automated control of postoperative hypertension: a prospective, randomized multicenter study

Automated Control of Postoperative Hypertension: A Prospective, Randomized Multicenter Study Delos M. Cosgrove 111, MD, John H. Petre, PhD, John L. Waller, MD, Jonathan V. Roth, MD, Carol Shepherd, RN, and Lawrence FI. Cohn, MD Departments of Thoracic and Cardiovascular Surgery and Cardiothoracic Anesthesia, The Cleveland Clinic Foundation, Cleveland, Ohio; Department of Anesthesia, Emory Clinic, Atlanta, Georgia; and Division of Cardiothcracic Surgery, Brigham and Women's Hospital, Boston, Massachusetts

Hypertension after a cardiac operation is a frequent phenomenon. Complications resulting from this include bleeding, disruption of vascular suture lines, subendocardial ischemia, and possible cerebrovascular accidents. Treatment with sodium nitroprusside has become accepted practice to prevent these complications. To improve control of arterial blood pressure, a closed-loop system for sodium nitroprusside administration was developed. A prospective, randomized multicenter study was carried out postoperatively in 180 cardiac surgical patients to evaluate the performance of this system compared with manual control of infusion. Adherence of

ypertension after a cardiac operation is a well- H recognized and described phenomenon. The incidence of hypertension varies with cardiac pathology, operative procedure, and definition of hypertension. Af- ter valve replacement, the reported incidence ranges between 8% and 12% [l]. After myocardial revascularization, a higher incidence, ranging from 17% to 7370, is found [24 ] . Complications resulting from postoperative hypertension include bleeding, disruption of vascular suture lines, subendocardial ischemia, and possible cerebrovascular accidents. Because this morbidity and mortal- ity are theoretically avoidable, treatment of hypertension has become accepted practice.

Postoperative hypertension is relatively brief in nature, lasting several hours. Sodium nitroprusside is the drug most commonly administered to control transient hypertension because it decreases systemic vascular resistance and has a rapid onset and short duration of action. To avoid complications, rapid and consistent control of hypertension is essential. The quickly changing hemody- namics in a patient after a surgical procedure demand frequent adjustments of the nitroprusside infusion rate to maintain desired mean arterial blood pressure. This re- quires a substantial amount of time and effort on the part of the nurses [ 5 ] .

To improve regulation of arterial blood pressure when using sodium nitroprusside, several authors [6-14] have

Presented at the Twenty-fourth Annual Meeting ( i f The Society of Thoracic Surgeons, New Orleans, LA, Sep 2 6 2 8 , 1988.

Address reprint requests to Dr Cosgrove, Department of Thor'icic ciiid Cardiovascular Surgerv, The Cleveland Clinic Foundation, One Clinic Circle, 9500 Euclid Ave, Cleveland, OH 44195-5066.

mean arterial blood pressure to +lo% of the target blood pressure occurred 8!i% of the time with the automatic system and 61% of the time with manual regulation ( p < 0.0001). With the aubomatic system, there was less hypertension (9% versus 2:!%;p < 0.0001) and hypotension (6% versus 22%; p < 0.0001). The superior control of hypertension was achievedl more rapidly with less requirement for nurse regulation of infusion rate. The superior control of blood pressure resulted in less chest tube drainage in the automatic mode (720 mL versus 840 mL; p < 0.05).

(Ann Thorac Surg 1989;47:678-83)

proposed the use of (an automated closed-loop system to obtain better control of hypertension, avoid extremes of hypotension and hypertension, and reduce the demands on the time of the intensive care nurses.

Stimulated by the pioneering efforts of Sheppard [7] and Slate [8], The Cleveland Clinic initiated an indepen- dent effort in 1979, which resulted in a clinically useful and reliable closed-loop system for the administration of sodium nitroprusside [15]. In 1982, a joint venture was begun with the 1VA.C Corporation, which resulted in additional modifications and improvement in the performance of the control system. This system was demonstrated effective in animals [9]. To evaluate the clinical efficacy, safety, and usefulness of the resultant system, a randomized prospeclive study was carried out at The Cleveland Clinic, Emory University, and Brigham and Women's Hospital. The patient study compared the results of the automated system with those obtained manually by trained, experienced intensive care nurses.

Material and Methsods The IVAC Titrator model 10K closed-loop sodium nitroprusside system is a dedicated microprocessor-based in- strument that controls the infusion rate of a standard WAC variable-pressure volumetric pump, model 560i, connected by an interface cable. The Titrator 10K uses two separate algorithms: the first is active during the process of lowering the patient's mean blood pressure to a desired level (transient control mode); the second is an enhanced proportional, integral, and derivative algorithm designed to maintain the mean pressure at a desired level (linear control mode).

(C 1989 by The Society of Thoracic Surgeons 0003-4975/89/$3.50

Ann Thorac Surg 1989;47678-83

COSGROVE ET AL 679 AUTOMATED CONTROL OF HYPERTENSION

Table 1. Surgical Procedures

Procedure Automatic Mode Manual Mode

(n = 90) (n = 90)

Myocardial revascularization 66 75 Aortic valve replacement 9 5 Repeat myocardial 6 3

Mitral valve replacement 4 2 revascularization

Other 5 5

I- z W

Lu 2 a

The system ”learns” from the patient’s response to nitroprusside, and uses ”learned information” in calcu- lating subsequent infusion rate adjustments. The arterial blood pressure is sensed every %nu of a second. The infusion rate is adjusted every 10 seconds. Should blood pressure drop rapidly, an aggressive infusion rate reduction program prevents overdosing of the patient, often reducing the infusion rate even when mean arterial pressure is greater than the target pressure. The Titrator can be placed in manual mode, thus reverting control of the model 560i pump to the operator through the front panel switches as though the Titrator were not present.

A magnetic tape recording device (Memodyne) was connected to the system and documented, on a real-time basis for all study patients, incoming transducer signal, infusion pump variables, and Titrator functions, such as the control mode (automatic or manual). The Memodyne device is a data-recording tool used for research purposes only; it was not an integral part of the IVAC closed-loop regulator system.

The protocol was reviewed by the respective institu- tional human investigation review boards, and informed consent was obtained. One hundred eighty patients were prospectively assigned in a random fashion according to the odd-even status of the hospital number to have the administration of nitroprusside regulated manually or in the automatic mode. There were 90 patients in each of the two study groups. The average age was 59.3 * 9.6 years (* the standard error of the mean) in the automatic mode group and 65.2 * 9.2 years in the manual mode group. In

loo[ 80

60

40

20 t 1 , NS,, & n -

> 20% I I - 20%

0 AUTOMATIC

n

*lo% 11-20% >20% BELOW ABOVE

Fig 1. Distribution of blood pressure determinations for the first ten inin utes.

‘OOE 80

L/! 20

0 > 20% I1-20%

pc.0001 ,-, OAUTOMATIC

f 10%

MANUAL

P<.OOO1 p<.ooo1

LEuL I I-20% > 20%

BELOW ABOVE Fig 2 . Distribution of blood pressure deterininations for the second ten minutes.

the automatic mode group, 84.4% of the patients were male and in the manual mode group, 78.9%. The surgical procedures performed are listed in Table 1. Blood pressures were recorded by the Memodyne recorder at 10- second intervals. There were 216,210 determinations in the automatic mode group and 228,726 determinations in the manual mode group. The investigation logged 600 total hours of recorded data in the automatic mode and 635 hours under manual control.

An analysis compared the two methods at various time intervals and demonstrated the range of the resultant mean pressure relative to the desired mean blood pressure. Blood pressures were grouped in five categories: more than 20% less than the desired mean; 11% to 20% less than the desired mean; within *lo% of the desired mean; 11% to 20% greater than the desired mean; and more than 20% greater than the desired mean.

Chest tube drainage was recorded from the time of operation until 7:OO AM the following morning.

Each patient was analyzed individually for each time frame. The time frames were divided into six windows of 0 to 10 minutes, 10 to 20 minutes, 20 to 30 minutes, 30 to 45 minutes, 45 to 60 minutes, and remaining time. A mean percent error ([actual - desired]/desired x 100) was calculated for each patient for each time interval. These values were then compared for the two groups (automatic mode and manual mode) with Student’s t test. A p value of 0.01 or less was considered significant.

Results Because of the potential for large oscillations in blood pressure at the beginning of treatment with nitroprusside, the first three 10-minute intervals were analyzed separately. Automatic control was superior to manual control in regulating hypertension. More pressures remained within the desired blood pressure range while avoiding the extremes of hypotension and hypertension (Figs 1-3). During this period, there was a dramatic improvement in the control of blood pressure between the first two periods in the automatic mode with little change in the manual mode.

To evaluate the ability of the automatic system to

680 COSGROVE ET AL AUTOMATED CONTROL OF HYPERTENSION

Ann Thorac Surg 1989;4767%33

l O O r 100 r t P<.OOOl

0

40

0

ps.02 p<.05 m

>20% I1-20% BELOW

f 10%

0 AUTOMATIC

MANUAL

p<.ooo1

p<.ooo1 I4-L I I -20% > 20%

ABOVE Fig 3 . Distribution of blood pressure determinations for the third ten minutes.

control blood pressure during a period when the requirements for nitroprusside were stable, the next two 15- minute periods were separately analyzed. A significantly greater percentage of blood pressure determinations fell within 210% of the desired mean with the automatic system during these two periods. Analysis also demonstrated less hypotension and hypertension in the automatic mode group (Figs 4, 5 ) .

To evaluate the period when requirements for nitroprusside were decreased and the infusion rate was ta- pered, the remaining time intervals were analyzed. The desired range for mean arterial blood pressure was obtained a higher percentage of the time with the automatic system than with the manual system, and hypertension and hypotension were less common (Fig 6) .

When all blood pressure determinations for the entire study were analyzed, a significantly greater percentage of determinations fell within the +lo% range for the automatic system, with substantially fewer determinations of hypertension and hypotension (Fig 7).

To further evaluate the ability of the two techniques to achieve regulation at a desired mean arterial blood pressure, the mean percent error from the target blood pressure of the entire group was calculated. The mean percent error for the automatic mode group was lower than that for the manual mode group ( p < 0.0001) (Fig 8).

2 40

20

O >20% 11-20?40 f 10% 11-20% >20% BELOW ABOVE

Fig 5. Distribution oj blood pressure determinations from 45 to 60 minutes.

The data were analyzed at individual study sites to determine the influence caused by a divergent patient population, intemive care personnel, and surgical and anesthesia practices. For the overall time period, the mean percent error (the percentage of deviation from the desired blood pressure) ranged from 4.9% to 5.8% for the automatic mode group and 10.1% to 10.9% for the manual mode group (Fig 9). The automatic control system had nearly identical results at each institution and produced significantly superior regulation of the mean arterial blood pressure.

To evaluate the rapidity by which hypertension was controlled by each of the two methods, mean percent error from the desired blood pressure was evaluated at various time intervals. During the first ten minutes of control, performance of the automated system was better than that of the manual system. The mean percent error was stabilized to less than 10% within the first ten minutes. This level of control was not achieved until after 45 minutes in the manual control mode (Fig 10).

When the automated mode was used, there was a significant reduction in the amount of time nurses spent to adjust the infusion rate of nitroprusside. In the automated mode, a mean of 0.45 change per hour was experienced by the nursing staff, whereas 6.03 changes per hour occurred in the manual mode ( p < 0.0001).

l O O r 100 r

I:[ 20 0

N S ,

>20% I1-20% BELOW

p<.ooo1

f 10%

0 AUTOMATIC

MANUAL

p<.ooo1

11-20% >20% ABOVE

80

60 0

40

20

O >20% 11-2096 *lo% 11-20~/0 >20% BELOW ABOVE

Fig 4 . Distribution of blood pressure deterrnimtions jroni 30 to 45 in in u tes .

Fig 6 . Distribution of blood pressure determinations for the remainder oj the time.


20

15

n:

n: z


- AUTOMATIC

MANUAL --*--- pco.ooo1

Q\ - '\, '. o-----o'

'\

\ '.

loo[ 80

I- z w 0 [r W a I 20 0

pc.0001 n 0 AUTOMATIC

MANUAL

> 20% I1-20% f 10% I I-20% > 20% ~. ~

BELOW ABOVE

Fig 7 . Distribution of blood pressure determinations for the entire study period.

Chest tube drainage was less in patients under automatic control (720 ? 312 mL) compared with those controlled in the manual mode (840 ? 462 mL) ( p < 0.05).

Comment The study of cardiac surgical patients after operation to evaluate an automated closed-loop system that controls the infusion of nitroprusside is appropriate for several reasons. First, hypertension is common in this setting, thus allowing controlled studies. Second, the hemody- namics are extremely labile, providing a rigorous test for any system. Finally, the complications associated with hypertension, such as blood loss, are measurable, resulting in an end point that can be measured.

The superiority of the automatic system for administer- ing nitroprusside is evident in this study. Adherence to within 210% of a target blood pressure was superior at all time intervals measured for the automatic system. Simi- larly, control of hypertension was dramatically better at all time intervals. In the entire study, 20% of the blood pressure determinations in the manual control mode were more than 10% greater than target pressure compared with 6% for the automatic system ( p < 0.0001). Hypoten- sion was equally well eliminated by the automatic system. Patients in the manual mode group had 22% of their determinations more than 10% less than the target pres-

2o r AUTOMATIC

MANUAL pc.0001

Fig 8. The mean percent error from desired blood pressure for the automatic mode group was less than that for the manual mode group.

pc.0001

0 AUTOMATIC

MANUAL

pc.0001 pc.0001

CLEVELAND EMORY BRIGHAM 8 WOMEN'S

Fig 9. The autornntic system had a loruer mean percent error from desired blood pressure nt all three study sites.

sure whereas only 9% of the automatic mode group were in this range ( p < 0.0001). The evidence that blood pressure control is superior in the automatic mode is unequivocal.

To be optimally effective and to minimize complications, control of hypertension must be rapid. The automatic system demonstrated this characteristic by reducing the mean percent error to less than 10% in the first ten minutes. This same degree of control required 45 minutes for patients with nurse-controlled infusions. Rapid response in the automatic mode is a direct result of specif- ically written algorithms that have a transient control mode.

Several features of this device enhance patient safety during the administration of nitroprusside. To avoid the dangerous complication of hypotension, the algorithm is designed in such a way that should the blood pressure begin to drop rapidly, an aggressive infusion rate reduction program rapidly reduces the infusion rate. This

'0-10 10-20 20-30 30-45 45-60 REMAINING TI ME

TIME (MINUTES) Fig 10. Hypertension 7uas more rapidly controlled with the automatic system, which reduced the mean percent error to less than 10% in the first ten minutes.

682 COSGROVE ET AL AUTOMATED CONTROL OF HYPERTENSION

Ann Thorac Surg 1989;47678-83

capability prevents overdosing of the patient even in instances where the mean arterial blood pressure is greater than desired. To avoid nitroprusside toxicity, infusion rate limits of 299 mL/h and 10 pg/kg/min and a total dose of 4 mg/kg of nitroprusside administered are included in the design. Other safety features include an automatic internal system for systematic checks during power-up and routine use, and an infusion shutoff when the mean arterial pressure signal is severely deranged for two minutes or if an internal component fault is detected. Further safety features include circuitry that rejects arti- facts, such as motion of the pressure tubing and loss of signal because of blood sampling. This maintains the previously determined infusion rate and prevents inap- propriate infusion rate adjustment during such activities.

The differences between institutions in patient population, intensive care personnel, and surgical and anesthesia practices provided test environments to determine the capability of the automated system to control blood pressure under diverse circumstances. The versatility of the system was demonstrated by the almost identical performance at all three institutions. This performance bodes well for the wide application of this technology.

It is well recognized that a substantial percentage of the time and attention of intensive care nurses is required to tightly control the blood pressure of a patient receiving an infusion of nitroprusside. One study [5] demonstrated that 40% of patients undergoing a cardiac surgical procedure received nitroprusside postoperatively and that 16% of nursing time was spent regulating the infusion of this drug. The effectiveness of this study demonstrates that infusion rate manipulation by nurses can be reduced 15-fold. This device has the potential of allowing nursing personnel to spend more time on other patient-related activities.

With an increased interest in reducing homologous blood transfusions to minimize the risk of acquired im- munodeficiency syndrome and hepatitis, a major benefit of the Titrator system may be the potential for reducing postoperative chest tube drainage. The observed reduction of chest tube drainage and bleeding indirectly provided by the automated system appears to offer a major opportunity to reduce blood transfusions. The potential to further reduce chest tube drainage by controlling blood pressure at lower than usual levels has yet to be investi- gated. This system provides the opportunity to attempt this in a highly controlled and safe fashion.

Although this device was developed solely for use in adults after cardiac operations, its use in hypotensive anesthesia has been explored [9]. Neurosurgery, hyper- tensive crisis, aortic surgery, and cardiopulmonary bypass are logical areas of potential application in the future.

Cardiac surgery has directly benefited very little from the advancements associated with computer technology and artificial intelligence. This infusion control system

represents a unique application of advanced technology to solve a specific clinical problem. It is envisioned that as additional sensors and algorithms are developed, closed- loop systems for controlling serum glucose, calcium, blood volume, cardiac output, oxygen saturation, cardiopulmonary bypa:js, ventilators, and anesthesia machines will become clinical realities that will increase the safety of cardiac operations. This system represents the development of a new drug-delivery system, but it is only the first step in the utilization of automated control technology, which holds vast potential.

References 1. Estafanous FG, Tarazi RC, Buckley S, Taylor PC. Arterial

hypertension in immediate postoperative period after valve replacement. Br Heart J 1978;40:718-24.

2. Hoar PF, Hickey RF, Ullyot DJ. Systemic hypertension following myocardial revascularization: a method of treatment using epidural anesthesia. J Thorac Cardiovasc Surg 1976;71: 85944.

3. Reves JG, Lell WA, McCracken LD Jr, et al. Comparison of ketamine and morphine anesthesia for coronary surgery. South Med J 197€,;71:33-6.

4. Wallach R, Karp RB, Reves JG, et al. Mechanism of hypertension after saphenous vein bypass surgery [Abstract]. Circulation 1977;55, 56(Suppl 3):141-2.

5. Mitchell RR. The need for closed-loop therapy. Crit Care Med 1982;10:8314.

6. de Asla RA, Benis AM, Jurado RA, Litwak RS. Management of postcardiotom,y hypertension by microcomputer-controlled administration of sodium nitroprusside. J Thorac Cardiovasc Surg 1985;89:115-20.

7. Sheppard LC. Computer control of the infusion of vasoactive drugs. Ann Biomed Eng 1980;8:4314.

8. Slate JR. Model-based design of a controller for infusing sodium nitroprusside during postsurgical hypotension [PhD Dissertation]. Madison, WI: University of Wisconsin, 1980.

9. Meline LJ, Westenskow DR, Pace NL, Bodily MN. Comput- er-controlled regulation of sodium nitroprusside infusion. Anesth Analg 1985;64:3842.

10. Stern KS, Chizeck 131, Walker BK, Krishnaprasad PS, Dau- chot PJ, Katona PG. The self-tuning controller: comparison with human performance in the control of arterial pressure. Ann Biomed Eng 1985;13:341-57.

11. He WG, Kaufman H, Roy R. Multiple model adaptive control procedure for blood pressure control. IEEE Trans Biomed Eng 1986;33:1G9.

12. McInnis BC, Deng L.Z. Automatic control of blood pressures with multiple drug inputs. Ann Biomed Eng 1985;13:217-25.

13. Koivo AJ, Larnard D, Gray R. Digital control of mean arterial blood pressure in dogs by injecting a vasodilator drug. Ann Biomed Eng 1981;9:185-97.

14. Smith NT, Quinn ML., Flick J, Fukui Y, Fleming R, Coles JR. Automatic control i r anesthesia: a comparison in performance between the anesthetist and the machine. Anesth Analg 1984;63:71522.

15. Petre JH, Cosgrove DM, Estafanous FG. Closed loop com- puterized control of sodium nitroprusside. Trans Am SOC Artif Intern Organs 1983;29:501-5.



DISCUSSION

DR FRANCIS ROBICSEK (Charlotte, NC): I thank The Society for asking me to review this paper, and I congratulate Delos Cosgrove and his associates on their very well constructed and carefully executed study, and at the same time, give proper credit to the technical developer.

The question one should ask after the presentation is not why, what, and how, but probably why so late? In the early 1970s, following John Kirklin’s footsteps, we “closed the loop” and introduced automated infusion and transfusion in eight intensive care beds at our department. I t is hard to forget that when we first approached computer engineers with our plans, their initial respect turned first into amazement and then into amusement over our clinical methods, which were certainly primitive compared with industrial sophistication. When we inquired whether they could measure urine output and chest tube drainage electronically, and together with left atrial pressure use them as an algorithm for volume replacement, one of them burst out laughing and said, “Doc, this machine can send this cup of urine to the moon, measure it there, and bring it back to run your ‘drip, drip, drip‘ machine.”

In 1988, it is an anachronism that while we trust our life to an autopilot to fly us to New Orleans, we let our patient’s blood pressure rise and fall like a roller coaster because a nurse, among the multiplicity of his or her chores and exhaustible attention span, is simply unable to keep track of it.

I have only a few questions. First, do you believe that the application of this technique will allow you to decrease the nurse/ patient ratio in your intensive care unit? We were hoping this would occur, but in our practice, it never materialized.

Second, the classic indication for nitroprusside infusion is not hypertension but elevated circulatory resistance. Do you intend to make your system more sophisticated to include automatic feed-in of cardiac output and calculation of resistance?

Finally, 1 am somewhat surprised that after you were so successful in keeping the blood pressure down, you are not turning your program ”in reverse” and using it to administer vasopressors (in the case of hypotension) or dopamine hydrochloride based on automatic cardiac output measurements. Do you plan to do this?

Again, I compliment Dr Cosgrove and associates for bringing the technique of intravenous drug therapy closer to the modern age.

DR KENNETH M. TAYLOR (London, England): I also compliment Dr Cosgrove and his co-workers for this important study. At the Hammersmith Hospital, we have had the opportunity to study the IVAC Titrator system, and we have used it in a prospective study of our cardiac surgical patients.

Almost 100,000 data points were collected, and the superiority of control of blood pressure was essentially identical to that presented by Dr Cosgrove‘s group. The automatic mode achieved target blood pressures during the study period in 90% of patients, whereas the manual mode group with our experienced intensive care nurses was at around 45%. We found consistently that the nurses tended to err on the side of allowing the pressure to be slightly on the high side of the target range, and there is no significant difference in the low incidence of hypotensive episodes described.

Like Dr Cosgrove‘s group, we found that the time necessary to get the blood pressure into the target range was on the order of 20 minutes in the automatic mode compared with more than two hours for the patients being controlled manually.

We were particularly impressed with some of the safety features built into the algorithm of this nitroprusside infusion system. For example, in 1 patient, the target level of blood pressure had been achieved well but at the expense, if you like,

of a consistently rising infusion rate for nitroprusside. Adminis- tration of a bolus of intravenous vasodilators by the surgical staff resulted in a fall in the blood pressure. When this sequence of events occurs, the algorithm of the system will shut off the Nipride (sodium nitroprusside) infusion totally if the blood pressure falls to less than 75% of the target level.

The interesting thing is that if such a patient were being managed manually by the intensive care nurse, the nurse is unlikely to restart the Nipride infusion until the blood pressure returns to target levels. The ability of this system to sense rates of change and adjust appropriately means that it switches the Nipride infusion back on again while the blood pressure is still low but climbing, and achieves restoration of the target control levels much more quickly than might be anticipated with manual control. Our experience with the Titrator system in Hammersmith, therefore, is entirely in accord with the data presented by Dr Cosgrove and his group.

What was the response of the nursing staff in the various institutions in your study, Dr Cosgrove? We found that after some initial and I think inevitable skepticism, the nursing staff was thoroughly convinced of the efficacy and value of the Titrator system, particularly in the management of difficult patients. We also believe that this is an important first step in the application of computer technology to data collection and to patient management after routine cardiac procedures.

DR COSGROVE: Dr Taylor alluded to nursing staff acceptance of this device. We noted that the nurses were initially cautious in their acceptance, but by the time they had worked with it for approxi- mately 2 weeks, they found it preferable to manual titration.

We also hoped that we would be able to reduce the nurse/ patient ratio, but found that this is not possible. We think that we are going to be able to maintain our current nurselpatient ratio as nurses do increasingly complex work, for example, running patients on left ventricular assist devices and caring for much more acutely ill patients.

Dr Robicsek asked if we have not directed our nitroprusside therapy at the wrong end point. Should we be controlling systemic vascular resistance as opposed to controlling blood pressure? It became clear to us that we needed to have continu- ous on-line cardiac output to derive systemic vascular resistance and, in fact, titrate nitroprusside to achieve the optimal systemic vascular resistance. Currently, this project is well underway and promises to bear fruit in the next few years.

Finally, I caution people not to proceed directly to the use of vasopressors based solely on hypotension in this sort of closed- loop device. The reason that we give vasopressors and all cardiotonic drugs is not just to raise blood pressure but, more importantly, to increase cardiac output. Until such time as we have an on-line sensor device to continuously measure cardiac output, 1 do not think i t is reasonable to expect that we can use dopamine, dobutamine hydrochloride, or Levophed (norepi- nephrine bitartrate) in a closed-loop system.

Cardiac surgery has not benefited directly from the advances associated with computer technology and artificial intelligence. This infusion control system represents a unique application of advanced technology to solve a specific clinical problem. It is envisioned that as additional sensors and algorithms are developed, closed-loop systems for controlling serum glucose, calcium, blood volume, cardiac output, oxygen saturation, cardiopulmonary bypass, venhlators, and anesthesia machines will become clinical realities that will increase the safety of cardiac operations.

This device represents the development of a new drug- administration system, but it is only the first step in the utilization of automated control technology, which holds vast potential.

automated control of postoperative hypertension: a prospective, randomized multicenter study

Documents