We have previously described the maintenance of blood pressure (BP) during spinal anesthesia for cesarean delivery using closed-loop feedback computer-controlled infusion of phenylephrine.1,2 This method of drug administration enables BP to be controlled automatically with precision that is equal to or better than manually controlled infusion.1 Our system was designed to operating using a standard noninvasive BP monitor. However, this results in an imperfect match because although the rate of a continuous vasopressor infusion can be adjusted on a second-by-second basis, noninvasive BP measurement is only performed intermittently, usually no more frequently than every 1 minute.
We postulated that the delivery of calculated doses of vasopressor by rapid boluses given immediately after the completion of each BP measurement would result in a faster response and therefore could improve control compared with adjustment of a continuous infusion rate. Accordingly, we hypothesized that computer-controlled administration of phenylephrine by intermittent boluses would result in greater precision of BP control compared with computer-controlled continuous infusion. This hypothesis was tested in the present study in which patients having spinal anesthesia for elective cesarean delivery were randomly assigned to have their BP controlled by 1 of 2 computer-controlled systems with performance of the 2 systems compared using performance error calculations. Precision of BP control was compared using performance error calculations, with the primary outcome-defining precision chosen as the median absolute performance error (MDAPE, defined in Statistical Analysis).
This was a randomized, 2-arm, parallel, single-blinded controlled trial. Approval for the study was obtained from the Joint Chinese University of Hong Kong—New Territories East Cluster Clinical Research Ethics Committee, Shatin, Hong Kong, China—and the study was registered in the Chinese Clinical Trial Registry (registration No. ChiCTR-TRC-12002418). All patients included gave written, informed consent. A total of 214 patients scheduled for elective cesarean delivery under spinal anesthesia at the Prince of Wales Hospital, Shatin, Hong Kong, China, were enrolled. Inclusion criteria were singleton pregnancy, age ≥18 years, weight ≥50 kg, height 140 to 180 cm, gestation ≥36 weeks, and ability to give informed consent. Exclusion criteria were American Society of Anesthesiologists physical status ≥3, pre-existing or pregnancy-induced hypertension, cardiovascular or cerebrovascular disease, known fetal abnormality, and any signs of onset of labor.
Standard antacid premedication of oral famotidine and sodium citrate was given. On arrival in the operating room, patients were allowed to rest in the left-tilted supine position, and standard monitoring was attached (Infinity C500, Dräger Medical AG & Co KG, Germany). BP was measured noninvasively at 1-minute intervals, and after a brief stabilizing period, baseline values were recorded as the mean of 3 consecutive measurements with a difference of no more than 10%. A wide-bore intravenous (IV) cannula was then inserted into a forearm vein under local anesthesia, but no prehydration was given. Spinal anesthesia was induced with the patient in the right lateral position. After skin disinfection and skin infiltration with lidocaine 1% w/v, a 25-gauge Whitacre spinal needle was inserted via an introducer at the estimated L3–L4 or L4–L5 vertebral interspace. After confirmation of free-flow of cerebrospinal fluid, 2.2 mL of hyperbaric bupivacaine 0.5% w/v and fentanyl 15 µg were injected intrathecally. The patient was then returned to the left-tilted supine position. Supplemental oxygen was not given unless the pulse oximeter reading was <95%.
Immediately after intrathecal injection, IV cohydration of up to 2 L of warmed Hartmann’s solution was started by fully opening the clamp of the infusion set. The total amount of IV fluid given up to the time of uterine incision was estimated by visual inspection of the bag. The noninvasive BP monitor was set to measure at 1-minute intervals starting 1 minute after the completion of intrathecal injection.
BP was maintained using phenylephrine administered by closed-loop feedback computer control that was started immediately after induction of spinal anesthesia. A standardized preparation of phenylephrine 100 µg/mL1 was prepared in a 50-mL syringe that was connected via narrow-bore tubing to a 3-way stopcock attached to the IV cannula through which IV fluid was continuously administered. The phenylephrine was delivered by a syringe pump (Graseby 3500 Anaesthesia Pump, Graseby Medical Ltd, Watford, Herts, UK) that was controlled by a laptop computer running 1 of 2 algorithms. The computer programs were developed by one of the authors (Y.H.T.) using Microsoft Visual Studio 6.0 (Microsoft Corporation, Redmond, WA) using Visual C++. The general basis of the system has been described previously.1,2
Patients were randomly assigned by the principal investigator (W.N.) at a 1:1 ratio to 1 of 2 groups according to a computer-generated random number code that had been prepared without blocking by one of the secretarial staff using Microsoft Office Excel 2010 (Microsoft Corporation, Redmond, WA). The codes were kept in sequentially numbered sealed envelopes that were opened when the patient arrived in the operating room. In both groups, the calculated dose of phenylephrine per minute (I) was determined according to the following algorithm:
where error% = (measured systolic BP – baseline systolic BP) / baseline systolic BP × 100, and the value of I was constrained to be within the limits 0 to 1 mL. This regimen has previously been shown to be effective for maintaining systolic BP near baseline.1
In the infusion group, the syringe pump was controlled to deliver phenylephrine at the rate of I mL/min. The infusion rate was adjusted after the completion of each automated BP measurement. In the bolus group, the syringe pump was controlled to deliver a rapid bolus of I mL of phenylephrine given at a rate of 1200 mL/h started after the completion of each automated BP measurement. The principal investigator was continuously present during operation and had discretion to inactivate or override the system if deemed necessary. The senior investigator was aware of the group to which patients were assigned but patients were blinded.
BP and heart rate (HR) were recorded after each automated measurement. In addition, cardiac output (CO) was measured noninvasively using a suprasternal Doppler technique (USCOM 1A cardiac output monitor, USCOM Ltd, Sydney, NSW, Australia) as we have previously described.3,4 These measurements were made by the same investigator (S.L.) who was blinded to the group of the patient at baseline and at 5-minute intervals after induction of anesthesia until delivery. The incidences of hypotension (defined as systolic BP < 80% of baseline), hypertension (defined as systolic BP >120% of baseline), bradycardia (defined as HR <50 beats/min), and nausea or vomiting were recorded.
Computer-controlled administration of phenylephrine was continued until the time of uterine incision, after which the study was terminated, and further hemodynamic management was at the discretion of the attending anesthesiologist. Apgar scores and umbilical cord blood gases were measured according to usual practice.
The performance of the 2 systems for controlling BP was assessed using performance error calculations as we have previously described.1,2 The following parameters were calculated: (1) percentage performance error (PE; defined as the difference between each measured value of systolic BP and the baseline value, expressed as a percentage of the baseline value); (2) median performance error (MDPE; defined as the median of all values of PE for each patient); (3) MDAPE (defined as the median of the absolute values of PE [│PE│] for each patient); (4) wobble (a measure of the variability of PE around MDPE for each patient, calculated as the median value of the differences between each value of PE and MDPE for each patient); and (5) divergence (a measure of the trend of change in │PE│ with time for each patient). Derivation of these parameters has been described previously.5 Precision of BP control was compared between groups by comparing values of the primary outcome MDAPE, with smaller values indicating greater precision. MDPE, wobble, and divergence were compared as secondary assessments of BP control, with smaller values considered indicative of better control. The performance error calculations were made using Microsoft Office Excel 2010.
The primary outcome was defined as MDAPE. An a priori power analysis was performed based on data from our previous study in which closed-loop feedback computer-controlled phenylephrine infusion resulted in mean (SD) values for MDAPE of 4.82% (2.01).1 In order to determine a difference of 20% in MDAPE between groups with 80% power at an α level of .05, it was calculated that a sample size of 103 patients per group was required. To allow for an estimated dropout rate of 5%, the sample size was increased to 107 patients per group.
Data for CO values were normalized to percentage of baseline values. For each patient, the area under the curve (AUC) for these values plotted against time were calculated using the trapezium rule.6 Because the number of data points recorded was variable among patients because of varying surgical times, standardized values were derived by dividing the values for AUC by the number of data points recorded for each patient.6 Standardized values were then compared between groups as a measure of overall change in CO.
Continuous data were checked for normal distribution using the Kolmogorov–Smirnov test, and intergroup comparisons were performed using the Student t test or the Mann-Whitney U test as appropriate. Nominal data were compared using the χ2 test. Statistical comparisons were made using IBM SPSS Statistics version 20 (IBM SPSS Inc, Chicago, IL). Values of P < .05 were considered statistically significant, except for comparison of performance error calculations for which, in order to account for multiple comparisons, a Bonferroni correction was applied to maintain α at .05 such that the significance criterion was P < .0125 (.05/4).
Patient recruitment and flow are shown in Figure 1. A total of 214 patients were enrolled in the study between August 2012 and September 2014. Data could not be collected for 10 patients who were excluded from the analysis for the following reasons: fault with syringe pump or tubing connections (4 patients), severe shivering preventing accurate BP measurement (3 patients), inadequate spinal anesthesia (2 patients), and fault in BP monitor (1 patient). After exclusions, 102 patients in the bolus group and 102 patients in the infusion group completed the study and had data analyzed for the primary outcome. In no case was it necessary for the senior investigator to inactivate or override the computer-controlled system. Because the investigator responsible for CO measurements had a change in employment status during the study, she was only able to be present and make measurements for 43 patients in the bolus group and 39 patients in the infusion group. Because CO was a secondary outcome of the study, it was considered acceptable to continue the study with this limitation.
Patient characteristics are shown in Table 1. Anesthetic details and surgical times are shown in Table 2. The total dose of phenylephrine and the rate of phenylephrine administration up to the time of uterine incision were greater in the infusion group versus the bolus group (both P < .001).
Changes in systolic BP over time for all patients are shown in Figure 2. Results for performance error calculations are shown in Table 3. The primary outcome MDAPE was smaller in the bolus group (median 4.38% [quartiles 3.22, 6.25]) versus the infusion group (5.39% [4.12, 7.04]; P < .001; Figure 3). MDPE was smaller in the bolus group (−0.21% [−2.82, 1.95]) versus the infusion group (3.72% [0.43, 5.84]; P = .008). Wobble and divergence were not different between groups.
Changes in CO over time, normalized to percentage of baseline, are shown in Figure 4. The standardized AUC was not different in the bolus group (median 587%.min [quartiles 549, 622]) versus the infusion group (552%.min [525, 596]; P = .16).
The incidences of hypotension, hypertension, bradycardia, and nausea or vomiting are shown in Table 4. More patients in the infusion group had one or more episodes of hypertension (P = .007).
Neonatal outcome is summarized in Table 5. Insufficient umbilical arterial blood was obtained for blood gas analysis in 9 patients in the bolus group and 4 patients in the infusion group. Insufficient umbilical venous blood was obtained for blood gas analysis in 3 patients in the bolus group and 5 patients in the infusion group. The UA PO2 was less than the lower limit of detection of the blood gas analyzer (10 mm Hg) in 8 patients in the bolus group and 5 patients in the infusion group; for these analyses, the data values were entered as constant values equal to the lower limit of detection divided by √2,7 and the values were then analyzed by ranks. There was no difference between groups.
The results of this study showed that when using closed-loop feedback computer control to administer phenylephrine to maintain BP during spinal anesthesia for cesarean delivery, use of intermittent boluses resulted in more precise control of BP compared with continuous infusion. This was evidenced by smaller values for MDAPE, a standard parameter used for assessing inaccuracy of closed-loop systems,5 in the bolus group compared with the infusion group. However, the difference between groups was modest, and there was no difference in clinical outcomes between groups.
The reason why the computer-controlled system performed better using intermittent boluses compared with continuous infusion probably relates to the method of BP measurement. We used a standard noninvasive monitor set to cycle at 1-minute intervals that was the highest frequency we considered practical. Administering the calculated vasopressor dose by a rapid bolus after the completion of each BP measurement may have resulted in a more rapid response compared with adjustment of the rate of infusion. Closed-loop administration of the vasopressor by intermittent boluses may thus be a better match to intermittent measurement of BP, which is the usual method of BP management in normal clinical practice. Although Sia et al8 recently described the use of a continuous noninvasive BP monitor in a computer-controlled double-vasopressor system,9, 10 continuous noninvasive monitors may not yet be available in every unit. Further investigation to determine the relative performance of vasopressor administration by boluses versus infusion in computer-controlled systems when continuous BP monitoring is used would be of interest.
Our results showed that MDPE was greater in the infusion group compared with the bolus group. MDPE is a measure of bias,5 and our results indicate that BP on average was maintained at a higher level in the infusion group (3.72% above baseline) compared with the bolus group (0.21% below baseline). Of note, this was associated with a greater rate of phenylephrine consumption in the infusion group compared with the bolus group. Although the 2 computer-controlled algorithms were designed to deliver the same amount of phenylephrine per minute, the algorithms did not account for the time required to complete each BP measurement. Each cycled measurement required a finite and variable period to complete. In addition, there were occasional further delays to measurement because of motion artifacts or other interferences that required the monitor to recycle. As a result, the average actual time between BP measurements was usually greater than 1 minute. This biased toward a higher rate of phenylephrine administration in the infusion group because, in this group, the rate of infusion was continuous and independent of the BP measurement time, whereas in the bolus group, the equivalent of a 1-minute dose was only delivered after each BP measurement regardless of the actual time taken.
Intermittent failures of measurement often occur in awake patients, and this is especially important for obstetric patients because of the high incidence of intraoperative shivering during neuraxial anesthesia.11 Because, in a bolus system each dose of vasopressor is only administered after a BP measurement is successfully completed, whereas in a continuous infusion system vasopressor administration continues regardless of any delays in BP measurement, unnecessary or overly large doses are less likely to be administered using a bolus system; this suggests a potential safety advantage of systems using boluses. Although it would be possible to modify the infusion system by incorporating a time limit for the infusion period, the simplicity of the bolus system remains attractive.
Previously it has been shown that administration of phenylephrine in high doses during spinal anesthesia was associated with dose-dependent decreases in CO, which is thought to be primarily related to baroreceptor-mediated decreases in HR.12 In our study, phenylephrine consumption was greater in the infusion group compared with the bolus group, and this was associated with the maintenance of BP at a higher average level. Furthermore, the incidence of hypertension was greater in the infusion group compared with the bolus group. However, despite this, no difference in CO changes was found between groups. This suggests that the mean rate of phenylephrine administration in the infusion group (36.9 µg/min) was not excessive. However, because logistic reasons prevented CO measurement in all patients, a type II statistical error for this outcome cannot be excluded.
An important limitation of our study is that the performance of the intermittent bolus algorithm was dependent on the measurement of BP at 1-minute intervals. This was appropriate for the limited period of time between the induction of anesthesia and uterine incision to which the system was applied in this study. However, the measurement of BP at 1-minute intervals may not be practical for longer-duration settings. In these situations, it is interesting to consider whether a hybrid bolus plus infusion system could be used. Furthermore, it would be of interest to determine whether a similar closed-loop vasopressor system would be useful in other clinical settings (eg, in the intensive care unit).
Finally, it should be noted that our comparison of automated boluses versus infusion was specific to the algorithms used in this study. It may not be valid to generalize our findings to the use of other closed-loop algorithms.
The authors thank the midwives of the labor ward at Prince of Wales Hospital, Shatin, Hong Kong, China, for their assistance and cooperation.
Name: Warwick D. Ngan Kee, MD, FANZCA, FHKCA.
Contribution: This author helped design and conduct the study, collect and analyze the data, and prepare the manuscript.
Name: Yuk-Ho Tam, BSc, MPhil.
Contribution: This author helped design the study and prepare the manuscript.
Name: Kim S. Khaw, MD, FRCA, FHKCA.
Contribution: This author helped design the study and prepare the manuscript.
Name: Floria F. Ng, RN, BASc.
Contribution: This author helped design the study, collect the data, and prepare the manuscript.
Name: Shara W. Y. Lee, PhD.
Contribution: This author helped design the study, collect the data, and prepare the manuscript.
This manuscript was handled by: Maxime Cannesson, MD, PhD.
1. Ngan Kee WD, Khaw KS, Ng FF, Tam YH. Randomized comparison of closed-loop feedback computer-controlled with manual-controlled infusion of phenylephrine for maintaining arterial pressure during spinal anaesthesia for caesarean delivery. Br J Anaesth. 2013;110:59–65.
2. Ngan Kee WD, Tam YH, Khaw KS, Ng FF, Critchley LA, Karmakar MK. Closed-loop feedback computer-controlled infusion of phenylephrine for maintaining blood pressure during spinal anaesthesia for caesarean section: a preliminary descriptive study. Anaesthesia. 2007;62:1251–1256.
3. Ngan Kee WD, Lee SW, Khaw KS, Ng FF. Haemodynamic effects of glycopyrrolate pre-treatment before phenylephrine infusion during spinal anaesthesia for caesarean delivery. Int J Obstet Anesth. 2013;22:179–187.
4. Ngan Kee WD, Lee SW, Ng FF, Tan PE, Khaw KS. Randomized double-blinded comparison of norepinephrine and phenylephrine for maintenance of blood pressure during spinal anesthesia for cesarean delivery. Anesthesiology. 2015;122:736–745.
5. Varvel JR, Donoho DL, Shafer SL. Measuring the predictive performance of computer-controlled infusion pumps. J Pharmacokinet Biopharm. 1992;20:63–94.
6. Peacock JL, Peacock PJ. Oxford Textbook of Medical Statistics. 2011. Oxford: Oxford University Press
7. Hornung RW, Reed LD. Estimation of average concentration in the presence of nondetectable values. Appl Occup Environ Hyg. 1990;5:46–51.
8. Sia AT, Tan HS, Sng BL. Closed-loop double-vasopressor automated system to treat hypotension during spinal anaesthesia for caesarean section: a preliminary study. Anaesthesia. 2012;67:1348–1355.
9. Sng BL, Tan HS, Sia AT. Closed-loop double-vasopressor automated system vs manual bolus vasopressor to treat hypotension during spinal anaesthesia for caesarean section: a randomised controlled trial. Anaesthesia. 2014;69:37–45.
10. Sng BL, Wang H, Assam PN, Sia AT. Assessment of an updated double-vasopressor automated system using Nexfin for the maintenance of haemodynamic stability to improve peri-operative outcome during spinal anaesthesia for caesarean section. Anaesthesia. 2015;70:691–698.
11. Crowley LJ, Buggy DJ. Shivering and neuraxial anesthesia. Reg Anesth Pain Med. 2008;33:241–252.
© 2017 International Anesthesia Research Society
12. Stewart A, Fernando R, McDonald S, Hignett R, Jones T, Columb M. The dose-dependent effects of phenylephrine for elective cesarean delivery under spinal anesthesia. Anesth Analg. 2010;111:1230–1237.