Original Articles
EVALUATIONOFA REGIONALOXYGENSATURATION CATHETERFORMONITORINGSJV02IN HEAD INJUREDPATIENTS
Ritter AM, Gopinath SP, Contant C, Narayan RK, Robertson CS. Evaluation of a regional oxygen saturation catheter for monitoring SjvO2 in head injured patients. J Clin Monit 1996; 12:285-291
Ann M. Ritter, MD, Shankar P. Gopinath, MD, Charles Contant, PhD, Raj K. Narayan, MD, and Claudia S. Robertson, M D
ABSTRACT. Objectives. Monitoring jugular venous oxygen saturation (SjvO2) has been useful for the early identification and treatment of cerebral ischemia in patients with severe head injury. However, the catheters that have been used for this purpose have not performed optimally. The purpose of this study was to evaluate the performance of a new regional oxygen saturation catheter for monitoring SjvO2. Methods. Eighteen regional oxygen saturation catheters, 4-Fr in diameter (Baxter Healthcare Corporation, Edward Critical Care), were used in this study. Each catheter was inserted percutaneously into the dominant jugular vein and the catheter's tip position in the jugular bulb was verified by radiograph. The catheter was calibrated in vitro prior to insertion using the optic calibrator provided by the manufacturer. The catheter was recalibrated every 8 to 12 hours by comparing the oxygen saturation value from the catheter with that measured by a cooximeter in a blood sample drawn through the catheter. Results. In vitro calibration using the optic calibrator was not always successful. Five catheters could not be calibrated. The remaining 13 catheters could all be calibrated, but only 9 provided a value that was within 4% of the oxygen saturation derived from the blood sample. After the first in vivo calibration, the correlation between the catheter and the blood sample values was improved. A total of 196 comparisons were made. The median, 25th, and 75th quartile differences between the catheter and the blood sample measurement of SjvO2 were 0.00, -1.15, and 1.25%, respectively. Using longitudinal data regression, the overall slope of the regression between the catheter and blood values was 0.997 (p = 0.001). Conclusions. The new regional oxygen saturation catheter provided reliable measurement of SjvO2 83% of the time when the signal quality index was <3, and may be useful for continuous monitoring of SjvO2.
KEY WORDS, Head injury, oxygen saturation, jugular bulb catheter.
INTRODUCTION
From the Department of Neurosurgery, Baylor College of Medicine, Houston, Texas. Received May 22,1995, and in revised form Oct 17, 1995. Accepted for publication Jan 10,1996. Address correspondence to Claudia S. Robertson, MD, Department of Neurosurgery, Baylor College of Medicine, 6560 Fannin St. Suite #944, Houston, TX 77030, U.S.A.
JournalofClinicalMonitoring12: 285-291, 1996. © 1996 KluwerAcademicPublishers.Printedin theNetherlands.
M o n i t o r i n g jugular venous o x y g e n saturation (SjvO2) provides information that is useful in the m a n a g e m e n t o f head-injured patients, A decrease in SjvO2 is an early manifestation o f decreasing cerebral blood flow and, therefore, can be used as a m o n i t o r for global ischemic insults [1]. SjvO2 is directly related to cerebral perfusion pressure (CPP) in head-injured patients and can be used to determine the optimal C P P for individual patients [2]. SjvO2 is directly related to the partial pressure o f arterial carbon dioxide (PCO2) and can be used to optimize hyperventilation as a treatment o f intracranial hypertension [3].
286 Journalof Clinical Monitoring Vo112 No 4 July 1996
SjvOa can be measured by placing a single lumen catheter in the jugular bulb and intermittently measuring oxygen saturation in a blood sample drawn through the catheter. Alternatively, SjvO2 can be monitored continuously by placing a fiberoptic oxygen saturation catheter in the jugular bulb. For monitoring for ischemic insults, the continuous measure appears to be advantageous. In a study o f 119 patients, 60% o f episodes of jugular venous desaturation were less than 1 hour in duration and would probably have been missed by intermittent measurements [1]. The purpose o f this study was to evaluate the performance of a new catheter (Model 94-040-4F, Baxter Healthcare Corporation, Edward-Critical Care, Irvine, CA), which has been specifically designed for monitoring oxygen saturation in regional vasculatures such as the jugular vein. Both the accuracy of the in vitro calibration, and the in vivo accuracy o f the SjvO2 measurements were examined.
Monitor
to Sa02 sensor
Optical module Cathode connector Blood sampling cannula
Blood channel
lmm
METHODS
Patient characteristics Eighteen catheters were used in 17 patients with severe head injury or subarachnoid hemorrhage (Glasgow Coma Scale _<8) who were admitted to the Ben Tanb General Hospital Neurosurgical Intensive Care Unit between September 1993 and January 1994.1 Twelve patients had sustained blunt head injury: four had cortical contusions, five had extracerebral hematomas, and three had diffuse axonal injury. Four patients sustained penetrating injuries to the head, and one patient had an aneurysmal bleed.
Double lumen 02 saturation catheter
Transmitting and reeeiving optical fibers
Fig 1. Diagramdepictingthe SjvO2 catheter. manually entered. SjvO 2 is displayed as a percentage o f oxygenated hemoglobin to total hemoglobin. The displayed SjvO2 is updated every 2 seconds. After a step change in SjvO2, complete equilibration occurs within 15 seconds.
Insertion of the catheter Description of the oxygen saturation catheter The principle underlying the measurement o f oxygen saturation in the blood with the fiberoptic catheter is that oxyhemoglobin absorbs light differently than deoxyhemoglobin. Light of two specific wavelengths is pulsed down one fiberoptic cable contained within the catheter. Remitted light is transmitted back up the second fiberoptic cable contained within the catheter to a photosensor, which detects the absorbance light at the two wavelengths (Fig 1). The SjvO2 is determined from these absorbances and from the patient's hemoglobin concentration, which is 1 This research was reviewed by the Baylor IRB and approved initially on January 19,1993and renewed on January 19,1994.
The patients were placed supine, with the head turned gently away from the dominant jugular vein, which was determined by the intracranial pressure response to compression of the jugular vein. The side o f the neck and upper thorax were aseptically scrubbed and draped. Local anesthesia was achieved with 1% xylocaine. A 20gauge needle was inserted at the apex o f the triangle formed by the two heads o f the sternocleidomastoid muscle. The needle was advanced cephalad in a line with the ipsilateral nipple in a 30 ° angle from the coronal plane. When venous blood was flowing freely through the hub o f the needle, a J-wire was passed. A 4.5-Fr peelaway introducer (Cook Critical Care) was inserted over the J-wire. Finally, the catheter was inserted through the introducer until resistance was met at the jugular bulb, and was then withdrawn 0.5 cm. The catheter was firmly
Ritter et al: Sjv02 monitoring in Head-Injured Patients 287
sutured to the skin before the introducer was peeled away, and the distance from the bulb to the skin was documented. Lateral skull radiograph verified the position of the catheter tip.
Calibration of the catheters Before the catheters were inserted, an initial in vitro calibration was performed using the optic calibrator provided by the manufacturer. The optic calibrator is constructed of a polyethylene material that completely encompasses the catheter tip, allowing no light to enter the system. The in vitro calibration value should be 80 42% in order for the calibration to be accepted; otherwise, a "cal error" message is displayed. An in vivo calibration was done 10 to 15 minutes after insertion o f the catheter, when the catheter readings had equilibrated. Calibration was checked every 8 to 12 hours, and recalibration was performed if the difference between the catheter SjvO2 and the oxygen saturation of the blood sample drawn through the catheter and measured by the co-oximeter (Model IL-282, Instrumentation Laboratory) was >4%. Each routine calibration check was performed at a time when the patient was physiologically stable. The quality o f the fiberoptic catheter measurement was recorded at each calibration by the signal quality index (SQI) on the monitor. The SQI is a composite of indicators that monitors the infrared intensity signal and the signal fluctuation and records this data on a scale from 1 to 4. Infrared light is used for this purpose because it is relatively insensitive to changes in oxygen saturation. Large changes in the baseline intensity suggests hematocfit changes, damage to the optical fibers by a kink, or thrombus accumulation over the catheter tip. Increases in signal fluctuation suggests vessel wall interference, while decreases suggest the presence o f a thrombus or migration of the catheter tip into a small vessel. An SQI of 1 to 2 indicates optimal conditions for measurement of SjvO2. An SQI of 3 - 4 indicates one of the problems described above.
Data collection Two aspects of catheter performance were tested: (1) the accuracy o f the in vitro optic calibrator, and (2) the longterm accuracy of the oxygen saturation measurements after the first in vivo calibration. In both circumstances, the catheter oxygen saturation value was compared with oxygen saturation in a blood sample, drawn anaerobically from the catheter and measured on a co-oximeter
(Model IL-282, Instrumentation Laboratory). The cooximeter was calibrated on each day of the study using commercial standards (Reference Standard for IL C O Oximeter, Baxter Scientific Products, McGraw Park, IL). O x y g e n saturation was measured in control samples (Certain Advance, Ciba-Corning Diagnostics C0rp, Medfield, MA) with each measurement of oxygen saturation in the blood samples. The accuracy of oxygen saturation measurements by the co-oximeter was 1% full-scale, and the precision was 0.5% full-scale.
Statistical analysis Standard statistical methods for analyzing performance of physiological monitors include: (1) Bias (or mean difference between catheter and cooximeter values); the bias should be close to zero. Bias over the range of physiological values can also be examined by plotting the difference between the two values against the co-oximeter values. In this case, the mean difference should remain close to zero throughout the physiological range. (2) Precision (or standard deviation o f the difference between catheter and co-oximeter values). As with bias, precision should be close to zero. (3) Linear regression analysis, comparing the co-oximeter values as the independent variable and the blood sample values as the dependent variable. An r-value of 0.90 to 0.95 would be considered acceptable for clinical monitoring purposes. As discussed below, and in the results section, assumptions of these standard tests were violated by the current set of data, and therefore additional statistical methods were used. Bias and precision (mean and standard deviation) require that the data have a normal distribution. The distribution of the data was examined using the lettervalue display and robust estimates of the means as implemented in the STATA package (STATA, Version 2.1, Stata Corp, College Station, TX). Since these tests demonstrated that the distribution was not normal, the data were displayed as a frequency histogram and summarized by median and interquartile values. Linear regression analysis requires that all of the observations be independent, but in the current study, the multiple observations obtained from each individual were found to be correlated. Therefore, to analyze the relationship between the SjvO2 measured with the cooximeter and with the catheter, the method o f Laird and Ware was used [4]. The model allows for examination of the overall regression relationship and individual deviations from that relationship. The data were transformed
288 Journal of Clinical Monitoring Vo112 No 4 July 1996
so that each individual's mean for both the co-oximeter and catheter data was zero. Therefore, the intercept should be zero for both the overall and for the individual regressions, and the overall slope should be one. The individual deviations from that slope should be near zero. After fitting the model, the overall slope was tested against zero, and the individual intercepts and slopes were plotted. Large deviations would be indicative o f individual variations in the relationship of the co-oximeter and catheter readings.
40
30 o £L
2O o 0 Q.
0<-20-20 -18 ~16 -14 -12 *t0 -8 -8 -4 -2
4
6
8
10 12 14 46 18 20 >20
Difference B e t w e e n C a t h e t e r a n d Blood S j v O 2
RESULTS A total of 209 simultaneous measurements of SjvO2 were made by the oxygen saturation catheter and by the co-oximeter. Thirteen of these measurements were performed immediately after insertion of the catheter, and reflected the accuracy o f the in vitro calibration. The remaining 196 measurements were after the initial in vivo calibration, and were analyzed separately as a measure o f the in vivo performance of the catheter.
Fig 2. Distribution of the difference between the Sjv02 value obtained by measurement with the catheter and with the co-oximeter. Eighty-one percent of the 196 comparisons from all 18 catheters are within 4%.
60
40 o
o
20 a
o °n
Accuracy o f the in vitro calibration f o r all 18 catheters 0
An in vitro calibration using the manufacturer's optic calibrator was possible in only 13 o f the 18 catheter insertions. Five of the catheters gave a "cal error" message. Immediately after insertion o f these 13 catheters, a blood sample was obtained to check the accuracy o f the in vitro calibration. The median difference between the SjvO2 f r o m the catheter and the blood sample was 0.8%, but ranged f r o m - 2 4 . 2 to 32%. Four of the 13 catheters had to be recalibrated immediately after insertion because the difference between the catheter value and the blood sample values exceeded 4% on the first in vivo check. Therefore, only 9 (50%) of the 18 catheters provided an acceptable (within 4%) in vitro calibration.
In vivo performance o f all 18 S j v 0 2 catheters The in vivo performance of all 18 catheters was examined, regardless o f whether or not the in vitro calibration was successful. One hundred ninety-six simultaneous measurements o f SjvO2 from the 18 catheters were available for analysis after the first in vivo calibration check. The SjvO2 measured in the blood samples averaged 64.0 + 10.0, and ranged from a low of 27.6 to a high of 85.6%. O n 159 (81%) of the 196 comparisons, the catheter and blood sample SjvO2 were within 4% and recalibration of the catheter was not required.
-20
o°o o
o
o
o o
Smoothed
-40
Mean
o
i
-60 20
i
i
30
40
i
~
i
i
i
50
60
70
80
90
100
Oxygen Saturation in the Blood
Fig 3. Relationship between the difference in the catheter saturation and blood saturation plotted against the saturation in the blood. The open circles indicate the actual reading, while the dark line shows the smoothed estimation of the relationship between the two values.
The differences between the two measurements were not normally distributed, as shown in Figure 2. The median o f the distribution was 0.00 with 25th and 75th quartiles of -1.15 and 1.25%, respectively. Outlying values tended to occur more often in the negative direction (blood sample > catheter value) than in the positive. To determine if these outlying values represented a significant bias in the catheter values, the relationship between the difference in the catheter saturation and blood saturation was plotted against the saturation in the blood (Fig 3). The lack o f a slope in the smoothed mean and absence of a large deviation from the zero line indicates there is little bias or offset in the catheter readings.
Ritter et al: Sjv02 monitoring in Head-Injured Patients
The relationship between the catheter and co-oximeter values were plotted for each patient. The correlation coefficient for the individual regressions ranged from -0.35 to 0.98. Nine patients had a good correlation (r > 0.9), and nine had a poor correlation (r < 0.9). For these reasons, linear regression analysis was not appropriate for the pooled data, and longitudinal data regression was used. The overall slope of the regression between the catheter and the blood values was 0.997 (comparison of observed slope to zero slope, p < 0.001).
289
50
v)
4O
o
r~ 30
o 20 rl
10
ili i{ iliili <-20-20 -18 -16 -14 -12 -10 -8
-6
-4
-2
2
4
6
8
10
12
14 16
18 20 >20
Difference Between Catheter and Blood SjvO 2
Factors that affected catheter performance Since the performance of the catheters varied significantly from patient to patient, several factors that might have affected catheter performance, including position of the catheter in the jugular vein, accuracy of the in vitro calibration, and intensity of the reflected light, were examined.
Fig 4A. Frequency histogram of the difference between catheter and blood sample Sjv02 in the nine patients with successful i n v i t r o catheter calibrations (n = 107 observations). Ninety-two percent of the observations are within 4%.
5o
v)
t0 i
Catheter position
n 30 i
The optimal catheter position for monitoring SjvO2 is controversial To ensure that the catheter is measuring oxygen saturation in cerebral venous blood, the catheter tip should be positioned as high in the jugular bulb as possible. Studies have demonstrated that when the catheter tip is more than 2 cm below the base of the skull, significant extracerebral contamination may occur [5]. Others have suggested that placing the catheter above the lower border o f the C1 vertebral body [6] or at the level of the C2 vertebral body [7] reduces the chance that the catheter tip will abut against the base of the skull or the vessel wall and improves performance of the catheter. After insertion, the catheter position was examined by obtaining a lateral radiograph, and was classified as (1) tip at the base of the skull, (2) tip at C1 vertebral body, or (3) tip at C2 vertebral body. Although the number o f patients in each category was small, the catheter performance appeared best when the catheter was positioned at the base of the skull or at C2 (Fisher's exact test, p = .03). In all three of the patients with the catheter tip at the base o f the skull, the correlation between the catheter and the co-oximeter values was good. In the 13 patients in w h o m the catheter tip was at the level of C1, only four had good correlations. Both o f the two patients with the catheter tip at the level o f C2 had good correlations.
0 20 10
<-20-20 -18 -16 -14 -12 -10 -8
-6
-4
-2
2
4
6
8
10
t2
14
16
18 20 >20
Difference Between Catheter and Blood SjvO 2
Fig 4B. Frequency histo2ram of the difference between catheter and blood sample Sjv02 in the nine patients with unsuccessful i n v i t r o catheter calibrations (n = 102 observations). Only 63% of the observations are within 4%.
Accuracy of the in vitro calibration The in vitro calibration was successful (within 4% of the co-oximeter value on the first in vivo check) in only nine o f the catheters. In five catheters, the in vitro calibration could not be completed, and in an additional four catheters, the in vitro calibration gave a value that was more than 4% from the co-oximeter value on the first in vivo check. A successful in vitro calibration predicted subsequent good performance. O f the nine patients with a successful in vitro calibration, seven had good correlation (r > 0.9) between the catheter and the co-oximeter values, and 92% o f the comparisons were within 4% (Fig 4a). O f the nine patients with an unsuccessful in vitro calibration, only two had good correlation, and only 63% of the comparisons were within 4% (Fig 4b).
290 Journalof Clinical Monitoring Vo112 No 4 July 1996
Signal quality index
In vitro optic calibrator
SQI was 1, 2, 3, or 4 on 59, 52, 34, and 38 occasions, respectively. SQI was not recorded on 14 occasions. The median differences between the catheter and blood sampie measurements of SjvO2 when the SQI was 1, 2, 3, and 4 were -0.1, -0.1, -0.2, and 0.45%, respectively. The number o f outlying data points tended to be larger in the SQI 4 group when the letter-value displays were examined in each group. When the SQI was _<3, 83% (121/145) of the values had a difference o f < 4 % . When SQI was 4, 68% (26/38) of the values had a difference o f < 4 % (p = 0.06).
The in vitro optic calibrator will clearly have to be reworked by the manufacturer. Only 50% of the catheters had a satisfactory in vitro calibration (i.e., within 4% o f a blood sample measurement), and five o f the catheters gave a "calibration error" message. Examination o f the catheters that did not calibrate suggested that there might have been a light leak around the catheter. The in vitro calibration is not necessary for subsequent performance of the catheter, since the first in vivo calibration erases the initial calibration. However, it is a convenience to have a reading of SjvO2 immediately upon insertion of the catheter, and a successful in vitro calibration suggests that the catheter is in a position that is favorable for accurate recordings o f SjvO2.
DISCUSSION
Sjv02 monitoring after head injury In vivo performance of the catheter Cerebral ischemia is one o f the most c o m m o n causes o f secondary injury in patients with severe head injury. SjvO2 is theoretically a useful monitor for cerebral hypoxia/ischemia because it reflects the balance between oxygen delivery to and oxygen consumption by the brain. Both delivery and consumption parameters can be abnormal after a head injury, and the relative balance between these two parameters is often more valuable information than the absolute level o f either o f the parameters alone. Any disturbance that increases cerebral oxygen consumption (fever or seizures) or decreases oxygen delivery (hypotension, hypoxia, hypocapnia, intracranial hypertension) may decrease SjvO2. Studies using SjvO2 to monitor patients with severe head injury have identified transient episodes of jugular desaturation in nearly 40% o f patients monitored, and the causes are often treatable systemic causes [1]. The occurrence o f one or more episodes of jugular desaturation are strongly associated with a poor neurological outcome, suggesting that prevention or early treatment o f cerebral ischemia may improve outcome after head injury. The previous catheters that have been used for continuous SjvO2 measurements were not manufactured for use in the internal jugular vein, but were instead designed for oxygen saturation monitoring of the pulmonary artery or neonatal umbilical artery. Many problems have been encountered in adapting these catheters to the jugular bulb, and the accuracy and stability of the saturation values have been questioned. The catheter evaluated in the present study was intended for monitoring oxygen saturation in regional vasculatures such as the jugular vein.
Although this was not a prospective comparison of the new catheter with the previously available catheters, a number of previous observations on the performance of fiberoptic oxygen saturation catheters are available for comparison. As in the present study, the catheters in these published studies were calibrated in vitro prior to inserting the catheter and then in vivo after insertion, and the performance of the two calibrations has been reported to vary markedly. Andrews et al. [6], in a study of 25 patients, reported a bias o f -10.82% (oxygen saturation from the blood sample compared with that from the catheter; a correlation coefficient o f 0.61 after the in vitro calibration; but a bias o f only 0.85% and a correlation coefficient of 0.91 when only values after the first in vivo calibration were considered. The performance deteriorated after 12 hours and the authors recommended routine calibration of the catheter at least every 12 hours. Sheinberg et al. [8] found a similar correlation coefficient o f 0.87 for the relationship between blood and catheter oxygen saturation after the first in vivo calibration and when light intensity indicated good positioning o f the catheter in a study of 45 patients. The calibration o f the catheter was checked every 8 hours in this study. On 40% o f these checks, the catheter and blood oxygen saturation values differed by > 4 % , requiring recalibration o f the catheter. Cruz [3] reported a similar correlation coefficient o f 0.92 in a study o f 69 patients. Fortune et al. [7] reported a correlation coefficient o f 0.58 in a study of 14 patients. The calibration was checked every 4 to 8 hours and on 38% of the occasions, the difference between catheter and blood oxygen saturation values was > 6 % , requiring recalibration. In contrast to these previously published studies where
Ritter et al: SjvOe monitoring in Head-Injured Patients
recalibration was required in 38 to 40% of the comparisons, the new catheter in the present study required calibration in only 19% o f the comparisons. The criteria for recalibrating the catheter and the frequency of the checking of the calibration varied in these studies, so caution should be used in making such comparisons. Nevertheless, the in vivo performance of the new catheter appears to equal or exceed that o f the previously available catheters. This work was supported by NIH grant -¢~PO1-NS26716.
REFERENCES 1. Gopinath SP, Robertson CS, Contant CF, Hayes C, Feldman Z, Narayan RK, Grossman RG. Jugular venous desaturation and outcome after head injury. J Neurol Neurosurg Psych •994; 57:717-723 2. Chan KH, MillerJD, Dearden NM, Andrews PJ, Midgley S. The effect of changes in cerebral perfusion pressure upon middle cerebral artery blood flow velocity and jugular bulb venous oxygen saturation after severe brain injury. J Neurosurg 1992; 77:55-61 3. CruzJ. Combined continuous monitoring of systemic and cerebral oxygenation in acute brain injury: Preliminary observations. Crit Care Med 1993; 21:1225-1232 4. Laird NM, WareJH. Random-effects models for longitudinal data. Biometrics 1982; 38:963-975 5. Jakobsen M, Enevoldsen E. Retrograde catheterization of the right internal jugular vein for serial measurements of cerebral venous oxygen content. J Cereb Blood Flow Metab 1989; 9: 717-720. 6. Andrews PJD, Dearden NM, Miller JD. Jugular bulb cannulation: Description of a cannulation technique and validation of a new continuous monitor. BrJ Anaesth 1991; 67:553-558 7. FortuneJB, Feustel PJ, Weigle CGM, PoppJA. Continuous measuremen of jugular venous oxygen saturation in response to transient elevations of blood pressure in headinjured patients. J Neurosurg •994; 80:461-468 8. Sheinberg M, Kanter MJ, Robertson CS, Contant CF, Narayan RK, Grossman RG. Continuous monitoring of jugular venous oxygen saturation in head-iNured patients. J Neurosurg 1992; 76:212-217
29