Secondary Logo

Journal Logo

Neurosurgical Anesthesia: Research Report

Measuring Hemoglobin Oxygen Saturation During Graded Hypoxic Hypoxia in Rat Striatum

Julien, Cécile PhD*; Bradu, Adrian PhD; Sablong, Raphaël PhD; Grillon, Emmanuelle MSc*; Remy, Chantal PhD*; Derouard, Jacques PhD; Payen, Jean-François MD, PhD*‡

Author Information
doi: 10.1213/01.ane.0000185039.85886.e3

Monitoring cerebral oxygenation is critical in patients with acute cerebral disorders, as cerebral ischemia is one of the most important causes of secondary brain damage. Measuring jugular venous oxygen saturation (Sjvo2) is an accepted method for monitoring global cerebral oxygenation, although it may not detect all causes of tissue hypoxia (1). Measurement of brain tissue oxygen tension (Pbro2) permits localized monitoring of ischemic insults in patients. This technique is increasingly being used for monitoring cerebral oxygenation in patients with intracranial disorders, such as traumatic brain injury or subarachnoid hemorrhage (2,3). Another method can be the measurement of cerebral hemoglobin oxygen saturation. Bedside measurements of cerebral hemoglobin oxygen saturation can be estimated with near-infrared spectroscopy (NIRS) (4,5). Because visible light (520–660 nm) is much more strongly absorbed by biological tissue, the near-infrared region (700–1000 nm) is chosen to illuminate deep structures by means of a superficial source. Because the absorption of light varies with the relative concentrations of the tissular chromophores (oxyhemoglobin, deoxyhemoglobin, cytochrome oxidase), NIRS is suited for noninvasively evaluating global brain oxygenation. However, several studies have pointed out technical problems using NIRS, primarily because of the source-detector distance and the extracerebral contamination of the signal (6–9).

Reflectance spectroscopy of visible light is also an established method for measuring cerebral hemoglobin oxygen saturation, although its catchment area is usually restricted to the tissue surface (10). To overcome this technical limitation, we developed a prototype based on the reflectance spectroscopy of the visible light as a method to assess in vivo local hemoglobin oxygen saturation using optical fiber probes inserted in the striatum (SstrO2). Simultaneous measurements of SstrO2 and of local cerebral blood flow (LCBF) were obtained in the rat during incremental reduction of the fraction of inspired oxygen (Fio2): 0.35, 0.25, 0.15, 0.12, and 0.10, followed by a reoxygenation period. Our goal was to determine the feasibility of SstrO2 measurement in rat striatum using this technique and to monitor the respective changes in SstrO2 and LCBF during graded hypoxic hypoxia. To distinguish between arterial and venous contributions to SstrO2 measurements, another group of rats was studied to measure sagittal sinus blood hemoglobin saturation (SssO2) during hypoxic hypoxia.

Methods

Two groups of fed Wistar female rats (200-220 g) were studied. Group 1 (n = 7 rats) was used for the simultaneous determination of SstrO2 and LCBF in the striatum during graded hypoxic hypoxia using reflectance spectroscopy and laser Doppler flowmetry (LDF), respectively. Group 2 (n = 6 rats) was studied to measure SssO2 via a catheter inserted into the superior sagittal sinus.

Preparation of animals was similar in the two groups and conformed to the guidelines of the French Government (decree N° 87-848 of October 19, 1987, licenses 006683 and A38071) and to the European Guidelines for the Protection of the Vertebrate Animals (ETS N° 123, Strasbourg March 18, 1986). In Group 1, a microdialysis cannula was chronically implanted in each corpus striatum to guide the insertion of probes used for reflectance spectroscopy and for LDF. For this purpose, rats were anesthetized with an intraperitoneal injection of chloral hydrate (400 mg/kg) and stereotaxically positioned (Harvard Apparatus, Les Ulis, France). After exposure of the parietal bones and the drilling of a hole in the cranium, a 1-mm diameter cannula (CMA12; Phymep, Paris, France) was implanted in each corpus striatum (0 mm to the bregma, 3 mm to the midline, 2.5 mm depth from the dura) and fixed to the skull using methyl acrylic cement.

The experiment was performed 1–2 wk after cannula implantation. Anesthesia was induced with 4% halothane and then maintained with an intraperitoneal injection of thiopental (40 mg/kg). Lidocaine 1% was injected subcutaneously for local anesthesia at all surgical sites. After tracheostomy, rats were mechanically ventilated with 65% nitrous oxide: 35% oxygen using a rodent ventilator (Model 683; Harvard Apparatus Inc, South Natick, MA). Ventilation was adjusted to maintain partial arterial CO2 pressure (Paco2) at 35 mm Hg. The Fio2 was continuously monitored (MiniOX I analyzer; Catalyst Research Corporation, Owings Mills, MD). A 0.7-mm indwelling catheter was inserted into the left femoral artery to monitor mean arterial blood pressure (MABP) via a chart recorder (8000S; Gould Electronic, Ballainvilliers, France). Blood gases (Pao2 and Paco2), arterial oxygen saturation of hemoglobin (Sao2), arterial pH (pHa), and hemoglobin content (tHba) were determined from <0.1 mL arterial blood samples (ABL 510; Radiometer, Copenhagen, Denmark). Another 0.7-mm indwelling catheter was inserted into the left femoral vein to continuously infuse a normal saline solution containing epinephrine (25 ng/min) and sodium bicarbonate (0.4 μmol/min) at a rate of 2 mL/h throughout the study. Epinephrine was required to protect from the adverse effects of both anesthesia and hypoxic hypoxia on the cardiovascular system. Sodium bicarbonate was used to prevent arterial acidosis. Rectal temperature was maintained at 37.5°C ± 0.5°C by using a heating pad placed under the abdomen.

In Group 2, the anesthesia protocol was similar. The posterior part of the superior sagittal sinus was gently exposed to allow insertion of a 0.7-mm indwelling catheter. Simultaneous arterial and venous blood gases (PssO2 and PssCO2), venous oxygen saturation of hemoglobin (SssO2) and venous pH (pHss) were determined from <0.1 mL blood samples (ABL 510; Radiometer). Cerebral arteriovenous difference in oxygen content (AvDO2) was calculated from measurements of arterial and sagittal sinus O2 contents.

LCBF within each corpus striatum was monitored using LDF (Periflux 4001 Master; Perimed, Stockholm, Sweden) and a flowprobe (MT B500; Perimed) with a tip diameter of 0.5 mm inserted into the cannula at a depth of 3.5 mm from the dura. The probe position was kept unchanged throughout the experiment. The striatal signal was digitized via an acquisition card system (National Instruments, Austin, TX) and continuously recorded on Labview (National Instruments) with a sampling time of 300 ms. Each measurement during hypoxia and reoxygenation was averaged over a 2-min period during steady-state conditions. The LCBF response was expressed as a percentage of the LCBF signal averaged over a 5-min period during the control period.

Reflectance spectroscopy within each corpus striatum was performed with a prototype instrument. Each reflectance probe consisted of 2 fibers of 100-μm diameter glued together using enamel and inserted with the LDF probe within each corpus striatum. Reflectance and LDF probes were separated by 300-μm distance within the microdialysis cannula, and the volumetric area of brain tissue insonated by the reflectance probe was approximately 1 mm3, overlapping that of the LDF probe. The emitter fiber of the reflectance probe was connected to white light from a stabilized tungsten halogen lamp. The receptor fiber directed the reflected light to a grating spectrometer (Roper Scientific Princeton Instruments, Trenton, NJ) equipped with a cooled, high dynamic range (16 bits) camera (Roper Scientific Princeton Instruments) and a computer with Winspec software (Roper Scientific Princeton Instruments). The spectral window contained all wavelengths (λ) between 520 and 590 nm. The reflectance spectrum R(λ) was determined according to the formula:

where Iref(λ) was a reference spectrum with the probe immersed in a white scattering medium (suspension of polystyrene microspheres of 430-nm diameter, concentration of 92% in distilled water) modeling cerebral tissue with no hemoglobin, I0(λ) was a baseline spectrum from artifacts of the probe immersed in distilled water, and I(λ) was the in vivo spectrum. Signals were acquired with a 1.0-s sampling time, including 300-ms exposure time (30,000 counts per pixel) and 700-ms transfer time. Each spectrum was then averaged over a 1-min accumulation period during steady-state conditions. Four spectra were acquired during hypoxia and reoxygenation (4-min acquisition), and 5 spectra during the control period (5-min acquisition).

The logarithm of the reflectance spectrum was expressed as a combination of the absorption spectra of oxyhemoglobin and deoxyhemoglobin, in accordance with the Beer-Lambert law:

where εHb(λ) and εHbO2(λ) are the absorption coefficients of deoxygenated and oxygenated hemoglobin and [Hb] and [HbO2] their tissular concentrations, respectively. L is the path length of the light in the tissue. The fitted parameters are [Hb]L, [HbO2]L, A and B. The baseline A+Bl accounts for differences in the scattering properties of the tissue and reference medium. Excellent agreement between observed and fitted parameters was declared when the tissue absorbency, as estimated by lnR(λ) - (A+Bλ), was <30%. The SstrO2 was then calculated according to the formula:

Group 1 was subjected to a stepwise decrease of Fio2: control (Fio2 of 0.35), normoxia (Fio2 of 0.25), hypoxia (Fio2 of 0.15, 0.12, and 0.10), and reoxygenation (Fio2 of 0.35) while the rat was lying in prone position. Group 2 was studied during the following conditions: control (Fio2 of 0.35) and hypoxia (Fio2 of 0.15, 0.12, and 0.10). In both groups of rats, the basic cycle was started after 30 min of equilibration at Fio2 of 0.35 (control period). The criteria for exclusion established a priori from the study were: MABP <100 mm Hg, pHa <7.30, Pao2 <100 mm Hg, arterial hemoglobin content <10 g/dL. Subsequent episodes were then first induced by decreasing the inhaled oxygen to Fio2 = 0.25, then by replacing the oxygen with air (Fio2 of 0.15, 0.12, and 0.10). During these episodes, fractions of inspired nitrous oxide were 0.65, 0.75, 0.25, 0.40, and 0.50, respectively. Reoxygenation was obtained by restoring the oxygen in the gas mixture. If MABP was <70 mm Hg, pHa <7.30, or Paco2 <27 mm Hg at Fio2 of 0.25, 0.15, or 0.12, the animal was excluded from the study. These exclusion criteria were not applied at Fio2 of 0.10 (severe hypoxia) and during the reoxygenation episode. If no LCBF changes were detected during hypoxia in Group 1, it was considered that the probe was not properly positioned, and the ensuing results were not considered for analysis. Each hypoxic episode lasted 10 min: a 4-min equilibrium period followed by LDF (2 min) and reflectance spectroscopy (4 min) acquisition and determination of MABP and arterial blood sampling. To avoid a possible interference between optical measurements, LDF and SstrO2 signals were acquired alternately. At the end of the measurement cycle, rats were euthanized by administration of an overdose of thiopental (50 mg/kg). The brain was excised from the skull, frozen in isopentane at −50°C, and stored at −80°C. The brains were sliced at the site of cannula implantation in the coronal plane using a cryotome (10-μm thickness). Brain sections were stained using hematoxylin-erythrosin-safran to verify the absence of hemorrhage at the site of the LCBF and SstrO2 measurements.

Data were expressed as mean ± sd. Analysis for statistical significance of changes during the successive episodes was performed using one-way analysis of variance for repeated measurements (StatView SE program; SAS, Cary, NC). Each value at a given episode was compared to that obtained at the control period using the Scheffé post hoc test. Difference in the LCBF and SstrO2 measurements between the two sides was tested using a paired Student’s t-test. If no significant difference was found between the two hemispheres, the two results were averaged to provide a mean estimate of changes. Multiple regression analysis was used to estimate the influence of Sao2 and of other factors (MABP, Paco2, tHb) on SstrO2 changes for each Fio2 episode. Physiological data in Group 2 were compared with those of Group 1 using an unpaired Student’s t-test. Statistical significance was declared at P < 0.05.

Results

Physiological and biochemical data of Group 1 are shown in Table 1. Hypoxic hypoxia caused a significant decrease in MABP at Fio2 of 0.12 and in Paco2 at Fio2 of 0.15. These changes were reversed during reoxygenation. A slight but significant decrease in tHb was also noted throughout the experiment. No histological evidence of hemorrhage was found at the site of LCBF and SstrO2 measurements. There was no statistical difference for LCBF and SstrO2 measurements between left and right corpus striatum throughout the experiment. Only one side from one animal was not further analyzed for LCBF because of a lack of LCBF changes during hypoxia. Hypoxic hypoxia resulted in a significant, graded, decrease in SstrO2, accompanied by an increase in LCBF (Table 1). Changes in LCBF and SstrO2 were reversed during reoxygenation. Multiple regression analysis showed that Sao2 was the only variable having a significant relationship with SstrO2 changes; adjustment for other variables (MABP, Paco2, tHb) did not affect this relationship.

Table 1
Table 1:
Physiological and Biochemical Data in Group 1, with the Determination of Hemoglobin Oxygen Saturation in Striatum (SstrO2) and of Changes in Local Cerebral Blood Flow (LCBF) (n = 7 rats) During Successive Fio2 Episodes

In Group 2, venous blood gases were obtained in 3 rats only at Fio2 of 0.10. Hypoxic hypoxia resulted in a progressive decrease in AvDO2 because of reductions in arterial oxygen content and, to a lesser extent, in superior sagittal sinus O2 content (Table 2). To distinguish between arterial and venous contribution to SstrO2 measurements, SstrO2 and SssO2 were plotted against Sao2 during control and hypoxic episodes. A comparable relationship of these 2 parameters (SstrO2, SssO2) with the degree of hypoxia was found, both having similar slope of changes per unit of SaO2: SstrO2 = −6.9 + 0.43 Sao2 (r2 = 0.49; P < 0.05) and SssO2 = −6.4 + 0.43 Sao2 (r2 = 0.50; P < 0.05) (Fig. 1). In addition, SstrO2 in Group 1 was not statistically different from SssO2 in Group 2 at control (Fio2 0.35): 38% ± 17% versus 38% ± 7%, respectively. Arterial blood gas, arterial pH, and Sao2 were comparable to the corresponding values obtained in Group 1 during the control period (data not shown). Conversely, no linear relationship was found between hypoxia-induced changes in LCBF and SsstrO2 (Fig. 2).

Table 2
Table 2:
Physiological and Biochemical Data in Group 2, with the Determination of Hemoglobin Oxygen Saturation in Superior Sagittal Sinus (SssO2) and of Cerebral Arteriovenous Difference in O2 Content (AvDO2) During Successive Fio2 Episodes
Figure 1
Figure 1:
Figure 1.
Figure 2
Figure 2:
Figure 2.

Discussion

The present study indicates that reflectance spectroscopy of visible light permits continuous and localized hemoglobin oxygen saturation measurements in the rat striatum, providing values that are strongly related to the degree of hypoxic hypoxia. The corresponding increase in LCBF at the same location confirms the tight regulation of oxygen delivery to the brain tissue when arterial oxygen content is altered. However, the close relation between changes in SssO2 and SstrO2 with the degree of hypoxia indicate that SstrO2 reflects primarily brain venous oxygenation.

Localized SstrO2 response to graded hypoxic hypoxia was determined using reflectance spectroscopy based on spectral analysis for wavelengths ranging between 520 and 590 nm. There have been few reports measuring cerebral hemoglobin oxygen saturation during ischemic/hypoxic challenges using this approach (11–13). In these studies, probes were placed in optical contact with the cortical surface and then the investigators examined surface vessel structures. Nevertheless, reflectance spectroscopy demonstrated a great sensitivity to detect critical conditions in the brain. One of the main advantages of this approach is that it circumvents the interoptode distance required when assessing the mean optical path length with NIRS. Some methodological issues need to be addressed. First, we used whole reflectance spectra, rather than three or four selected wavelengths, to improve agreement between observed and fitted values, as suggested by others (14,15). Second, each reflectance spectrum was corrected by two reference spectra corresponding to maximal and minimal reflectance (Equation 1). The variability in SstrO2 measurements attributable to light scattering was thus limited. Third, local tissue heating with the use of a tungsten halogen lamp was negligible, considering the very low optical power of the probe (0.0003 W spread over the whole emitted spectrum), the limited tissue volume area sampled by the detected radiation (1 mm3), and the fraction of light absorption in this tissue volume (<10%).

The absorption of reflected visible light during hypoxic hypoxia can be affected mainly by three factors that also interfere with the PbrO2 measurements (3): local hematocrit, vessel diameter, and the ratio of the sizes of venous and arterial compartments. Any influence of local hematocrit on SstrO2 measurements with the progression of hypoxia is unlikely because hematocrit measured in various brain regions was not altered during hypoxic hypoxia (16). Hypoxic hypoxia is accompanied by an increase in CBF, so that delivery of oxygen tends to be maintained (17). This increase is attributable to the dilation of cerebral arterioles, resulting in increased cerebral blood volume (CBV) (18). We previously measured local changes in CBV (corpus striatum, neocortex, cerebellum) using susceptibility contrast magnetic resonance imaging during experimental conditions similar to those applied in this study (19,20). With the use of indocyanine green and O2 as intravascular tracers, NIRS has been also proposed to measure CBV and CBF (5). However, in newborn piglets subjected to inhaled CO2 alterations and to hypoxic hypoxia, NIRS measurements of CBV did not correlate well with radioactive measurements but reliably monitored changes in cerebral tissue oxygenation (SssO2) (21). In dogs, NIRS was an inaccurate measure of CBF during changes in Paco2 (6). If changes in optical signals (NIRS) do not accurately reflect those in CBV or CBF during vessel diameter alterations (CO2 challenges), SstrO2 measurements using reflectance spectroscopy should be sensitive to phenomena other than changes in CBV during hypoxic hypoxia.

Data obtained in the two groups of rats indicate that SstrO2 values reflect primarily brain venous oxygenation. This is not surprising because approximately 70% of CBV is distributed in the venous compartment. Our PssO2 values at control (Group 2) compare favorably with others (PssO2 = 33 ± 4 mm Hg) in ventilated rats with Fio2 of 0.28 (22). During graded hypoxic hypoxia, we found changes in SstrO2 closely related with those in SssO2 (Fig. 1). If cerebral metabolic rate of O2 (CMRO2) is assumed to remain unchanged during hypoxic hypoxia (23,24), then according to the Fick equation (CMRO2 = CBF × arteriovenous O2 content difference [AvDO2]), AvDO2 will change proportionally but in the opposite direction of the LCBF changes. In terms of relative changes in AvDO2 induced by hypoxia, those in the venous compartment should be of smaller magnitude than those in the arterial one, to decrease AvDO2. This was indeed found in this study (Table 2), suggesting that SstrO2 reflects venous cerebral O2 oxygenation.

One of theoretical advantages of the measurement of localized hemoglobin oxygen saturation can be its great sensitivity to detect brain hypoxia. We found linear changes of SstrO2 with the degree of hypoxia, beginning to be significant at a threshold value of Pao2 55 mm Hg, in accordance with the cerebrovascular response to hypoxia in the rat (23). Potential confounding factors such as progressive arterial hypotension and hypocapnia in the present study were found to have no major influence on this relationship. Conversely, PbrO2 measurement can be significantly influenced by several factors such as variations of cerebral perfusion pressure and of Paco2 (25). It was even suggested that correct interpretation of low PbrO2 values should include simultaneous measurements of LCBF and/or of tissue oxygen extraction (AvDO2) to distinguish between arterial and venous contributions to PbrO2 measurements (26). At present, it is beyond the scope of this study to claim the superiority of one technique over another.

In summary, the present study shows that localized SstrO2 measurements using reflectance spectroscopy are possible during graded hypoxic hypoxia. SstrO2 signal mainly reflects brain venous oxygenation. The simultaneous measurements of LCBF and SstrO2 give the opportunity to assess circulatory and metabolic responses of a brain structure to hypoxic hypoxia.

The authors thank Christoph Segebarth for helpful comments on the manuscript.

References

1. Gopinath SP, Valadka AB, Uzura M, Robertson CS. Comparison of jugular venous oxygen saturation and brain tissue Po2 as monitors of cerebral ischemia after head injury. Crit Care Med 1999;27:2337–45.
2. van den Brink WA, van Santbrink,H Steyerberg EW, et al. Brain oxygen tension in severe head injury. Neurosurgery 2000;46:868–76.
3. Haitsma IK, Maas AI. Advanced monitoring in the intensive care unit: brain tissue oxygen tension. Curr Opin Crit Care 2002;8:115–20.
4. Owen-Reece H, Smith M, Elwell CE, Goldstone JC. Near infrared spectroscopy. Br J Anaesth 1999;82:418–26.
5. Madsen PL, Secher NH. Near-infrared oximetry of the brain. Prog Neurobiol 1999;58:541–60.
6. Newton CR, Wilson DA, Gunnoe E, et al. Measurement of cerebral blood flow in dogs with near infrared spectroscopy in the reflectance mode is invalid. J Cereb Blood Flow Metab 1997;17:695–703.
7. Germon TJ, Kane NM, Manara AR, Nelson RJ. Near-infrared spectroscopy in adults: effects of extracranial ischaemia and intracranial hypoxia on estimation of cerebral oxygenation. Br J Anaesth 1994;73:503–6.
8. Pollard V, Prough DS, DeMelo AE, et al. The influence of carbon dioxide and body position on near-infrared spectroscopic assessment of cerebral hemoglobin oxygen saturation. Anesth Analg 1996;82:278–87.
9. Ter Minassian A, Poirier N, Pierrot M, et al. Correlation between cerebral oxygen saturation measured by near-infrared spectroscopy and jugular oxygen saturation in patients with severe closed head injury. Anesthesiology 1999;91:985–90.
10. Siegemund M, van Bommel J, Ince C. Assessment of regional tissue oxygenation. Intensive Care Med 1999;25:1044–60.
11. Rampil IJ, Litt L, Mayevsky A. Correlated, simultaneous, multiple-wavelength optical monitoring in vivo of localized cerebrocortical NADH and brain microvessel hemoglobin oxygen saturation. J Clin Monit 1992;8:216–25.
12. Nakase H, Heimann A, Kempski O. Alterations of regional cerebral blood flow and oxygen saturation in a rat sinus-vein thrombosis model. Stroke 1996;27:720–7.
13. Nakase H, Zhenquan S, Kotani A, et al. Cerebral blood flow and tissue oxygen saturation in immediate and progressive ischemia in rat brain. Neurol Res 2001;23:875–80.
14. Takashima S, Ando Y. Reflectance spectrophotometry, cerebral blood flow and congestion in young rabbit brain. Brain Dev 1988;10:20–3.
15. Frank KH, Kessler M, Appelbaum K, Dummler W. The Erlangen micro-lightguide spectrophotometer EMPHO I. Phys Med Biol 1989;34:1883–900.
16. Bereczki D, Wei L, Otsuka T, et al. Hypoxia increases velocity of blood flow through parenchymal microvascular systems in rat brain. J Cereb Blood Flow Metab 1993;13:475–86.
17. Brown MM, Wade JP, Marshall J. Fundamental importance of arterial oxygen content in the regulation of cerebral blood flow in man. Brain 1985;108:81–93.
18. Shockley RP, LaManna JC. Determination of rat cerebral cortical blood volume changes by capillary mean transit time analysis during hypoxia, hypercapnia and hyperventilation. Brain Res 1988;454:170–8.
19. Julien-Dolbec C, Tropres I, Montigon O, et al. Regional response of cerebral blood volume to graded hypoxic hypoxia in rat brain. Br J Anaesth 2002;89:287–93.
20. Julien C, Payen JF, Tropres I, et al. Assessment of vascular reactivity in rat brain glioma by measuring regional blood volume during graded hypoxic hypoxia. Br J Cancer 2004;91:374–80.
21. Brun NC, Moen A, Borch K, et al. Near-infrared monitoring of cerebral tissue oxygen saturation and blood volume in newborn piglets. Am J Physiol 1997;273:H682–6.
22. Nwaigwe CI, Roche MA, Grinberg O, Dunn JF. Effect of hyperventilation on brain tissue oxygenation and cerebrovenous PO2 in rats. Brain Res 2000;868:150–6.
23. Johannsson H, Siesjo BK. Cerebral blood flow and oxygen consumption in the rat in hypoxic hypoxia. Acta Physiol Scand 1975;93:269–76.
24. Ulatowski JA, Bucci E, Razynska A, et al. Cerebral blood flow during hypoxic hypoxia with plasma-based hemoglobin at reduced hematocrit. Am J Physiol 1998;274:H1933–42.
25. Hemphill JC, Knudson MM, Derugin N, et al. Carbon dioxide reactivity and pressure autoregulation of brain tissue oxygen. Neurosurgery 2001;48:377–83.
26. Scheufler KM, Rohrborn HJ, Zentner J. Does tissue oxygen-tension reliably reflect cerebral oxygen delivery and consumption? Anesth Analg 2002;95:1042–8.
© 2006 International Anesthesia Research Society