Secondary Logo

Journal Logo

Original Article

Hypoxaemia after cardiac surgery: clinical application of a model of pulmonary gas exchange

Kjærgaard, S.*; Rees, S. E.; Grønlund, J.; Nielsen, E. M.*; Lambert, P.*; Thorgaard, P.*; Toft, E.; Andreassen, S.

Author Information
European Journal of Anaesthesiology: April 2004 - Volume 21 - Issue 4 - p 296-301


Pulmonary dysfunction leading to significant arterial hypoxaemia is well documented after coronary artery bypass grafting (CABG) [1-3], and occurs in 30-60% of patients after otherwise uncomplicated CABG. Hypoxaemia may be caused by the effects of surgery and anaesthesia upon breathing patterns and pulmonary gas exchange [2], and may contribute to postoperative complications [4-7]. Therefore, we need a method that can be applied in daily clinical practice to describe hypoxaemia.

Patients with abnormal breathing patterns and impaired pulmonary gas exchange may present with almost normal arterial oxygen saturation (SaO2) on breathing air. However, the combination of abnormal breathing patterns and impaired pulmonary gas exchange may lead to episodes of severe arterial hypoxaemia, e.g. provoked by sleep apnoea [8]. Therefore, single measurements of SaO2 do not provide sufficient detail to decide on specific therapeutic interventions, e.g. prolonged oxygen supplementation.

Performing an oxygen titration test enables plotting SaO2 against end-tidal oxygen fraction (FEO2). The shape and position of the FEO2/SaO2 curve can be quantified by fitting a mathematical model of pulmonary gas exchange with two parameters, shunt and ventilation/perfusion (V/Q) mismatch, to the data [1,9-12]. The distinction between shunt and V/Q mismatch may be important, because V/Q mismatch causes a rightward shift of the FEO2/SaO2 curve which implies an increased vulnerability to decreases in alveolar oxygen fraction.

This study investigated patients 2-4 h after CABG, and then on days 2, 3 and 7 after surgery. At each time point an FEO2/SaO2 curve was obtained, and shunt and V/Q mismatch were estimated. The aim was to investigate whether the method can be applied in a clinical setting, and if changes in pulmonary gas exchange after CABG can be described by a rightward shift of the FEO2/SaO2 curve, i.e. increased V/Q mismatch, or shunt.


Ethics approval for the study was obtained from the Ethics Committee of North Jutland and Viborg Counties. Informed written and oral consent was obtained from all patients. Fourteen patients, thirteen males and one female, scheduled for CABG, were recruited (preoperative data are given in Table 1). Inclusion criteria were: scheduled heart surgery requiring extracorporeal circulation, age >18 yr, a preoperative left ventricular ejection fraction >0.5, normal preoperative chest radiograph. None of the patients included in the study required medication for pre-existing chronic pulmonary disease. Postoperative exclusion criteria were: cardiac index <2.0 L min−1 m−2, mixed venous oxygen saturation <60%, signs of myocardial ischaemia, haematocrit <0.25, unstable acid-base status (i.e. pH < 7.3 or pH > 7.45 or base excess < −5 mmol L−1), complicated surgery and anaesthesia (e.g. bleeding leading to reoperation, need for continuous inotropic support, postoperative respiratory distress syndrome).

Table 1
Table 1:
Preoperative patient characteristics.

All patients received morphine and scopolamine (SAD, Copenhagen, Denmark) intramuscularly as premedication. Anaesthesia was induced with midazolam 0.03-0.1 mg kg−1 (Roche, Basel, Switzerland) and fentanyl 7.5-12.5 μg kg−1 (Janssen Pharmaceutical, Beerse, Belgium). Endotracheal intubation was facilitated by pancuronium (Organon Teknika, Boxtel, Holland). Anaesthesia was maintained with enflurane (Abbott Laboratories, Gentofte, Denmark) and supplemental doses of fentanyl 2-5 μg kg−1 h−1 and midazolam 0.1 mg kg−1 h−1. The lungs were re-expanded after extracorporeal circulation by four to six inflations to an airway pressure of 30 cmH2O for 30 s. All patients were ventilated postoperatively (Servo® 900C; Siemens, Solna, Sweden) with a positive end-expiratory pressure (PEEP) of 5 cmH2O. Propofol (Leiras, Turku, Finland) was used for postoperative sedation in all cases. Morphine and acetaminophen (paracetamol) were given for postoperative analgesia. Surgery was uncomplicated. All patients received internal mammary artery grafting, and 12 patients received additional saphenous vein grafts. Cardioplegia was a 4:1 mixture of blood and St. Thomas' Hospital solution with KCl 100 mmol L−1 added. Intraoperative data are given in Table 2.

Table 2
Table 2:
Intraoperative patient characteristics.

Routine monitoring of haemodynamics during surgery and in the intensive care unit (ICU) included a thermodilution pulmonary artery catheter and an arterial catheter. After tracheal extubation in the ICU all patients were given intermittent continuous positive airway pressure (CPAP) and humidified oxygen supplementation during the first 2 postoperative days. All chest tubes were removed during the first 24 h postoperatively.

To enable plotting of the FEO2/SaO2 curve, FiO2 was lowered and subsequently increased in four to six steps to achieve SaO2 in the range from 90% to 100%. After each change in FiO2, 5 min were allowed for oxygen equilibration [13]. In order to ensure a stable inspired oxygen concentration, patients' lungs were either ventilated via a lung ventilator or breathing through a tightly fitting facemask with a three-way valve separating inspired and expired gases. Nitrogen mixed with air was used when it proved necessary to deliver subatmospheric oxygen fractions in order to lower SaO2 as indicated above.

The patients were investigated at four time points: 2-4 h after surgery, when they were still intubated and ventilated, and then on days 2, 3 and 7 after surgery. Eight patients were studied on all four occasions. Six patients had been discharged from the hospital before postoperative day 7 and only completed the first three measurements.

Measurements were performed at each FiO2 level after equilibration. The measurements included: minute ventilation, using values either from the respirator (Servo® 900C) or a volumemeter (Elkro Gas, Salerno, Italy); respiratory frequency recorded by side-stream capnography (average over 3 min); end-expired carbon dioxide fraction (FeCO2), FiO2, FEO2 (Datex AS-3®; Datex-Engström, Helsinki, Finland). In addition, SaO2 was measured either from arterial blood samples at each FiO2 level (at 2-4 h after surgery), or from pulse oximetry (SPO2) (Datex AS-3®), supplemented by a single arterial blood sample (SaO2), which was used to calibrate the pulse oximeter. The following calibration procedure was used: for each patient a single graph of SPO2 vs. SaO2 was constructed for the data measured on all occasions. Calibration was performed by adjusting the pulse oximetry readings using a linear regression fit to the SPO2/SaO2 data. In all cases blood samples were analysed to obtain blood-gases, acid-base status and concentrations of haemoglobins (pH, PaO2, PaCO2, base excess, Hb, methaemoglobin and CO-haemoglobin) (Radiometer ABL 525®; Copenhagen, Denmark).

Oxygen consumption (O2) was calculated according to the Fick equation (O2 = Q(CaO2 − CmvO2)). CaO2 and CmvO2 represent oxygen content in arterial and mixed venous blood, respectively. Cardiac output (Q) at 2-4 h after surgery was determined by the thermodilution method as the mean of four valid measurements. For the experiments performed when the pulmonary artery catheter had been removed (days 2, 3 and 7 after surgery), a Q = 5 L min−1 was assumed [14]. Then, O2 was calculated from the measurements of respiratory frequency (f), tidal volume (VT), and mixed expired oxygen fraction (FĒO2) taken at FiO2 = 0.21 using a Douglas bag, i.e. O2 = fVT (FiO2 − FĒO2). Anatomical dead space was assumed to be 150 mL when the patients were intubated, and 200 mL when breathing through the face mask [15].

Measurements of FEO2 were subsequently plotted against the corresponding SaO2 values, enabling construction of an FEO2/SaO2 curve. The measurements of ventilation, circulation and arterial blood were inserted into the equations of the mathematical model of oxygen transport published previously [12]. This model includes two parameters describing gas exchange, pulmonary shunt (shunt) and a measure of V/Q mismatch (fA2). 'fA2' is the fraction of ventilation to an alveolar compartment that is perfused by 90% of the non-shunted blood flow, i.e. fA2 = 0.9 indicates a perfect match between ventilation and perfusion. This parameter can be used to calculate the effects of the V/Q mismatch on ΔPO2, defined as the difference in oxygen pressure between gas leaving the mixed alveolar compartment (PaO2) and the oxygen pressure in the lung capillary blood before mixing with shunted blood (PcO2). We will use ΔPO2 rather than fA2, to quantify the severity of the V/Q mismatch, since ΔPO2 is a direct measure of the right-ward shift of the FEO2/SaO2 curve. Shunt and ΔPO2 were estimated for each patient on each measurement occasion so as to fit the measured SaO2 (2-4 h after surgery) or calibrated SPO2 (days 2, 3 and 7 after surgery) using the weighted least squares method.

All continuous variables are reported as medians and ranges. P values were obtained from non-parametric ANOVA.


Comparing the measurements of arterial oxygenation using either an arterial blood sample (SaO2) or the pulse oximeter (SPO2) by a linear regression fit across all patients gives a correlation coefficient of r = 0.87.

SaO2 (at FiO2 = 0.21) decreased significantly during postoperatives days 2 (P < 0.05) and 3 (P < 0.05) (Table 3). Arterial hypoxaemia (SaO2 < 92%) [7] was found in 11 patients in that time period as assessed by single measurements of SaO2 at FiO2 = 0.21. On postoperative day 7, SaO2 had improved significantly (P < 0.05), as compared to postoperative day 3.

Table 3
Table 3:
Shunt, ΔPO2 and SaO2 at FiO2 = 0.21 for 14 patients 2-4 h, days 2, 3 and 7 after surgery.

Measurements of arterial oxygen saturation (SaO2) were plotted against the end-tidal oxygen fraction (FEO2). Figure 1 illustrates the plots of SaO2 against FEO2 for patients on postoperative day 3 (crosses) and day 7 (circles). Curves illustrate the fit of the two-parameter model of oxygen transport [12] to data. The shift to the right on day 3 after surgery indicates desaturation of arterial blood at higher values of FEO2. In Table 3, shunt and ΔPO2 according to the model is given for all 14 patients at all four time points. Pulmonary shunt increased slightly during the study period, but this did not reach statistical significance.

Figure 1
Figure 1:
Measured arterial oxygen saturation (SaO2) plotted against the end-tidal oxygen fraction (FEO2) 3 days (+) and 7 days after surgery (○) for patient 6. Values for shunt and ΔPO2 at the two time points are also given.

ΔPO2 increased significantly from the measurement 2-4 h after surgery to postoperative days 2 (P < 0.01) and 3 (P < 0.01). On the postoperative day 7, ΔPO2 decreased significantly compared to the value on day 3 after surgery (P < 0.01).

The correlation between ΔPO2 and SaO2 measured at FiO2 = 0.21 was found to be significant (P < 0.05), however with only a modest coefficient of correlation, r = 0.62.

The changes in ΔPO2 were not related to clinical data such as body mass index (BMI), age, NYHA class, intraoperative fluid gain, time on extracorporeal circulation, chest radiography, pulmonary capillary wedge pressure or cardiac output.


In the present study we performed an oxygen titration test by lowering the FiO2 and monitoring SaO2 in postoperative patients after CABG. This procedure decreases alveolar oxygen fraction, and allows plotting of the FEO2/SaO2 curve. It has been shown previously that the shape and position of this curve is affected by disorders in pulmonary gas exchange [1,10]. We found that the patient group as a whole had increased values of shunt and ΔPO2. This is consistent with findings in other studies [16,17]. Values of shunt were, however, only moderately increased and gave only a minor contribution to the oxygenation problem in most cases. The shunt did not change significantly during the course of the study. ΔPO2 quantifies the rightward shift of the FEO2/SaO2 curve, caused by V/Q mismatch, and increased significantly on days 2 and 3.

Decreased oxygenation is well documented after CABG [2,18], and may be due to impaired gas exchange in combination with abnormal breathing patterns. Impaired gas exchange may be related to activation of an inflammatory response caused by the surgical trauma and extracorporeal circulation leading to pulmonary capillary leakage [19-21], continued atelectasis and airway closure [22], loss of surfactant during surgery [23] or diaphragmatic dysfunction [24]. Atelectases are frequently seen after CABG [15], especially in dependent lung regions and in particular of the left lower lobe [22]. It has been shown that the area of the atelectatic lung region correlates significantly with increased shunt fraction [25]. Breathing patterns are abnormal in the days after surgery and anaesthesia when rebound of rapid eye movement sleep (REM-sleep) is common. This correlates with the occurrence of frequent sleep apnoeas [26].

Apnoeas may lead to episodes of severe arterial hypoxaemia. This can be conceptualized by the following examples: assuming normal values of O2 = 300 mL min−1 and lung volume (FRC) = 3000 mL, the alveolar oxygen fraction will fall by 1% every 6 s during apnoea. This means that an apnoea with a relatively modest duration of 30 s will reduce the alveolar oxygen fraction by 5%. The effects of this reduction are illustrated in Figure 2, where the model has been used to simulate the effects of shunt and V/Q mismatch on SaO2 in three typical cases: Case 1 gives the normal situation (shunt = 3%, ΔPO2 = 0 kPa). Case 2 (shunt = 12%, ΔPO2 = 0 kPa) illustrates that an increase in shunt shifts the FEO2/SaO2 curve downwards over a wide range of FEO2. A modest reduction in FEO2 from 0.20 to 0.15 will result in a drop in SaO2 from 95.8% to 94.3% in this case. Case 3 (shunt = 3%, ΔPO2 = 8.3 kPa) shows that increased V/Q mismatch shifts the FEO2/SaO2 curve to the right by an amount equal to ΔPO2. The same reduction in FEO2 as in Case 2 will result in a drop in SaO2 from 95.8% to 84.7%. It is not possible to simulate the right shift of the FEO2/SaO2 curves using models that only use shunt to explain hypoxaemia [1].

Figure 2
Figure 2:
Arterial oxygen saturation (SaO2) as a function of the end-tidal oxygen fraction (FEO2) calculated using the model for three hypothetical cases: 1, normal subject with shunt = 3% and ΔPO2 = 0 kPa (――――); 2, patient with shunt = 12% and ΔPO2 = 0 kPa (―); 3, patient with shunt = 3% and ΔPO2 = 8.3 kPa (– –). A cardiac output of 5 L min−1 and an oxygen consumption of 0.25 L min−1 were assumed for the calculations. The effects of lowering FEO2 from 0.24 to 0.18 are illustrated with circles and values of calculated SaO2 for each case.

This means that the episodic hypoxaemia observed in postoperative patients with rebound of REM-sleep and associated sleep apnoeas [26] may be due to the increased vulnerability to hypoxaemia in patients with postoperatively increased ΔPO2. Patients may have a clinically significant disorder in pulmonary gas exchange, but may nevertheless present with an almost normal SaO2 when breathing air [27], because of compensatory hyperventilation. It is impossible to distinguish between increased shunt, which does not put the patient at risk of episodic hypoxaemia, and increased ΔPO2, which may represent a risk, even in patients with a reduced SaO2. Therefore, measurement of SaO2 at FiO2 = 0.21 can only give an indication of the oxygenation problem. If the oxygenation problem is described using shunt and ΔPO2 as in this study, the compensatory changes seen in ventilatory volume are irrelevant, since this method is independent of changes in ventilatory pattern.

The method presented here for estimation of shunt and ΔPO2 can be made with equipment routinely available in departments of anaesthesia and intensive care medicine. The experiment can be performed quickly since oxygen equilibrates on average 2-3 min after changes in FiO2[13,28]. Studies including four measurement points, where arterial blood oxygen saturation is measured using pulse oximetry, can therefore be performed in approximately 10-15 min [28]. The need for adding nitrogen to the gas mixture might represent a practical problem. However, subatmospheric oxygen fractions are only required in cases with non-significant disorders in pulmonary gas exchange, in this study for the measurements 7 days after surgery.

Further studies may establish this method as a noninvasive clinical tool for identifying patients at risk of postoperative hypoxaemia before sending them from the post-anaesthesia care unit to the surgical ward where continuous monitoring and oxygen supplementation are less routinely used. In conclusion, we have described oxygenation in 14 patients 2-4 h and then 2, 3 and 7 days after uncomplicated CABG. Significantly decreased oxygenation was found on postoperative days 2 and 3. Oxygenation improved on the seventh postoperative day. By performing a simple oxygen titration experiment and plotting an FEO2/SaO2 curve it was possible to estimate the parameters of a mathematical model of pulmonary gas exchange using shunt and ΔPO2. ΔPO2 quantifies the rightward shift of the FEO2/SaO2 curves seen in these patients and hence describes the vulnerability to decreases in alveolar oxygen fraction. The mathematical model may as such find clinical application, since ΔPO2 may serve as an index of the risk of hypoxaemic events after CABG.


This work was partially supported by grants awarded by the Danish Heart Foundation, the Danish Research Academy under the DANVIS program and by the IT-committee under the Danish Technical Research Council. Thanks to Charlie Pedersen, Medical Technician, Department of Neonatology, Aalborg Hospital, for his invaluable assistance.


1. Andreassen S, Rees SE, Kjærgaard S, et al. Hypoxemia after coronary bypass surgery modeled by resistance to oxygen diffusion. Crit Care Med 1999; 27: 2445-2453.
2. Taggart DP, el Fiky M, Carter R, Bowman A, Wheatley DJ. Respiratory dysfunction after uncomplicated cardiopulmonary bypass. Ann Thorac Surg 1993; 56: 1123-1128.
3. Hachenberg T, Tenling A, Nystrom SO, Tyden H, Hedenstierna G. Ventilation-perfusion inequality in patients undergoing cardiac surgery. Anesthesiology 1994; 80: 509-519.
4. Aakerlund LP, Rosenberg J. Postoperative delirium: treatment with supplementary oxygen. Br J Anaesth 1994; 72: 286-290.
5. Gill NP, Wright B, Reilly CS. Relationship between hypoxaemic and cardiac ischaemic events in the perioperative period. Br J Anaesth 1992; 68: 471-473.
6. Jönsson K, Goodson WH, Scheuenstuhl H, West J, Hopf HW, Hunt TK. Tissue oxygenation, anemia, and perfusion in relation to wound healing in surgical patients. Ann Surg 1991; 214: 605-613.
7. Greif R, Akca O, Horn EP, Kurz, A, Sessler DI. Supplemental perioperative oxygen to reduce the incidence of surgical-wound infection. New Engl J Med 2000; 342: 161-167.
8. Rosenberg J, Wildschiødtz G, Pedersen MH, von Jessen F, Kehlet H. Late postoperative nocturnal episodic hypoxaemia and associated sleep pattern. Br J Anaesth 1994; 72: 145-150.
9. Roe PG, Gadelrab R, Sapsford D, Jones JG. Intra-operative gas exchange and post-operative hypoxaemia. Eur J Anaesthesiol 1997; 14: 203-210.
10. Sapsford DJ, Jones JG. The PIO2 vs. SPO2 diagram: a noninvasive measure of pulmonary oxygen exchange. Eur J Anaesthesiol 1995; 12: 375-386.
11. de Gray L, Rush EM, Jones JG. A noninvasive method for evaluating the effect of thoracotomy on shunt and ventilation perfusion inequality. Anaesthesia 1997; 52: 630-635.
12. Kjærgaard S, Rees SE, Nielsen JA, Freundlich M, Thorgaard P, Andreassen S. Modelling of hypoxaemia after gynecological laparotomy. Acta Anaesthesiol Scand 2001; 45: 349-356.
13. Nunn JF. Nunn's Applied Respiratory Physiology. Oxford, UK: Butterworth-Heinemann, 1993.
14. Lentner C. Geigy Scientific Tables, Volume 5, Heart and Circulation. Basel, Switzerland: Ciba-Geigy, 1990.
15. Benumof JL. Respiratory physiology and respiratory function during anesthesia. In: Miller RD, ed. Anesthesia. New York: Churchill Livingstone Inc, 1990: 523.
16. Hachenberg T, Brüssel T, Roos N, et al. Gas exchange impairment and pulmonary densities after cardiac surgery. Acta Anaesthesiol Scand 1992; 36: 800-805.
17. Tenling A, Hachenberg T, Tydén H, Wegenius G, Hedenstierna G. Atelectasis and gas exchange after cardiac surgery. Anesthesiology 1998; 89: 371-378.
18. Kirklin JK, Westaby S, Blackstone EH, Kirklin JW, Chenoweth DE, Pacifico AD. Complement and the damaging effects of cardiopulmonary bypass. J Thorac Cardiovasc Surg 1983; 86: 845-857.
19. Royston D, Minty BD, Higenbottam TW, Wallwork J, Jones GJ. The effect of surgery with cardiopulmonary bypass an alveolo-capillary barrier function in human beings. Ann Thorac Surg 1985; 40: 139-143.
20. Morioka K, Muraoka R, Chiba Y, et al. Leucocyte and platelet depletion with a blood cell separator: effects on lung injury after cardiac surgery with cardiopulmonary bypass. J Thorac Cardiovasc Surg 1996; 111: 45-54.
21. Zehr KJ, Poston RS, Lee PC, et al. Platelet activating factor inhibition reduces lung injury after cardiopulmonary bypass. Ann Thorac Surg 1995; 59: 328-335.
22. Hachenberg T, Tenling A, Nyström SO, Tydén H, Hedenstierna G. Ventilation-perfusion inequality in patients undergoing cardiac surgery. Anesthesiology 1994; 80: 509-519.
23. Ellison LT, Duke JF, Ellison RG. Pulmonary compliance following open-heart surgery and its relation to ventilation and gas exchange. Circulation 1967; 35: I217-I225.
24. Diehl JL, Lofaso F, Deleuze P, Similowski T, Lemaire F, Brochard L. Clinically relevant diaphragmatic dysfunction after cardiac operations. J Thorac Cardiovasc Surg 1994; 107: 487-498.
25. Tokics L, Hedenstierna G, Svensson L, et al. V/Q distribution and correlation to atelectasis in anesthetized paralyzed humans. J Appl Physiol 1996; 81: 1822-1833.
26. Rosenberg J, Rosenberg Adamsen S, Kehlet H. Postoperative sleep disturbance: causes, factors and effects on outcome. Eur J Anaesthesiol 1995; 10 (Suppl): 28-30.
27. Rosenberg J, Wildschiødtz G, Pedersen MH, von Jessen F, Kehlet H. Late postoperative nocturnal episodic hypoxaemia and associated sleep pattern. Br J Anaesth 1994; 72: 145-150.
28. Rees SE, Kjærgaard S, Thorgaard P, Toft E, Andreassen S. The automatic lung parameter estimator (ALPE) system: non-invasive estimation of pulmonary gas exchange parameters in 10-15 minutes. J Clin Monit 2002; 17: 43-52.

CARDIAC SURGICAL PROCEDURES; MODELS, BIOLOGICAL, models, cardiovascular; MONITORING, PHYSIOLOGICAL; POSTOPERATIVE COMPLICATIONS; RESPIRATION, respiratory transport, pulmonary gas exchange, ventilation/perfusion ratio.

© 2004 European Academy of Anaesthesiology