You can respond to this article at http://www.anaesthesiacorrespondence.com
Closed-loop double-vasopressor automated system vs manual bolus vasopressor to treat hypotension during spinal anaesthesia for caesarean section: a randomised controlled trial
Article first published online: 20 NOV 2013
© 2013 The Association of Anaesthetists of Great Britain and Ireland
Volume 69, Issue 1, pages 37–45, January 2014
How to Cite
Sng, B. L., Tan, H. S. and Sia, A. T. H. (2014), Closed-loop double-vasopressor automated system vs manual bolus vasopressor to treat hypotension during spinal anaesthesia for caesarean section: a randomised controlled trial. Anaesthesia, 69: 37–45. doi: 10.1111/anae.12460
- Issue published online: 9 DEC 2013
- Article first published online: 20 NOV 2013
- Manuscript Accepted: 6 SEP 2013
- National Medical Research Council. Grant Number: EDG/1003/2010
Hypotension necessitating vasopressor administration occurs commonly during caesarean section under spinal anaesthesia. We developed a novel vasopressor delivery system that automatically administers phenylephrine or ephedrine based on continuous non-invasive arterial pressure monitoring. A phenylephrine bolus of 50 μg was given at 30-s intervals when systolic blood pressure fell < 90% of baseline; an ephedrine bolus of 4 mg was given instead if systolic pressure fell < 90% of baseline together with a heart rate < 60 beats.min−1. The control group used manual boluses of either phenylephrine 100 μg or ephedrine 8 mg, administered at 1-min intervals based on the same thresholds for systolic pressure and heart rate. This randomised, controlled, double-blinded trial involved 213 healthy women who underwent elective caesarean delivery under spinal anaesthesia using 11 mg hyperbaric bupivacaine with 15 μg fentanyl and 100 μg morphine. The automated vasopressor group had better systolic pressure control, with 37/106 (34.9%) having any beat-to-beat systolic pressure reading < 80% of baseline compared with 63/107 (58.9%) in the control group (p < 0.001). There was no difference in the incidence of reactive hypertension, defined as systolic pressure > 120% of baseline, with 8/106 (7.5%) in the automated vasopressor group vs 14/107 (13.1%) in the control group, or total dose of vasopressors. The automated vasopressor group had lower median absolute performance error of 8.5% vs control of 9.8% (p = 0.013), and reduced incidence of nausea (1/106 (0.9%) vs 11/107 (10.3%), p = 0.005). Neonatal umbilical cord pH, umbilical lactate and Apgar scores were similar. Hence, our system afforded better control of maternal blood pressure and reduced nausea with no increase in reactive hypertension when compared with manual boluses.
Peri-operative hypotension during spinal anaesthesia for caesarean section is common owing to the loss of sympathetic tone and the physiological predisposition of pregnant women to hypotension [1-4]. The consequences of maternal hypotension are potentially severe, affecting both the mother and child. Maternal effects include nausea, vomiting and headache attributable to inadequate organ perfusion; fetal adverse effects may include fetal acidosis and poor neurological outcomes because of a lack of placental vascular autoregulation, which renders its perfusion pressure-dependent . While several strategies, both pharmacological and non-pharmacological, have been used to treat hypotension, the optimal treatment is still debatable. Non-pharmacological methods alone are often not effective ; hence, the use of vasopressors is often required. Recent data support the use of alpha-1 adrenergic agonists such as phenylephrine [7, 8], although ephedrine (a mixed alpha- and beta-adrenergic agonist) may be useful in the event of concomitant hypotension with bradycardia.
The optimal timing and dose of vasopressors is dependent on accurate and timely blood pressure (BP) monitoring. The current standard of care involves non-invasive oscillometric BP monitoring, which is commonly limited by its inability to provide accurate cyclical readings at less than 1-min intervals. This may be inadequate to guide vasopressor administration, whether by intermittent bolus or continuous infusion, owing to the rapid changes in BP that commonly occur during spinal anaesthesia. In addition, manual control of vasopressor administration is labour intensive and requires the full attention of the anaesthetist who is often also involved in attending to the other needs of the awake patient and her partner during surgery.
In a preliminary study, we described the safety and efficacy of a closed-loop dual-vasopressor automated system that integrates beat-to-beat systolic BP data from a continuous non-invasive arterial pressure (CNAP) device using a novel customised algorithm that controls the infusion of phenylephrine and ephedrine . Continuous non-invasive arterial pressure was found in recent studies to be superior to intermittent oscillometric measurements in detecting both rapid BP changes and intra-operative hypotension during spinal anaesthesia for caesarean section , and its accuracy and precision have been compared favourably to invasive intra-arterial BP monitoring [11, 12]. With continuous non-invasive BP monitoring, our system is able to respond efficiently to rapid BP fluctuations and tailor vasopressor administration according to the patient's haemodynamic profile. We use phenylephrine as the first-line vasopressor for hypotension, whilst ephedrine is utilised to treat low BP with bradycardia. This closely mimics clinical practice at our institution and allows us to capitalise on the advantages of phenylephrine, namely, its ease of titration and minimal fetal effects, while reserving the use of ephedrine for when phenylephrine is not appropriate. The automation of the system ameliorates the need for our current manual dosing and titration by the anaesthetist.
We compared our automated system against manual vasopressor boluses in this double-blinded, randomised control trial. We aimed to provide a tighter control of maternal blood pressure by reducing the incidence of reactive hypertension, defined as systolic BP > 120% of baseline, and hypotension, defined as any beat-to-beat reading of systolic BP < 80% of baseline. Clinically relevant outcomes such as maternal nausea and neonatal outcomes were also examined.
This study was a randomised, double-blinded controlled study to compare our automated vasopressor system against our existing practice of managing hypotension after spinal anaesthesia. Approval was obtained from the Centralised Institutional Review Board.
After obtaining written informed consent, we recruited women of ASA physical status 1–2 with singleton full-term pregnancies who presented for elective caesarean section under spinal anaesthesia. Inclusion criteria were age between 21 and 45 years, weight 40–99 kg and height 145–170 cm. We did not include women with an allergy to drugs used in the study, contraindications to spinal anaesthesia, obstetric complications such as pre-eclampsia and placenta praevia, and those with uncontrolled medical conditions such as hypertension, diabetes mellitus and cardiovascular disease.
All the enrolled patients received ranitidine and sodium citrate pre-operatively. The woman rested supine with left uterine displacement provided by a wedge for 15 min in a quiet environment. Baseline heart rate and non-invasive BP taken from the right brachial artery were calculated as the mean of three consecutive readings recorded at 1-min intervals. An 18-G cannula was inserted into a forearm vein and additional standard monitors (electrocardiogram and pulse oximetry) were applied. Spinal anaesthesia was performed in the sitting position at the L2-3 or L3-4 vertebral interspace using a 27-G pencil point needle (Espocan; B. Braun, Melsungen, Germany) with 20-G introducer. After ensuring free flow of cerebrospinal fluid, 11 mg hyperbaric bupivacaine, 0.5% with 15 μg fentanyl and 100 μg morphine was injected over 15–20 s. The patient was then positioned supine with left lateral table tilt and 10–15 ml.kg−1 Ringer's lactate intravenous co-hydration commenced, whilst the vasopressor system was initiated.
The patients were randomly assigned to two groups using a computer-generated code that was placed in sealed opaque sequentially numbered envelopes. The investigators prepared all study medications. In the automated vasopressor group, phenylephrine (100 μg.ml−1) and ephedrine (8 mg.ml−1) were prepared in separate 50-ml syringes connected via fine-bore extension tubes to the intravenous cannula by three-way stopcocks. The automated system provided BP measurements continuously in real time, and activated one of two syringe pumps (Perfusor Compact S230-2; B. Braun) when systolic BP was < 90% of the baseline systolic BP. The algorithm administered 50 μg phenylephrine if systolic BP, integrated over 10 s, fell to < 90% of the baseline. If hypotension occurred with bradycardia < 60 beats.min−1, 4 mg ephedrine was administered instead. Vasopressor administration required 10 s, followed by a 10-s lockout period to permit the vasopressor to take effect. Each administration cycle lasted 30 s, allowing two cycles per min. A schematic of the algorithm is shown in Fig. 1. The goal was to maintain systolic BP within 10% of baseline and heart rate > 60 beats.min−1.
The attending anaesthetist in the automated vasopressor group, blinded to group allocation, was given two separate 10-ml syringes containing saline but labelled as 100 μg.ml−1 phenylephrine and 8 mg.ml−1 ephedrine, and instructed to administer the drugs accordingly: 1 ml phenylephrine if systolic BP < 90% baseline, or 1 ml ephedrine if systolic BP < 90% baseline and heart rate < 60 beats.min−1 based on CNAP readings cycled at 1-min intervals.
In the control group, systolic BP measurements at 1-min intervals using CNAP were provided to the attending blinded anaesthetist, who administered 100 μg phenylephrine or 8 mg ephedrine boluses at a maximum of 1 bolus every minute to maintain systolic BP > 90% baseline and heart rate > 60 beats.min−1. The double-pump system was set up as for the active group, but the 50-ml syringes contained saline. The patients were instructed to report nausea, vomiting and headache.
The spinal block height was measured by the attending anaesthetist 5 min after initiation of spinal anaesthesia using loss of sensation to cold. We recorded the timing of spinal anaesthesia, surgical incision and delivery, the presence of nausea, vomiting, or headache, total volume of intravenous fluid given up to delivery, and any technical problems during surgery. The study duration lasted from the administration of spinal anaesthesia to delivery. The attending midwife or neonatologist assessed Apgar scores at 1 and 5 min after delivery, and neonatal birth weight and umbilical cord blood pH and lactate were measured.
Our target cohort of 200 was powered for a 20% reduction in reactive hypertension reported previously in 47% of subjects who received phenylephrine infusion with spinal anaesthesia for caesarean section, with alpha error = 0.05 and power = 0.80 . We allowed for a 10% dropout rate in this study.
Patient's data were first analysed visually for normal distribution, followed by independent-sample, two-tailed t-test if normally distributed, or Mann–Whitney U-test for non-continuous data. The chi-squared test was used for categorical data. A p value < 0.05 was considered significant. All statistical analyses were performed using Statistical Package for Social Services (v20; SPSS, Armonk, NY, USA). We utilised the pooled-data approach to assess the performance of our system as previously described [14, 15].
Percentage performance error
Performance error (PE) was defined as the percentage difference between each measured value of systolic BP from the baseline. The PE for the ith patient at the jth minute was calculated as follows:
Median absolute performance error
Median absolute performance error (MDAPE) is a measure of inaccuracy and indicates the absolute magnitudes of the differences between measured and baseline BPs.
with Ni being the number of values of |PE| for the ith patient and M being the number of patients in the study.
Median performance error
Median performance error (MDPE) is a measure of bias and indicates whether the differences between measured BPs were systematically above or below baseline BPs.
Wobble measures how much PE fluctuates around the MDPE with time, for each patient (i.e. intrasubject PE variability).
All calculations were performed using Microsoft Office Excel, 2010 (Microsoft Corporation, Redmond, WA, USA).
Two hundred and sixteen women fulfilled the inclusion criteria for the trial and consented. Three were not taken into the study owing to failed administration of spinal anaesthesia as a result of technical difficulty necessitating general anaesthesia. In total, 213 women completed the study and did not require supplemental analgesia. The CONSORT diagram is shown in Fig. 2.
Patients’ baseline characteristics and haemodynamic parameters were similar and are summarised in Table 1.
|Automated vasopressor (n = 106)||Control (n = 107)|
|Age; years||32.4 (5.4)||33.5 (4.6)|
|Weight; kg||72.4 (12.3)||70.9 (11.7)|
|Height; cm||158.5 (6.7)||158.2 (5.6)|
|Baseline systolic BP; mmHg||122.7 (10.6)||124.4 (11.1)|
|Baseline heart rate; beats.min−1||89.1 (12.8)||88.9 (13.6)|
|Anaesthetic block; dermatome||T4 (T4–T4 [T1–T6])||T4 (T4–T4 [T1–T6])|
|Spinal to delivery time; min||17 (14–23 [9–47])||18 (14–24 [6–48])|
The study haemodynamic outcomes are summarised in Table 2. The majority of systolic BP measurements from both groups fell within 20% variation of baseline, 81 580 out of 90 824 (89.8%) for automated vasopressor vs 77 236 out of 88 256 (87.5%) for control. In the automated vasopressor group, 8 out of 106 (7.5%) patients were hypertensive, accounting for 1245 out of 90 824 (1.4%) total readings. In comparison, 14 out of 107 (13.1%) patients were hypertensive in the control group, comprising 1816 out of 88 256 total readings (2.1%). Conversely, 37 (34.9%) patients in the automated vasopressor group were hypotensive (8.8% of total readings). This was significantly lower than the control group with 63 (58.9%) hypotensive patients (10.4% of total readings) at p = 0.001. These results included any patient with any beat-to-beat systolic BP <80% of baseline, regardless of whether it was sustained over any period of time. Hence, this would overestimate the incidence of hypotension detected by the intermittent oscillometric technique in most clinical settings. There was no difference between the automated vasopressor and control groups in terms of median highest recorded systolic BP, lowest systolic BP, highest heart rate and lowest heart rate.
|Automated vasopressor (n = 106)||Control (n = 107)||p value|
|Total fluid volume given; l||1.0 (1.0–1.0 [0.5–1.5])||1.0 (1.0–1.0 [0.5–1.5])||0.973|
|Readings with hypertension||1245 (1.4%)||1816 (2.1%)||0.411|
|Readings with hypotension||7999 (8.8%)||9204 (10.4%)||0.104|
|Patients with hypertension||8 (7.5%)||14 (13.1%)||0.260|
|Patients with hypotension||37 (34.9%)||63 (58.9%)||0.001|
|Highest systolic BP; mmHg||139 (130–149 [110–192])||142 (133–153 [108–206])||0.081|
|Lowest systolic BP; mmHg||90 (79–98 [45–128])||86 (75–98 [43–141])||0.471|
|Highest heart rate; beats.min−1||110 (99–121 [77–174])||109 (98–120 [72–162])||0.415|
|Lowest heart rate; beats.min−1||59 (54–64 [36–87])||60 (54–68 [42–88])||0.340|
|Median absolute performance error (MDAPE); %||8.5 (6.5–11.00 [2.8–34.2])||9.8 (7.6–12.5 [3.4–33.3])||0.013|
|Median performance error (MDPE); %||−5.7 (−9.5 to −2.5 [−34.2 to 17.7])||−7.5 (−9.7 to −2.4 [−33.3 to 21.3])||0.264|
|Wobble; %||6.3 (4.0–8.1 [2.0–23.6])||6.2 (4.6–8.5 [1.5–21.3])||0.684|
With regard to performance errors in the system, the MDAPE in the automated vasopressor group was reduced compared with the controls. The other performance measures were comparable across the groups.
No additional phenylephrine, ephedrine or atropine was administered by the investigators before the pumps were set up or during the operation. In all cases of hypertension, the blood pressure decreased spontaneously towards the baseline without intervention within 3 min.
Maternal outcomes are summarised in Table 3. The automated vasopressor group had a significant reduction in the incidence of nausea compared with the control group.
|Automated vasopressor (n = 106)||Control (n = 107)||p value|
|Nausea||1 (0.9%)||11 (10.3%)||0.005|
|Vomiting||0 (0%)||3 (2.8%)||0.246|
|Headache||0 (0%)||1 (0.9%)||1.000|
|Total dose phenylephrine; μg||400 (250–650 [0–1600])||400 (300–700 [0–1550])||0.727|
|Total dose ephedrine; mg||0 (0–0 [0–44])||0 (0–0 [0–20])||0.947|
Almost all patients required phenylephrine at some point between initiation of anaesthesia and delivery, 105 (99.1%) in the automated vasopressor group and 101 (94.4%) in the control group. Seventeen (16.0%) of the automated vasopressor group and 17 (15.9%) of the control group received ephedrine.
Neonatal outcomes are summarised in Table 4. There was no significant difference in umbilical artery-vein pH difference and lactate difference between the automated vasopressor and control groups, respectively. In the automated vasopressor group, three neonates had an Apgar score at 1 min of 8, vs two neonates in the control group with a score of 8 and one of 3. All neonates had Apgar scores of 9 at 5 min.
|Automated vasopressor (n = 106)||Control (n = 107)||p value|
|pH difference (umbilical artery – vein)||0.072 (0.057–0.095 [−0.069 to 0.237])||0.077 (0.048–0.097 [−0.009 to 0.199])||0.813|
|Lactate difference (umbilical artery – vein); mmol.l−1||0.34 (0.06–0.62 [−0.62 to 2.71])||0.39 (0.07–0.78 [−1.72 to 3.26])||0.389|
|Apgar score; 1 min||9 (9–9 [8–9])||9 (9–9 [3–9])||0.991|
|Apgar score; 5 min||9 (9–9 [9–9])||9 (9–9 [9–9])||1.000|
In this study, we found that the use of our novel vasopressor administration system resulted in a significantly lower incidence of hypotension and lower MDAPE compared with the control group, with no increase in the incidence of reactive hypertension. The system also reduced maternal nausea and had similar neonatal outcomes to those of controls.
A previous study by Ngan Kee et al. using a phenylephrine infusion that was commenced when systolic BP decreased below baseline showed that hypotension could be practically abolished, but at the expense of increased phenylephrine consumption and a 47% incidence of reactive hypertension . The impact of reactive hypertension with reflex bradycardia on the mother and neonate is still not fully understood, but may not be innocuous, especially in high-risk patients with potential placental insufficiency. Recently, a closed-loop feedback computer-controlled infusion with a variable algorithm administering phenylephrine has been described by Ngan Kee et al. that demonstrated superiority in BP stability over a manual-controlled infusion . We set the threshold for starting the vasopressor administration at 90% of baseline systolic BP to enable a more permissive treatment of hypotension while reducing the risk of reactive hypertension. In this study, only 8.8% of all systolic BP readings in the automated vasopressor group were < 80% of baseline and no episode of hypotension lasted more than 1 min; reactive hypertension occurred in only 1.4% of readings. Future studies to refine our system should address the need to control these values even more tightly.
We chose a bolus regimen rather than infusion for vasopressor delivery. Doherty et al.  supported the use of phenylephrine bolus rather than infusion. Their bolus regimen maintained BP closer to the baseline during the initial minutes after spinal injection and required a lower total dose of phenylephrine to maintain BP compared with the infusion regimen. However, some studies suggest similar outcomes between phenylephrine bolus and infusion techniques using conventional BP monitoring [19, 20]. Allen et al. reported the incidence of intra-operative nausea in women undergoing caesarean section under spinal anaesthesia as 32% with 100 μg.ml−1 phenylephrine infusion and 35% with 100-μg phenylephrine boluses .
We assessed the performance of our system based on measures previously utilised to evaluate similar closed-loop systems [14, 15], but employed the more complicated pooled-data method that assigns a weight to each BP measurement according to the total number of measurements obtained for each patient. We did this because, as a result of variability in systolic BP between patients and the large number of BP measurements obtained for each patient, averaging the measurements would have resulted in an overemphasis on patients with fewer BP readings and underemphasis on others with greater numbers of measurements. Our method allowed us to assign more weight to measurements from patients in which our system performance is known to have produced a greater ‘accuracy’ (i.e. in those with a greater number of BP measurements), and vice versa . The negative MDPE indicates that our system maintained systolic BP slightly below baseline values by a median of 5.7%, with the Wobble indicating that the systolic BP fluctuated around this MDPE by 6.3%. The MDAPE variable reveals that our system resulted in systolic BP values that differed from baseline by a median of 8.5%. Of these, MDAPE was judged to be the most important parameter as it indicates the accuracy of the system with respect to baseline BP, and our system showed significantly reduced MDAPE compared with controls (Table 2). An MDAPE of 9.8% in our control group suggests that BP is maintained within 10 mmHg from a baseline of 100 mmHg, which, in our opinion, represents sufficiently tight control. The ability to enhance the system's stability and accuracy is arguably more crucial in high-risk pregnancies when a tighter control of haemodynamic function is necessary. Nevertheless, these results should enable us to modify our system to improve performance further in future studies. For instance, the MDPE and Wobble in our study suggest that we can increase the threshold for vasopressor administration closer to the baseline systolic BP; this should improve both MDPE as well as MDAPE.
Our system also resulted in better maternal outcome in terms of a reduction in intra-operative nausea without increasing the amount of phenylephrine and ephedrine required. The role of our automated system to improve vasopressor administration and thereby neonatal and maternal outcomes warrants further investigation. Indeed, our system is able to detect and respond to hypotension instantaneously owing to the use of CNAP, hence restoring normal BP earlier compared with conventional 1-min interval BP readings, while avoiding the complications of intra-arterial BP monitoring. Additionally, as BP measurements were integrated over 10-s intervals, artefacts or loss of single readings due to movement or shivering has less impact. The ability to respond to rapid fluctuations in BP and to monitor the effect of each bolus would be potentially favourable compared with intermittent BP monitoring, as the trough of the hypotensive episode using intermittent BP monitoring will be missed owing to the non-continuous measurement technique.
Our system is, to the best of our knowledge, the only one that utilises a double-vasopressor system to harness the positive haemodynamic effects of phenylephrine and ephedrine. Our results showed that phenylephrine alone was adequate to maintain BP in the majority of the cases. Phenylephrine, which is the currently recommended vasopressor of choice in this context, also has the advantages of high efficacy, ease of titration and minimal fetal acidaemia, although reflex bradycardia may occur with its use . Reflex bradycardia may result in a reduction of maternal cardiac output, although its true clinical impact is undetermined . Additionally, 16% of cases had hypotension with concomitant bradycardia, which might indicate blockade of cardiac sympathetic accelerator afferents; this supports the addition of ephedrine with its alpha- and beta-sympathomimetic effects. However, sparing use of ephedrine is recommended as it is associated with an increased risk of fetal acidosis when used in high dose as a sole agent . Despite the slower onset and longer duration of action of ephedrine, making careful titration difficult, ephedrine may have a role as a ‘second-line’ vasopressor in the event of bradycardia.
As our study design involved the administration of vasopressors only if systolic BP fell to < 90% of baseline, it might have resulted in a higher incidence of hypotension than a strategy of administration when systolic BP fell below baseline, as described by Ngan Kee et al. . However, we found that our system reduced the incidence of reactive hypertension. Within the set limits, we have demonstrated the value of the closed-loop double-vasopressor automated system as an effective and efficient method to treat hypotension during spinal anaesthesia. Furthermore, the accuracy and precision of CNAP, which is the pivotal ‘monitoring arm’ of the closed-loop double-vasopressor automated system, have been shown to be comparable with that of intra-arterial BP monitoring [11, 12]; CNAP automatically recalibrates against oscillometric measurements, reducing potential errors.
In conclusion, we found that the use of an automated, closed-loop, algorithm-based administration of vasopressors in conjunction with the use of CNAP resulted in reduced hypotension and maternal nausea when compared with the manual bolus regimen with intermittent BP monitoring for women undergoing elective caesarean section. The automated system was also able to maintain BP nearer to the baseline values. Future research is needed to refine the administration of vasopressors in the management of hypotension in the obstetric population.
We thank Mr Philip Cheong (system analyst) and Ms Agnes Teo (research nurse) for their assistance during this trial.
No competing interests declared. The authors acknowledge the funding from the National Medical Research Council, Singapore (NMRC: EDG/1003/2010) for this trial.
- 6Techniques for preventing hypotension during spinal anaesthesia for caesarean section. Cochrane Database of Systematic Reviews 2006; 4: CD002251., , , , .
- 10Continuous non-invasive blood pressure monitoring during spinal anesthesia for cesarean section. ASA Annual Meeting, 2009; A827. http://www.asaabstracts.com/strands/asaabstracts/abstract.htm;jsessionid=5F43A6D4D70C215A3B8A64D2FC427F46?year=2009&index=14&absnum=445 (accessed 19/10/2013)., , , , .