Continuous

Intra-arterial R.Haessler,

Oximetry, Pulse Oximetry, Cardiac Surgery

P

ULSE oximetry has become a standard method of monitoring in the perioperative period.‘J Capillary oxygen saturation, which is measured by pulse oximetry (SpO& represents arterial oxygen saturation (SaO,). Sa02 multiplied by the concentration of hemoglobin represents arterial oxygen content, if the dissolved quantity of oxygen is neglected. A critical reduction of content (=hypoxemia) results in hypoxia of the tissues. Hypoxia can develop rapidly, especially in patients with cardiopulmonary disease, whenever ventilation or perfusion is altered (eg, by surgical intervention or by general anesthesia” or by decreases of cardiac output (CO).4) Therefore, there is a need for a reliable and continuous method of measurement of saturation, particularly in patients with cardiopulmonary disease. In healthy patients pulse oximetry is a reliable method; however, failure of pulse oximetry in critically ill patients is well documented.t,5-‘o An association has been found between failure of pulse oximetry and low CO, peripheral vasoconstriction, high systemic vascular resistance (SVR), and hypothermia. These conditions are commonly present in patients undergoing cardiothoracic surgery. Therefore, an alternative method was sought for obtaining precise and reliable oxygen saturation values in this patient population. From a theoretical point of view, intra-arterial catheter oximetry seemed to be a suitable alternative method. The aim of this study was to evaluate intra-arterial catheter oximetry and pulse oximetry in comparison to co-oximetry in critically ill patients during surgery. METHODS After consent,

obtaining institutional approval and written informed eight adult patients of ASA physical status class III-IV

From the Institute of Anesthesiology, University ofMunich, Klinikum Grosshadem, Munich, Germany. Reprints are not available. Copyright 0 I992 by W.B. Saunders Company 1053-0770/92/0606-0005$03.0000/0 668

During

MD, F. Brandl, MD, M. Zeller, MD, J. Briegel, MD, and K. Peter, MD

This study evaluated arterial catheter oximetry versus pulse oximetry in eight patients (ASA III-IV) who underwent cardiac surgery. Co-oximeter saturation values served as the standard. Arterial oxygen saturation was determined simultaneously with these three methods at 162 prospectively defined points of measurement before, during, and after cardiopulmonary bypass (CPB). At the same times before and after CPB, arterial, pulmonary arterial, and central venous pressures, and cardiac output determinations were recorded. Saturation readings were obtained in more than 99% of measurements with catheter oximetry and in only 59% to 84% of measurements with pulse oximetry. Failure of pulse oximetry correlated with low mean arterial pressures and low cardiac outputs, but not with high systemic vascular resistance. The mean saturation values determined by catheter oximetry as well as by pulse oximetry differed from the mean values obtained by co-oximetry by less than 1%

cardiac

and Co-oximetry

(=bias). The standard deviations of the individual differences between readings of catheter or pulse oximetry and readings of co-oximetry (=precision) were *0.5% to -cl.O% for catheter oximetry and *l.O% to *1.2% for pulse oximetry. In summary, catheter oximetry was superior to pulse oximetry with regard to obtaining readings and to reliability of the obtained readings. Invasiveness and high costs influence the decision as to whether to use catheter oximetry, but if reliable and precise measurements of saturation are important at any time during surgery, pulse oximetry is an insufficient method and co-oximetry is a time-consuming method of analysis, whereas catheter oximetry is quick, reliable, and precise. Copyright o 1992 by W.B. Saunders Company KEY WORDS: oximetry, cardiac, cardiothoracic

surgery

were investigated. Six of eight patients had cardiac surgery in the supine position; 2 patients had combined cardiac and pulmonary surgery in the lateral decubitus position. The mean preoperative cardiac index (CI) was 2.2 k 0.5 L/minim* (mean ? SD), the mean left ventricular end-diastolic pressure was 17 k 7 mmHg. and the mean SVR was 2,480 2 640 dynes. sec. cme5. Prior to induction of anesthesia, a Nellcor finger probe (Nellcor. Haywood, CA) for a Siemens-Sirecust pulse oximeter (Siemens, Germany) was placed on the left index finger. The hand was covered during surgery and neither infrared warming devices nor dyes that could interfere with pulse oximetry was used during surgery. Under local anesthesia the right femoral artery was cannulated with a 4.5-Fr introducer set (USCI-HP Medica. Germany). A U425C oximetry catheter (Abbott, Germany, 4 French = 1.33 mm) was placed through the introducer. These catheters contained fiberoptics and a lumen for blood sampling and pressure measurements, and had been calibrated in vitro just before insertion. No catheter was recalibrated in vivo. They were attached to an Oximetrix 3 SOzICO Computer (Abbott). After induction of anesthesia with etomidate and fentanyl, a 7.5.Fr pulmonary artery catheter (Spectramed, Germany) was introduced for hemodynamic measurements. Heart rate, arterial, pulmonary arterial, and central venous pressures were recorded continuously with a sixchannel recorder (Siemens, Germany). Anesthesia was maintained with fentanyl combined with flunitrazepam or propofol. Pancuronium was given for muscle relaxation. Oxygen/medical air was used for ventilation. The times for measurements were prospectively defined, because in addition to evaluation of the accuracy of saturation readings, the frequency of failure to obtain saturation data at prospectively defined points was evaluated. Altogether, at 162 prospectively defined times during general anesthesia, heparinized arterial blood samples were drawn for measurements of oxygenated hemoglobin (02.Hb). methemoglobin (Met-Hb), and carboxyhemoglobin (CO-Hb). The samples were analyzed immediately on a calibrated IL 282 co-oximeter (Instrumentation, Lexington, MA). On 85 of these occasions, before or after cardiopulmonary bypass (CPB), determinations of CO were performed in quadruplicate and the SVR was calculated. On all 162 occasions, saturation was intended to be measured by means of catheter oximetry using the intra-arterial U425C catheter. Eighty-five saturation readings were obtained before/after CPB. 29 readings during partial CPB with pulsatile flow under normothermic rectal temperature (temperature >34”C), 43 values during CPB with

Journalof Cardiothoracic and VascularAnesthesla, Vol6, No 6 (December),

1992: pp 668-673

OXIMETRY DURING CARDIAC SURGERY

669

nonpulsatile flow (aorta cross-clamped) under hypothermia (rectal temperature I 34°C); five additional readings obtained during normothermic CPB could not be analyzed because of a technical failure that had occurred independently of catheter oximetry. On 119/162 occasions, readings of pulse oximetry were recorded: 85 measurements before/after CPB, and 34 measurements during partial CPB with pulsatile flow under normothermic conditions; during CPB with nonpulsatile flow, readings of pulse oximetry were not recorded. The points of measurement before and after CPB were assigned to two groups for statistical analysis with the hemodynamic results (Fig 1): group 1 included all points of measurement at which saturation signals could be obtained by pulse oximetry; group 2 included all points at which no pulse oximetry signals were delivered. The mean values of arterial pressure, CO, and SVR of the two groups were compared by means of the Mann-Whitney test. Both catheter oximetry and pulse oximetxy cannot determine the amount of CO-Hb and Met-Hb; with co-oximetry, the exact amounts of CO-Hb, Met-Hb, and Oz-Hb are measured selectively and indicated.“,‘* Compared to co-oximetry, higher readings of saturation are obtained by both pulse oximetty and catheter oximetry.12 The values of pulse oximetty (y) or catheter oximetry (y) were, therefore, corrected by a mathematical procedure for comparison with co-oximetry (x): % Ox-Hb(x) = % Oz-Hb(y) x [l-(% CO-Hb + % Met-Hb)/lOO]. Agreement among catheter oximetry, pulse oximetry, and cooximetry was studied by linear regression analysis. In addition, mean differences (=bias) and standard deviations of the individual differences (=precision) of catheter and pulse oximetry were calculated in relation to co-oximeter readings.13,14 With bias and precision, a 95% interval of confidence was calculated, which gave evidence about the range in which 95% of catheter or pulse oximeter readings were located, if the co-oximeter readings were known. In order to focus attention on the magnitude and characteristics of the differences, the data are plotted as the individual means for the two measurement techniques versus the individual differences in Figs 2 through 5. In Figs 6 and 7 the individual differences are plotted against the body temperature.

mean arterial pressure (mmtig)

(-

p34”C) with pulsatile flow, SpOz was only obtained at 59% of the points of measurement (Table 1). In contrast to this, with catheter oximetry before, during, and after CPB, saturation signals were delivered at more than 99% of the previously defined points of measurement, regardless of whether pulsatile or nonpulsatile flow and normothermia or hypothermia were present. Figure 1 shows the hemodynamics at times when signals could be obtained with pulse oximetry in comparison to times when no signals were delivered. When readings could be obtained, the mean arterial pressure (MAP) (84 k 16 (2) difference between %-SO2 (intra-

(1) mean for %-SO2

I -5 1 90

91

92

93

94

95

96

97

Fig 5. Intra-arterial and co-oximetry pass. See Fig 2 for abbreviations.

98

99

(i&arterial) and %-S02(cooximeby)

during cardiopulmonary

by

mmHg, range, 53 to 133) was significantly higher (P < 0.01) than when no readings were obtained (70 t 14 mmHg, range, 51 to 94). Furthermore, when signals were delivered, CO (5.0 2 1.1 Limin, range, 2.8 to 8.1) was significantly (P < 0.05) higher compared to the times when no signals were delivered (4.4 t 0.8 L/min, range, 3.4 to 5.7). The mean values of systemic vascular resistance (1,334 t 447 dynes. set cm-“, range, 634 to 2,662 and 1,238 k 354 dynes sec. cmm5, range, 716 to 1873) were not significantly different in the two groups. The analysis of linear regression between values of the independent variable co-oximetry and values of the depen-

(2) difference between %-SO2 (intraarterial) and %-SO2 (cooximetry) 5

3

arterial) and %-SO2 (cooximetry) 5T

2 (5)

4

41 I

3 1

_2

-----------

(5)

__--__---_-------.-

- _ - - - - _ - _ ,.-

----‘-

(6)

-4

(1) mean for %-SO2 (inusarterial) and

-5 90

91

92

93

94

95

96

97

98

99

-51 34

35

36

37

38

temperature (degrees Celsius)

%-S02(ccxkmetry)

Fig 4. Intra-arterial and co-oximetry before and after cardiopulmonary bypass. See Fig 2 for abbreviations.

Fig 6. Intra-arterial oximetry during normothermic cardiopulmonary bypass (partial bypass, pulsable flow). See Fig 2 for abbreviations.

671

OXIMETRY DURING CARDIAC SURGERY

Table 1. Characteristics

of the Investigated Methods of Oximetry

Pointsof Measurement

Method

Before or after cardiopulmonary bypass (CPB)

Pulse oximetry

Values Obtained/ Performed Measurements n = 71185 84% successful

During CPB, pulsatile flow, normothermia

n = 20/34 59% successful

Intraarterial oximetry

n = 84185

Before or after CPB

99% successful During CPB, pulsatile flow, normothermia

n = 29/29 100% successful

During CPB, nonpulsatile flow hypothermia

n = 43143 100% successful

dant variable pulse oximetry, which were all obtained before and after CPB, resulted in a regression coefficient of r = 0.75 (P < 0.001). Saturation readings were between 90% and 100%. The mean saturation measured by pulse oximetry was 0.1% lower than the mean saturation measured by co-oximetry (=bias). The standard deviation of the individual differences was 2 1.2% (=precision). The mean of the 95% interval of confidence was, therefore, 0.1% lower than the corresponding mean co-oximeter reading. The range of the interval was 22.4% (Table 1, Fig 2). When comparing the pulse oximeter values obtained during CPB and pulsatile flow with values obtained by co-oximetry, a regression coefficient r = 0.59 (P < 0.01) resulted. Saturation values were delivered only at 59% of prospective points of measurement. Bias of -0.8% and a precision of *l.O% were found (Table 1, Fig 3). All 20 saturation values were measured during partial CPB with pulsatile flow under normothermia. Rectal temperatures below 34°C were only present during total CPB with a cross-clamped aorta. Because nonpulsatile flow was present at the same time,

(2) difference between %-SO2 (intraarterial) and %-SO2 (cooximetry) 5 4 3 i 2-l-’

____------.

0_-

;’ .I’:-

_l_

_

_

‘, -

-

Y c

(4)

-- _

(3) (6)

-3 -4 1 -5 1 26

R = 0.75

~o.l%+l.z%

(p < 0.001) R = 0.59

~O.a~/Oti.O%

(p < 0.01) R = 0.82

0.4%? 1 .O%

(p < 0.001) R = 0.84

-0.3%~0.6%

(p < 0.001) R = 0.79

0.1%?0.5%

(p < 0.001)

readings delivered by pulse oximetry were not statistically analyzed for this period of cardiac surgery. However, a trend analysis produced a continuous line of pulse oximeter readings for this period. Only 8% (mean) of the length of this line was within a saturation range between 60% and 100%. Comparison of values obtained by intra-arterial catheter oximetry and co-oximetry in the periods before and after CPB resulted in a regression coefficient of r = 0.82 (P < 0.001). The mean value delivered by catheter oximetry was 0.4% higher than the mean value obtained by co-oximetry (bias). The standard deviation of the individual differences was ?l.O% (precision) (Table 1, Fig 4). In contrast to pulse oximetry during CPB, with the intraarterial catheter, saturation readings were indicated at all prospectively defined points of measurement. The regression coefficient concerning comparison with co-oximetry was r = 0.80 (P < 0.001). The statistical analysis resulted in a bias of -0.1% and a precision of &0.6% (Fig 5). The measurements during CPB were performed during normothermia (rectal temperature >34”C) and during hypothermia (T I 34°C). The comparison of catheter oximeter values, which were recorded under normothermic conditions with corresponding co-oximeter values resulted in a regression coefficient of r = 0.84 (P < O.OOl), a bias of -0.3%, and a precision of ?0.6% (Fig 6). Comparison of corresponding values that were obtained under hypothermic conditions resulted in a regression coefficient of r = 0.79, a bias of O.l%, and a precision of ?0.5% (Fig 7). DISCUSSION

# 5 “_, F, 7’ -..: - Z_*- -

.

24

Bias f Precision (systematicaland randomerror)

_(5)

-2 --

22

Regression Coefficient

28

30

32

temperature 34 (degrees Celsius)

Fig 7. Intra-arterial oximetry during hypothermic cardiopulmonary bypass (nonpulsatile flow). See Fig 2 for abbreviations.

Agreement between saturation values obtained by catheter and pulse oximetry with values obtained by co-oximetry was studied by linear regression analysis and by calculation of bias and precision. Strictly speaking, regression analysis requires that the standard or independent variable, which is co-oximetry, be equivalent to a “gold standard.” For the IL 282 co-oximeter, the 95% interval of confidence of the random error is 2 l%‘* provided that pH is between 7.0 and 7.4, the fraction of Met-Hb between 0 and lo%, and the concentration of hemoglobin between 12 and 16 g/dL. Because a more precise standard does not exist, a systematic error cannot be addressed for co-oximetry; therefore, and because of the random error of tl%, co-oximetry is

672

ruled out as a gold standard in a strict sense and the most important prerequisite for the application of linear regression analysis cannot be fulfilled.*5 In spite of this, regression analysis has been carried out and regression coefficients have been named, because this makes a comparison with earlier publications possible. However, the significance of another kind of analysis was considered more favorably: the calculation and interpretation of mean differences (bias) and of the standard deviations (precision) of the individual differences between the readings obtained by the different methods. This approach, proposed by Altmann and Bland,13J4 is more informative, especially if both methods of monitoring are imprecise and neither one is a true gold standard. Bias serves as a measure of the systematic difference between the hvo methods, whereas precision acts as a measure of random error. If readings of the standard method, ie, co-oximetry, are known, bias and the interval of two standard deviations on either side of the mean difference result in an interval in which 95% of readings delivered by the method of comparison, ie, catheter oximetry or pulse oximetry, are situated. This interval is called the “limit of agreement” between two methods.14 If the differences between individual corresponding values are plotted against the means for individual corresponding values, the distribution of the points shows whether there is a connection between the bias and the absolute dimension of the values, eg, a great systematic error for high saturation values. This kind of presentation is used in Figs 2 through 5 in which two methods are compared. Figures 2 through 5 reveal that, for mean saturation values lower than 97%, catheter readings as well as these obtained from pulse oximetry tended to be lower than co-oximeter readings. In Figs 6 and 7, no such tendency was seen for different body temperatures ranging from 22 to 38°C. The manufacturers of most pulse oximeters indicate their 95% interval of confidence as being lower than *4% for saturation values over 80%.6 The 95% interval of confidence of the study pulse oximeter was ~2.4% for saturation values between 90% and 100%. Bias and precision in this investigation are in the same range as in the study of Gabrielczyk and Buist,7 who evaluated pulse oximetry after cardiac surgery in hypothermic patients. They indicated that failure to obtain readings was a rare event. In contrast, the pulse oximeter used in the present study delivered no signals in more than 15% of the prospectively defined points of measurement. This is in accordance with the results of Tremper et a1,9who could not measure saturation values with pulse oximetry at 57 of 383 points of measurement; in 43% of these cases the CI was lower than 2.5 L/min/m2, in 16% core temperature was lower than 35”C, in 9% Hb concentration was lower than 8 g/dL, and in 35% the index of systemic vascular resistance (SVRI) was greater than 2,600 dynes. sec. cm-5 . m2. Furthermore, high SVRI values and low temperature values were associated with great systematic error in this study. Clayton et al6 studied 20 different pulse oximeters under conditions of poor tissue perfusion and could not obtain saturation readings at about 10% of the points of measurement. In an earlier study, where they had applied the corresponding

HAESSLER ET AL

methods in healthy patients with normal tissue perfusion, signals could not be obtained at only 0.3% of points of measurement. Pglve et alx found that measurements with pulse oximetry were possible provided that mean CI was greater than 2.4 Liminlm’, mean peripheral temperature greater than 26.5”C, and mean SVRI lower than 2,930 dynes set cme5 m*. Thus, these studies showed a connection between failure of pulse oximetry and low CO. low central body temperature, low peripheral temperature. low Hb-concentration, and high SVR. In the present study, failure was associated with a low MAP and CO. but not with high SVR. Not only failure to obtain readings at all, but also the reliability of obtained saturation readings is of importance. Desiderio et al5 evaluated pulse oximetry in patients with preexisting lung pathology undergoing thoracic surgery and concluded that pulse oximetry could not replace arterial blood gas sampling for the intraoperative management because pulse oximeter saturation values were not precise enough. The oximetry catheter used was designed for in vivo measurement of oxygen saturation in the umbilical artery of newborn infants. Wilkinson and coworkers’h,‘7 evaluated the catheter in newborn infants and found a correlation coefficient of r = 0.976 in relation to co-oximetry for saturation readings that were between 40% and lOO%, which was interpreted as close agreement. Bias and precision have not been calculated. In this study investigating critically ill adult patients with cardiac or cardiopulmonary diseases, a close agreement between both catheter and co-oximeter readings in terms of bias and precision was found for all periods before, during, and after CPB. A bias of 0.4% and a precision of + 1.0% are remarkably low and, therefore, encouraging if it is taken into consideration that the standard deviation of the IL 282 co-oximeter, which contributes to the statistical analysis as a random error, is ?0.5%. During CPB greater absolute values of saturation were present compared to the periods before and after CPB. These high saturation values may be responsible for the low bias (0.1%) and precision (+0.6%) during CPB with nonpulsatile flow. Furthermore, hypothermia proved to bo no problem for catheter oximetry. During hypothermia, bias and precision were even lower than during normothermic conditions, and there was no association between body temperature and random or systematic error. The most convincing advantage of catheter oximetry is the fact that readings in more than 99% of the prospectively defined points of measurement can be obtained (cvcn under circumstances that have caused pulse oximetry to fail). Precise saturation values can be obtained at any given time by co-oximetry as well; however, sampling of arterial blood and a time-consuming analysis are mandatory. In contrast, the catheter oximeter continuously delivers saturation values without having to conduct in vitro analyses. At first glance, it seems to be a disadvantage that the catheter has to be introduced before starting surgery. However. the authors have applied the catheter in two unexpected. prolonged emergency situations, with marginal saturation values, which had developed after CPB. The catheters were

OXIMETRY

673

DURING CARDIAC SURGERY

introduced through a 16G-“Abbocath” (Abbot) into the brachial artery and provided excellent results. Invasiveness and high costs may affect the decision as to whether to use the catheter; however, if reliable saturation values are

necessary at any given time of an operation for managing a critically ill patient, catheter oximetry is the method of choice compared to pulse oximetry and even compared to co-oximetry.

REFERENCES

1. Barker SJ, Tremper KK: Pulse oximetry: Applications and limitations. Int Anesthesiol Clin 25:155-176, 1987 2. Tremper KK, Barker SJ: Pulse oximetry. Anesthesiology 70:98-108, 1989 3. Dueck R, Young I, Clausen J, et al: Altered distribution of pulmonary ventilation and blood flow following induction of inhalational anesthesia. Anesthesiology 52:113-118, 1980 4. West JB: Ventilation-perfusion relationships, state of the art. Am Rev Resp Dis 116:919-943,1977 5. Desiderio DP, Wong G, Shah NK, et al: A clinical evaluation of pulse oximetry during thoracic surgery. J Cardiothorac Anesth 4:30-34,199o 6. Clayton DG, Webb RK, Ralston AC, et al: A comparison of the performance of 20 pulse oximeters under conditions of poor perfusion. Anaesthesia 46:3-10, 1991 7. Gabrielczyk MR, Buist RJ: Pulse oximetry and postoperative hypothermia. Anaesthesia 43:402-404, 1988 8. Palve H, Vuori A: Pulse oximetry during low cardiac output and hypothermia states immediately after open heart surgery. Crit Care Med 17:66-69, 1989 9. Tremper KK, Hufstedler SM, Barker SJ: Accuracy of a pulse oximeter in the critically ill adult: Effect of temperature and hemodynamics. Anesthesiology 63:A175,1985

10. Wilkins CJ, Moores M, Hanning CD: Comparison of pulse oximeters: Effects of vasoconstriction and venous engorgement. Br J Anaesth 62:439-444,1989 11. Barker SJ, Tremper KK, Hyatt J: Effects of methemoglobinemia on pulse oximetry and mixed venous oximetry. Anesthesiology 70:112-117,1989 12. IL 282 Operator Manual-part no. 79282, Lexington, MA: Instrumentation Laboratories, 1977 13. Altman DG, Bland JM: Measurement in medicine: The analysis of method comparison studies. Statistician 32:307-317, 1983 14. Bland JM, Altman DG: Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1:307-310,1986 15. LaMantia KR, O’Connor T, Barash PG: Comparing methods of measurement: An alternative approach. Anesthesiology 72:781-783,199O 16. Wilkinson AR, Phibbs RH, Gregory GA: Continuous measurement of oxygen saturation in sick newborn infants. J Pediatr 93:1016-1020,1978 17. Wilkinson AR, Phibbs RH, Gregory GA: Continuous in vivo oxygen saturation in newborn infants with pulmonary disease-A new fiberoptic catheter oximeter. Crit Care Med 7:232-236,1979

Continuous intra-arterial oximetry, pulse oximetry, and co-oximetry during cardiac surgery.

This study evaluated arterial catheter oximetry versus pulse oximetry in eight patients (ASA III-IV) who underwent cardiac surgery. Co-oximeter satura...
689KB Sizes 0 Downloads 0 Views