British Journal of Anaesthesia 1992; 69: 607-610
EVALUATION OF CLOSED LOOP CONTROL OF ARTERIAL PRESSURE DURING HYPOTENSIVE ANAESTHESIA FOR LOCAL RESECTION OF INTRAOCULAR MELANOMA S. CHAUDHRI, J. R. COLVIN, J. G. TODD AND G. N. C. KENNY PATIENTS AND METHODS
SUMMARY
KEY WORDS Anaesthetic techniques • hypotensive
anaesthesia, manual
control, closed-loop computer control.
Patients undergoing local resection of intraocular melanoma require intraoperative arterial pressure control to a systolic pressure of 40 ± 5 mm Hg during the period of choroid dissection. Hypotension to this degree is required to prevent intraocular haemorrhage from the vascular choroid, which may affect visual outcome [1,2]. Precise control of arterial pressure during this period is essential, combined with meticulous monitoring of die adequacy of cerebral and myocardial perfusion. Control of arterial pressure by closed-loop infusion of vasodilators has been shown to compare favourably with the control achieved by intensive therapy staff in postoperative cardiac surgical patients [3-6]. Automatic control of arterial pressure has also been used during operation and was shown to compare favourably with manual control by the anaesthetist[7-10}. We have compared the quality of induced hypotension by computer-controlled infusion with that achieved manually by an experienced anaesthetist during local resection of choroidal melanoma.
We have studied 20 patients aged between 39 and 66 yr undergoing local resection of intraocular melanoma and considered suitable for hypotensive anaesthesia. Hospital Ethics Committee approval was obtained and informed consent given by all patients. All patients received the standard anaesthetic for this procedure in this unit [7], including oral atenolol 50-100 mg 90 min before operation, followed by i.m. pethidine 50-100 mg and promethazine 25-50 mg 1 h before surgery. Anaesthesia was induced with thiopentone 5 mg kg"1 and suxamethonium -1 1.5mgkg was given to facilitate tracheal intubation. Anaesthesia was maintained with nitrous oxide and 1 % enflurane in oxygen. Controlled ventilation to an end-tidal carbon dioxide concentration of 3.8-4.2 kPa was facilitated by tubocurarine and a 10° head-up tilt was used. The dose of tubocurarine was titrated using the Datex Relaxograph to maintain the first twitch response to a train-of-four stimulation (Tl) at less than 15% of control. Antagonism of neuromuscular block was achieved with neostigmine and glycopyrronium. Systolic arterial pressure was controlled by an infusion of sodium nitroprusside (SNP) 200 ug ml"1 and trimetaphan camsylate (TMP) 1 mg ml"1 in 5 % glucose [11-12]. Patients were allocated randomly to two groups: in the manual group, the infusion rate was adjusted manually; in die computer group, the infusion rate was adjusted by a computer-controlled, closed-loop system which comprised an Imed 929 Volumetric pump linked by an Apple He microcomputer to the data output of a Datascope 2002 monitor. This system has been described and evaluated previously [3, 5] and consists of a Proportional Integral Derivative controller widi adaptive features. The initial infusion rate was calculated by the closed-loop system. After each new cal-
SOFIA CHAUDHRI*, M.B., CH.B., F.R.C.ANAES. J JOHN R. COLVTNt, M.B., CH.B., F.F.A.R.C.S.I., F.R.C.ANAES.; GAVIN N . C. KENNY, B.SC.,
M.D., F.R.C.ANAES. ; University Department of Anaesthesia, Glasgow Royal Infirmary, Glasgow G31 2ER. J. GORDON TODD, M.B., CH.B., F.R.C.ANAES., Division of Anaesthesia, Western Infirmary^Glasgow-G H-6NTrAccepted forPublication: October 10, 1991. Present addresses: •Western Infirmary, Glasgow Gil 6NT. fNinewells Hospital, Dundee DD1 95Y. Correspondence to S.C.*
Downloaded from http://bja.oxfordjournals.org/ at UQ Library on June 14, 2015
We have studied 20 patients undergoing local resection of intraocular melanoma during hypotensive anaesthesia, allocated randomly to receive either manual control by an experienced anaesthetist or closed-loop computer control of an infusion of a 5:1 mixture of trimetaphan camsylate (IMP) and sodium nitroprusside (SNP). There were no significant differences in the smallest systolic and diastolic arterial pressures obtained, heart rate or infusion requirements between the two groups, but the duration of both the infusion and the operation were significantly longer in the computer-controlled group (P < 0.05). The quality of control of arterial pressure was assessed by the percentage of time spent at pressures greater and less than the prescribed target values, and was satisfactory in both groups during the critical period of profound hypotension. We conclude that the computer-controlled infusion performed satisfactorily during profound hypotension compared with an experienced anaesthetist. (Br. J. Anaesth. 1992; 69: 607-610)
BRITISH JOURNAL OF ANAESTHESIA
608
RESULTS
There were no statistical differences between the two groups in age, sex or weight (table I) or in the duration of the period of hypotension, the systolic pressure at induction, the smallest systolic and diastolic pressures obtained, or the smallest mean arterial pressures and heart rates during the period of hypotension in the two groups (table I). The quality of arterial pressure control was assessed by measuring the percentage of time spent at > ±10% outside the prescribed target pressure. During the entire period of infusion of vasodilator (table II), there were no significant differences in the percentage of time spent at pressures greater than the prescribed target pressure. However, patients in the computer-controlled group spent more time at pressures 10% less than the target pressure (P < 0.05). During the period of profound hypotension, there were no significant differences between the two
groups in the percentage of time spent at pressures greater and less than the target (table II) and the quality of arterial pressure control was satisfactory in both groups. This may be explained by the smaller reductions in the prescribed target pressure as the critical hypotension was reached, and the corresponding small changes in infusion rate required to achieve the prescribed target. There were no significant differences in the total amount of SNP or TMP infused in the two groups. Both the duration of the operation and the duration of infusion of vasodilators were longer in the computer-controlled group (P < 0.05) (table III). The longer duration of die infusion could be explained by the computer increasing the infusion in smaller increments dian the experienced manual TABLE I. Patient characteristics {mean {range or SDJ) and operative details (median (range))
Age (yr) Weight (kg) Sex (M/F) Hypotension (min) Stan SAP (mm Hg) Smallest SAP (mm Hg) Smallest DAP (mm Hg) Smallest MAP (mm Hg) Heart rate (beat min"1)
Manual group (n = 10)
Computer group (n = 10)
51.4(31-68) 68.0(13.6) 4/6 90 (50-105) 135(100-200)
52.9 (39-68) 70.6(12.5) 6/4 97.5 (60-130) 127(100-170) 40(38-^12) 21.5(20-32) 27.5 (20-34)
40(36-^5)
21(18-25) 27.5 (25-31) 67 (54-90)
60 (50-75)
TABLE II. Quality of arterial pressure control throughout the infusion and during profound hypotension (median (range)), expressed as time spent at pressures greater than ±10% outside target pressures. *P < 0.05 between groups
Manual group (n = 10) During infusion Target+10% Target - 1 0 % During hypotension Target+10% Target-10%
7.69(2.1-22.4) 10.3 (0-22.4) 2.3 (0-6.0) 6.6(0-31.8)
Computer group (n = 10) 8.7(5.4-13.9) 22.3(5.3-27.6)* 4.6(0-14.5) 2.6 (0-27.7)
TABLE III. Infusion requirements and intraoperative complications (median (range) or number). *P < 0.05 between groups
Manual group (n = 10) Infusion requirements Total SNP (ml) 8.8(2.2-13.8) Total TMP (ml) 44(11-69) Duration of 158(120-202) infusion (min) Duration of 198(160-250) operation (min) Intraoperative complications Oculocardiac reflex: 2 Early 0 Late 1 Both ST depression > 1 mrrl 0 1 Tachycardia 1 Increased IOP 0 Bleeding during choroid dissection
Computer group (n = 10) 13.5(6-41.8) 67.8 (30-209) ' 204 (100-268)* 248(150-285)*
3 0 1 0 0 0 4
Downloaded from http://bja.oxfordjournals.org/ at UQ Library on June 14, 2015
culation of the infusion rate, the system tracked the rate of change in pressure so that excessive or inadequate decrease caused a further calculation to be made. The raw arterial waveform was scanned by a background subroutine which allowed detection of zeroing, flushing, damping or disturbance of the line. A warning was provided, with an appropriate message for any computer-pump interface errors. Standard monitoring for this procedure comprised a continuous ECG using the CM5 lead, direct arterial pressure from the radial artery with the transducer at the level of the external auditory meatus, neuromuscular function and end-tidal carbon dioxide concentration. The mean amplitude and zero cross frequency of the EEG were displayed on the Datex Anaesthesia and Brain Activity Monitor (BAM). During the first 60-90 min, the systolic arterial pressure was reduced gradually to 70-80 mm Hg to allow fashioning of a superficial scleral flap and a vitrectomy to be performed. Profound hypotension to 40 + 5 mm Hg was then induced before final dissection of a deep scleral flap and incision of the choroid. During this period, ST segment changes and the zero cross frequency of the BAM were used as indicators of global myocardial and cerebral perfusion, respectively. If ST segment depression of greater than 1 mm or a sustained zero cross frequency of less than 5 Hz occurred, the arterial pressure was increased, although this could result in loss of the eye. Nitrous oxide was discontinued 20 min before closure of the scleral flap. After closure of the scleral flap and injection of sulphur hexafluoride gas, the arterial pressure was allowed to increase slowly by less than 5 mm Hg min"1 to avoid variations in cerebral blood flow and cerebral oedema. At the end of the procedure anaesthesia was discontinued, neuromuscular block antagonized and the trachea extubated. Patients were kept in the recovery area until obeying commands; i.v. fluids and oxygen were administered until the following day. Results were analysed using chi-square and the Mann-Whitney U tests.
CLOSED LOOP CONTROL OF ARTERIAL PRESSURE
DISCUSSION
Closed-loop, computer-controlled infusion of vasodilators has been used successfully in the control of hypertension after cardiac surgery [3—6], in which the quality of control achieved by the computercontrolled system has been shown to compare favourably with that achieved manually by nursing staff [3]. Closed-loop computer control of arterial pressure using either volatile agents or vasodilators such as SNP has also been used successfully during operation [7-10], and the control achieved compared favourably with that achieved by the anaesthetist. We are not aware of any other study in which a combination of SNP and TMP has been infused by a closed-loop, computer-controlled system. The system we used had been used previously to achieve normotension after operation. We have shown that this system can also be used during operation for the production of induced hypotension. Because of variations in operating stages, it was not possible to standardize the time intervals at which changes in target arterial pressure were made or the magnitude of changes in target pressure, and changes were made in response to prevailing surgical conditions. The period of steady-state control of arterial pressure is critical, and the quality of control of arterial pressure during the period of profound hypotension required for dissection of the tumour from the choroid was equally satisfactory in both groups. However, the anaesthetist controlling the infusion in the manually controlled group of patients was experienced in the provision of anaesthesia for this type of surgery and results achieved by a less experienced operator may have compared less favourably with the computer-controlled group. "Despite^atiffactory control oFafferial pressure~in both groups during the period of profound hypotension, four patients in the computer-controlled group were reported to have some minor bleeding during dissection of the tumour from the vascular choroid. This was a subjective assessment made by
the surgeon, who was unaware to which group the patient had been allocated. The cause was unclear, but may be related to other factors such as size and position of the tumour or difficulty in dissection, as it was also noted that the duration of operation and infusion of vasodilators was significantly greater in the computer-controlled group. Retrospective analysis of the patient data by the surgeon suggests that, in those patients in whom bleeding was noted, the tumour may have been more difficult to operate upon. Loss of the eye as a consequence of poor operating conditions did not occur in any of the patients. This was a small study, however, and it was not possible to study long-term outcome. Further studies would be required to evaluate this complication in greater detail. There were no adverse changes on the ECG or the BAM during the operation in either group. All patients were followed up at 6 weeks, 6 months and 1 year. There were no sequelae attributable to hypotension or anaesthesia. Detailed psychometric testing was not carried out, but one patient in the manually controlled group reported mild memory impairment in the early postoperative period. This showed rapid improvement and by 6 weeks the patient was unaware of any impairment of memory function. All patients returned to their previous functional level within the limits of their visual outcome. We believe that automated control of arterial pressure should only be used under supervision and that careful validation of the input signal must be made. We have found that analysis of the raw arterial waveform allows detection of potential errors such as flushing or disturbance of the arterial line and prevents the system infusing vasodilator in response to an inappropriate signal. In addition, an infusion pump designed for use by computer control should be used to reduce the possibility of incorrect transfer of commands from the computer. With these safeguards, we conclude that a closed-loop, computer-controlled system designed to infuse vasodilators for the control of postoperative hypertension provided satisfactory control of arterial pressure during induced hypotension when compared with a manually controlled infusion.
REFERENCES Holloway KB. Control of the eye during general anaesthesia for intraocular surgery. British Journal of Anaesthesia 1980; 52: 671-679. Foulds WS, Damato BE, Burton RL. Local resection versus enucleation in the management of choroidal melanoma. Eye 1987; 1: 676-679. Reid JAj Kenny GNC. Evaluation of closed-loop control of arterial pressure after cardiopulmonary bypass. British Journal of Anaesthesia 1987; 59: 247-255. Shcppard LC, Shorts JF, Robertson NF, Wallace FD, Kouchoukos NT. Computer controlled infusion of vasoactive drugs in post-cardiac surgical patients. In: Proceedings of The New York State Society of Anesthesiologists Inc. 35th Postgraduate Assembly. New York: New York State Society of Anesthesiologists Inc., 1981; 12-16. De Asia RA, Benis AM, Jurado RA, Litwak RS. Management of postcardiotomy hypertension by microcomputer controlled administration of sodium nitroprusside. Journal of Thoracic and Cardiovascular Surgery 1985; 89: 115-120.
Downloaded from http://bja.oxfordjournals.org/ at UQ Library on June 14, 2015
operator, and therefore taking longer to achieve any particular target at either induction or offset of the hypotension. However, when the hypotensive target was achieved, the quality and duration of the control were not significantly different from those of the manual control. The incidence of intraoperative complications was compared in the two groups (table III). There were no significant differences in the incidence of the oculocardiac reflex in the two groups. Adverse changes in the ECG or the ABM did not occur in either group. One patient in the manually-controlled group developed tachycardia requiring treatment with labetalol; an increased intraocular pressure requiring treatment with a 50 % aqueous solution of sucrose also occurred in this patient. Despite equally satisfactory control of arterial pressure in both groups during the critical period of profound hypotension, four patients in the computercontrolled group were noted to have some bleeding during the period of choroid dissection. This did not occur in the manually-controlled group.
609
610 6. Colvin JR, Kenny GNC. Automatic control of arterial pressure after cardiac surgery: Evaluation of a microcomputer based control system using glyceryl trinitrate and sodium nitroprusside. Anaesthesia 1989; 44: yi-\\. 7. Monk CR, Millard RK, Hutton P, Prys-Roberts C. Automatic arterial pressure regulation using isoflurane: comparison with manual control. British Journal of Anaesthesia' 1989; 63: 22-30. 8. Smith NT, Quinn ML, Flick J, Fukui Y, Fleming R, Coles JR. Automatic control in anesthesia: a comparison in performance between the anesthetist and the machine. Anesthesia and Analgesia 1984; 63: 715-722. 9. Westenskow DR, Meline L, Pace NL. Controlled hypo-
BRITISH JOURNAL OF ANAESTHESIA tension with SNP; anesthesiologist versus computer. Journal of Clinical Monitoring 1987; 3: 80-86. 10. Pereira E, Millard RK, Prys-Roberts C, Hutton P. Intraoperative pressure regulation using adaptive closed-loop control with sodium nitroprusside. British Journal of Anaesthesia 1986; 58: 815P-816P. 11. Todd JG, Colvin JR. Hypotensive anaesthesia in ophthalmic surgery. In: MacRae WK, Wildsmith JAW, eds. Monographs in Anaesthesiology—Induced Hypotension. Amsterdam: Elsevier, 1992; in press. 12. MacRae WR, Wildsmith JAW, Dale BAB. Induced hypotension with a mixture of sodium nitroprusside and trimetaphan camsylate. Anaesthesia 1981; 36: 312-315.
Downloaded from http://bja.oxfordjournals.org/ at UQ Library on June 14, 2015
EVALUATION OF CLOSED LOOP CONTROL OF ARTERIAL PRESSURE DURING HYPOTENSIVE ANAESTHESIA FOR LOCAL RESECTION OF INTRAOCULAR MELANOMA S. CHAUDHRI, J. R. COLVIN, J. G. TODD AND G. N. C. KENNY PATIENTS AND METHODS
SUMMARY
KEY WORDS Anaesthetic techniques • hypotensive
anaesthesia, manual
control, closed-loop computer control.
Patients undergoing local resection of intraocular melanoma require intraoperative arterial pressure control to a systolic pressure of 40 ± 5 mm Hg during the period of choroid dissection. Hypotension to this degree is required to prevent intraocular haemorrhage from the vascular choroid, which may affect visual outcome [1,2]. Precise control of arterial pressure during this period is essential, combined with meticulous monitoring of die adequacy of cerebral and myocardial perfusion. Control of arterial pressure by closed-loop infusion of vasodilators has been shown to compare favourably with the control achieved by intensive therapy staff in postoperative cardiac surgical patients [3-6]. Automatic control of arterial pressure has also been used during operation and was shown to compare favourably with manual control by the anaesthetist[7-10}. We have compared the quality of induced hypotension by computer-controlled infusion with that achieved manually by an experienced anaesthetist during local resection of choroidal melanoma.
We have studied 20 patients aged between 39 and 66 yr undergoing local resection of intraocular melanoma and considered suitable for hypotensive anaesthesia. Hospital Ethics Committee approval was obtained and informed consent given by all patients. All patients received the standard anaesthetic for this procedure in this unit [7], including oral atenolol 50-100 mg 90 min before operation, followed by i.m. pethidine 50-100 mg and promethazine 25-50 mg 1 h before surgery. Anaesthesia was induced with thiopentone 5 mg kg"1 and suxamethonium -1 1.5mgkg was given to facilitate tracheal intubation. Anaesthesia was maintained with nitrous oxide and 1 % enflurane in oxygen. Controlled ventilation to an end-tidal carbon dioxide concentration of 3.8-4.2 kPa was facilitated by tubocurarine and a 10° head-up tilt was used. The dose of tubocurarine was titrated using the Datex Relaxograph to maintain the first twitch response to a train-of-four stimulation (Tl) at less than 15% of control. Antagonism of neuromuscular block was achieved with neostigmine and glycopyrronium. Systolic arterial pressure was controlled by an infusion of sodium nitroprusside (SNP) 200 ug ml"1 and trimetaphan camsylate (TMP) 1 mg ml"1 in 5 % glucose [11-12]. Patients were allocated randomly to two groups: in the manual group, the infusion rate was adjusted manually; in die computer group, the infusion rate was adjusted by a computer-controlled, closed-loop system which comprised an Imed 929 Volumetric pump linked by an Apple He microcomputer to the data output of a Datascope 2002 monitor. This system has been described and evaluated previously [3, 5] and consists of a Proportional Integral Derivative controller widi adaptive features. The initial infusion rate was calculated by the closed-loop system. After each new cal-
SOFIA CHAUDHRI*, M.B., CH.B., F.R.C.ANAES. J JOHN R. COLVTNt, M.B., CH.B., F.F.A.R.C.S.I., F.R.C.ANAES.; GAVIN N . C. KENNY, B.SC.,
M.D., F.R.C.ANAES. ; University Department of Anaesthesia, Glasgow Royal Infirmary, Glasgow G31 2ER. J. GORDON TODD, M.B., CH.B., F.R.C.ANAES., Division of Anaesthesia, Western Infirmary^Glasgow-G H-6NTrAccepted forPublication: October 10, 1991. Present addresses: •Western Infirmary, Glasgow Gil 6NT. fNinewells Hospital, Dundee DD1 95Y. Correspondence to S.C.*
Downloaded from http://bja.oxfordjournals.org/ at UQ Library on June 14, 2015
We have studied 20 patients undergoing local resection of intraocular melanoma during hypotensive anaesthesia, allocated randomly to receive either manual control by an experienced anaesthetist or closed-loop computer control of an infusion of a 5:1 mixture of trimetaphan camsylate (IMP) and sodium nitroprusside (SNP). There were no significant differences in the smallest systolic and diastolic arterial pressures obtained, heart rate or infusion requirements between the two groups, but the duration of both the infusion and the operation were significantly longer in the computer-controlled group (P < 0.05). The quality of control of arterial pressure was assessed by the percentage of time spent at pressures greater and less than the prescribed target values, and was satisfactory in both groups during the critical period of profound hypotension. We conclude that the computer-controlled infusion performed satisfactorily during profound hypotension compared with an experienced anaesthetist. (Br. J. Anaesth. 1992; 69: 607-610)
BRITISH JOURNAL OF ANAESTHESIA
608
RESULTS
There were no statistical differences between the two groups in age, sex or weight (table I) or in the duration of the period of hypotension, the systolic pressure at induction, the smallest systolic and diastolic pressures obtained, or the smallest mean arterial pressures and heart rates during the period of hypotension in the two groups (table I). The quality of arterial pressure control was assessed by measuring the percentage of time spent at > ±10% outside the prescribed target pressure. During the entire period of infusion of vasodilator (table II), there were no significant differences in the percentage of time spent at pressures greater than the prescribed target pressure. However, patients in the computer-controlled group spent more time at pressures 10% less than the target pressure (P < 0.05). During the period of profound hypotension, there were no significant differences between the two
groups in the percentage of time spent at pressures greater and less than the target (table II) and the quality of arterial pressure control was satisfactory in both groups. This may be explained by the smaller reductions in the prescribed target pressure as the critical hypotension was reached, and the corresponding small changes in infusion rate required to achieve the prescribed target. There were no significant differences in the total amount of SNP or TMP infused in the two groups. Both the duration of the operation and the duration of infusion of vasodilators were longer in the computer-controlled group (P < 0.05) (table III). The longer duration of die infusion could be explained by the computer increasing the infusion in smaller increments dian the experienced manual TABLE I. Patient characteristics {mean {range or SDJ) and operative details (median (range))
Age (yr) Weight (kg) Sex (M/F) Hypotension (min) Stan SAP (mm Hg) Smallest SAP (mm Hg) Smallest DAP (mm Hg) Smallest MAP (mm Hg) Heart rate (beat min"1)
Manual group (n = 10)
Computer group (n = 10)
51.4(31-68) 68.0(13.6) 4/6 90 (50-105) 135(100-200)
52.9 (39-68) 70.6(12.5) 6/4 97.5 (60-130) 127(100-170) 40(38-^12) 21.5(20-32) 27.5 (20-34)
40(36-^5)
21(18-25) 27.5 (25-31) 67 (54-90)
60 (50-75)
TABLE II. Quality of arterial pressure control throughout the infusion and during profound hypotension (median (range)), expressed as time spent at pressures greater than ±10% outside target pressures. *P < 0.05 between groups
Manual group (n = 10) During infusion Target+10% Target - 1 0 % During hypotension Target+10% Target-10%
7.69(2.1-22.4) 10.3 (0-22.4) 2.3 (0-6.0) 6.6(0-31.8)
Computer group (n = 10) 8.7(5.4-13.9) 22.3(5.3-27.6)* 4.6(0-14.5) 2.6 (0-27.7)
TABLE III. Infusion requirements and intraoperative complications (median (range) or number). *P < 0.05 between groups
Manual group (n = 10) Infusion requirements Total SNP (ml) 8.8(2.2-13.8) Total TMP (ml) 44(11-69) Duration of 158(120-202) infusion (min) Duration of 198(160-250) operation (min) Intraoperative complications Oculocardiac reflex: 2 Early 0 Late 1 Both ST depression > 1 mrrl 0 1 Tachycardia 1 Increased IOP 0 Bleeding during choroid dissection
Computer group (n = 10) 13.5(6-41.8) 67.8 (30-209) ' 204 (100-268)* 248(150-285)*
3 0 1 0 0 0 4
Downloaded from http://bja.oxfordjournals.org/ at UQ Library on June 14, 2015
culation of the infusion rate, the system tracked the rate of change in pressure so that excessive or inadequate decrease caused a further calculation to be made. The raw arterial waveform was scanned by a background subroutine which allowed detection of zeroing, flushing, damping or disturbance of the line. A warning was provided, with an appropriate message for any computer-pump interface errors. Standard monitoring for this procedure comprised a continuous ECG using the CM5 lead, direct arterial pressure from the radial artery with the transducer at the level of the external auditory meatus, neuromuscular function and end-tidal carbon dioxide concentration. The mean amplitude and zero cross frequency of the EEG were displayed on the Datex Anaesthesia and Brain Activity Monitor (BAM). During the first 60-90 min, the systolic arterial pressure was reduced gradually to 70-80 mm Hg to allow fashioning of a superficial scleral flap and a vitrectomy to be performed. Profound hypotension to 40 + 5 mm Hg was then induced before final dissection of a deep scleral flap and incision of the choroid. During this period, ST segment changes and the zero cross frequency of the BAM were used as indicators of global myocardial and cerebral perfusion, respectively. If ST segment depression of greater than 1 mm or a sustained zero cross frequency of less than 5 Hz occurred, the arterial pressure was increased, although this could result in loss of the eye. Nitrous oxide was discontinued 20 min before closure of the scleral flap. After closure of the scleral flap and injection of sulphur hexafluoride gas, the arterial pressure was allowed to increase slowly by less than 5 mm Hg min"1 to avoid variations in cerebral blood flow and cerebral oedema. At the end of the procedure anaesthesia was discontinued, neuromuscular block antagonized and the trachea extubated. Patients were kept in the recovery area until obeying commands; i.v. fluids and oxygen were administered until the following day. Results were analysed using chi-square and the Mann-Whitney U tests.
CLOSED LOOP CONTROL OF ARTERIAL PRESSURE
DISCUSSION
Closed-loop, computer-controlled infusion of vasodilators has been used successfully in the control of hypertension after cardiac surgery [3—6], in which the quality of control achieved by the computercontrolled system has been shown to compare favourably with that achieved manually by nursing staff [3]. Closed-loop computer control of arterial pressure using either volatile agents or vasodilators such as SNP has also been used successfully during operation [7-10], and the control achieved compared favourably with that achieved by the anaesthetist. We are not aware of any other study in which a combination of SNP and TMP has been infused by a closed-loop, computer-controlled system. The system we used had been used previously to achieve normotension after operation. We have shown that this system can also be used during operation for the production of induced hypotension. Because of variations in operating stages, it was not possible to standardize the time intervals at which changes in target arterial pressure were made or the magnitude of changes in target pressure, and changes were made in response to prevailing surgical conditions. The period of steady-state control of arterial pressure is critical, and the quality of control of arterial pressure during the period of profound hypotension required for dissection of the tumour from the choroid was equally satisfactory in both groups. However, the anaesthetist controlling the infusion in the manually controlled group of patients was experienced in the provision of anaesthesia for this type of surgery and results achieved by a less experienced operator may have compared less favourably with the computer-controlled group. "Despite^atiffactory control oFafferial pressure~in both groups during the period of profound hypotension, four patients in the computer-controlled group were reported to have some minor bleeding during dissection of the tumour from the vascular choroid. This was a subjective assessment made by
the surgeon, who was unaware to which group the patient had been allocated. The cause was unclear, but may be related to other factors such as size and position of the tumour or difficulty in dissection, as it was also noted that the duration of operation and infusion of vasodilators was significantly greater in the computer-controlled group. Retrospective analysis of the patient data by the surgeon suggests that, in those patients in whom bleeding was noted, the tumour may have been more difficult to operate upon. Loss of the eye as a consequence of poor operating conditions did not occur in any of the patients. This was a small study, however, and it was not possible to study long-term outcome. Further studies would be required to evaluate this complication in greater detail. There were no adverse changes on the ECG or the BAM during the operation in either group. All patients were followed up at 6 weeks, 6 months and 1 year. There were no sequelae attributable to hypotension or anaesthesia. Detailed psychometric testing was not carried out, but one patient in the manually controlled group reported mild memory impairment in the early postoperative period. This showed rapid improvement and by 6 weeks the patient was unaware of any impairment of memory function. All patients returned to their previous functional level within the limits of their visual outcome. We believe that automated control of arterial pressure should only be used under supervision and that careful validation of the input signal must be made. We have found that analysis of the raw arterial waveform allows detection of potential errors such as flushing or disturbance of the arterial line and prevents the system infusing vasodilator in response to an inappropriate signal. In addition, an infusion pump designed for use by computer control should be used to reduce the possibility of incorrect transfer of commands from the computer. With these safeguards, we conclude that a closed-loop, computer-controlled system designed to infuse vasodilators for the control of postoperative hypertension provided satisfactory control of arterial pressure during induced hypotension when compared with a manually controlled infusion.
REFERENCES Holloway KB. Control of the eye during general anaesthesia for intraocular surgery. British Journal of Anaesthesia 1980; 52: 671-679. Foulds WS, Damato BE, Burton RL. Local resection versus enucleation in the management of choroidal melanoma. Eye 1987; 1: 676-679. Reid JAj Kenny GNC. Evaluation of closed-loop control of arterial pressure after cardiopulmonary bypass. British Journal of Anaesthesia 1987; 59: 247-255. Shcppard LC, Shorts JF, Robertson NF, Wallace FD, Kouchoukos NT. Computer controlled infusion of vasoactive drugs in post-cardiac surgical patients. In: Proceedings of The New York State Society of Anesthesiologists Inc. 35th Postgraduate Assembly. New York: New York State Society of Anesthesiologists Inc., 1981; 12-16. De Asia RA, Benis AM, Jurado RA, Litwak RS. Management of postcardiotomy hypertension by microcomputer controlled administration of sodium nitroprusside. Journal of Thoracic and Cardiovascular Surgery 1985; 89: 115-120.
Downloaded from http://bja.oxfordjournals.org/ at UQ Library on June 14, 2015
operator, and therefore taking longer to achieve any particular target at either induction or offset of the hypotension. However, when the hypotensive target was achieved, the quality and duration of the control were not significantly different from those of the manual control. The incidence of intraoperative complications was compared in the two groups (table III). There were no significant differences in the incidence of the oculocardiac reflex in the two groups. Adverse changes in the ECG or the ABM did not occur in either group. One patient in the manually-controlled group developed tachycardia requiring treatment with labetalol; an increased intraocular pressure requiring treatment with a 50 % aqueous solution of sucrose also occurred in this patient. Despite equally satisfactory control of arterial pressure in both groups during the critical period of profound hypotension, four patients in the computercontrolled group were noted to have some bleeding during the period of choroid dissection. This did not occur in the manually-controlled group.
609
610 6. Colvin JR, Kenny GNC. Automatic control of arterial pressure after cardiac surgery: Evaluation of a microcomputer based control system using glyceryl trinitrate and sodium nitroprusside. Anaesthesia 1989; 44: yi-\\. 7. Monk CR, Millard RK, Hutton P, Prys-Roberts C. Automatic arterial pressure regulation using isoflurane: comparison with manual control. British Journal of Anaesthesia' 1989; 63: 22-30. 8. Smith NT, Quinn ML, Flick J, Fukui Y, Fleming R, Coles JR. Automatic control in anesthesia: a comparison in performance between the anesthetist and the machine. Anesthesia and Analgesia 1984; 63: 715-722. 9. Westenskow DR, Meline L, Pace NL. Controlled hypo-
BRITISH JOURNAL OF ANAESTHESIA tension with SNP; anesthesiologist versus computer. Journal of Clinical Monitoring 1987; 3: 80-86. 10. Pereira E, Millard RK, Prys-Roberts C, Hutton P. Intraoperative pressure regulation using adaptive closed-loop control with sodium nitroprusside. British Journal of Anaesthesia 1986; 58: 815P-816P. 11. Todd JG, Colvin JR. Hypotensive anaesthesia in ophthalmic surgery. In: MacRae WK, Wildsmith JAW, eds. Monographs in Anaesthesiology—Induced Hypotension. Amsterdam: Elsevier, 1992; in press. 12. MacRae WR, Wildsmith JAW, Dale BAB. Induced hypotension with a mixture of sodium nitroprusside and trimetaphan camsylate. Anaesthesia 1981; 36: 312-315.
Downloaded from http://bja.oxfordjournals.org/ at UQ Library on June 14, 2015
