- Email: [email protected]
The effect of sepsis and reduced colloid osmotic pressure on pulmonary edema
JOURNAL
OF SURGICAL
RESEARCH
27, 347-351 (1%‘)
The Effect of Sepsis and Reduced Colloid on Pulmonary Edema CHARLES Department
Osmotic
Pressure
L. RICE, M.D., PAUL M. DOROGHAZI, M.D., JOHN P. KOHLER, C,HRISTOPHER K. ZARINS, M.D., AND GERALD S. Moss, M.D. of Surgery,
Michael
Reese Hospital Submitted
and the University
for publication
of Chicago,
Chicago,
M.D.,
Illinois
60616
June 20, 1979
randomly assigned to one of three experimental groups: P-plasmapheresis alone, The interrelationship of hydrostatic presS-sepsis alone, P + S-both sepsis and sure and osmotic forces across the capillary plasmapheresis.Each animal was sedatedwith membrane are described by the Starling 0.1 mg/kg phencyclindme-HCl (Semylan) equation [lo]. Increasing pulmonary capilwith supplemental doses administered every lary hydrostatic pressure is a cause of l-2 hr during the study. A cuffed endopulmonary edema 131.Pulmonary edema detracheal tube was placed and the animals velops at a lower hydrostatic pressure when were ventilated with a volume ventilator plasma colloid oncotic pressure is lowered at 15 ml/kg delivered at a fractional oxygen suggesting that the gradient between colloid concentration of 0.5. The rate was adjusted oncotic pressure (COP) and pulmonary so that the arterial carbon dioxide was artery wedge pressure (PAW) is an im30-40 mm Hg and was not changed thereportant factor in the development of pulafter for the duration of the experiment. monary edema [3]. Decrease in plasma onIncisions were made in both groins under cotic pressure (COP) can also result in a local lidocaine anesthesia and catheters reduction in the COP-PAW gradient. Pulwere positioned in the thoracic aorta, inmonary edema has been attributed to the ferior vena cava and pulmonary artery reduction of the COP alone [7,8]. We have (Edwards Laboratories). The pulmonary previously shown that reduction of the COP artery wedge pressure and arterial preswithout elevation of the pulmonary artery sure were monitored using Statham P-23 wedge pressure does not result in pultransducers and a Brush 4-channel recorder. monary edema [14, 163. Blood gases were measured using the InSepsis is a common denominator in pastrumentation Laboratories Model 8 13 tients with respiratory failure [2] and it has blood gas analyzer. Hemoglobin and oxygen been suggested that altered pulmonary saturation were measured using the Instrucapillary permeability is a possible cause for mentation Laboratories 282 Cooximeter. the development of pulmonary edema [ 11. Intrapulmonary shunt was calculated using The purpose of this investigation is to the standard shunt equation [13]. The coldetermine the effect of both reduced COP loid oncotic pressure was measured by a and sepsis on the development of pultransducer membrane system (Instrumentamonary edema and pulmonary dysfunction tion Laboratories Model 186). in baboons. Sepsis was induced by the intravenous infusion of live Escherichia coli organisms MATERIALS AND METHODS over a 2-hr period. The E. coli were isolated Nine male baboons (Pupio anubis) weighfrom a neonate who died of sepsis and were ing between 18 and 27 kg (mean 23 kg) were further subcultured in trypticase soy broth, INTRODUCTION
347
0022~4804/79/ 110347-05$0 1.00/o CopyrQht All rights
Q 1979 by Academic Press, Inc. of reproduction in any form reserved.
348
JOURNAL
OF SURGICAL
RESEARCH:
VOL. 27, NO. 5, NOVEMBER
1979
The organism count was adjusted by optical monary shunts between groups and the coldensity measurement and organisms were loid oncotic pressure and the pulmonary infused at a rate of 2.7 to 5.4 x IO9 or- artery wedge pressure. The multiple linear ganisms per hour. regression analysis of variance was used to Plasmapheresis was accomplished by re- determine the factors influencing the intramoval of 25% of estimated blood volume pulmonary shunt. and reinfusion of packed red blood cells while pulmonary artery wedge pressure was RESULTS maintained at a constant level by the intravenous infusion of lactated Ringer’s solution. Blood cultures were positive at 24 hr in all animals that received E. cofi infusion. In Plasmapheresis was repeated three times. During the first 2 hr of the experiment, Group S, one of three animals died within 24 E. coli were infused in Groups S and P + S hr while in Groups P + S, two animals while Group P received an infusion of died within 24 hr. normal saline. During the second 2 hr, Pulmonary artery wedge pressure reGroups P and P + S underwent plasma- mained unchanged throughout the course pheresis while Group S received a saline of the experiment (Fig. 1). infusion to maintain pulmonary artery Plasmapheresis alone (Group P) caused a wedge pressure. All animals were observed 47% decrease in COP. In the group with for a final 4 hr. sepsis and plasmapheresis (Group S and P), After completion of the experiment the there was a 60% reduction. In the group animals were returned to their cages. Blood that was septic alone (Group S) there was a cultures were taken at 24 hr and the ani- 32% decrease of the COP without plasmamals were sacrificed. The lungs were re- pheresis (Fig. 2). moved immediately and dried in a hot air The COP-PAW gradient in the group oven at 55°C until the weight was stable and with plasmapheresis alone (Group P) dethe wet-to-dry lung weight ratio was cal- creased from 12.7 + 4.5 to 1.2 + 1.7 mm culated. COP, PAWP, COP-PAW, Q/et, Hg (P < 0.05). In the group with sepsis followed by plasmapheresis the COP-PAW A-aD02, and wet-to-dry lung weights were gradient was reduced from 14.7 rt 4.2 to compared using one way analysis of variance. The Wilcoxon’s rank sum test was 1.1 & 1.1 mm Hg (P < 0.05). The COP-PAW gradient fell from 17.3 used to determine differences in intrapul-
After
FIG. 1. The pulmonary experiment.
Infusion
After
Plasmapheresis
artery wedge pressure did not change in any group at any time during the
RICE ET AL.: SEPSIS AND REDUCED COP
Baseline
After
Infusion
After
349
Plarmapheresis
FIG. 2. The colloid oncotic pressure in mm Hg in the three groups plasmapheresis resulted in a significant reduction of colloid oncotic pressure.
+ 0.9 to 8.4 +- 3.2 mm Hg (P < 0.05) in the group with sepsis alone (Group S) (Fig. 3). While this reduction in Group S was significantly lower than the baseline, it was also higher than the COP-PAW gradients in the other two groups (P < 0.05). Intrapulmonary shunt did not change in any group during the course of the experiment (Fig. 4). Linear regression analysis revealed that the magnitude of the shunt was unrelated to the COP or COP-PAW gradient. The calculated wet-to-dry lung weight ratios were quite similar in all three groups (P: 3.5 k 0.5; S: 2.4 k 0.1; P and S: 3.3 k 0.7). When these data were examined
using the one-way analysis of variance, there was no significant differences between the groups. DISCUSSION The most common form of clinical pulmonary edema is that associated with left ventricular failure. It is due to an increase in the hydrostatic pressure as demonstrated by Guyton and Lindsey [3]. The hydrostatic pressure must be raised to a threshold pressure of 25 mm Hg before pulmonary edema ensued. Guyton and Lindsey [3] also found that if the level of plasma colloids
302520-75lo-
5-
Easeline
After
Infusion
After
Plasmapheresis
FIG. 3. The colloid oncotic pressure in pulmonary artery wedge pressure gradient was significantly reduced with Plasmapheresis.
350
JOURNAL OF SURGICAL RESEARCH: VOL. 27, NO. 5, NOVEMBER
Baseline
After
Infusion
After
1979
Plasmapheresls
FIG. 4. There was no significant change in Qs/et at any time in the experiment.
were reduced by half, the threshold pressure necessary for formation of pulmonary edema fell to 1I- 13 mm Hg. Reduced colloids did have an effect on the development of pulmonary edema, but it was still necessary for hydrostatic pressure to be raised above a threshold for pulmonary edema to occur. This observation of Guyton and Lindsey 131was applied to the treatment of respiratory failure by Skillman [9]. He advocated raising the colloid level with infusion of colloid, but there was no striking and consistent improvement in pulmonary function with this treatment. Mere reduction of the COP does not lead to pulmonary dysfunction or pulmonary edema. Zarins [ 161reduced the colloid oncotic pressure in baboons to near 0 mm Hg through plasmapheresis without the development of pulmonary edema or increase in the intrapulmonary shunt. In his experiment, the pulmonary artery wedge pressure was kept below 10 mm of mercury which is below the threshold of Guyton. Thus reduction of the colloid oncotic pressure without an increase in intravascular pressure does not lead to pulmonary edema. Morissette and Weil [7] and Rackow and Fein [8] have stated that reduction of the COP and the COP-PAW gradient is an important prognostic indicator for the development of pulmonary edema and subsequent
cardiopulmonary death. However, in their studies they used radiographic data, usually insensitive to early changes and not readily quantifiable, for the determination of pulmonary edema. Their colloid oncotic pressure determinations were made at unspecified times after admission to the ICU, and it is difficult to single out decreased COP as the primary predisposing cause of pulmonary edema. Brigham [l] showed that with the infusion of live Pseudomonas into sheep, there was a drop in the arterial oxygen tension and an increase in the pulmonary extravascular water measured noninvasively. However, at the end of the experiment, there was no gravimetric evidence for pulmonary edema with sepsis. This lack of clear evidence for pulmonary edema is reiterated in our experiment. The lung weight ratios were well below the normal baboon wet-dry lung weight of 5.5 for this laboratory. It is possible that interstitial pulmonary edema may have been detected by a more sensitive technique. This discrepancy with a strong clinical association of sepsis with respiratory failure and previously reported primate studies by Hinshaw [4] may be the result of differences in the strain of E. coli and in the dosage. At present, we cannot ascribe pulmonary dysfunction or pulmonary edema to sepsis. In the present experiment, there is no synergistic relationship between the ef-
RICE ET AL.: SEPSIS AND REDUCED COP
feet of sepsis and reduction of the colloid oncotic pressure. However, reduction of the colloid oncotic pressure and colloid oncotic pressure-pulmonary artery wedge gradient to extremely low levels did not result in pulmonary dysfunction. REFERENCES 1. Brigham, C. L., Woolverton, W. C., and Staub, N. C. Increased capillary vascular permeability after Pseudomonas aeruginosa bacteremia in unanesthetized sheep. Fed. Proc. 32: 440, 1973. 2. Fulton, R. L., and Jones, C. E. The cause of post-traumatic pulmonary insufficiency in man. Surg. Gynec. Obstet. 140: 179, 1975. 3. Guyton, A. C., and Lindsey, A. W. Effect of elevated left atrial pressure and decreased plasma protein concentration on the development of pulmonary edema. Circ. Res. 7: 649, 1959. 4. Hinshaw, L. B., Benjamin, B., Holmes, D., Beller, B., Archer, T., Coalson, J. J., and Whitsett, T. Responses of the baboon to live Escherichia coli organisms and endotoxin. Surg. Gynec. Obstet. 145: 1, 1977. 5. Holcroft, J. W., Blaisdell, F. W., Trunkey, D. D., and Lim, R. C. Intravascular coagulation in pulmonary edema in the septic baboon. J. Surg. Res. 22: 209, 1977. 6. Horowitz, D. K., Moquin, R. B., and Herman,
C. M. Coagulation changes of septic shock in the subhuman primate and the relationship to hemodynamic changes. Ann. Surg. 175: 417, 1972. 7. Morissette, M., Weil, M. H., and Shubin, H. Reduction in colloid osmotic pressure associated with
351
fatal progression of cardiopulmonary failure. Crit. Care Med. 3: 115, 1975. 8. Rackow, C., Fein, I. A., and Leppo, J. Colloid osmotic pressure as a prognostic indicator of pulmonary edema and mortality in the critically ill. Chest 72: 709, 1977. 9. Skillman, J. J., Parikh, B. M., and Tannenbaum, B. J. Pulmonary arteriovenous admixture: Improvement with albumin and diuretics. Amer. J. Surg. 119: 440, 1970.
10. Starling, E. H. On the absorption of fluids from the connective tissue spaces. .I. Physiol. (London) 19: 312, 1896.
11. Staub, N. C. Pulmonary edema. Physiol. Rev. 54: 678, 1974.
12. Staub, N. C., Nagano, J., and Pearce, M. L. Pulmonary edema in dogs, especially the sequence of fluid accumulation in the lungs. J. Appl. Physiol. 22: 227, 1%7. 13. Thal, A. B., Brown, E. B., Jr., Hermrack, A. S., and Bell, H. H. In Shock, Physiological Basis for Treatment. Chicago: Year Book Medical Publishers, 1971, P. 200. 14. Virgilio, R. W., Rice, C. L., Smith, D. E., James, D. R., Zarins, C. K., Hobelman, C. F., and Peters, R. M. Crystalloid vs colloid resuscitation is one better? Surgery 85: 179, 1979. 15. Virgilio, R. W., Smith, D. E., Rice, C. L., Hobelman, C. L., Zarins, C. K., James, D. R., and Peters, R. M. Effect of colloid osmotic pressure and pulmonary capillary wedge pressure on intrapulmonary shunt. Surg. Forum 27: 168, 1976. 16. Zarins, C. K., Rice, C. L., Peters, R. M., and Virgilio, R. W. Lymph and pulmonary response to isobaric reduction in plasma oncotic pressure. Circ. Res. 43: 925, 1978.
OF SURGICAL
RESEARCH
27, 347-351 (1%‘)
The Effect of Sepsis and Reduced Colloid on Pulmonary Edema CHARLES Department
Osmotic
Pressure
L. RICE, M.D., PAUL M. DOROGHAZI, M.D., JOHN P. KOHLER, C,HRISTOPHER K. ZARINS, M.D., AND GERALD S. Moss, M.D. of Surgery,
Michael
Reese Hospital Submitted
and the University
for publication
of Chicago,
Chicago,
M.D.,
Illinois
60616
June 20, 1979
randomly assigned to one of three experimental groups: P-plasmapheresis alone, The interrelationship of hydrostatic presS-sepsis alone, P + S-both sepsis and sure and osmotic forces across the capillary plasmapheresis.Each animal was sedatedwith membrane are described by the Starling 0.1 mg/kg phencyclindme-HCl (Semylan) equation [lo]. Increasing pulmonary capilwith supplemental doses administered every lary hydrostatic pressure is a cause of l-2 hr during the study. A cuffed endopulmonary edema 131.Pulmonary edema detracheal tube was placed and the animals velops at a lower hydrostatic pressure when were ventilated with a volume ventilator plasma colloid oncotic pressure is lowered at 15 ml/kg delivered at a fractional oxygen suggesting that the gradient between colloid concentration of 0.5. The rate was adjusted oncotic pressure (COP) and pulmonary so that the arterial carbon dioxide was artery wedge pressure (PAW) is an im30-40 mm Hg and was not changed thereportant factor in the development of pulafter for the duration of the experiment. monary edema [3]. Decrease in plasma onIncisions were made in both groins under cotic pressure (COP) can also result in a local lidocaine anesthesia and catheters reduction in the COP-PAW gradient. Pulwere positioned in the thoracic aorta, inmonary edema has been attributed to the ferior vena cava and pulmonary artery reduction of the COP alone [7,8]. We have (Edwards Laboratories). The pulmonary previously shown that reduction of the COP artery wedge pressure and arterial preswithout elevation of the pulmonary artery sure were monitored using Statham P-23 wedge pressure does not result in pultransducers and a Brush 4-channel recorder. monary edema [14, 163. Blood gases were measured using the InSepsis is a common denominator in pastrumentation Laboratories Model 8 13 tients with respiratory failure [2] and it has blood gas analyzer. Hemoglobin and oxygen been suggested that altered pulmonary saturation were measured using the Instrucapillary permeability is a possible cause for mentation Laboratories 282 Cooximeter. the development of pulmonary edema [ 11. Intrapulmonary shunt was calculated using The purpose of this investigation is to the standard shunt equation [13]. The coldetermine the effect of both reduced COP loid oncotic pressure was measured by a and sepsis on the development of pultransducer membrane system (Instrumentamonary edema and pulmonary dysfunction tion Laboratories Model 186). in baboons. Sepsis was induced by the intravenous infusion of live Escherichia coli organisms MATERIALS AND METHODS over a 2-hr period. The E. coli were isolated Nine male baboons (Pupio anubis) weighfrom a neonate who died of sepsis and were ing between 18 and 27 kg (mean 23 kg) were further subcultured in trypticase soy broth, INTRODUCTION
347
0022~4804/79/ 110347-05$0 1.00/o CopyrQht All rights
Q 1979 by Academic Press, Inc. of reproduction in any form reserved.
348
JOURNAL
OF SURGICAL
RESEARCH:
VOL. 27, NO. 5, NOVEMBER
1979
The organism count was adjusted by optical monary shunts between groups and the coldensity measurement and organisms were loid oncotic pressure and the pulmonary infused at a rate of 2.7 to 5.4 x IO9 or- artery wedge pressure. The multiple linear ganisms per hour. regression analysis of variance was used to Plasmapheresis was accomplished by re- determine the factors influencing the intramoval of 25% of estimated blood volume pulmonary shunt. and reinfusion of packed red blood cells while pulmonary artery wedge pressure was RESULTS maintained at a constant level by the intravenous infusion of lactated Ringer’s solution. Blood cultures were positive at 24 hr in all animals that received E. cofi infusion. In Plasmapheresis was repeated three times. During the first 2 hr of the experiment, Group S, one of three animals died within 24 E. coli were infused in Groups S and P + S hr while in Groups P + S, two animals while Group P received an infusion of died within 24 hr. normal saline. During the second 2 hr, Pulmonary artery wedge pressure reGroups P and P + S underwent plasma- mained unchanged throughout the course pheresis while Group S received a saline of the experiment (Fig. 1). infusion to maintain pulmonary artery Plasmapheresis alone (Group P) caused a wedge pressure. All animals were observed 47% decrease in COP. In the group with for a final 4 hr. sepsis and plasmapheresis (Group S and P), After completion of the experiment the there was a 60% reduction. In the group animals were returned to their cages. Blood that was septic alone (Group S) there was a cultures were taken at 24 hr and the ani- 32% decrease of the COP without plasmamals were sacrificed. The lungs were re- pheresis (Fig. 2). moved immediately and dried in a hot air The COP-PAW gradient in the group oven at 55°C until the weight was stable and with plasmapheresis alone (Group P) dethe wet-to-dry lung weight ratio was cal- creased from 12.7 + 4.5 to 1.2 + 1.7 mm culated. COP, PAWP, COP-PAW, Q/et, Hg (P < 0.05). In the group with sepsis followed by plasmapheresis the COP-PAW A-aD02, and wet-to-dry lung weights were gradient was reduced from 14.7 rt 4.2 to compared using one way analysis of variance. The Wilcoxon’s rank sum test was 1.1 & 1.1 mm Hg (P < 0.05). The COP-PAW gradient fell from 17.3 used to determine differences in intrapul-
After
FIG. 1. The pulmonary experiment.
Infusion
After
Plasmapheresis
artery wedge pressure did not change in any group at any time during the
RICE ET AL.: SEPSIS AND REDUCED COP
Baseline
After
Infusion
After
349
Plarmapheresis
FIG. 2. The colloid oncotic pressure in mm Hg in the three groups plasmapheresis resulted in a significant reduction of colloid oncotic pressure.
+ 0.9 to 8.4 +- 3.2 mm Hg (P < 0.05) in the group with sepsis alone (Group S) (Fig. 3). While this reduction in Group S was significantly lower than the baseline, it was also higher than the COP-PAW gradients in the other two groups (P < 0.05). Intrapulmonary shunt did not change in any group during the course of the experiment (Fig. 4). Linear regression analysis revealed that the magnitude of the shunt was unrelated to the COP or COP-PAW gradient. The calculated wet-to-dry lung weight ratios were quite similar in all three groups (P: 3.5 k 0.5; S: 2.4 k 0.1; P and S: 3.3 k 0.7). When these data were examined
using the one-way analysis of variance, there was no significant differences between the groups. DISCUSSION The most common form of clinical pulmonary edema is that associated with left ventricular failure. It is due to an increase in the hydrostatic pressure as demonstrated by Guyton and Lindsey [3]. The hydrostatic pressure must be raised to a threshold pressure of 25 mm Hg before pulmonary edema ensued. Guyton and Lindsey [3] also found that if the level of plasma colloids
302520-75lo-
5-
Easeline
After
Infusion
After
Plasmapheresis
FIG. 3. The colloid oncotic pressure in pulmonary artery wedge pressure gradient was significantly reduced with Plasmapheresis.
350
JOURNAL OF SURGICAL RESEARCH: VOL. 27, NO. 5, NOVEMBER
Baseline
After
Infusion
After
1979
Plasmapheresls
FIG. 4. There was no significant change in Qs/et at any time in the experiment.
were reduced by half, the threshold pressure necessary for formation of pulmonary edema fell to 1I- 13 mm Hg. Reduced colloids did have an effect on the development of pulmonary edema, but it was still necessary for hydrostatic pressure to be raised above a threshold for pulmonary edema to occur. This observation of Guyton and Lindsey 131was applied to the treatment of respiratory failure by Skillman [9]. He advocated raising the colloid level with infusion of colloid, but there was no striking and consistent improvement in pulmonary function with this treatment. Mere reduction of the COP does not lead to pulmonary dysfunction or pulmonary edema. Zarins [ 161reduced the colloid oncotic pressure in baboons to near 0 mm Hg through plasmapheresis without the development of pulmonary edema or increase in the intrapulmonary shunt. In his experiment, the pulmonary artery wedge pressure was kept below 10 mm of mercury which is below the threshold of Guyton. Thus reduction of the colloid oncotic pressure without an increase in intravascular pressure does not lead to pulmonary edema. Morissette and Weil [7] and Rackow and Fein [8] have stated that reduction of the COP and the COP-PAW gradient is an important prognostic indicator for the development of pulmonary edema and subsequent
cardiopulmonary death. However, in their studies they used radiographic data, usually insensitive to early changes and not readily quantifiable, for the determination of pulmonary edema. Their colloid oncotic pressure determinations were made at unspecified times after admission to the ICU, and it is difficult to single out decreased COP as the primary predisposing cause of pulmonary edema. Brigham [l] showed that with the infusion of live Pseudomonas into sheep, there was a drop in the arterial oxygen tension and an increase in the pulmonary extravascular water measured noninvasively. However, at the end of the experiment, there was no gravimetric evidence for pulmonary edema with sepsis. This lack of clear evidence for pulmonary edema is reiterated in our experiment. The lung weight ratios were well below the normal baboon wet-dry lung weight of 5.5 for this laboratory. It is possible that interstitial pulmonary edema may have been detected by a more sensitive technique. This discrepancy with a strong clinical association of sepsis with respiratory failure and previously reported primate studies by Hinshaw [4] may be the result of differences in the strain of E. coli and in the dosage. At present, we cannot ascribe pulmonary dysfunction or pulmonary edema to sepsis. In the present experiment, there is no synergistic relationship between the ef-
RICE ET AL.: SEPSIS AND REDUCED COP
feet of sepsis and reduction of the colloid oncotic pressure. However, reduction of the colloid oncotic pressure and colloid oncotic pressure-pulmonary artery wedge gradient to extremely low levels did not result in pulmonary dysfunction. REFERENCES 1. Brigham, C. L., Woolverton, W. C., and Staub, N. C. Increased capillary vascular permeability after Pseudomonas aeruginosa bacteremia in unanesthetized sheep. Fed. Proc. 32: 440, 1973. 2. Fulton, R. L., and Jones, C. E. The cause of post-traumatic pulmonary insufficiency in man. Surg. Gynec. Obstet. 140: 179, 1975. 3. Guyton, A. C., and Lindsey, A. W. Effect of elevated left atrial pressure and decreased plasma protein concentration on the development of pulmonary edema. Circ. Res. 7: 649, 1959. 4. Hinshaw, L. B., Benjamin, B., Holmes, D., Beller, B., Archer, T., Coalson, J. J., and Whitsett, T. Responses of the baboon to live Escherichia coli organisms and endotoxin. Surg. Gynec. Obstet. 145: 1, 1977. 5. Holcroft, J. W., Blaisdell, F. W., Trunkey, D. D., and Lim, R. C. Intravascular coagulation in pulmonary edema in the septic baboon. J. Surg. Res. 22: 209, 1977. 6. Horowitz, D. K., Moquin, R. B., and Herman,
C. M. Coagulation changes of septic shock in the subhuman primate and the relationship to hemodynamic changes. Ann. Surg. 175: 417, 1972. 7. Morissette, M., Weil, M. H., and Shubin, H. Reduction in colloid osmotic pressure associated with
351
fatal progression of cardiopulmonary failure. Crit. Care Med. 3: 115, 1975. 8. Rackow, C., Fein, I. A., and Leppo, J. Colloid osmotic pressure as a prognostic indicator of pulmonary edema and mortality in the critically ill. Chest 72: 709, 1977. 9. Skillman, J. J., Parikh, B. M., and Tannenbaum, B. J. Pulmonary arteriovenous admixture: Improvement with albumin and diuretics. Amer. J. Surg. 119: 440, 1970.
10. Starling, E. H. On the absorption of fluids from the connective tissue spaces. .I. Physiol. (London) 19: 312, 1896.
11. Staub, N. C. Pulmonary edema. Physiol. Rev. 54: 678, 1974.
12. Staub, N. C., Nagano, J., and Pearce, M. L. Pulmonary edema in dogs, especially the sequence of fluid accumulation in the lungs. J. Appl. Physiol. 22: 227, 1%7. 13. Thal, A. B., Brown, E. B., Jr., Hermrack, A. S., and Bell, H. H. In Shock, Physiological Basis for Treatment. Chicago: Year Book Medical Publishers, 1971, P. 200. 14. Virgilio, R. W., Rice, C. L., Smith, D. E., James, D. R., Zarins, C. K., Hobelman, C. F., and Peters, R. M. Crystalloid vs colloid resuscitation is one better? Surgery 85: 179, 1979. 15. Virgilio, R. W., Smith, D. E., Rice, C. L., Hobelman, C. L., Zarins, C. K., James, D. R., and Peters, R. M. Effect of colloid osmotic pressure and pulmonary capillary wedge pressure on intrapulmonary shunt. Surg. Forum 27: 168, 1976. 16. Zarins, C. K., Rice, C. L., Peters, R. M., and Virgilio, R. W. Lymph and pulmonary response to isobaric reduction in plasma oncotic pressure. Circ. Res. 43: 925, 1978.