EFFECTOF HEPARINIZATIONOF CATHETERS ON PULMONARYARTERYOXlMETRY Gwendolyn M. Stritter, MD, Ronald G. Pearl, MD, and Frederick G. Mihm, MD
Stritter GM, Pearl RG, Mihm FG. Effect of heparinization of catheters on pulmonary artery oximetry. J Clan Monit 1988;4:204-209 ~b~'. A clinical study was performed in two phases to determine whether pulmonary artery oximeter catheters that were impregnated or bonded with heparin would affect the accuracy of measurements of in vivo mixed venous oxygen saturation (S'~O2). In phase 1, 40 patients were catheterized with either a heparin-impregnated or a plain pulmonary artery catheter. Blood was sampled at random times to correlate in vivo with in vitro S~O2 measurements. In phase 2, 16 patients who were not receiving systemic heparin therapy or aspirin and who had no coagulopathies were catheterized with either a heparin-bonded or a plain pulmonary artery catheter in a blinded order. In phase 1, a total of 364 blood samples were obtained from 40 patients. Linear regression analysis of the pooled data demonstrated y = 0.98x - 0.0t, r = 0.93, P < 0.001, and n = 141 with heparin-impregnated catheters; and y = 0.87x + 8.0, r = 0.81, P < 0.001, and n = 223 with plain catheters. The mean difference (in vivo minus in vitro) revealed a similar error ( - 1.3 -+ 0.4 versus - 1.4 -+ 0.4, respectively, mean +-- SE). The 95% confidence limits of an individual value (+ 8.1 versus +- 12.3) suggested slightly greater accuracy with heparin-impregnated catheters. In phase 2, a total of 134 blood samples were obtained from 16 patients. Linear regression analysis showed nearly equal performance with heparin-bonded and plain catheters (r = 0.97 versus r = 0.98, respectively) with similar slopes (1.0 versus 1.1, respectively) but different intercepts ( - 0 . 6 versus -8.4, respectively). Analysis of the mean difference revealed a measurement error o f 0.4 -+ 0.3 versus - 1.3 -+ 0.3 with similar 95% confidence limits of individual values (+ 5.0 versus + 4.8, respectively). These differences do not appear clinically important. These data suggest that heparinization only minimally enhanced accuracy with a pulmonary artery oximetry system, which accurately measured S~702 to within 3 to 4% of true values. KEYWORDS.Blood: anticoagulants; heparin. Complications:
blood clots. Equipment: catheters, pulmonary artery. Measurement techniques: oximetry. Monitoring: oxygen saturation, mixed venous.
From the Department of Anesthesia, Stanford University School of Medicine, Stanford, CA. Received Mar 16, 1987, and in revised form Nov 20. Accepted for publication Nov 30, 1987. Address correspondence to Dr Mihm, Department of Anesthesia, Room $278, Stanford University Medical Center, Stanford, CA 94305.
204 Copyright © 1988 by Little, Brown and Company
Solid-state o x i m e t r y catheters were introduced into clinical practice in the mid-1970s. T h e accuracy and utility o f these catheters for continuous m o n i t o r i n g o f o x y gen saturation (So2) have been d e m o n s t r a t e d in a variety o f clinical settings [1-9]. H o w e v e r , clot deposition on the catheter, an unpredictable event, produces falsely l o w So2 readings in vitro [2,4] and also can interfere with in vivo measurements o f m i x e d venous Soz (S~702) [4]. In the early 1980s, heparinized p u l m o n a r y artery catheters were introduced. These catheters are less t h r o m b o g e n i c than their nonheparinized counterparts [6,10-12]; h o w e v e r , it is n o t k n o w n h o w the heparinization o f p u l m o n a r y artery catheters m a y affect the continuous m e a s u r e m e n t o f S'FOz b y oximeters. W e corn-
Stritter et al: Heparin Effects on Pulmonary Artery Oximetry
pared the effects ofheparinized and plain catheters on in vivo S~¢O2 monitoring in critically ill patients. METHODSAND MATERIALS
Phase 1 In 40 patients admitted to the intensive care unit (ICU) for whom pulmonary artery catheterization was clinically indicated, fiberoptic, 7-French pulmonary artery oximetry catheters (Opticath P7110, Oximetrix, Mountain View, CA) were used. For half the patients, plain catheters were used; for the other half, catheters were used that had the distal 10 cm (including balloon) impregnated with heparin. The manufacturer used a solvent (toluene) that causes the polyvinyl chloride catheter to swell and thus entrap a tridodecylmethylammonium chloride-heparin complex [13]. Radioisotopic analysis of these catheters demonstrated they had approximately 15 Units of total heparin activity that eluted over time; only 1 to 2 units of heparin activity remained after the catheters had stood for 24 hours in a saline bath. All catheters were calibrated in vitro according to the manufacturer's recommendations before insertion but were not recalibrated in vivo during the course of the study. The catheter was inserted via an 8.5-French valved introducer with sidearm (Arrow AK-09802) into the pulmonary artery so that pulmonary artery occlusion pressures could be obtained by inflation of the balloon with 0.8 to 1.5 cc of air. Proximal and distal lumens of all catheters were continuously flushed (Sorenson Intraflow) with heparinized saline (2 units/ ml) at a rate of 3 ml/hr. In vivo S~¢O2was recorded, and blood was sampled from the distal catheter lumen into glass syringes, placed on ice, and analyzed in triplicate in an IL 282 CO-Oximeter. Blood was sampled in both groups at random times.
rently the industry standard. (The total amount ofheparin applied may vary between catheter manufacturers.) Plain, 7-French pulmonary artery catheters were used in the other 8 patients. Before placement, the catheter type was assigned in a random, double-blind fashion, and blood was sampled for blood urea nitrogen (BUN), prothrombin (PT) and partial thromboplastin (PTT) times, and platelet count. The protocol was the same as that for phase 1 except that blood was sampled from the distal catheter lumen 5 minutes and 1 hour after insertion of the catheter, and then every 6 hours until the catheter was removed. Both phases of the study were approved by the Human Subjects Committee. Informed consent was obtained from each subject or his or her family. Statistical Analysis The value of So2 provided by the IL 282 CO-Oximeter is expressed as a percentage of the total hemoglobin (%O2Hbtotal), which included oxygenated and deoxygenated hemoglobin (O2Hb and deoxyHb) as well as carboxyhemoglobin (COHb) and methemoglobin (MetHb) species: %O2Hbtot~i = [O2Hb/(O2Hb + deoxyHb + MetHb + COHb)] x 100%. Since the Oximetrix system reports S02 as a percentage of available hemoglobin (%O2Hbavailable), where Hbavaihble is equal to oxygenated and deoxygenated hemoglobin only, %O2Hbavailable = [O2Hb/(O2Hb + deoxyHb)] x 100%. The following equation was used to transform % O2Hbtotal to % O2Hbavaihble before comparisons were made: %O2Hbavailable =
Phase 2 In phase 2, 16 ICU patients were studied in whom pulmonary artery catheterization was indicated and who had no clinical signs of bleeding and had not had heparin therapy in the past 24 hours or aspirin or persantine in the previous 2 weeks. Fiberoptic, 7-French pulmonary artery catheters (Opticath P7110-E) that had been "heparin-bonded" over their entire surface by being dipped in a solvent (isopropyl alcohol) containing a benzalkonium-heparin complex [14] were used in 8 patients. Heparin, approximately 300 units, was applied to each bonded catheter by this technique, which is cur-
205
%O2Hb~o~j[1 - (%COHb + %MetHb)/100].
In vivo and in vitro measurements of SVO2 were compared in each group of patients by using linear regression analysis. The correlation coefficients, slopes, and intercepts were compared with a t statistic [t5]. Error was more specifically evaluated by comparing the mean differences between in vitro and in vivo SVO2 measurements and expressing these results as the mean difference +-- standard error with the 95% confidence interval of an individual value [16]. In phase 2, mean duration of catheterization, B U N , PT, PTT, and platelet count for heparin-bonded and plain catheter groups were corn-
206 Journal of CtinicaI Monitoring Vol 4 No 3 July 1988
100
Heparinized Catheters
80 S~ 0 2
60
Oximetrix Opticath
(%)
./
" "
N o n - Heparinized
Catheters
"
°
's s
40 ,
s.
.
.87x + 8.0 r=.81 n= 223 p< .O01
r=.93 n=141 p< .001 I
i
I
I
20
40
60
80 $902
/
I
100 IL 282
Fig 1. Comparison of pooled datafor in vivo SYOz (Oximetrix Opticath catheter) with in vitro SVOe (IL 282 Co-Oximeter) in 40 patients (phase 1) catheterized with either heparin-impregnated pulmonary artery catheters or plain pulmonary artery catheters. Lines of identity (solid) and regression (broken) are shown. pared by using an unpaired t test. Data are expressed as mean -+ SE. RESULTS
Phase 1 Heparin-impregnated catheters were inserted in 21 patients and plain catheters in 19 patients; a total o f 364 blood samples were analyzed. Mean duration o f catheterization varied widely and did not differ between the groups (51 +- 44 hours, n = 40). In both groups, the linear correlation between in vivo and in vitro $v¢O2 measurements was significant (P < 0.001) (Fig 1). H o w ever, data from heparinized compared with plain catheters generated not only better correlation coefficients (0.93 versus 0.81, respectively, P < 0.001), but also a slope (0.98 versus 0.87, respectively, P < 0.05) and yintercept ( - 0 . 0 1 versus 8.0, respectively, P < 0.05)that were significantly different and closer to the line o f identity. Examination o f the average measurement error (in vivo minus in vitro $v¢O2) in both groups revealed that measurements with heparinized catheters differed from laboratory measurements o f SWO2 by - 1 . 3 + 0.4 (mean -+ SE). The measurement error for nonheparinized catheters was - 1 . 4 + 0.4 (mean + SE), The 95% confidence interval for a single measurement was -+8.1 versus ---12.3 for the two groups, respectively.
s
,s
" y=.gSx-,Ot
20
i
I
I
I
20
40
60
Co-Oxirneter
I
80
I
100
(%)
Phase 2 Heparin-bonded catheters were used in 8 patients and plain catheters in 8. Duration o f catheterization did not differ significantly between the groups (43 + 5 hours, n = 16) nor did B U N , PT, P T T , or platelet count (P > 0.05); no patient had a B U N greater than 30 mg/dl, a P T greater than 14 seconds, or a P T T greater than 40 seconds. A total o f 134 blood samples were obtained. Linear regression analysis o f in vivo and in vitro S~02 yielded highly significant correlations in both groups (P < 0.001) (Fig 2). While the slopes o f both equations (1.0 and 1.1 for heparin-bonded and plain catheters, respectively) did not differ significantly from the line o f identity, the y-intercept o f the nonheparin group ( - 8 . 4 ) was significantly different from that o f the heparin group ( - 0 . 6 ) as well as from the line o f identity (P < 0.001). Linear regression analysis o f individual catheter performance revealed very little within-subject variability; all individual r values in both groups were greater than 0.85. Examination o f the average measurement error (in vivo minus in vitro S'VO2) indicated that measurements o f S~O2 with heparin-bonded catheters differed from laboratory data by 0.4 -+ 0.3 (mean + SE); the measurement error for plain catheters was - 1 . 3 +- 0.3 (mean - SE). The mean error for heparin-bonded catheters did not differ significant!y from zero, while the mean error for the plain catheters did differ significantly from zero (P < 0.001) and from the heparin-bonded group. The 95% confidence interval for an individual measurement was --- 5.0 versus +- 4.8, respectively, for the two groups. When data were pooled for analysis by linear regres-
Stritter et al: Heparin Effects on Pulmonary Artery Oximetry
Heparin Bonded Catheters
Non-Heparin
/
"/
Bonded
/
.,//
Catheters
90
S~O 2 Oximetrix Opticath
(%)
70
,'"= . 9 " ~ -,o6~
50
o=~s
/, -
r= . g 8
3O I
I
I
I
30
50
70
90
$902
I
I
30
IL 282
I
50
Co-Oximeter
I
70
90
(%)
Fig 2. Comparison of pooted data for in vivo S~'02 (Oximetrix Opticath catheter) with in vitro S~02 (IL 282 Co-Oximeter) in 16 patients (phase 2) catheterized with either a heparin-bonded pulmonary artery catheter or a plain pulmonary artery catheter according to random double-blind assignment. Lines of identity (solid) and regression (broken) are shown.
Hep~;~eBt°nded 4°~ OxAmSTr02(%)
Non-Heparin Bonded Catheters
/
40"
seo2 (°/4
30-
20 t
•
1 0 _- [~" e~ e ~
I
//
~Sv02
10 I" -40
•
/
/
/
,.;0
-10
-20 -30
~°
(%)
Co-( IL 282 Co-Oximeter
y=0.9x + 0.2 n=60 r-0.92
-4o
Fig 3. Comparison of the change in sequential measurements of i n vivo Sff02 (Oximetrix Opticath catheter) versus the change in in vitro S~02 (IL 282 Co-Oximeter) of pooled data from 16 patients (phase 2) catheterized with either a heparin-bonded pulmonary artery catheter or a plain pulmonary artery catheter, according to random, double-blind assignment. The lines of identity (solid) and regression (broken) are shown.
"/
./.,
/_//
Ib.~_/
fO Jl
1.10 J
/
Opticath/
Oximetrix
~s~°2(°/o) IL 282
I
10
_ 'l' L I 20
Co-Oximeter
I
30
t
I
40
-10
-20 "
.oi
-40
y=0.9x + 0.03 n=56 r=0.94
207
208 Journal of Clinical Monitoring Vol 4 No 3 July 1988
sion to compare the change (difference between sequential measurements) of in vivo S'~O2 with the change of in vitro Sr¢O2 (Fig 3), the results for both groups were similar. DISCUSSION
The results of phase 1 suggested that heparinization of pulmonary artery catheters might improve pulmonary artery oximetry, but the heparin effect was small. These results might have been biased because of the nonblinded design, or influenced by other uncontrolled factors (e.g., coagulopathies, systemic heparinization, other anticoagulants, or thrombocytopenia). The blinded phase 2 study more clearly demonstrated the effect of heparinization of pulmonary artery catheters. Plain catheters in phase 2 showed greater accuracy (see Figs 1 and 2) than plain catheters in phase 1; correlation coefficients and slopes were similar to those obtained with the heparin-bonded catheters in phase 2. By regression analysis, the only significant difference found was in the y-intercepts, which continued to be closer to zero (line of identity) with the heparinized catheters (P < 0.001). With the mean difference analysis, greater accuracy was demonstrated with the heparin-bonded catheters, although the degree of error ( - 1.3) with the plain catheters, while statistically significant, probably is not clinically important. The accuracy of the pulmonary artery oximeter system has been addressed previously by studies of correlation values [1,3,4,6,8,10,17]. In all but two of these studies, the regression equation was not provided, and in only one was the line of identity even shown on the graphed data. High correlation values prove association, but not necessarily agreement, even when the regression line is shown not to differ significantly from the line of identity [16]. Determining agreement by analysis of mean differences is statistically sound and provides a simple and clearer answer to the question of accuracy [161. Our data indicate that pulmonary artery oximetry is accurate enough for clinical use, that is, the ptflmonary artery oximeter tested in phase 2 will measure SVO2 to within 5% (saturation units) of actual values with 95% confidence. This error is more likely to be 3 to 4% if it is accepted that some small variability (1 to 2%) exists in the ability of the IL 282 CO-Oximeter to measure actual S'VO2, and that daily in-vivo calibrations (not performed in this study) would also improve the accuracy of this monitoring system. The differences in the results of the phase 1 and 2 studies are primarily due to greater accuracy with plain catheters, and not to less accuracy with heparinized
catheters. There are two possible explanations for the greater accuracy with plain catheters. First, the phase 1 study lacked blind randomization, and variables affecting coagulation were not controlled. Second, there was roughly a 2-year interval between the manufacturing of the two types of heparinized catheters. During this time, coating, surface smoothing, and flexibility of plain pulmonary artery catheters were improved, which could have resulted in less thrombogenicity and greater accuracy with nonheparinized catheters in phase 2. If heparinization of catheters can improve pulmonary artery oximetry, a possible explanation for our negative results with heparin-bonded catheters is that the heparin may have been removed inadvertently when the catheter was passed through the tight-fitting latex valve that is now a standard part of the pulmonary artery catheter introducers [14]; in fact, studies demonstrating dramatic reduction of clot deposition on heparinized pulmonary artery catheters used valveless introducers [11,12]. The current costs or methods of heparinization of pulmonary catheters may need to be reexamined if it can be shown that valved introducers are indeed causing deheparinization. It is possible that catheter design affects clot deposition at the very tip of the catheter. For instance, with increased flexibility of the catheters, the end of the catheter might spend less time in contact with the pulmonary artery vessel wall, which may decrease thrombogenesis. It may also be possible that clot formation primarily occurs ~/long the longitudinal axis of these catheters, sparing the tip, although this seems unlikely. We have demonstrated with appropriate statistical analysis that pulmonary artery oximetry is an acceptably accurate technique. Our results also show that, when catheters are placed according to current clinical standards (i.e., via valved introducers), improvement in the accuracy of in vivo pulmonary artery oximetry with heparinized catheters is not clinically significant.
We thank Oximetrix Corporation for supplying the catheters in this study.
REFERENCES
1. Cole JS, Martin WE, Cheung PW, Johnson CC. Clinical studies with a solid state fiberoptic oximeter. Am J Cardiol 1972;29:383-388 2. Taylor JB, Lown B, Polanyi M. In vivo monitoring with a fiberoptic catheter. JAMA 1972;221:667-673 3. Krauss XH, Verdouw PD, Hugenholtz PG, et al. Con-
Stritter et al: Heparin Effects on Pulmonary Artery Oximetry
4. 5. 6. 7. 8. 9. 10. 11. 12. 13.
14. 15. 16. 17.
tinuous monitoring of 02 saturation in the intensive care unit. Circulation 1972;46(suppl II):178. Abstract Martin WE, Cheung PW, Johnson CC, Wong KC. Continuous monitoring of mixed venous oxygen saturation in man. Anesth Analg 1973;52:784-793 Krauss XH, Verdouw PD, Hugenholtz PG, NautaJ. Online monitoring of mixed venous oxygen saturation after cardiothoracic surgery. Thorax 1975;30:636-643 Waller JL, Kaplan JA, Bauman DI, Craver JM. Clinical evaluation of a new fiberoptic catheter oximeter during cardiac surgery. Anesth Analg 1982;61:676-679 Schmidt CR, Starr NJ. Evaluation of a continuous SVO2 monitoring system. Anesthesiology 1981;55:A125. Abstract Baele PL, McMichan JC, Marsh HM, et al. Continuous monitoring of mixed venous oxygen saturation in critically ill patients. Anesth Analg 1982;61:513-517 Jamieson WR, Turnbull KW, Larrieu AJ, et at. Continuous monitoring of mixed venous oxygen saturation in cardiac surgery. Can J Surg 1982;25:538-543 Fahey PJ, Harris K, Vanderwarf C. Clinical experience with continuous monitoring of mixed venous oxygen saturation in respiratory failure. Chest 1984;86:748-752 Hoar PF. Heparin bonding reduces the thrombogenicity of pulmonary artery catheters. N Engl J Med 1981; 305:993-995 Kido DK, Paulin S, AlenghatJA, et al. Thrombogenicity ofheparin and non-heparin-coated catheters: clinical trial. AJR 1982;139:957-961 Grode GA, Anderson FJ, Grotta HM, Fallo RD. Nonthrombogenic materials via a simple coating process. Transactions of the American Society of Artificial Internal Organs 1969;15:1-6 Amplatz K. A simple non-thrombogenic coating. Invest Radiol 1971;6:280-289 Sokal RR, RohlfFJ. Biometry: the principles and practice of statistics in biological research. San Francisco: Freeman, 1981:588-589 Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1:307-310 Weston GA, Ledingham IM, Douglas IH, MacArthur KJ. Evaluation of a new fiberoptic pulmonary artery catheter in intensive care. A preliminary study. Anaesthesia 1984;39:272-276
209