Multicenter Trial of Automated Nitroprusside Infusion for Postoperative Hypertension W. Randolph Chitwood, Jr, MD, Delos M. Cosgrove 111, MD, Robert M. Lust, PhD, and the Titrator Multicenter Study Group Division of Cardiothoracic Surgery, East Carolina University School of Medicine, Greenville, North Carolina, and Department of Thoracic and Cardiothoracic Surgery, The Cleveland Clinic Foundation, Cleveland, Ohio
Hypertension is common after a cardiac operation and may result in postoperative hemorrhagic and other complications. Most often this problem has been treated using manually controlled doses of intravenous sodium nitroprusside. To evaluate the clinical impact of an automated closed-loop administration system on patients after cardiotomy, a prospective trial was conducted at nine clinical centers. Patients with hypertension were managed by either manual nitroprusside titration (n = 532) or a closed-loop automated titration system (n = 557). Patient groups were not significantly different in age, weight, or height. Moreover, the types of surgical procedures were comparable: primary coronary artery bypass grafting, 59.2% and 58.9%, manual group versus automated group; repeat coronary artery bypass grafting, 10.5% and 8.6%, respectively; valve procedures, 11.3% and 15.1%, respectively; and other cardiac procedures, 19.0% and 17.4%, respectively (all p = not significant). The automated group showed a significant reduction in the number of hypertensive episodes per patient (1.8 f 0.2 versus 0.6 f 0.07; p = 0.0001. At the same time, the
H
ypertension after cardiovascular surgical procedures is common, occurring in an estimated one third to one half of all patients recovering from procedures involving cardiopulmonary bypass [l-31. Often the cause remains unclear. Factors include withdrawal from /3 blockers, hypothermia, elevated levels of circulating catecholamines, increased renin-angiotensin activity, and emergence from anesthesia [4-81. Untreated hypertension after a cardiac operation may result in complications including increased bleeding, disruption of suture lines, arrhythmias, subendocardial ischemia, and cerebrovascular hemorrhage. Moreover, cardiac ischemia and depressed ventricular performance have been demonstrated during hypertensive episodes [9-111. Precise control of blood pressure may be necessary to avoid these complications. Sodium nitroprusside dilates both venous capacitance (preload) and arteriolar resistance (afterload) vessels withAccepted for publication Feb 10, 1992. Address reprint requests to Dr Chitwood, Division of Cardiothoracic Surgery, East Carolina University School of Medicine, Greenville, NC 27858-4354.
0 1992 by The Society of Thoracic Surgeons
number of hypotensive episodes per patient was reduced with automated closed-loop titration (0.40 f 0.05 versus 0.30 f 0.03; p = 0.02). Chest tube drainage (866 & 37 mL versus 693 2 23 mL [mean & standard error of the mean]; p = O.OOOl), percentage of patients receiving transfusion (40.0% versus 33.0%; p = 0.02), and total amount transfused (2.4 & 0.12 units versus 2.0 & 0.10 units;p = 0.0003) were all reduced significantly by the use of an automated titration system. Finally, of the 3.3% of all patients in the series who required reexploration for bleeding, almost twice as many had been managed by manual infusion compared with automated titration. These data suggest that the automated control of nitroprusside infusion may result in improved patient outcome through more effective control of hypertension, a decreased incidence of therapy-induced hypotension, a reduction in postoperative bleeding, decreased transfusion requirements, and a probable decreased incidence of reexploration for postoperative bleeding. (Ann Thorac Surg 1992;54:517-22)
out major cardiotonic effects [12]. The rapid onset and the short duration of pharmacologic activity have made nitroprusside the most widely accepted treatment for postoperative hypertension after cardiopulmonary bypass [ll, 15161. Hypotension, induced by antihypertensive therapy, also can produce unwanted consequences. A decrease in diastolic pressure may lead to myocardial underperfusion and ischemia. For this reason, the rate of nitroprusside infusion must be monitored closely and adjusted accurately to maintain blood pressures within the desired limits, making manual adjustment to physiologic variables a difficult and time-consuming process. To facilitate this process, a closed-loop bedside controller system has been developed to sense blood pressure and automatically regulate intravenous sodium nitroprusside delivery based on variables measured a few moments earlier. To evaluate the comparative effects on hypertension and to elucidate the differences between closed-loop automated infusion and conventional manual titration of sodium nitroprusside, a prospective clinical trial was designed and completed in nine separate cardiac surgical centers. 0003-4975/92/$5.00
518
CHITWOOD ET AL CLOSED-LOOP NITROPRUSSIDE INFUSION
Materia.1 and Methods The principal investigators who participated in this study aye listed in Appendix 1 and are referred to as the Titrator Multicenter Study Group. The closed-loop drug delivery system used in this study was the Titrator module model 10K (IVAC Corporation, San Diego, CA). Other studies [5, 17, 181 have shown this device to be safe and to render rapid, effective control of postoperative hypertension. With this system, a microprocessor module senses the mean arterial pressure every 0.01 second from a disposable transducer attached to a radial intraarterial line. The microprocessor has artifact rejection capabilities and determines both patient sensitivity and responses to the drug wiihin the first 5 minutes of therapy. The titrator system includes the microprocessor linked to a variable pressure volumetric infusion pump (model 560i; IVAC corporation) through which the delivery of sodium nitroprusside was adjusted automatically every 10 seconds. The desired mean arterial blood pressure between 60 and 120 mm Hg could be preselected with nitroprusside infusion rates displayed continuously. Approximately 4 weeks before the study, individual familiarization programs were conducted to allow nursing staff to become acquainted with the device and data collection plan. Subsequently, all patients eligible for the study were entered sequentially at each site over a 12- to 14-week period. All patients in whom hypertension developed after a cardiovascular procedure and who required sodium nitroprusside for blood pressure control were included in the study. Hypertension and desired mean arterial pressure were defined by each center, and therapy was initiated accordingly. Intraoperative anesthesia, surgical management, administration of postoperative medications, and therapy for bleeding complications were accomplished by the standard practices of each institution. In all instances, nitroprusside therapy was initiated in the intensive care unit (ICU). Patients arriving from the operating room with nitroprusside infusions already begun were excluded. Data were collected from all patients, the starting point being the initiation of nitroprusside infusion in the ICU and the ending point, drug discontinuation. A single individual at each institution was designated to collect the data. Information was obtained from the nursing flow sheets and shift data records maintained by the bedside nurse. Collected data end points included: (1) length of ICU stay, (2) reexploration based on bleeding indications, (3) amount of chest tube drainage, (4) amount of blood products transfused, (5) number of hypertensive or hypotimsive episodes, (6) number of manual infusion rate changes, and (7) number of times the physician was notified ior problems related to blood pressure control. The data manager at each institution collected this informatian using a standardized database. The data were downloaded by modem to a central data bank, where patient information from all the centers was collated in a blinded fashion. Statistical methods used depended on the data set being antdyzed. Student's t test, one-way and two-way
Ann Thorac Surg 1992;54:517-22
Table 1. Patient Characteristics and Procedures" Manual (n = 532)
Variable Characteristics Age (Y) Weight (kg) Height (cm) Sex (ME) (%) Prior operation (%) Procedures First CABG (%) Repeat CABG (%) Valve (%)h Other I%)'
Titrator (n = 557)
*
62.5 0.5 62.6 ? 0.5 77.1 ? 0.7 78.0 ? 0.8 168.0 ? 0.7 166.0 2 0.7 76/24 67133 20.9 19.4 59.2 10.5 11.3 19.0
58.9 8.6 15.1 17.4
p Value Test 0.9 t test t test 0.4 0.15 t test 0.001 x2 0.544 x z 0.17 0.17 0.16 0.16
x2 x2 x2 x2
Where applicable, data are shown as the mean ? the standard error of the mean. This includes mitral valve replacement, aortic valve re' This placement, double-valve replacement, and double-valve repair. includes aortic aneurysm repair, Bentall grafts, repair of aortic dissections, transplantation, and repair of atrial septal defects.
a
CABG
=
coronary artery bypass grafting
analysis of variance, and x2 analysis each were used when appropriate. Unless otherwise indicated, all data are exthe standard error of the mean, pressed as the mean with significance determined at an a level of 0.05.
*
Results A total of 1,089 patients were enrolled in this study; 532 constituted the manual control group and 557, the automated titration cohort. A comparison of patient characteristics by group is provided in Table 1. No significant differences existed between groups for age, height, weight, or incidence of prior operation. However, the automated group was composed of approximately 7% more women than the manual infusion group (p < 0.001). There was a similar incidence of patients undergoing primary coronary artery bypass grafting, reoperative coronary artery bypass grafting, valve operations, and other procedures. Valve operations included mitral valve replacement, aortic valve replacement, double-valve replacement, and double-valve repair. Other procedures included grafting of ascending aortic aneurysms, repair of aortic dissections, repair of atrial septal defects, and cardiac transplantation. The incidence of hypertension was reduced signifi0.2 episodes per cantly by automated titration: 1.8 patient versus 0.6 0.07 episode per patient (p = 0.0001) (Fig 1). In addition, the frequency of hypotensive episodes resulting from nitroprusside therapy was reduced: 0.40 2 0.05 versus 0.30 0.03 episode per patient (p = 0.02). Although there was variability from site to site, overall chest tube drainage in the automated group was reduced by an average of 170 mL per patient ( p = 0.0001) compared with the manual cohort (Fig 2). As seen in Figures 3 and 4, the total blood products (whole blood and packed red blood cells) transfused per patient and the percentage of patients requiring transfusion were reduced
*
*
*
CHITWOOD ET AL CLOSED-LOOP NITROPRUSSIDE INFUSION
Ann Thorac Surg 1992;54:517-22
519
7
0 Manual I
.
._ c a U
$
3.0-
*
pco.01
2
* p=O.O001 ** p=0.02
Lo
.-P
E
.-c
Titrator
2.0
Site
EZ Titrator
o Manual
5 v)
3
c
Manual
Titrator
1
1.8ic 26
1.7f 2 1
2
2.8f .56
3.0 f .65
3
3.0f .42
2.0-
o*o 2.5 f 2 2
C
E
4.0 f .58
c
1.0
T Hypertensive
(0 c ._
5
**
Hypotensive
Fig 3 . Total blood products (whole blood, packed red blood cells) transfused by cohort and participating Titrator Multicenter Study Group institution, listed in blinded fashion. Data are shown as the mean and the standard error of the mean.
Fig 1. Number of hypertensive and hypotensive episodes per patient. Data are shown as the mean and the standard error of the mean.
significantly by closed-loop titration ( p = 0.01 and p = 0.0007, respectively). However, transfusion requirements vaned widely depending on the center. For example, Figures 3 and 4 demonstrate that when the titrator was used, blood product usage and transfusion patterns decreased at five centers (3, 4, 5, 6, and 8), did not change at two centers (1 and 7), and actually increased at two centers (2 and 9). On the other hand, examination of site data in Figure 2 demonstrates that chest tube drainage was decreased at both centers with increased transfusion requirements in the automated group. Thus, although an increase in blood product usage was demonstrated, it could not be correlated to a bleeding indication. Additional factors besides chest tube drainage, and possibly blood pressure, also variably influenced the usage of blood products in this study. These findings highlight problems inherent in a multicenter trial, particularly when a standardized protocol for the study is not established. At the same time, the fact that an overall difference could be demonstrated, despite these limitations, suggests strongly that the merits of the titration approach are valid. Reexploration for bleeding was almost twice as likely in the manually titrated patients (Fig 5). Overall, 36 (3.3%)of all 1,089 patients had reoperation for bleeding; nearly two thirds of them came from the manually controlled group
( p = 0.066). Although not significant for this population, the trend certainly is noteworthy. Automated nitroprusside titration reduced the average length of ICU stay by approximately 5 hours ( p = 0.038), virtually eliminated the need for manual infusion rate changes ( p = O.OOOl), reduced the average duration of nitroprusside therapy by about 3 hours ( p = 0.027), and reduced by one third ( p = 0.0001) the number of times the physician had to be contacted for problems related to blood pressure control (Fig 6).
Comment Estafanous and colleagues [l] first described hypertension as a consequence of aortocoronary bypass graft procedures. Generally, increases in blood pressure are encountered within the first 6 hours after cardiopulmonary bypass and may last 12 to 36 hours. The incidence of postoperative hypertension has been reported to range between 8% and 32% for valve replacements and from 8% to 73% after coronary operations [l-31. Although elevated blood pressure after cardiovascular procedures is common, management guidelines remain unclear. Flaherty 60
1000 nnn
800 c
0 Manual
rn Titrator
Site
Manual
Titrator
2
463ic138 1312+166
587 + 43 769+120 830 + 75
4
1705+429 629 k 156
1530 + 306 611 k63 407 f 149
1157f172
669 f 25 1039f144
834f185
300 200
100
* p=o.oooi
8
2.5k .50
1.0-
* p=0.0007
o Manual m Titrator
I
Site
Manual
43% 89%
5
46% 78% 55% 53% 1 9%
6
43%
1 7%
7
30%
27%
8
44% 6%
35%
40 30
~
9
Titrator
45% 0%
19%
586+138
Fig 2 . Chest tube drainage by cohort and participating Titrator Multicenter Study Group institution, listed in blinded fashion. Data are shown as the mean and the standard error of the mean.
Fig 4. Percentage of patients having transfusion by cohort and participating Titrator Multicenter Study Group institution, listed in blinded fashion. Data are shown as the mean and the standard error of the mean.
520
CHITWOOD ET AL CLOSED-LOOP NITROPRUSSIDE INFUSION
6
r
0 Manual
m Titrator
* p=0.066 (Chi-square) % of total
re-explored
96.7%
group distribution
Fig 5. Frequency of reexploration for bleeding by cohort and as a function of the total study population. (M = manual; T = titrutor.)
and colleagues [13] suggested that a sudden, persistent increase in mean arterial pressure of 20 mm Hg or more, in the presence of a mean pressure of at least 95 mm Hg, should be considered a hypertensive episode. By this criterion, 61% of all patients undergoing bypass grafting in their study displayed postoperative hypertension. Fremes and co-workers [ll]proposed that patients not be considered hypertensive postoperatively until the mean arterial pressure exceeds 105 mm Hg. Even by this liberal definition, 42% of their patients demonstrated systemic hypertension after bypass grafting. In the present study,
E
c
n
*I
Ann Thorac Surg 1992:54:517-22
a specific definition of hypertension was not preselected. Members from each institution were allowed to use their own variables to define systemic hypertension. Although the number of hypertensive episodes is known for each center, precise blood pressure levels were not recorded and varied from site to site. Although @-blockadewithdrawal, increased catecholamine levels, increased renin-angiotensin activity, emergence from anesthesia, and withdrawal from hypothermia all have been proposed as etiologic factors for systemic hypertension after bypass, mechanisms often remain obscure [4, 6-81. Irrespective of the cause, potential consequences may be serious and include increased bleeding, additional transfusion requirements, disruption of vascular anastomoses, arrhythmias, subendocardial ischemia, and possibly cerebrovascular hemorrhage [9, 101. The cost of transfusions and reoperation adds substantially to health care expenses [19]. In the present study, no attempt was made to define the cause of the hypertension. However, improved blood pressure control was evident using the closed-loop system and was reflected by a decreased number of hypertensive episodes, a reduced amount of bleeding, a reduced number of patients having transfusion, and a reduced volume of blood products needed in those patients requiring transfusion. Sodium nitroprusside remains the therapy of choice for
Length of ICU Stay
Hrs on SNP
* l *
20
p=0.027
40'
c
20'
Manual
Titrator
Manual
Manual Infusion Rate Changes
*
10
2.0
-
1.3
-
p=o.o001
Titrator
M.D. Consults re: BP
*
p=o.o001
* Manual
Titrator
Manual
Titrator
Fig 6 . Len,gth of intensive cure unit (ICU) stay, duration of nitroprusside (SNP) therapy, frequency of manual infusion rate change, and number of consults for blood pressure (BP) problems, by cohort.
Ann Thorac Surg 1992;54:517-22
management of systemic hypertension after bypass procedures. The major advantages of this drug are rapid onset of action, short duration, relatively minor direct cardiac effects, and effective afterload reduction by vasodilation of both venous capacitance and arteriolar resistance beds [ll, 1>16]. However, many of the characteristics that make nitroprusside effective and attractive also can be disadvantageous. Severe afterload reduction may decrease diastolic perfusion pressure below physiologic ranges and produce subendocardial ischemia. Also, a reflex tachycardia may occur and produce unwanted increases in myocardial oxygen consumption. As the drug is short acting, the rate of infusion must be adjusted frequently, causing manual blood pressure control to be extremely time-consuming for the nursing staff. The present study demonstrated that hypertension was better controlled by automated titration and that the incidence of unwanted hypotensive episodes was reduced. At the same time, the necessity for manual changes in infusion rate was almost eliminated. Although not directly evaluated, the ability to automate antihypertensive therapy suggests improved patient outcome by permitting the bedside nurse to concentrate more completely on other facets of critical care. In fact, the reduction in nursing time and stress associated with antihypertensive therapy may be the major benefit in utilizing the system. This ingenious technology now may allow accurate and sustained blood pressure control in the cardiac surgery ICU and has great potential for all critical-care environments. The closed-loop controller monitors blood pressure continuously and compares the instantaneous measurements with the operator-selected level. Incremental adjustments in infusions are made automatically at 10second intervals, allowing more precise control. Moreover, the controller compares blood pressure responses with the ”history” of previous infusion rates and pressure changes and then can extrapolate future blood pressure trend changes. This cumulative experience contributes to closer regulation of arterial pressure over time. In essence, the longer the controller regulates blood pressure, the better the hemodynamic stability becomes. Drs Sheppard and Kirklin promoted an interest in automated intensive care at the University of Alabama in the early 1970s. Subsequently, cardiac surgeons have continued to show an interest and have realized both the utility and future promise of these methods. Cosgrove and Petre, working at The Cleveland Clinic, developed the prototype closed-loop infusion system to control hypertension after a cardiac operation. At that institution, an upgraded system has been effective for routine clinical use since 1982. The development of microchip technology and the application of closed-loop theory to the bedside delivery of drugs have enabled clinicians to develop new potentials for the critical care environment. Sheppard [20] pioneered the application of computer control to the infusion of vasoactive drugs. Within the last 5 years, several groups [21-241 have developed excellent custom systems for microcomputer control of blood pressure and nitroprusside infusion. In 1989, a device approved by the Food and Drug Administration for the
CHITWOOD ET AL CLOSED-LOOP NITROPRUSSIDE INFUSION
521
infusion of sodium nitroprusside in patients hypertensive after bypass became available commercially (Titrator module, model 1OK; IVAC Corporation). Since that time, three clinical studies have been conducted using this device. Keogh and associates [25] compared 12 patients whose blood pressure was controlled by manual adjustments of nitroprusside with 22 patients controlled by automated titration. The desired preselected mean arterial pressure was 70 mm Hg for all patients. Throughout the study, the mean arterial pressure was maintained within 10% of the desired value 46% of the time in the manual group and 90% of the time in the automated group. Cosgrove and colleagues [17] also conducted a prospective, randomized study of 180 patients having a cardiac operation in whom manual infusion was compared with automated titration. In that study, mean arterial pressure was maintained within 10% of a selected target value 61% of the time with manual infusion versus 85% of the time with automated titration. Moreover, the incidence of unwanted hypertensive and hypotensive episodes during drug infusion was reduced from more than 20% in the manual group to less than 10% in the automated group. Bednarski and colleagues [18] compared automated control in 20 patients with manual control in 10 patients. During the first 10 minutes, blood pressure was within 10% of the desired level in 60% of the patients from both groups. However, over the next 140 minutes, blood pressure control within 10% of targeted values improved in 90% of patients in the automated group but remained only 60% for patients in the manual group. The present study confirms and extends the findings of Keogh [25], Cosgrove [17], Bednarski “1, and their associates. However, the patient sample was much larger and represented multiple centers. Moreover, additional end points included transfusion requirements, number of patients having transfusion, quantity of chest tube drainage, duration of antihypertensive therapy, and rate of reoperation for bleeding. Use of the automated titration system reduced the incidence of reoperation, hypertensive episodes, hypotensive episodes, and transfusions. When a transfusion became necessary, automated titration reduced the amount of blood products used. Improved control of blood pressure led to an earlier discontinuation of antihypertensive therapy and shorter stays in the ICU. Although the study shows improved blood pressure control using the closed-loop system, some improvements may have resulted from the increased attention to patients that nurses could provide when freed from frequent manual titrations. At present, commercial access to this technology is relatively expensive. As with most new technology, initial costs are high but can be expected to decrease as volume increases and additional applications for the technology are identified. For example, there remains great potential for the use of this system in the delivery of vasopressor and inotropic agents. Presently, the device is not approved for this use; however, in Australia this system has been shown effective with cardiotonic and vasopressor drugs [26]. The advent of continuous cardiac output measurements
522
CHITWOOD ET AL
Ann Thorac Surg 1992;54:517-22
CLOSED-LOOP NITROPRUSSIDE INFUSION
should help expand the utility of closed-loop systems w i t h these and other drugs. In coinclusion, the use of a commercially available closed-loop delivery system i n the administration of nitroprusside significantly improved patient outcome. In particular, there w a s a strong trend toward a reduced incidence of reoperation for bleeding indications. We recognize the assistance of Maureen Fallon, Sharon Gallagher, Cathy Galvin, Debbie Hall, Suzanne Provost, Janette Sherwood, Julie Shinn, and Pat Young in the collection of the patient data at each of the clinical centers. The assistance of Dr Mary Ann Christopher in the statistical analysis of the data is also appreciated.
Ref erenices 1. Estafanous FG, Tarazi RC, Viljoen JF, El Tawil MY. Systemic hypertension following myocardial revascularization. Am Heart J 1973;85:732-8. 2. Roberts AJ, Niarchos AP, Subramanian VA, et al. Systemic hypertension associated with coronary artery bypass surgery. Predisposing factors, hemodynamic characteristics, humoral profile, and treatment. J Thorac Cardiovasc Surg 1977;;'4:846-57. 3. Hoar PF, Hickey RF, Ullyot DJ. Systemic hypertension following myocardial revascularization. J Thorac Cardiovasc Surg 1976;71:859-64. 4. Matthews HR, Meade JB, Evans CC. Peripheral vasoconstriction after open-heart surgery. Thorax 1974;29:33842. 5. Taylor KM, Morton IJ, Brown JJ, et al. Hypertension and the renin-angiotensin system following open-heart surgery. J Thorac Cardiovasc Surg 1977;7484&5. 6. Fouacl FM, Estafanous FG, Bravo EL, et al. Possible role of cardioaortic reflexes in postcoronary bypass hypertension. Am J Cardiol 1979;44:866-72. 7. Wallach R, Karp RB, Reves JG, et al. Pathogenesis of paroxysmal hypertension developing during and after coronary bypass surgery. A study of hemodynamic and humoral factors. Am J Cardiol 1980;46:559-65. 8. Reves JG, Karp RB, Buttner EE, et al. Neuronal and adrenomedullary catecholamine release in response to cardiopulmonary bypass in man. Circulation 1982;66:49-55. 9. Viljoen JF, Estafanous FG, Tarazi RC. Acute hypertension immefdiately after coronary artery surgery. J Thorac Cardiovasc Surg 1976;71:548-50. 10. Estafanous FG, Tarazi RC. Systemic arterial hypertension associated with cardiac surgery. Am J Cardiol1980;46:68594. 11. Fremes SE, Weisel RD, Baird RJ, et al. Effects of postoperative hypertension and its treatment. J Thorac Cardiovasc Surg 1983;8,6:47-56. 12. Palmer RF, Lasseter KC. Sodium nitroprusside. N Engl J Med 1975;2.92:294-7. 13. Flaherty JT, Magee PA, Gardner TS, et al. Comparison of intravenous nitroglycerin and sodium nitroprusside for treatment of acute hypertension developing after coronary bypass surgeiy. Circulation 1982;65:1072-7. 14. Bixler TJ, Gardner TJ, Donahoo JS, Brawley RK, Potter A, Gott VL. Improved myocardial performance in postoperative cardiac surgical patients with sodium nitroprusside. Ann Thorac Surg 1978;25:44&8.
15. Stinson EB, Hollaway EL, Derby GC, et al. Control of myocardial performance early after open-heart operations by vasodilator treatment. J Thorac Cardiovasc Surg 1977;73: 523-9. 16. Gall WE, Clarke WR, Doty DR. Vasomotor dynamics associated with cardiac operations. I. Venous tone and the effects of vasodilators. J Thorac Cardiovasc Surg 1982;83:724-1. 17. Cosgrove DM 111, Petre JH, Waller JL, Roth JV, Shepherd C, Cohn LH. Automated control of postoperative hypertension: a prospective, randomized multicenter study. Ann Thorac Surg 1989;47:678-83. 18. Bednarski P, Siclair F, Voight A, Demertzis S, Lau G. Use of a computerized closed-loop sodium nitroprusside titration system for antihypertensive treatment after open heart surgery. Crit Care Med 1990;18:1061-5. 19. Taylor GJ, Mike11 FL, Moses HW, et al. Determinants of hospital charges for coronary artery bypass surgery: the economic consequences of postoperative complications. Am J Cardiol 1990;65:309-13. 20. Sheppard LC. Computer control of the infusion of vasoactive drugs. Ann Biomed Eng 1980;8:4314. 21. Pomer S, Elert 0, Satter P. Effects of the automated management of hypertension after open heart surgery. Int J Clin Pharmacol Ther Toxicol 1984;22:207-13. 22. De Asla RA, Benis AM, Jurado RA, Litwak RS. Management of postcardiotomy hypertension by microcomputer-controlled administration of sodium nitroprusside. J Thorac Cardiovasc Surg 1985;89:115-20. 23. Reid JA, Kenny GNC. Evaluation of closed-loop control of arterial pressure after cardiopulmonary bypass. Br J Anaesth 1987;59:247-55. 24. Rosenfeldt FL, Chang V, Grigg M, et al. A closed loop microprocessor controller for treatment of hypertension after cardiac surgery. Anaesth Intensive Care 1986;14:158-62. 25. Keogh BE, Jacobs J, Royston D, Taylor KM. Microprocessorcontrolled hemodynamics: a step towards improved efficiency and safety. J Cardiothorac Anesth 1989;3(1):4-9. 26. McKinley S, Cade JF, Siganporia R, Evans OM, Mason OG, Packer JS. Clinical evaluation of closed-loop control of blood pressure in seriously ill patients. Crit Care Med 1991;19: 166-70.
Appendix 1. Participating Centers and Investigators Boston University Hospital The Cleveland Clinic Good Samaritan Hospital (Los Angeles) The Johns Hopkins University Hospital Mt. Sinai Medical Center (New York) Stanford University Hospital Texas Medical Center Toronto General Hospital University of Kentucky Hospital
Richard J. Shemin, MD Delos M. Cosgrove 111, MD Gregory L. Kay, MD Bruce A. Reitz, MD, and William A. Baumgartner, MD Steven L. Lansman, MD D. Craig Miller, MD James J. Livesay, MD Richard D. Weisel, MD W. Randolph Chitwood, Jr, MD
Hypertension is common after a cardiac operation and may result in postoperative hemorrhagic and other complications. Most often this problem has been treated using manually controlled doses of intravenous sodium nitroprusside. To evaluate the clinical impact of an automated closed-loop administration system on patients after cardiotomy, a prospective trial was conducted at nine clinical centers. Patients with hypertension were managed by either manual nitroprusside titration (n = 532) or a closed-loop automated titration system (n = 557). Patient groups were not significantly different in age, weight, or height. Moreover, the types of surgical procedures were comparable: primary coronary artery bypass grafting, 59.2% and 58.9%, manual group versus automated group; repeat coronary artery bypass grafting, 10.5% and 8.6%, respectively; valve procedures, 11.3% and 15.1%, respectively; and other cardiac procedures, 19.0% and 17.4%, respectively (all p = not significant). The automated group showed a significant reduction in the number of hypertensive episodes per patient (1.8 f 0.2 versus 0.6 f 0.07; p = 0.0001. At the same time, the
H
ypertension after cardiovascular surgical procedures is common, occurring in an estimated one third to one half of all patients recovering from procedures involving cardiopulmonary bypass [l-31. Often the cause remains unclear. Factors include withdrawal from /3 blockers, hypothermia, elevated levels of circulating catecholamines, increased renin-angiotensin activity, and emergence from anesthesia [4-81. Untreated hypertension after a cardiac operation may result in complications including increased bleeding, disruption of suture lines, arrhythmias, subendocardial ischemia, and cerebrovascular hemorrhage. Moreover, cardiac ischemia and depressed ventricular performance have been demonstrated during hypertensive episodes [9-111. Precise control of blood pressure may be necessary to avoid these complications. Sodium nitroprusside dilates both venous capacitance (preload) and arteriolar resistance (afterload) vessels withAccepted for publication Feb 10, 1992. Address reprint requests to Dr Chitwood, Division of Cardiothoracic Surgery, East Carolina University School of Medicine, Greenville, NC 27858-4354.
0 1992 by The Society of Thoracic Surgeons
number of hypotensive episodes per patient was reduced with automated closed-loop titration (0.40 f 0.05 versus 0.30 f 0.03; p = 0.02). Chest tube drainage (866 & 37 mL versus 693 2 23 mL [mean & standard error of the mean]; p = O.OOOl), percentage of patients receiving transfusion (40.0% versus 33.0%; p = 0.02), and total amount transfused (2.4 & 0.12 units versus 2.0 & 0.10 units;p = 0.0003) were all reduced significantly by the use of an automated titration system. Finally, of the 3.3% of all patients in the series who required reexploration for bleeding, almost twice as many had been managed by manual infusion compared with automated titration. These data suggest that the automated control of nitroprusside infusion may result in improved patient outcome through more effective control of hypertension, a decreased incidence of therapy-induced hypotension, a reduction in postoperative bleeding, decreased transfusion requirements, and a probable decreased incidence of reexploration for postoperative bleeding. (Ann Thorac Surg 1992;54:517-22)
out major cardiotonic effects [12]. The rapid onset and the short duration of pharmacologic activity have made nitroprusside the most widely accepted treatment for postoperative hypertension after cardiopulmonary bypass [ll, 15161. Hypotension, induced by antihypertensive therapy, also can produce unwanted consequences. A decrease in diastolic pressure may lead to myocardial underperfusion and ischemia. For this reason, the rate of nitroprusside infusion must be monitored closely and adjusted accurately to maintain blood pressures within the desired limits, making manual adjustment to physiologic variables a difficult and time-consuming process. To facilitate this process, a closed-loop bedside controller system has been developed to sense blood pressure and automatically regulate intravenous sodium nitroprusside delivery based on variables measured a few moments earlier. To evaluate the comparative effects on hypertension and to elucidate the differences between closed-loop automated infusion and conventional manual titration of sodium nitroprusside, a prospective clinical trial was designed and completed in nine separate cardiac surgical centers. 0003-4975/92/$5.00
518
CHITWOOD ET AL CLOSED-LOOP NITROPRUSSIDE INFUSION
Materia.1 and Methods The principal investigators who participated in this study aye listed in Appendix 1 and are referred to as the Titrator Multicenter Study Group. The closed-loop drug delivery system used in this study was the Titrator module model 10K (IVAC Corporation, San Diego, CA). Other studies [5, 17, 181 have shown this device to be safe and to render rapid, effective control of postoperative hypertension. With this system, a microprocessor module senses the mean arterial pressure every 0.01 second from a disposable transducer attached to a radial intraarterial line. The microprocessor has artifact rejection capabilities and determines both patient sensitivity and responses to the drug wiihin the first 5 minutes of therapy. The titrator system includes the microprocessor linked to a variable pressure volumetric infusion pump (model 560i; IVAC corporation) through which the delivery of sodium nitroprusside was adjusted automatically every 10 seconds. The desired mean arterial blood pressure between 60 and 120 mm Hg could be preselected with nitroprusside infusion rates displayed continuously. Approximately 4 weeks before the study, individual familiarization programs were conducted to allow nursing staff to become acquainted with the device and data collection plan. Subsequently, all patients eligible for the study were entered sequentially at each site over a 12- to 14-week period. All patients in whom hypertension developed after a cardiovascular procedure and who required sodium nitroprusside for blood pressure control were included in the study. Hypertension and desired mean arterial pressure were defined by each center, and therapy was initiated accordingly. Intraoperative anesthesia, surgical management, administration of postoperative medications, and therapy for bleeding complications were accomplished by the standard practices of each institution. In all instances, nitroprusside therapy was initiated in the intensive care unit (ICU). Patients arriving from the operating room with nitroprusside infusions already begun were excluded. Data were collected from all patients, the starting point being the initiation of nitroprusside infusion in the ICU and the ending point, drug discontinuation. A single individual at each institution was designated to collect the data. Information was obtained from the nursing flow sheets and shift data records maintained by the bedside nurse. Collected data end points included: (1) length of ICU stay, (2) reexploration based on bleeding indications, (3) amount of chest tube drainage, (4) amount of blood products transfused, (5) number of hypertensive or hypotimsive episodes, (6) number of manual infusion rate changes, and (7) number of times the physician was notified ior problems related to blood pressure control. The data manager at each institution collected this informatian using a standardized database. The data were downloaded by modem to a central data bank, where patient information from all the centers was collated in a blinded fashion. Statistical methods used depended on the data set being antdyzed. Student's t test, one-way and two-way
Ann Thorac Surg 1992;54:517-22
Table 1. Patient Characteristics and Procedures" Manual (n = 532)
Variable Characteristics Age (Y) Weight (kg) Height (cm) Sex (ME) (%) Prior operation (%) Procedures First CABG (%) Repeat CABG (%) Valve (%)h Other I%)'
Titrator (n = 557)
*
62.5 0.5 62.6 ? 0.5 77.1 ? 0.7 78.0 ? 0.8 168.0 ? 0.7 166.0 2 0.7 76/24 67133 20.9 19.4 59.2 10.5 11.3 19.0
58.9 8.6 15.1 17.4
p Value Test 0.9 t test t test 0.4 0.15 t test 0.001 x2 0.544 x z 0.17 0.17 0.16 0.16
x2 x2 x2 x2
Where applicable, data are shown as the mean ? the standard error of the mean. This includes mitral valve replacement, aortic valve re' This placement, double-valve replacement, and double-valve repair. includes aortic aneurysm repair, Bentall grafts, repair of aortic dissections, transplantation, and repair of atrial septal defects.
a
CABG
=
coronary artery bypass grafting
analysis of variance, and x2 analysis each were used when appropriate. Unless otherwise indicated, all data are exthe standard error of the mean, pressed as the mean with significance determined at an a level of 0.05.
*
Results A total of 1,089 patients were enrolled in this study; 532 constituted the manual control group and 557, the automated titration cohort. A comparison of patient characteristics by group is provided in Table 1. No significant differences existed between groups for age, height, weight, or incidence of prior operation. However, the automated group was composed of approximately 7% more women than the manual infusion group (p < 0.001). There was a similar incidence of patients undergoing primary coronary artery bypass grafting, reoperative coronary artery bypass grafting, valve operations, and other procedures. Valve operations included mitral valve replacement, aortic valve replacement, double-valve replacement, and double-valve repair. Other procedures included grafting of ascending aortic aneurysms, repair of aortic dissections, repair of atrial septal defects, and cardiac transplantation. The incidence of hypertension was reduced signifi0.2 episodes per cantly by automated titration: 1.8 patient versus 0.6 0.07 episode per patient (p = 0.0001) (Fig 1). In addition, the frequency of hypotensive episodes resulting from nitroprusside therapy was reduced: 0.40 2 0.05 versus 0.30 0.03 episode per patient (p = 0.02). Although there was variability from site to site, overall chest tube drainage in the automated group was reduced by an average of 170 mL per patient ( p = 0.0001) compared with the manual cohort (Fig 2). As seen in Figures 3 and 4, the total blood products (whole blood and packed red blood cells) transfused per patient and the percentage of patients requiring transfusion were reduced
*
*
*
CHITWOOD ET AL CLOSED-LOOP NITROPRUSSIDE INFUSION
Ann Thorac Surg 1992;54:517-22
519
7
0 Manual I
.
._ c a U
$
3.0-
*
pco.01
2
* p=O.O001 ** p=0.02
Lo
.-P
E
.-c
Titrator
2.0
Site
EZ Titrator
o Manual
5 v)
3
c
Manual
Titrator
1
1.8ic 26
1.7f 2 1
2
2.8f .56
3.0 f .65
3
3.0f .42
2.0-
o*o 2.5 f 2 2
C
E
4.0 f .58
c
1.0
T Hypertensive
(0 c ._
5
**
Hypotensive
Fig 3 . Total blood products (whole blood, packed red blood cells) transfused by cohort and participating Titrator Multicenter Study Group institution, listed in blinded fashion. Data are shown as the mean and the standard error of the mean.
Fig 1. Number of hypertensive and hypotensive episodes per patient. Data are shown as the mean and the standard error of the mean.
significantly by closed-loop titration ( p = 0.01 and p = 0.0007, respectively). However, transfusion requirements vaned widely depending on the center. For example, Figures 3 and 4 demonstrate that when the titrator was used, blood product usage and transfusion patterns decreased at five centers (3, 4, 5, 6, and 8), did not change at two centers (1 and 7), and actually increased at two centers (2 and 9). On the other hand, examination of site data in Figure 2 demonstrates that chest tube drainage was decreased at both centers with increased transfusion requirements in the automated group. Thus, although an increase in blood product usage was demonstrated, it could not be correlated to a bleeding indication. Additional factors besides chest tube drainage, and possibly blood pressure, also variably influenced the usage of blood products in this study. These findings highlight problems inherent in a multicenter trial, particularly when a standardized protocol for the study is not established. At the same time, the fact that an overall difference could be demonstrated, despite these limitations, suggests strongly that the merits of the titration approach are valid. Reexploration for bleeding was almost twice as likely in the manually titrated patients (Fig 5). Overall, 36 (3.3%)of all 1,089 patients had reoperation for bleeding; nearly two thirds of them came from the manually controlled group
( p = 0.066). Although not significant for this population, the trend certainly is noteworthy. Automated nitroprusside titration reduced the average length of ICU stay by approximately 5 hours ( p = 0.038), virtually eliminated the need for manual infusion rate changes ( p = O.OOOl), reduced the average duration of nitroprusside therapy by about 3 hours ( p = 0.027), and reduced by one third ( p = 0.0001) the number of times the physician had to be contacted for problems related to blood pressure control (Fig 6).
Comment Estafanous and colleagues [l] first described hypertension as a consequence of aortocoronary bypass graft procedures. Generally, increases in blood pressure are encountered within the first 6 hours after cardiopulmonary bypass and may last 12 to 36 hours. The incidence of postoperative hypertension has been reported to range between 8% and 32% for valve replacements and from 8% to 73% after coronary operations [l-31. Although elevated blood pressure after cardiovascular procedures is common, management guidelines remain unclear. Flaherty 60
1000 nnn
800 c
0 Manual
rn Titrator
Site
Manual
Titrator
2
463ic138 1312+166
587 + 43 769+120 830 + 75
4
1705+429 629 k 156
1530 + 306 611 k63 407 f 149
1157f172
669 f 25 1039f144
834f185
300 200
100
* p=o.oooi
8
2.5k .50
1.0-
* p=0.0007
o Manual m Titrator
I
Site
Manual
43% 89%
5
46% 78% 55% 53% 1 9%
6
43%
1 7%
7
30%
27%
8
44% 6%
35%
40 30
~
9
Titrator
45% 0%
19%
586+138
Fig 2 . Chest tube drainage by cohort and participating Titrator Multicenter Study Group institution, listed in blinded fashion. Data are shown as the mean and the standard error of the mean.
Fig 4. Percentage of patients having transfusion by cohort and participating Titrator Multicenter Study Group institution, listed in blinded fashion. Data are shown as the mean and the standard error of the mean.
520
CHITWOOD ET AL CLOSED-LOOP NITROPRUSSIDE INFUSION
6
r
0 Manual
m Titrator
* p=0.066 (Chi-square) % of total
re-explored
96.7%
group distribution
Fig 5. Frequency of reexploration for bleeding by cohort and as a function of the total study population. (M = manual; T = titrutor.)
and colleagues [13] suggested that a sudden, persistent increase in mean arterial pressure of 20 mm Hg or more, in the presence of a mean pressure of at least 95 mm Hg, should be considered a hypertensive episode. By this criterion, 61% of all patients undergoing bypass grafting in their study displayed postoperative hypertension. Fremes and co-workers [ll]proposed that patients not be considered hypertensive postoperatively until the mean arterial pressure exceeds 105 mm Hg. Even by this liberal definition, 42% of their patients demonstrated systemic hypertension after bypass grafting. In the present study,
E
c
n
*I
Ann Thorac Surg 1992:54:517-22
a specific definition of hypertension was not preselected. Members from each institution were allowed to use their own variables to define systemic hypertension. Although the number of hypertensive episodes is known for each center, precise blood pressure levels were not recorded and varied from site to site. Although @-blockadewithdrawal, increased catecholamine levels, increased renin-angiotensin activity, emergence from anesthesia, and withdrawal from hypothermia all have been proposed as etiologic factors for systemic hypertension after bypass, mechanisms often remain obscure [4, 6-81. Irrespective of the cause, potential consequences may be serious and include increased bleeding, additional transfusion requirements, disruption of vascular anastomoses, arrhythmias, subendocardial ischemia, and possibly cerebrovascular hemorrhage [9, 101. The cost of transfusions and reoperation adds substantially to health care expenses [19]. In the present study, no attempt was made to define the cause of the hypertension. However, improved blood pressure control was evident using the closed-loop system and was reflected by a decreased number of hypertensive episodes, a reduced amount of bleeding, a reduced number of patients having transfusion, and a reduced volume of blood products needed in those patients requiring transfusion. Sodium nitroprusside remains the therapy of choice for
Length of ICU Stay
Hrs on SNP
* l *
20
p=0.027
40'
c
20'
Manual
Titrator
Manual
Manual Infusion Rate Changes
*
10
2.0
-
1.3
-
p=o.o001
Titrator
M.D. Consults re: BP
*
p=o.o001
* Manual
Titrator
Manual
Titrator
Fig 6 . Len,gth of intensive cure unit (ICU) stay, duration of nitroprusside (SNP) therapy, frequency of manual infusion rate change, and number of consults for blood pressure (BP) problems, by cohort.
Ann Thorac Surg 1992;54:517-22
management of systemic hypertension after bypass procedures. The major advantages of this drug are rapid onset of action, short duration, relatively minor direct cardiac effects, and effective afterload reduction by vasodilation of both venous capacitance and arteriolar resistance beds [ll, 1>16]. However, many of the characteristics that make nitroprusside effective and attractive also can be disadvantageous. Severe afterload reduction may decrease diastolic perfusion pressure below physiologic ranges and produce subendocardial ischemia. Also, a reflex tachycardia may occur and produce unwanted increases in myocardial oxygen consumption. As the drug is short acting, the rate of infusion must be adjusted frequently, causing manual blood pressure control to be extremely time-consuming for the nursing staff. The present study demonstrated that hypertension was better controlled by automated titration and that the incidence of unwanted hypotensive episodes was reduced. At the same time, the necessity for manual changes in infusion rate was almost eliminated. Although not directly evaluated, the ability to automate antihypertensive therapy suggests improved patient outcome by permitting the bedside nurse to concentrate more completely on other facets of critical care. In fact, the reduction in nursing time and stress associated with antihypertensive therapy may be the major benefit in utilizing the system. This ingenious technology now may allow accurate and sustained blood pressure control in the cardiac surgery ICU and has great potential for all critical-care environments. The closed-loop controller monitors blood pressure continuously and compares the instantaneous measurements with the operator-selected level. Incremental adjustments in infusions are made automatically at 10second intervals, allowing more precise control. Moreover, the controller compares blood pressure responses with the ”history” of previous infusion rates and pressure changes and then can extrapolate future blood pressure trend changes. This cumulative experience contributes to closer regulation of arterial pressure over time. In essence, the longer the controller regulates blood pressure, the better the hemodynamic stability becomes. Drs Sheppard and Kirklin promoted an interest in automated intensive care at the University of Alabama in the early 1970s. Subsequently, cardiac surgeons have continued to show an interest and have realized both the utility and future promise of these methods. Cosgrove and Petre, working at The Cleveland Clinic, developed the prototype closed-loop infusion system to control hypertension after a cardiac operation. At that institution, an upgraded system has been effective for routine clinical use since 1982. The development of microchip technology and the application of closed-loop theory to the bedside delivery of drugs have enabled clinicians to develop new potentials for the critical care environment. Sheppard [20] pioneered the application of computer control to the infusion of vasoactive drugs. Within the last 5 years, several groups [21-241 have developed excellent custom systems for microcomputer control of blood pressure and nitroprusside infusion. In 1989, a device approved by the Food and Drug Administration for the
CHITWOOD ET AL CLOSED-LOOP NITROPRUSSIDE INFUSION
521
infusion of sodium nitroprusside in patients hypertensive after bypass became available commercially (Titrator module, model 1OK; IVAC Corporation). Since that time, three clinical studies have been conducted using this device. Keogh and associates [25] compared 12 patients whose blood pressure was controlled by manual adjustments of nitroprusside with 22 patients controlled by automated titration. The desired preselected mean arterial pressure was 70 mm Hg for all patients. Throughout the study, the mean arterial pressure was maintained within 10% of the desired value 46% of the time in the manual group and 90% of the time in the automated group. Cosgrove and colleagues [17] also conducted a prospective, randomized study of 180 patients having a cardiac operation in whom manual infusion was compared with automated titration. In that study, mean arterial pressure was maintained within 10% of a selected target value 61% of the time with manual infusion versus 85% of the time with automated titration. Moreover, the incidence of unwanted hypertensive and hypotensive episodes during drug infusion was reduced from more than 20% in the manual group to less than 10% in the automated group. Bednarski and colleagues [18] compared automated control in 20 patients with manual control in 10 patients. During the first 10 minutes, blood pressure was within 10% of the desired level in 60% of the patients from both groups. However, over the next 140 minutes, blood pressure control within 10% of targeted values improved in 90% of patients in the automated group but remained only 60% for patients in the manual group. The present study confirms and extends the findings of Keogh [25], Cosgrove [17], Bednarski “1, and their associates. However, the patient sample was much larger and represented multiple centers. Moreover, additional end points included transfusion requirements, number of patients having transfusion, quantity of chest tube drainage, duration of antihypertensive therapy, and rate of reoperation for bleeding. Use of the automated titration system reduced the incidence of reoperation, hypertensive episodes, hypotensive episodes, and transfusions. When a transfusion became necessary, automated titration reduced the amount of blood products used. Improved control of blood pressure led to an earlier discontinuation of antihypertensive therapy and shorter stays in the ICU. Although the study shows improved blood pressure control using the closed-loop system, some improvements may have resulted from the increased attention to patients that nurses could provide when freed from frequent manual titrations. At present, commercial access to this technology is relatively expensive. As with most new technology, initial costs are high but can be expected to decrease as volume increases and additional applications for the technology are identified. For example, there remains great potential for the use of this system in the delivery of vasopressor and inotropic agents. Presently, the device is not approved for this use; however, in Australia this system has been shown effective with cardiotonic and vasopressor drugs [26]. The advent of continuous cardiac output measurements
522
CHITWOOD ET AL
Ann Thorac Surg 1992;54:517-22
CLOSED-LOOP NITROPRUSSIDE INFUSION
should help expand the utility of closed-loop systems w i t h these and other drugs. In coinclusion, the use of a commercially available closed-loop delivery system i n the administration of nitroprusside significantly improved patient outcome. In particular, there w a s a strong trend toward a reduced incidence of reoperation for bleeding indications. We recognize the assistance of Maureen Fallon, Sharon Gallagher, Cathy Galvin, Debbie Hall, Suzanne Provost, Janette Sherwood, Julie Shinn, and Pat Young in the collection of the patient data at each of the clinical centers. The assistance of Dr Mary Ann Christopher in the statistical analysis of the data is also appreciated.
Ref erenices 1. Estafanous FG, Tarazi RC, Viljoen JF, El Tawil MY. Systemic hypertension following myocardial revascularization. Am Heart J 1973;85:732-8. 2. Roberts AJ, Niarchos AP, Subramanian VA, et al. Systemic hypertension associated with coronary artery bypass surgery. Predisposing factors, hemodynamic characteristics, humoral profile, and treatment. J Thorac Cardiovasc Surg 1977;;'4:846-57. 3. Hoar PF, Hickey RF, Ullyot DJ. Systemic hypertension following myocardial revascularization. J Thorac Cardiovasc Surg 1976;71:859-64. 4. Matthews HR, Meade JB, Evans CC. Peripheral vasoconstriction after open-heart surgery. Thorax 1974;29:33842. 5. Taylor KM, Morton IJ, Brown JJ, et al. Hypertension and the renin-angiotensin system following open-heart surgery. J Thorac Cardiovasc Surg 1977;7484&5. 6. Fouacl FM, Estafanous FG, Bravo EL, et al. Possible role of cardioaortic reflexes in postcoronary bypass hypertension. Am J Cardiol 1979;44:866-72. 7. Wallach R, Karp RB, Reves JG, et al. Pathogenesis of paroxysmal hypertension developing during and after coronary bypass surgery. A study of hemodynamic and humoral factors. Am J Cardiol 1980;46:559-65. 8. Reves JG, Karp RB, Buttner EE, et al. Neuronal and adrenomedullary catecholamine release in response to cardiopulmonary bypass in man. Circulation 1982;66:49-55. 9. Viljoen JF, Estafanous FG, Tarazi RC. Acute hypertension immefdiately after coronary artery surgery. J Thorac Cardiovasc Surg 1976;71:548-50. 10. Estafanous FG, Tarazi RC. Systemic arterial hypertension associated with cardiac surgery. Am J Cardiol1980;46:68594. 11. Fremes SE, Weisel RD, Baird RJ, et al. Effects of postoperative hypertension and its treatment. J Thorac Cardiovasc Surg 1983;8,6:47-56. 12. Palmer RF, Lasseter KC. Sodium nitroprusside. N Engl J Med 1975;2.92:294-7. 13. Flaherty JT, Magee PA, Gardner TS, et al. Comparison of intravenous nitroglycerin and sodium nitroprusside for treatment of acute hypertension developing after coronary bypass surgeiy. Circulation 1982;65:1072-7. 14. Bixler TJ, Gardner TJ, Donahoo JS, Brawley RK, Potter A, Gott VL. Improved myocardial performance in postoperative cardiac surgical patients with sodium nitroprusside. Ann Thorac Surg 1978;25:44&8.
15. Stinson EB, Hollaway EL, Derby GC, et al. Control of myocardial performance early after open-heart operations by vasodilator treatment. J Thorac Cardiovasc Surg 1977;73: 523-9. 16. Gall WE, Clarke WR, Doty DR. Vasomotor dynamics associated with cardiac operations. I. Venous tone and the effects of vasodilators. J Thorac Cardiovasc Surg 1982;83:724-1. 17. Cosgrove DM 111, Petre JH, Waller JL, Roth JV, Shepherd C, Cohn LH. Automated control of postoperative hypertension: a prospective, randomized multicenter study. Ann Thorac Surg 1989;47:678-83. 18. Bednarski P, Siclair F, Voight A, Demertzis S, Lau G. Use of a computerized closed-loop sodium nitroprusside titration system for antihypertensive treatment after open heart surgery. Crit Care Med 1990;18:1061-5. 19. Taylor GJ, Mike11 FL, Moses HW, et al. Determinants of hospital charges for coronary artery bypass surgery: the economic consequences of postoperative complications. Am J Cardiol 1990;65:309-13. 20. Sheppard LC. Computer control of the infusion of vasoactive drugs. Ann Biomed Eng 1980;8:4314. 21. Pomer S, Elert 0, Satter P. Effects of the automated management of hypertension after open heart surgery. Int J Clin Pharmacol Ther Toxicol 1984;22:207-13. 22. De Asla RA, Benis AM, Jurado RA, Litwak RS. Management of postcardiotomy hypertension by microcomputer-controlled administration of sodium nitroprusside. J Thorac Cardiovasc Surg 1985;89:115-20. 23. Reid JA, Kenny GNC. Evaluation of closed-loop control of arterial pressure after cardiopulmonary bypass. Br J Anaesth 1987;59:247-55. 24. Rosenfeldt FL, Chang V, Grigg M, et al. A closed loop microprocessor controller for treatment of hypertension after cardiac surgery. Anaesth Intensive Care 1986;14:158-62. 25. Keogh BE, Jacobs J, Royston D, Taylor KM. Microprocessorcontrolled hemodynamics: a step towards improved efficiency and safety. J Cardiothorac Anesth 1989;3(1):4-9. 26. McKinley S, Cade JF, Siganporia R, Evans OM, Mason OG, Packer JS. Clinical evaluation of closed-loop control of blood pressure in seriously ill patients. Crit Care Med 1991;19: 166-70.
Appendix 1. Participating Centers and Investigators Boston University Hospital The Cleveland Clinic Good Samaritan Hospital (Los Angeles) The Johns Hopkins University Hospital Mt. Sinai Medical Center (New York) Stanford University Hospital Texas Medical Center Toronto General Hospital University of Kentucky Hospital
Richard J. Shemin, MD Delos M. Cosgrove 111, MD Gregory L. Kay, MD Bruce A. Reitz, MD, and William A. Baumgartner, MD Steven L. Lansman, MD D. Craig Miller, MD James J. Livesay, MD Richard D. Weisel, MD W. Randolph Chitwood, Jr, MD
