During elective caesarean section, it is usual to monitor blood pressure (BP) using an automated non-invasive blood pressure (NIBP) machine. During regional anaesthesia, these devices may give erratic readings or fail to record, e.g. due to shivering or excessive arm movement. In such situations, the anaesthetist may rely on clinical signs of adequate end-organ perfusion such as the patient’s colour and responsiveness, or may attempt to obtain an indication of the patient’s BP by feeling for the radial arterial pulse or observing the pulse oximeter whilst the NIBP cuff inflates or deflates, the disappearance/return of a palpable pulsation or oximetry trace indicating systolic arterial pressure. Limited data have been published on the accuracy of these methods for estimating BP [1, 2], and none that we are aware of in the obstetric setting or during regional anaesthesia. Our aim was therefore to study the usefulness of arterial palpation and pulse oximetry for estimating the NIBP machine-derived systolic pressure, in the context of obstetric regional anaesthesia.
We assessed the ability of palpating the radial arterial pulse and observing the oximeter trace to estimate the automated non-invasive systolic pressure reading in 20 healthy female volunteers and 20 parturients undergoing spinal anaesthesia for elective caesarean section. Using real-time values of cuff pressure during inflation/deflation, the pressure was recorded when the manually palpated radial arterial pulse or pulse oximeter waveform disappeared and reappeared. The actual measured systolic pressure was noted and the results compared using Bland–Altman analysis. In the volunteers, the bias/precision for radial arterial palpation was −12.9/22.1 mmHg (inflation) and −9.7/16.7 mmHg (deflation), and for oximetry 29.5/18.8 mmHg (inflation) and −20.7/21.7 mmHg (deflation). In the parturients, the bias/precision was −19.0/47.6 mmHg (inflation) and −15.5/51.0 mmHg (deflation) for arterial palpation, and 22.6/16.1 mmHg (inflation) and −14.2/19.9 mmHg (deflation) for oximetry. Our results suggest that neither method is accurate at estimating the non-invasive systolic pressure, with all except oximetry (inflation) underestimating it by approximately 10–20 mmHg and with poor precision.
After Local Research Ethics Committee approval the study was conducted in two groups, each of 20 women: healthy non-pregnant female volunteers and women undergoing spinal anaesthesia for elective caesarean section. Written informed consent was obtained from all participants. All women were aged 18–45 years and of ASA physical status 2–3. Exclusion criteria included any medical condition associated with high or low BP, treatment with drugs known to affect BP, any condition causing impaired circulation, and any respiratory disease affecting arterial oxygen saturation. For the parturients, gestation < 36 weeks and multiple pregnancy were additional exclusion criteria.
In both groups, the women lay supine on the operating table with a standardised left lateral tilt produced by a wedge. A NIBP cuff was applied to their right arm and a pulse oximeter probe to their right index finger (both Infinity®; Draeger Medical UK Ltd, Hemel Hempstead, Herts, UK). Women undergoing caesarean section also had an electrocardiograph and cardiotocograph applied, as per routine practice. The order of BP measurement methods (palpation and oximeter) was randomly assigned via computer generated random numbers. For radial artery palpation, an investigator palpated the pulse at the right radial artery and observed the dynamic pressure readings of the NIBP machine as the cuff inflated/deflated, recording the pressures displayed at the point when the radial pulsations disappeared/reappeared, respectively. The pulse oximeter waveform was hidden from the investigator during this part of the study. For pulse oximetry, the investigator observed the pulse oximeter trace, noting the cuff pressure readings when the waveform disappeared and reappeared. For both methods, the actual NIBP measurements of systolic, mean and diastolic BP obtained were hidden from this investigator but recorded by a second investigator. The same two investigators (AS and TS) took all the measurements to reduce observer variability.
For the parturients, measurements were made after establishment of spinal anaesthesia. After baseline measurements and placement of an intravenous cannula, women presenting for elective caesarean section received spinal anaesthesia in the sitting position, using heavy bupivacaine 0.5% with diamorphine (doses at the discretion of the anaesthetist in charge of the case). The women then were laid supine with a left lateral tilt, as before. Prevention and treatment of hypotension was left to the managing anaesthetist’s preference but was in fact always with 50–100 μg boluses of phenylephrine and intravenous crystalloid, the doses and volumes of which were recorded. The time taken for the height of the block to reach a level deemed suitable for caesarean section was recorded as was the final height of the block (to icy cold). Once the anaesthetist was satisfied that the patient was stable and the block was suitable for surgery, the palpation/oximetry/NIBP measurements were taken. The patient was then prepared and draped for surgery, and her involvement with the study ceased.
Bland–Altman plots  were constructed to assess the relationship between the arterial palpation/pulse oximeter estimations of BP and the measurements made by the NIBP machine, both for volunteers and patients, and during inflation and deflation of the cuff. In Bland–Altman analysis, the difference between the measurements (representing how closely the palpation/oximeter-derived BP matches the NIBP) is plotted against the average of the two measurements; the mean difference represents the ‘bias’ of one measuring method over the other, with 2 × SD of the differences representing the ‘precision’.
All participants completed the study. Mean (SD) body mass index was 23.6 (4.0) kg.m−2 in the volunteers and 23.0 (4.0) kg.m−2 (on booking) in the parturients. Indications for caesarean section were previous caesarean section (12), maternal request (4), breech delivery (2), placenta praevia (1) and previous myomectomy (1). Anaesthetic details are given in Table 1.
|Spinal bupivacaine 0.5%; ml||2.4 (2.3–2.4 [2.2–2.5])|
|Spinal diamorphine; μg||300 (300–350 [300–350])|
|Time between spinal and blood pressure assessments; min||9.5 (9.0–10.5 [6.3–16.0])|
|Phenylephrine requirements; μg||200 (50–363 [0–500])|
|Intravenous fluids requirements; ml||1000 (500–1063 [100–1500])|
|Height of block||T3 (T2–T3 [T2–T5])|
The bias/precision in the volunteers (Fig. 1) for radial arterial palpation was −12.9/22.1 mmHg (inflation) and −9.7/16.7 mmHg (deflation), and for oximetry 29.5/18.8 mmHg (inflation) and −20.7/21.7 mmHg (deflation). In the parturients (Fig. 2), the bias/precision was −19.0/47.6 mmHg (inflation) and −15.5/51.0 mmHg (deflation) for arterial palpation, and 22.6/16.1 mmHg (inflation) and −14.2/19.9 mmHg (deflation) for oximetry.
We found that neither radial arterial palpation nor observation of the pulse oximeter trace was accurate for estimating the systolic BP recorded by the NIBP machine. Apart from pulse oximetry during cuff inflation, which overestimated NIBP by a mean of 20–30 mmHg, the other methods underestimated NIBP by an average of 10–20 mmHg. Precision was also poor.
Radial arterial palpation is commonly performed to give a rough estimate of systolic BP, for example during emergencies or before a more formal measurement using Korotkoff sounds and a manometer. Indeed, when the BP cuff was introduced into US practice in the early 1900s, it was commonly used to measure systolic BP by occluding the palpable radial pulsation; this method was gradually replaced by auscultation, which allowed ready measurement of both diastolic and systolic pressures . In order to produce an accurate estimation of BP using the pulse occlusion method, it is important that the cuff pressure changes slowly enough to allow the ‘signal’ (in this case disappearance/appearance of the palpable pulsation) to be registered before the cuff pressure has changed further. Because the rate at which the NIBP cuff inflates/deflates cannot be altered, and is considerably faster than we would have chosen were we to use a manually controlled cuff with a manometer, it is likely that this is a major factor in the relatively poor performance of the palpation method to estimate BP in both our groups.
Similarly, the pulse oximeter trace can be an accurate indicator of the presence or absence of pulsatile flow distal to a cuff, so long as sufficient time is allowed when inflating/deflating the cuff . Additional delay is added to the oximeter method by the processing of the oximetry signal, although the photoplethysmogram trace can also give information about the shape of the arterial waveform and the volaemic status of the patient [5, 6]. A possible explanation for the positive bias with oximetry during inflation is that when the cuff inflates rapidly, there is a delay after obliteration of the radial pulse before the oximeter display becomes flat, due to processing of the signal. This results in the estimated BP’s exceeding the NIBP. During deflation, the delay results in the opposite effect. Previous studies examining the usefulness of pulse oximetry to indicate BP have found good correlation between BP estimated this way and readings obtained by direct arterial cannulation , Doppler ultrasound  and manual palpation  – however, Bland and Altman have pointed out the folly of assessing measuring devices in this manner, and it is no surprise that the pressures are correlated as they are measuring the same thing. We are unaware of any studies that have applied Bland and Altman analysis, which measures the accuracy and usefulness of one measuring method compared with another, to either the arterial palpation or the pulse oximetry method.
Non-invasive BP machines are now widespread not only in operating theatres but throughout clinical areas . The manufacturers of the device we used do not provide details of the algorithm used or the magnitude/speed of the steps in pressure during inflation/deflation on their written material, and we were unable to obtain further information even on request. Each model has its own particular features and the algorithms used to calculate systolic and diastolic BP may vary between devices, as may the speed of inflation/deflation. We observed that in the particular device used for this study, cuff inflation was more rapid than deflation; furthermore, the speed of inflation was not linear, being greatest at the start of measurement. Thus our results may not be precisely transferable to other devices. We suggest that the ability to control the speed of cuff inflation/deflation manually, to enable more precise estimation of BP using either radial arterial palpation or pulse oximetry, would be a useful feature of such devices – especially if the patient is shivering, as may sometimes be the case during obstetric regional anaesthesia .
The altered haemodynamics of pregnancy are well known to affect the accuracy of NIBP machines in parturients compared with non-pregnant women , hence our including a group of healthy, non-pregnant volunteers. The effect of regional anaesthesia on the circulation in pregnancy is also well described . We found similar results in both groups, suggesting that the general poor performance of the palpation and oximeter methods during regional anaesthesia is a feature of the methods rather than due to pregnancy or to regional anaesthesia, although we did not include groups of parturients without regional anaesthesia, or non-pregnant patients undergoing regional anaesthesia. We also did not examine any possible effect of intravenous fluids and/or vasoconstrictors, choosing instead to observe routine practice. Furthermore, we did not record heart rate and this might influence all the indirect methods we assessed since blood pressure may change more between slow pulsations than between fast ones. Finally, we did not compare the BP estimated using arterial palpation or oximetry with what many would consider the true‘gold standard’, i.e. direct arterial measurement, since it is known that NIBP machines themselves are not perfectly accurate, especially in pregnancy . However, our intention was not to estimate the accuracy of the palpation/oximetry methods for indicating actual arterial pressure per se, but to estimate their usefulness in indicating what the NIBP reading would be, since NIBP is the standard measurement that forms part of our routine intra-operative monitoring.
In conclusion, both arterial palpation and pulse oximetry may be used to indicate systolic BP during caesarean section under regional anaesthesia but both methods are relatively inaccurate and imprecise. Further work would be required to investigate whether their performance can be improved by combining or repeating measurements.
SY is Editor-in-Chief of Anaesthesia and this manuscript has undergone an additional external review as a result. No external funding and no competing interests declared.