Initial data

MAP (mm Hg) 82.8 + 14.0 66.6 i 19.6’ 66.0 k 18.7+ 67.3 + 11.3’

SVRI (dynes/sec/cm-5/m2) 1,920 + 792 1,562 f 784 1,618 + 1,060 1,215+432

PHa 7.41 * 0.08 7.42 + 0.06 7.35 + 0.07.z 7.37 + 0.085

Lactate (mmol/L) 3.4 + 1.5 4.8 f 3.1 5.3 * 2.2 7.0 + 5.4

CaO, (mg/dL) 14.3 f 0.8 14.4 f 1.6 14.3 f 2.0 13.8 + 1.7 DO, (mUmin/m’) 504 f 191 518 k 182 500 + 213 548 f 165

VO, (mUmin/m2) 156 zt 49 175k62 153+53 176 + 54 Final Data

MAP (mm Hg) 91.2 + 16.4 82.5 5 13.5’ 72.7 f 18.1+ 67.6 f 16.0* SVRI (dynes/sec/cm-5/m2) 1,983 * 964 2,014 f 979 1,662 + 762 1,096 * 415fl

@+a 7.45 + 0.04 7.42 zt 0.08 7.36 2 0.08+§ 7.39 * 0.145

Lactate (mmol/L) 1.7 f 0.5 2.9 + 2.2 4.2 f 3.6+ 5.1 + 4.0+5 CaO, (mg/clL) 14.9 f 1.2 14.5 f 1.7 13.9 * 1.9 13.3 f 1.8+ DO, (mUmin/m*) 532 k 149 465 f 135 447 * 161 570 f 1911

VO, (mL/min/m2) 167 zt 60 173 + 60 141 + 365 199 + 62”

Only parameters showing sigmficant differences are presented *p ~0.01 versus I-2 Q <O.O5 versus I-2. -? ~0 Of versus 3-4. *P ~0.05 versus 3-4. cP co.05 versus 5-6. MAP = mean arterial pressure; SW = systemic vascular resistance index; CaO, = artenai oxygen content; DO2 = oxygen delivery; VO, = oxygen consumption.

negative organisms were recovered in 57 (66%) patients As compared to patients with low OF scores, patients with (Table II). The most common organisms cultured were high OF scores had a lower mean arterial pressure, a higher Escherichiacoli (n = 18), St~~~hylococctts (n = 13), Streotococcw. blood lactate level, and a lower pHa, both in the initial and (n = II), and Pseudomonas species (n = 8). Blood cultures final phases of septic shock (Table V, Figure 3). Oxygen- were positive in 52 (60%) patients, and gram-negative or- derived variables were not significantly correlated to the de- ganisms were cultured in 58% of them. No relation was found velopment of OF. If anything, initial and final DO, and VOz between source of infection or the bacteriologic results and were higher in the patients with most severe OF (OF scores the development of OF or survival. 7 and 8). Patients with a higher OF score had higher lac-

Of the 87 patients, 54 (62%) died and 33 (38%) survived. time and AUC (Figure 3). Multiple regression analysis re- On presentation, the nonsurvivors had a higher blood lactate vealed that lactime was the only significant discriminant of level than the survivors (6.6 + 4.4 mmol/L versus 4.7 f 2.5 the OF score. The addition of any other parameter did not mmol/L, P ~0.05). Of the 54 nonsurvivors, 13 (24%) died significantly improve the prediction of the OF score. within the first 24 hours of septic shock, and these early fatal- ities presented with a higher blood lactate level than those who COMMENTS died later (9.6 f 5.3 mmol/L versus 5.6 f 3.7 mmol/L, P ~0.05). The present study indicates that nc)t only the initial blood

The 74 other patients, including 33 survivors and 41 non- lactate levels, but also the duration of hyperlactatemia have survivors, were studied in more detail. The survivors pre- an important prognostic value in septic shock. Of all param- sented with a higher mean arterial pressure and a lower ar- eters, the duration of hyperlactatrmia (“lactime”) was the terial pH (pHa) than the nonsurvivors (Table III). Although best discriminant of survival and organ failure. It should be the differences in initial lactate levels did not reach statisti- emphasized however, that none of the parameters studied cal significance (4.7 f 2.5 mmol/L versus 5.6 f 3.7 mmol/L), could accurately predict outcome from septic shock, which the decrease in blood lactate levels during the first 24 hours can be influenced by numerous factors. was more significant in the survivors than in the nonsurvivors A difficulty in establishing a relation between the time (Figure 1). The survivors also had a significantly shorter lac- course of blood lactate levels and the prognosis is that the time and smaller AUC (Figure 2). The survivors had a sig patients who die early have a shorter period of observation, nificantly lower OF score than the nonsurvivors (4.0 f 2.0 thus, a shorter lactime and a smaller AUC. To test the hy- versus 5.3 2 1.5, P <O.OOl). Of all parameters studied, mul- pothesis that blood lactate levels are related not only to out- tiple regression analysis showed that lactime was the most come but also to the severity of organ failure, we excluded significant discriminant of survival, followed by age, heart patients who died during the first 24 hours following the on- rate, and mean arterial pressure (Table IV). DOz and VO, set of shock. In these patients, biochemical signs of organ had no significant prognostic value and were not correlated dysfunction had no time to develop, and the time course was with lactate levels. too short to yield a meaningful analysis.

224 THE AMERICAN JOURNAL OF SURGERY” VOLUME 171 FEBRUARY 1996

1 BLOOD LACTATE IN SEPTIC SHOCWBAKKER ET AL

For the entire population, initial blood lactate levels were significantly higher in the nonsurvivors than in the sur- vivors. After exclusion of the patients who died within 24 hours, these differences were no longer significant. However, the lactime and the AUC were significantly greater in the nonsurvivors than in the survivors. This was observed even though some patients still died relatively early. Several clin- ical studies have correlated blood lactate levels to the mor- tality rate in septic shock, like in other forms of circulatory shock’J2*‘4; however, most of these studies focused on initial blood lactate levels. In a previous study, Vincent et al” re- ported that the time course of blood lactate levels during the initial fluid resuscitation could be related to survival. Also Falk et al’” reported a delayed lactate clearance in the non- survivors from shock.

The multiple regression analysis of all measurements indi- cated that the lactime was the best discriminant of survival. This occurred despite the fact that lactate levels were de- termined at variable time intervals. Elevated blood lactate levels are supposed to reflect an imbalance between the oxy- gen needs of the cells and the oxygen supply to them. Even though the oxygen supply may be similar in the survivors and the nonsurvivors, the latter patients may have greater oxygen requirements associated with a more severe degree of illness. Although a greater degree of anaerobic metabo- lism in the nonsurvivors is likely to represent the most im- portant mechanism, other alterations may be incriminated, including more severe alterations in liver perfusion and/or function reducing lactate clearance or other cellular alter- ations involving pyruvate metabolism.” The more rapid res- olution of the lactic acidosis would reflect an improvement in cellular function in any case. As in other studies, the ini- tial mean arterial pressure also represented a good prognos- tic index,“J” but cardiac index and oxygen-derived variables did not.1 1.20,21

Our findings are compatible with a direct role of cellular hypoxia in the development of organ failure. Other recent studies related blood lactate levels to the development of or- gan failure and mortality following severe trauma.” Another report indicated that acutely ill surgical patients who de- velop postoperative complications had a more severe degree of estimated oxygen debt during the surgical procedure.”

Organ dysfunction following septic shock may stem from a complex interrelation between an excessive immune re- sponse and cellular hypoxia. Circulating levels of cytokines also have been related to the development of multiple or- gan failure and death.‘“J5 Cytokines can trigger a complex cascade of events susceptible to inducing cellular hypoxia by increasing the metabolic demand of the tissues, altering the oxygen extraction, and depressing myocardial contractility. Regional hypoxia may in turn activate the immune system, and lactic acid itself has also been found to activate human macrophages and to stimulate the release of cytokines.‘,”

Adequate oxygen supply should be provided to the tissues to prevent the development of organ failure following sep- tic shock. Rather than increasing oxygen supply to supra- normal levels in all critically ill patients,4,i we prefer to base the therapeutic options on measurable parameters. In this context, serial measurements of blood lactate levels may rep- resent a useful guide to therapy. The present study indicates

-**- r-*1

28 -i T

OFS l-2 OFS3-4 OFS5-6 OFS 7-8

OFS l-2 OFS3-4 OFS5-6 OFS7-8

-*- -*- d 1 1

OFSl-2 OFS3-4 OFS5-6 OFS7-8

Figure 3. Initial blood lactate level, time during which blood lac- tate exceeded 2.0 mmol/L (lactime), and area under the curve (AUC) according to the organ failure score (OFS) (mean + stan- dard error of the mean). l P ~0.05, **P ~0.01

that serial lactate measurements represent a helpful moni- toring systetn to estimate the risk of organ failure and death.

Surgeons continue to search for Rosetta-stone clues in the patho- genesis of multisystem organ failure, hoping to define high-risk groups that warrant special monitoring and/or therapy to present

the full-blown picture of m&system urgan failure. It is 1~) SW-

prise that elevated blood lactate levels correlate with a poor out-

THE AMERICAN JOURNAL OF SURGERY” VOLUME 171 FEBRUARY 1996 225

1 BLOOD LACTATE-

;

I

come; on the other hand, the emphusis on duration of hype&x- tatemia could be a useful clue.

REFERENCES 1. Parrillo JE, Parker MM, Natansnn C, et al. Septic shock in humans. Ann Intern Med. 1990;113:227-242. 2. Bihari D, Smithies M, Gimson A, Tinker J. The effects of vasodi- lation with prosracyclin cm oxygen delivery and uptake in critically 111 patients. NEIh4. 1987;317:397-403. 3. Vincent JL, Roman A, DrBacker D, Kahn RJ. Oxygen uptake/sup- ply dependency: effects of short-term dokutamine infusion. Am Rew Respir Dis. 1990;142:2-8. 4. Shoemaker WC, Appel PL, Kram HB, et al. Prospective trial of supranormal values of survivors as therapeutic goals in high-risk sur- gical patients. Chest. 1988;94:1176-1186. 5. Tuchschmidt J, Fried J, Astiz M, Rackow E. Elevation of cardiac output and oxygen delivery improves outcome in septic shock. Chest. 1992;102:216-220. 6. Deitch EA, Berg R, Specian R. Endotoxm promotes the tmnsloca- tion of bacteria from the gut. Arch Surg. 1987;122:185-192. 7. Maynard N, Bihari D, Beale R, et al. iissessment of splanchnic oxy- genation by gastric ronometry in patients with acute circulatory fail- ure. JAh4A. 1993;270:1203-1210. 8. Jensen JC, Buresh C, Norton JA. Lactic acidosis increases tumor necrosis factor secretion and transcription in vitro. J Surg Res. 1990;49:35&353. 9. Ghezzi I’, Dinarello CA, Bianchl M. Hypoxia increases IL-1 and TNF production by hurnan mononuclear cells. Crtokine. 1991;3:189-194. 10. Vincent JL, Berlot G. Cardiac effects of the mediators of sepsis. In: Lamy M, Thijs L, eds. Mediucors of Sepsis (Update cm Intense Care and Emtlr,qznq Medicine-16). New York: Springer-Verlag; 1992:255-266. 11. Bakker J, Coffernils M, Leon M, et al. Blood lactate levels are su- perior to oxygen derived variables in predicting outcome in human septic shock. Chest. 1991;99:956-962. 1.2. Vincent JL, Dufaye I’, Berre J, ct al. Serial lactate determinations during circulatory shock. Crit Care Med. 1983;11:449%451. 13. Weil MH, Afifi AA. Experimental and clinical studies on lactate

and pyruvate as indicators of the seventy of acute circulatory failure (shock). Circulation. 1970;41:989-1001. 14. Rashkin M, Boxkin C, Baughman R. Oxygen delivery in critically ill patients. Relatlonship to blood lactate and survival. Chest. 1985;87:580-584. 15. Gutierrez G, Dubin A. Cellular met&hsm in sepsis. In: Gutierrez G, Vincent JL, eds. Tissue Qq~en L;t&&m (U@ti in Intensive Cnre and Emgenq Medicine-l 2). New York: Springer-Verlag; 1991:227-241. 16. Vary TC, Siegel JH, Nakatani T, et al. Effect of sepsis on activity of pyruvate dehydrogenase complex in skeletal muscle and liver. .4m J Physiol. 1986;250:E634-E640. 17. Vincent JL, Roman A, Kahn RJ. Dobutamine administration in septic shock: addition to a standard protocol. Crit Care LMled. 1990;18: 689493. 18. Falk JL, Rackow EC, Leavy JL, et al. LIelayed lactate clearance in patients surviving circulatory shock. Acute Care. 1985;11:212-215. 19. Groeneveld ABJ, Kester ADM, Nduta JJP. Relation of arterial blood lactate to oxygen delivery and hemodynamic variables in hu- man shock states. Circ Shock. 1987;2?:?5%53. 20. Vincent JL, Gris I’, Coffemils M, et al. Myocardial depression char- acterizes the fatal course of septic shock. Surge?. 1992;111:660-667. 21. Marik PE. Gastric intramucosal pH. A better predictor of multi- organ dysfunction syndrome and death than oxygen-derived variables in patients with sepsis. Chest. 1993;104:225-229. 22. Roumen RM, Red1 H, Schlag G, et ,11. Scoring systems and blood lactate concentrations in relation to the development of adult respi- ratory distress syndrome and multlplc org,m failure in severely trau- matized patients. J Trauma. 1993;35:34%355. 23. Shoemaker WC, Appel PL, Kram Hf3. Tissue oxygen debt as a determinant of lethal and nonlethal postoperative organ failure. Crit CareMed. 1988;16:1117-1120. 24. Calandra T, Baumgartner JD, Grau I1G. et al. Prognostic values of tumor necrosis factor/cachectin, interleukm-1, interferon-alpha, and interferon-gamma in the serum of patients with septic shock. J Infect Dis. 1990;161:982-987. 25. Pinsky MR. Vincent JL, Deviere J. et al. Serum cytokine levels in human septic shock: relation to multiple-systems organ failure and mortality. Chest. 1993;103:565%575.

226 THE AMERICAN JOURNAL OF SURGEFW VOLUME 171 FEBRUARY 1996