Secondary Logo

Journal Logo

Original Article

Effects of sequential changes from conventional ventilation to high-frequency oscillatory ventilation at increasing mean airway pressures in an ovine model of combined lung and head injury

O'Rourke, J.*; Sheeran, P.; Heaney, M.; Talbot, R.; Geraghty, M.§; Costello, J.; McDonnell, C.**; Newell, J.††; Mannion, D.

Author Information
European Journal of Anaesthesiology: May 2007 - Volume 24 - Issue 5 - p 454-463
doi: 10.1017/S0265021506002006



Acute lung injury (ALI) occurs in up to 25% of patients with traumatic brain injury (TBI) and contributes to increased mortality in this group, from 15% in isolated TBI to 38% in combined TBI and ALI [1,2]. Management strategies for ALI comprising high positive end expiratory pressure (PEEP), permissive hypercarbia and lung recruitment to achieve the ‘open lung' are currently standard practice. High-frequency oscillator ventilation (HFOV) replicates this open lung strategy and has demonstrated a non-significant mortality benefit when used in patients with acute respiratory distress syndrome (ARDS) [36]. HFOV is rarely used in patients with combined lung and head injuries due to concerns that persistently increased intrathoracic pressures may impede cerebral venous drainage and therefore extend the brain injury.

The safety of HFOV in an animal model of decreased intracranial compliance has previously been examined. HFOV did not adversely affect the cardiovascular system or intracranial pressure (ICP) in a feline model of combined head and lung injury. However, the continuous distending pressures (CDP) used (6–8 cm H2O) were significantly less than those employed in clinical practice [7]. A similar study in six lung injured lambs examined the intracranial effects of conventional mechanical ventilation (CMV) and HFOV at mean airway pressures of up to 25 cm H2O. ICP volume curves were generated by increasing cerebrospinal fluid (CSF) volume with concurrent intraventricular pressure measurement. Both CMV and HFOV groups demonstrated similar intracranial compliance curves, similar decreases in mean arterial pressure (MAP) and cerebral perfusion pressure (CPP) and also similar increases in central venous pressures (CVP) [8]. One adult case series has examined the effects of converting from CMV to HFOV in five patients with subacute head injury and ARDS (>5 days after head injury). This is the only evidence published to date on the use of HFOV in TBI. In this series HFOV improved PaO2/FiO2 ratios in four of the five patients, the ICP decreased in three patients and CPP actually increased in these three patients. No HFOV-associated sequelae were reported during its use in these patients [9].

This experimental model was designed to examine the intracranial effects of short-term perturbations in intrathoracic pressure which typically occur when patients are converted from the conventional ventilator to the high-frequency oscillator. The literature suggests that non-compliant lungs are protective in terms of pressure transmission to the intracranial compartment. We hypothesized that when ICP is elevated, initial increases in intrathoracic pressure when lung volumes are low and lungs derecruited would not affect ICP. In addition, we postulated that high intrathoracic pressures are more likely to adversely affect ICP due to changes in pulmonary compliance as the lung is ‘opened'. Furthermore we hypothesized (on the basis of the three studies presented above [7–9]) that HFOV, would not result in significantly worse indices of cerebrovascular resistance and cerebral metabolism at equivalent mean airway pressures to CMV. In this study indices of cerebral metabolic activity were measured and correlated with transcranial Doppler profiles (TCD) through a range of intrathoracic pressures. We examined how changing modes of ventilation, but not mean airway pressures, impacted on cerebral, cardiovascular and respiratory parameters. Any change noted as a result of the treatment, i.e. mode of ventilation at equivalent mean intrathoracic pressure, was defined a ‘treatment effect'. Secondly, indices were analysed based on the response to the increase in intrathoracic pressure between 16 and 40 cm H2O. Any change seen as a result of increasing intrathoracic pressure was termed a ‘pressure effect'. The mean airway pressure was incrementally increased into the zone of pulmonary overdistension to determine whether critical ICP values or CPP values would occur and correlate with cerebral circulatory arrest.


After obtaining institutional research and Ethics Committee approval and under licence from the Governmental Department of Health and Children, Ireland, the following model of combined lung and head injury was instituted. The timeline of interventions is briefly presented in Figure 1.

Figure 1.
Figure 1.:
Flow diagram describing progression through experimental protocol. Central vertical arrow represents timeline. Full series of measurements (in parenthesis 1–11) taken before each transition point throughout protocol at 15min intervals.

Animal preparation

Eleven female sheep with a mean weight of 42.8 kg were anaesthetized in a consistent manner. All were fasted overnight and premedicated with intramuscular (i.m.) midazolam 0.5 mg kg−1 1 h preoperatively. Intravenous (i.v.) access was established and anaesthesia was induced with alphaxalone and alphadolone (Saffan, Pittman-Moore, Uxbridge, Middlesex, UK) 2 mg kg−1 and midazolam 0.5 mg kg−1. Total i.v. anaesthesia (TIVA) was commenced and continued for the duration of the experiment with infusions of saffan, morphine, midazolam and cisatracurium at rates of 14.1, 0.1, 0.2 and 0.1 mg kg−1 h−1, respectively. Each animal was orally intubated with a size 7.5 mm internal diameter customized cuffed endotracheal tube (Malincrodt Ltd., Athlone, Ireland). A distal pressure monitoring port allowed continuous measurement of airway pressures at the tip of the endotracheal tube with a bourdon gauge.

The animals were positioned laterally and ventilated with 100% oxygen, a ventilation rate of 12, a PEEP of 5 cm H2O and tidal volumes of 10 mL kg−1 with a volume cycled ventilator (Servo 900B, Siemens, Solna, Sweden). Intravascular lines were inserted comprising a femoral arterial line, a jugular central venous line and a 5 French thermistor-tipped pulmonary artery floatation catheter (American Edwards Laboratories, Irvine, CA, USA). Intravascular lines were zeroed at mid-thoracic level and transduced with fluid-filled pressure transducer systems (DTX disposable pressure transducer system, Ohmeda, Madison, WI, USA). The pulmonary artery catheter was calibrated in vivo, animal weights entered and a standard height of 100 cm was used for all. Continuous cardiac output (CO) data were calculated throughout the experiment (continuous CO monitor, Baxter Healthcare Laboratories, Ireland). An orogastric tube was also inserted. Lactated Ringer's solution was infused at a rate of 10 mL kg−1 h−1 throughout the experiment. Supplemental 250 mL boluses of lactated Ringer's solution were administered where necessary to maintain the mean arterial pressure (MAP) greater than 50 mmHg. Epinephrine was commenced if the MAP decreased to less than 50 mmHg with a concomitant CVP of greater than 8 cm H2O. Core temperature was measured via the pulmonary artery catheter and environmental adjustments made to maintain normothermia.

A midline burrhole was performed 15 mm anterior to the lambdoid suture. The saggital venous sinus was cannulated after making a 20-G puncture in the vein and advancing an 18-G epidural catheter approximately 2 cm. An 18-G puncture was made in the dura of the lower hemisphere through which an intraventricular multi-orifice brain catheter was inserted into the lateral ventricle. Clear CSF confirmed correct placement and the catheter was fixed in place. An 8 French foley catheter was placed extradurally over the contralateral (upper) hemisphere. Slow inflation of the catheter balloon with saline when scheduled during the experiment simulated an expanding extradural haematoma. Careful attention was paid to the manner in which the dura was dissected from the skull to avoid CSF leaks. In the absence of significant CSF leak, the expanded extradural balloon simulates a space occupying lesion.

Transcranial Doppler studies

Studies of the principal intracranial carotid artery branch were performed throughout using a 2-MHz probe at an insonation depth of 35–39 mm and a power of 60%. The gain was adjusted to optimize audio and colour flow Doppler images (Scimed TC 22, Siemens Transcranial Doppler). In sheep, a single carotid artery on each side supplies cerebral territories equivalent to those of the internal and external carotid arteries in human beings. Insonation was performed via the transorbital approach in a manner similar to that described previously [10]. A customized immobilization device minimized head movements and optimized the signal to noise ratio of the TCD during the procedure. The pulsatility index (PI), which is an index of cerebrovascular resistance, was calculated from the following equation:


Lung and head injury

Lung lavage was performed with repeated 500 mL aliquots of warmed 0.9% saline (37°). During each lavage the animal's position was changed to ensure optimal surfactant washout of both lungs. The animals were maintained on baseline ventilation parameters and blood gas analysis was performed after 20 min. All animals demonstrated a PaO2 less than 8 kPa (60 mmHg) as demonstrated in previous similar models of animal lung injury [11]. One hour after lung injury the PaO2 remained less than 13.3 kPa (100 mmHg) in 10 of the 11 animals. This PaO2 and interval have been reported in ovine models as indicative of significant lung injury [12,13]. With confirmation of lung injury, the extradural balloon catheter was inflated slowly until the ICP exceeded 50 mmHg. At this point, inflation was stopped and the outflow clamped. This model of head injury has been described previously with balloons so as to ascertain the exact volume of brain tissue displaced. Intracranial pressure changes are monitored independently with an intraventricular catheter [1416].

Experimental protocol

Intracranial and intravascular pressures, blood gas analysis and TCD tracings were taken at baseline, following induction of both lung and head injuries and before all transition points during the protocol. Each transition point represents the time when the animal was converted from one ventilation mode to the other during the protocol. This comprised 11 datasets in total. The first dataset was taken at baseline with blood gas analysis after the lung injury and TCD tracings after the head injury. The second dataset was collected at a mean airway pressure of 16 cm H2O after 15 min of CMV, the third at a mean airway pressure of 16 after 15 min of HFOV. The fourth and subsequent measurements were taken 15 min after each transition point, i.e. at a mean airway pressure of 22 on CMV and then 22 on HFOV. The airway pressure was sequentially increased to 28 with 15 min of CMV and HFOV, then 34 and finally 40 cm H2O. The arteriovenous oxygen difference (AVDO2) and arteriovenous lactate differences (AVDL) were used to calculate the lactate oxygen index (LOI) from the equations presented below.

Arteriovenous oxygen difference (AVDO2)


Lactate Oxygen index (LOI)


SjvO2=Jugular venous oxygen saturation, or saggital sinus oxygen saturation.

Ventilation strategies during experimental protocol

The animals were ventilated at mean airway pressures (CMV) or continuous distending pressures (HFOV) of 16, 22, 28, 34 and 40 cm H2O for periods of 15 min alternating between CMV and HFOV. The FiO2 was set at 1.0. The CMV strategy delivered a tidal volume of 8 mL kg−1, 12 times min−1. PEEP values were adjusted to achieve the desired target airway pressure. The high-frequency oscillator was set to its maximal power of 100 and minimal rate of 5 Hz. The inspiratory/expiratory ratio was set to 50% and CDP (the mean distending pressure as set on the oscillator) adjusted per protocol (3100 A, Sensorimedics, Yorba, Linda, CA). Maximal settings were augmented by deflating the endotracheal tube cuff during HFOV and increasing the bias flow to 30 L min−1 to enhance CO2 elimination while maintaining mean airway pressures. The functional residual capacity (FRC) was maintained between ventilation changes by clamping the endotracheal tube. Both ventilation strategies were kept constant throughout the experiment with the mean airway pressure adjusted per protocol.

At the end of the experiment each animal was euthanased with pentobarbital 150 mg kg−1. The brain of each animal was removed and sectioned to confirm correct placement of the intraventricular catheter.

Statistical analysis

The primary outcome variables were ICP, TCD PI and LOI as indices of cerebrovascular resistance and metabolism. The effects of mean airway pressure and ventilation mode on PaO2, PaCO2, CO, MAP and CVP were measured. All indices of intracranial, respiratory and cardiac performance were analysed based on whether the animal was being ventilated with CMV or HFOV. This was defined as the treatment effect. Secondly, all indices were analysed based on the response to the increase in intrathoracic pressure. Any change seen as a result of increasing the intrathoracic pressure was defined as the pressure effect.

A repeated measures analysis of variance (ANOVA) with fixed effect terms for treatment (i.e. CMV or HFOV), pressure (mean intrathoracic pressure 16, 22, 28, 34, 40 cm H2O) and a random effect to model between animal variability was used (using a significance level of 0.05). The random effect accounts for the repeated measures component of the design, namely that repeated measures were recorded for each sheep at the different pressures. Post hoc analysis examined whether significant breakpoints or points of major change occurred as the airway pressure was increased by 6 cm H2O. Tukey tests were used to determine whether a disproportionate change occurred at any individual airway pressure by using pairwise comparisons of variables at the lower and higher pressures.


The intracranial effects of changing from CMV to HFOV at increasing pressures are presented in Table 1. Although the PaCO2 significantly increased during periods of HFOV (P = 0.001) there was no difference in ICP values for both modes of ventilation at each pressure point.

Table 1
Table 1:
Intracranial, respiratory and cardiovascular effects of changing ventilation modes from CMV to HFOV at increasing mean airway pressures.

The arteriovenous lactate difference (AVDL) was not affected by mode of ventilation (no treatment effect); however, it did increase as mean airway pressure was increased (P = 0.001). In 7% (9/122) of AVDL measurements, the arterial lactate exceeded the saggital venous value, resulting in a negative AVDL. No further calculations were undertaken to compute the LOI when the AVDL was negative. Tukey's simultaneous tests determined that the AVDL increased significantly from baseline to an airway pressure of 22 cm H2O (difference of means 0.26, P = 0.02) with a further increase to 28 cm H2O (difference of means 0.39, P = 0.0001). AVDL calculations at the higher airway pressures of 34 cm H2O demonstrate a decrease in AVDL towards baseline.

There were four high baseline LOI values; each increased value was associated with a significant respiratory alkalosis (pH > 7.5). This resulted in a wide range of LOI values at baseline (0.02–2.4). Throughout the rest of the experiment, no significant treatment effect was observed when LOI values between CMV and HFOV were compared. The LOI increased as mean airway pressure increased and the difference became statistically significant at an airway pressure of 40 cm H2O (difference of means 4.15, P = 0.02). When the range of LOI values is examined, the extremely high values (LOI > 5) were observed with SjvO2 values in excess of 85% and decreased AVDO2 (data not presented, constituents of LOI calculation).

The TCD trace changed with the introduction of the head injury. In certain animals, a significant Cushing reflex followed inflation of the extradural balloon as demonstrated by a marked surge in mean blood pressure. In these animals the peak TCD systolic and diastolic velocities increased. In other animals the increase in ICP exceeded the CPP and inflation of the extradural balloon was followed by short periods of absent or reversed diastolic blood flow. The ICP decreased to between 40 and 50 mmHg 10 min after the induced head injury. Throughout the experiment, the ICP increased and the MAP decreased as mean airway pressure was increased. Cerebral circulatory arrest ensued in most animals. A consistent finding in all animals was a significant increase in the PI. The PI increased from a median baseline value (range) of 0.77 (0.4–2.3) to 2.36 (0.42–4.1) at a mean airway pressure of 34 cm H2O on CMV. While no absolute value correlated with cerebral circulatory arrest in all animals, there was a significant pressure effect with increasing intrathoracic pressure (P = 0.001).

The PaO2 increased with intrathoracic pressures (P = 0.001), but was not affected by mode of ventilation. Lung recruitment was seen at a distinct mean airway pressure and could be easily identified in each animal. It was observed in most animals at intrathoracic pressures of 22–28 cm H2O. In the post hoc analysis the increase in PaO2 was most marked between 22 and 34 cm H2O.

While a significant increase in the CVP was observed at each mean airway pressure, no difference in CVP was seen when CMV and HFOV were compared at equivalent mean airway pressures. As the mean airway pressure was increased, MAP decreased (P < 0.001). At equivalent mean airway pressures, the MAP was higher during periods of CMV (P = 0.05) but CO was higher during periods of HFOV (P = 0.04). The CPP decreased to almost zero in all animals in the latter stages of the experiment. At equivalent mean airway pressures, the CPP was lower during periods of HFOV when compared with CMV (P = 0.02).


This study compared CMV and HFOV strategies in TBI at increasing airway pressures and their respective effects on intracranial metabolic indices. The principal finding of this study is that the open lung strategy increased ICP in the setting of decreased pulmonary compliance (i.e. ALI). The respective mode of ventilation did not result in significant differences in ICP and cerebral metabolic indices were not significantly different between CMV and HFOV at equivalent mean airway pressures. However, HFOV demonstrated significantly lower CPP values at all measured airway pressures.

A main effects plot for each variable was constructed using Tukey's simultaneous tests in which the fitted means were plotted against pressures. Some variables exhibited a linear increase with airway pressure (ICP, CVP) or a linear decrease with airway pressure (CPP and MAP). Significant changes in oxygenation occurred as the airway pressure was increased between 16 and 28 cm H2O. The AVDL increased to an airway pressure of 28 cm H2O, subsequently decreasing with further increases of airway pressure. The LOI increased acutely at pressures higher than 22 cm H2O. The SjvO2 increased with increasing intrathoracic pressures until it approximated the arterial oxygen saturation at the point of cerebral circulatory arrest. Changes in the AVDL, LOI and SjvO2 indicated progressive cerebral ischaemia at higher mean airway pressures. Previous work has demonstrated absent oxygen extraction from jugular venous blood to be indicative of brain death [17,18]. The simultaneous TCD recordings demonstrated increasing cerebrovascular resistance as mean airway pressures were increased until tracings typical of cerebral circulatory arrest occurred. These observations may be correlated with metabolic and TCD patterns during ischaemia or cerebral infarction as detailed in Table 2.

Table 2
Table 2:
Transcranial Doppler data correlation with cerebral metabolic indices. An important observation is that ischaemia and infarction or circulatory arrest should be diagnosed based on multiple parameters due to confounding factors that may be manifest with individual recordings.

In circumstances where cerebral oxygen delivery is limited, lactic acid may be used as a metabolic fuel and pathologically high levels of cerebral lactate production are seen with TBI. Therefore, the AVDL is regarded as an index of anaerobic cerebral metabolism and, in this study, both the AVDL and LOI increased acutely as head injury was induced. The LOI relates the anaerobic index: the AVDL, to an index of aerobic metabolism, the cerebral oxygen extraction. The LOI is reliant on a stable haemoglobin, acid–base balance and temperature. Acute changes in one of these parameters may lead to incorrect results. Overventilation of four animals led to abnormally high LOI values at baseline, a finding previously reported by others [19]. While these high LOI values were the result of hyperventilation-induced ischaemia, a second factor responsible for the high LOI values was the Bohr effect. The Bohr effect describes how alkalosis shifts the oxyhaemoglobin dissociation curve to the left resulting in a higher oxygen saturation for any given PaO2. Since SaO2 and SjvO2 are major components of the AVDO2 calculation, an alkalosis induced increase in SjvO2 will increase the LOI. With a stable blood pH, an increase in SjvO2 is likely to be a result of increased cerebral blood flow or, as was the case later during the course of this study, reduced cerebral oxygen extraction due to infarction.

PEEP improves oxygenation by reversing atelectasis and increasing the FRC. In addition, PEEP may also decrease areas of steal within the brain and improve blood flow to areas with borderline perfusion. In these penumbral circumstances, PEEP increases venous pressure uniformly, resulting in decreased shunting and less preferential flow to vessels with lower resistance [20]. The factors that are most important in determining how PEEP affects the ICP are [2027]: (1) The level of PEEP applied: in this study, the higher the PEEP or CDP applied, the greater the detrimental effect on cerebral haemodynamics. (2) Patient positioning: improved venous drainage in the semirecumbent posture will counteract the hydrostatic effect of applied PEEP. Almost all of the human work studying PEEP and ICP has been conducted with patients semirecumbent at 30°. Animal work, on the other hand, has more consistently demonstrated an increase in ICP with PEEP and this may be explained by conduct of these studies in the horizontal position. (3) Presence or absence of ALI: non-compliant lungs demonstrate less pressure transmission to the vasculature, resulting in less systemic hypotension and less changes in ICP. However, in this animal model of decreased pulmonary compliance, increasing intrathoracic pressure resulted in systemic hypotension, increased CVP and increased intracranial pressures. (4) Baseline ICP: higher baseline ICP values are less affected by increasing PEEP. Although the intracranial pressure was 40–50 mmHg after the head injury, increasing intrathoracic pressures significantly increased the ICP. The findings in (3) and (4) may be explained by changes in pulmonary compliance as the lung progressively passes from the zone of atelectasis through the compliant portion of the pressure volume curve and into the zone of overdistension.

In the current study we did not demonstrate a difference between the intracranial effects of HFOV or CMV; however, three clinically relevant issues do emerge. Firstly, both modes of ventilation were equally detrimental at high intrathoracic pressures; therefore, distinction between these modes at mean airway pressures in excess of 30 cm H2O is likely irrelevant. Secondly, airway opening or recruitment occurred in most animals at mean airway pressures less than 28 cm H2O and, lastly, we demonstrated significant increases in PaCO2 each time HFOV was initiated despite all attempts to optimize PaCO2 elimination. Although HFOV did not increase the ICP, whether or not a protective mechanism exists is difficult to unmask in the presence of hypercarbia as observed during this experimental protocol.

When considering the limitations of this study, it would appear that the baseline ventilation strategy was excessive. This resulted in cerebral vasoconstriction and high LOI values in four of 11 animals. The animals' body weights were at the upper limits of manufacturer recommendations for this particular ventilator and significant fluctuations in PaCO2 were observed when converting from CMV to HFOV. While the changes in PaCO2 could be regarded as a significant limitation of this study, a complete search of the literature demonstrates that fluctuations in PaCO2 are frequently seen in this situation. Another limitation of this study is that the head injury alone was not characterized over time. As the balloon volume remains static, increments in intrathoracic pressure lead to increases in ICP. Although this model of TBI is well characterized [1416], the effects of increasing intrathoracic pressures were not separated from the contribution of cerebral oedema to ICP. This may have been circumvented by randomization of the five airway pressures or further control animals ventilated at a constant airway pressure. The predictive value of the AVDL in the detection of ischaemia has been previously questioned as regards how local production, consumption or transport of lactate is reflected globally [28]. While we acknowledge these issues, cerebral AVDL and LOI have been related to outcome after TBI in previous reports [19,2931].

In conclusion, the results of this study indicate that in this ovine model of combined lung and brain injury, increasing intrathoracic pressures increase cerebral markers of head injury as demonstrated by increased ICP, decreased CPP and increased TCD PI (all to the point of cerebral circulatory arrest) regardless of mode of ventilation. Although we did not demonstrate further intracranial injury with HFOV when compared to CMV, the decrease in MAP, and the significant decrease in CPP together with fluctuations in PaCO2 while on HFOV are of concern. Continuous PaCO2 monitoring is essential in situations where cerebral blood flow is critical and ventilation changes anticipated. Finally, this study demonstrates how TCD may be a useful adjunct to ICP and SjvO2 monitoring when major ventilation changes are anticipated in the setting of compromised intracranial compliance.


The results of this study were presented as an oral presentation at the 21st Conference on High Frequency Ventilation of Infants, Children, and Adults (31 March–3 April 2004, Snowbird, Utah). Funding for this study was provided through charitable donations to Our Lady's Hospital for Sick Children Research facility. We wish to acknowledge the expert veterinary assistance provided by Ms Phillippa Marks and Mr Peter Nowlan during the course of this study.


1. Zygun D. Non-neurological organ dysfunction in neurocritical care: impact on outcome and etiological considerations. Curr Opin Critical Care 2005; 11: 139–143.
2. Mascia L, Andrews PJ. Acute lung injury in head trauma patients. Intensive Care Med 1998; 24: 1115–1116.
3. Derdak SDO. High frequency oscillatory ventilation for acute respiratory distress syndrome in adult patients. Crit Care Med 2003; 31(Suppl 4): 317–323.
4. David M, Weiler N, Heinrichs W et al. High frequency oscillatory ventilation in adult acute respiratory distress syndrome. Intensive Care Med 2003; 29: 1656–1665.
5. Fort P, Farmer C, Westerman J et al. High frequency oscillatory ventilation for adult respiratory distress syndrome – a pilot study. Crit Care Med 1997; 25(6): 937–947.
6. Mehta S, Lapinsky SE, Hallett DC et al. Prospective Trial of high frequency oscillation in adults with acute respirtatory distress syndrome. Crit Care Med 2001; 29(7): 1360–1369.
7. Raju TN, Braverman B, Nadkarny U, Kim WD, Vidyasagar D. Intracranial pressure and cardiac output remain stable during high frequency oscillation. Crit Care Med 1983; 11(11): 856–858.
8. Walker AM, Brodecky VA, de Prue N, Ritchie BC. High frequency oscillatory ventilation compared with conventional mechanical ventilation in newborn lambs: Effects of increasing airway pressure on intracranial pressure. Pediatr Pulmonol 1992; 12: 11–16.
9. David M, Karmrodt J, Weiler N et al. High frequency oscillatory ventilation in adults with traumatic brain injury and acute respiratory distress syndrome. Acta Anaesthesiol Scand 2005; 49: 209–214.
10. Markus H, Loh A. Microscopic air embolism during cerebral angiography and strategies for its avoidance. Lancet 1993; 341: 784–787.
11. Lachmann B, Jan Van Daal G. Adult respiratory distress syndrome; Animal models (Chapter 26). In: Robertson B, Van Golde LMG, Batenburg JJ, eds. Pulmonary Surfactant: From Molecular Biology to Clinical Practice. Elsevier Science Publishers, 1992.
12. Kerr CL, McCaig LA, Veldhuizen RAW, Lewis JF. High frequency oscillation and exogenous surfactant in lung-injured sheep. Crit Care Med 2003; 31(10): 2520–2526.
13. Sedeek KA, Takeunchi M, Suchdolski K, Kacmarek RM. Determinants of tidal volume during high frequency oscillation. Crit Care Med 2003; 31(1): 227–231.
14. Fukushima A, Miyashita K, Shozo O et al. Evaluation of intracranial pressure by transcranial Doppler ultrasound in dogs with intracranial hypertension. J Vet Med Sci 2000; 62(3): 353–355.
15. Biros M. Experimental Head trauma models: A clinical perspective. Resuscitation 1991; 22: 283–293.
16. Finnie JW, Blumbergs PC. Animal models: Traumatic brain injury. Vet Pathol 2002; 39: 679–689.
17. Takasu A, Yagi K, Ischihara S, Okada Y. Combined continuous monitoring of systemic and cerebral oxygen metabolism after cardiac arrest. Resuscitation 1995; 29: 189–194.
18. Reganon GD, Minambres E, Holanda M et al. Usefullness of venous oxygen saturation in the jugular bulb for the diagnosis of brain death: report of 118 patients. Intensive Care Med 2002; 28: 1724–1728.
19. Murr R, Stummer W, Schurer L, Polasek J. Cerebral lactate production in relation to intracranial pressure, cranial computed tomography findings, and outcome in patients with severe head injury. Acta Neurochirurgica 1996; 138: 928–937.
20. Andrews PJD. Pressure, Flow and Occam's Razor: a matter of ‘steal'? Intensive Care Med 2005; 31: 323–325.
21. Aidinis SJ, Lafferty J, Shapiro HM. Intracranial responses to PEEP. Anesthesiology 1976; 45(3): 275–286.
22. Mascia L, Grasso S, Fiore T et al. Cerebro-pulmonary interactions during the application of low levels of positive end expiratory pressure. Intensive Care Med 2005; 31: 373–379.
23. Videtta W, Villarejo F, Cohen M et al. Effects of positive end expiratory pressure on intracranial pressure and cerebral perfusion pressure. Acta Neurochir Suppl 2002; 81: 93–97.
24. Caricato A, Conti G, Corte D et al. Effects of PEEP on the intracranial system of patients with head injury and subarachnoid hemorrhage: The role of respiratory system compliance. J Trauma 2005; 58(3): 571–576.
25. McGuire G, Crossley D, Richards J, Wong D. Effects of varying levels of positive end expiratory pressure on intracranial pressure and cerebral perfusion pressure. Crit Care Med 1997; 25(6): 1059–1062.
26. Londrini S, Montolivo M, Pluchino F, Borroni V. Positive end expiratory pressure in supine and sitting positions: Its effects on intrathoracic and intracranial pressures. Neurosurgery 1989; 24(6): 873–876.
27. Cooper KR, Boswell PA, Choi SC. Safe use of PEEP in patients with severe head injury. J Neurosurg 1985; 63: 552–555.
28. Poca MA, Sahuquillo J, Monforte R, Vilalta YA. Global systems for monitoring cerebral hemodynamics in the neurocritical patient: basic concepts, controversies and recent advances in measuring jugular bulb oxygenation. Neurocirugia 2005; 16: 301–322 (article in Spanish).
29. Artru F, Daillier F, Burel E et al. Assessment of jugular blood oxygen and lactate indices for detection of cerebral ischemia and prognosis. J Neurosurg Anesthesiol 2004; 16(3): 226–231.
30. Millar SA, Alston RP, Souter MJ, Andrews PJD. Aerobic, anaerobic, and combination estimates of cerebral hypoperfusion during and after cardiac surgery. Br J Anaesth 1999; 83(6): 936–939.
31. Robertson CS, Raj KN, Ziya L et al. Cerebral arteriovenous difference as an estimate of cerebral blood blow in comatose patients. J Neurosurg 1989; 70: 222–230.


© 2007 European Society of Anaesthesiology