Article Text


Use of pulse oximetry for blood pressure measurement after cardiac surgery
  1. A J Movius,
  2. S L Bratton,
  3. G K Sorensen
  1. Department of Anesthesiology, University of Washington, School of Medicine, Seattle, WA, USA
  1. Dr A J Movius, PO Box 5371, Seattle, WA 98105, USA.


Blood pressure measurement using pulse oximeter waveform change was compared with an oscillometric measurement and the gold standard, intra-arterial measurement, in children after cardiac surgery. Forty six patients were enrolled and divided into groups according to weight. Simultaneous blood pressure measurements were obtained from the arterial catheter, the oscillometric device, and the pulse oximeter. Pulse oximeter measurements were obtained with a blood pressure cuff proximal to the oximeter probe. The blood pressure measurements from the pulse oximeter method correlated better with intra-arterial measurements than those from the oscillometric device (0.77–0.96v 0.42–0.83). The absolute differences between the pulse oximeter and intra-arterial measurements were significantly smaller than between the oscillometric and intra-arterial measurements in children less than 15.0 kg. The pulse oximeter waveform change is an accurate and reliable way to measure blood pressure in children non-invasively, and is superior to the oscillometric method for small patients.

Statistics from

Automated oscillometric measurement of blood pressure has been the standard of care for non-invasive blood pressure assessment in children since the early 1980s.1-3 These devices transmit the pressure oscillations generated by arterial pulsation to a blood pressure cuff. The oscillations are then measured by a transducer and fed into a microprocessor.4 Oscillometric measurements have been shown repeatedly to correlate well with the gold standard of intra-arterial measurements, but the absolute accuracy of the measurements has been problematical. The 95% confidence interval for absolute accuracy has been shown repeatedly to exceed 15 mm Hg.5-8

Studies have also indicated that oscillometric measurements become less accurate when blood pressure is low, especially if mean arterial pressure is less than 40 mm Hg.5 9 10 Nevertheless, it has remained the preferred method for non-invasive blood pressure measurement because other options such as palpation, auscultation, and Doppler are either even less accurate or are technically difficult to undertake in children.

Recently, another method of measuring blood pressure non-invasively, using the pulse oximeter, has been reported. This requires a manual blood pressure cuff to be placed on the same extremity as the pulse oximeter probe. The pressure is recorded at disappearance and/or reappearance of the oximeter waveform display during cuff inflation/deflation, which approximates to systolic blood pressure. This was first described in 1987 after it was noted intraoperatively that the loss of the audio and visual oximeter (Nellcor Pulse Oximeter Model N-100) display correlated with the systolic blood pressure obtained by Doppler when the Doppler and pulse oximeter probes were on the same extremity.11 Subsequent studies have verified the usefulness of this methodology and have expanded it to use the disappearance and/or reappearance of the real time plethysmographic waveform available on many pulse oximeters.12-14

Our study investigates the accuracy of pulse oximeter blood pressure measurement using real time plethysmographic waveform disappearance and reappearance in term newborns and children immediately after cardiac surgery. The pulse oximeter measurements are compared with those obtained using an oscillometric device, and by using direct intra-arterial measurements as the gold standard.

Materials and methods

We enrolled patients admitted to the paediatric intensive care unit at the Children’s Hospital and Medical Center, Seattle after cardiac surgery with indwelling arterial catheters. Patients were divided into three groups on the basis of weight: 2.0 kg to  < 6.0 kg; 6.0 kg to 15.0 kg; and > 15.0 kg. A power analysis with α = 0.05 and β = 0.2 indicated that 15 patients with two sets of measurements would be adequate to demonstrate an absolute difference between the two indirect and intra-arterial measurements of 5 torr. Exclusion criteria were aortic arch anomalies or other anatomical abnormalities that could result in disparate four limb blood pressure (such as a Blalock-Taussig shunt), presence of persistent irregular cardiac rhythm, or death within 24 hours of surgery. The study protocol was approved by the Institutional Review Board and informed consent was obtained from parents or guardians.

The pulse oximeter probe and a blood pressure cuff were placed on the corresponding contralateral limb to the oscillometric (Dinamap 8100; Critikon, Inc, Tampa, Florida, USA) cuff site. For the oscillometric and oximetric blood pressure cuff selection, the bladder width was 40% to 60% of the patient’s arm circumference. The blood pressure cuffs were placed on extremities where there would be no interference with the intraoperatively placed arterial catheter; thus, each set of measurements was obtained from three different extremities. For example, if the arterial catheter was placed in a radial artery, the oscillometric blood pressure cuff would be placed on one calf and the oximeter cuff on the other calf.

At the time of each measurement, we evaluated the function of the pulse oximeter and the arterial catheter (Multi-parameter Module 90470; SpaceLabs, Inc, Redmond, Washington, USA). For the pulse oximeter, we assessed the characteristic plethysmographic waveform, heart rate, and oxygen saturation. Normal pulse oximeter function was assumed to be present when we observed a distinct waveform and when the pulse oximeter heart rate was equivalent to that obtained by continuous electrocardiography. We assessed the intra-arterial catheter with respect to calibration, waveform, blood return, and resistance to flushing. Presence of a distinct arterial tracing, recent calibration, easy flushing, and a brisk blood return were taken to indicate normal function. A nursing assessment rating the blood return, ease of flushing, and overall arterial catheter function was obtained at the time of each measurement, as was a print out of the arterial line tracing (Dual Channel Bedside Printer Module 90449; SpaceLabs Inc).

Simultaneous blood pressure measurements were obtained from the arterial line, the oscillometric device, and the pulse oximeter. The measurements were obtained twice for each patient during the first 48 hours after surgery. Time between sets of measurements varied from 15 minutes to 25 hours. Systolic blood pressure measurements were obtained from the pulse oximeter by inflating the blood pressure cuff, placed on the extremity with the pulse oximeter probe, in slow 2–5 mm Hg increments. We noted the pressure at which the pulse oximeter waveform became flat. The cuff was then quickly inflated an additional 20 mm Hg and slowly deflated, again by increments of 2–5 mm Hg, noting the pressure at which the pulse oximeter waveform returned. The blood pressures obtained at waveform disappearance and reappearance were also averaged for each measurement. Systolic, diastolic, and mean blood pressures were recorded from the oscillometric device and the arterial line.

When obtaining the oximeter measurements, an independent observer (the patient’s nurse) identified the time at which the pulse oximeter waveform became flat and returned. This individual also recorded the intra-arterial and oscillometric blood pressures when the pulse oximeter waveform disappeared. The principal investigator was responsible for manual blood pressure cuff inflation/deflation and recording of the pressure when the nurse indicated that the pulse oximeter waveform had disappeared and returned.

Pearson correlation coefficients were calculated between intra-arterial and the non-invasive blood pressure methods. Limits of agreement was used to evaluate the absolute agreement between the invasive and non-invasive techniques, and the differences between the indirect and intra-arterial measurement were compared with a paired ttest.15 A p value of  < 0.05 was considered significant.


There were 46 patients enrolled in the study: 15 in the < 6.0 kg group, 15 in the 6.0–15.0 kg group, and 16 in the > 15.0 kg group. One additional patient was enrolled in the >15.0 kg group to maintain a constant total number of measurements at 30 for each group. This was necessary because the arterial line had been discontinued before the second set of blood pressure measurements in two patients. There were no failures in obtaining oscillometric or oximetric measurements in any patient. The male:female ratio was 1:2.7 and was similar for all weight groups. Eighty nine per cent of measurements were performed by the end of postoperative day 1 and all were completed by postoperative day 2. Most patients (91%) had peripheral arterial catheters, all located in the upper extremities. All four central arterial catheters (9%) were located in femoral arteries. Many patients were receiving vasoactive medications. Table 1depicts these data for each patient group.

Table 1

Patient characteristics by weight group

Intra-arterial systolic blood pressure ranged from 64–100 mm Hg (mean, 77.7 mm Hg), 60–140 mm Hg (mean, 88.7 mm Hg), and 81–136 mm Hg (mean, 105.0 mm Hg) for the < 6.0 kg, 6.0–15.0 kg, and > 15.0 kg groups, respectively. The absolute mean differences for intra-arterial minus oximeter waveform return blood pressures were significantly less than for the systolic oscillometric blood pressures for the < 6.0 kg and 6.0–15.0 kg groups (p < 0.0001 and p < 0.01, respectively). This relation was also observed in the > 15.0 kg group, although statistical significance was not reached (p = 0.055). Of the various oximeter measurements analysed, waveform return provided the most accurate measurement of systolic blood pressure in all groups, although waveform average was also significant for patients < 15.0 kg and waveform disappearance for patients < 6.0 kg.

Correlation coefficients for intra-arterial blood pressure compared with oscillometric measurements ranged from 0.75–0.79 for systolic measurements, 0.68–0.83 for mean measurements, and 0.42–0.83 for diastolic measurements. Correlation coefficients for intra-arterial measurements compared with oximeter measurements ranged from 0.77–0.96 for waveform disappearance, 0.83–0.95 for waveform return, and 0.81–0.96 for waveform average. The limits of agreement between intra-arterial and pulse oximeter pressures at waveform return had narrower ranges than those obtained between intra-arterial and oscillometric systolic and mean blood pressure measurements (table2).

Table 2

Correlation (r) and limits of agreement (LOA) between direct and indirect blood pressure measurement


We found the reappearance of the pulse oximeter plethysmographic waveform during blood pressure cuff deflation to be a useful method of non-invasively measuring blood pressure after cardiac surgery in children of all ages. The oximeter measurements, when compared with direct intra-arterial systolic pressure, were superior to the oscillometric measurements for systolic, diastolic, or mean blood pressure.

These results agree with those of other studies investigating the usefulness of the pulse oximeter signal disappearance and reappearance as an alternative method of non-invasive blood pressure measurement in adults and neonates. Since the initial report in 1987,11two additional studies in adult patients have also reported increased accuracy when the pulse oximeter method was compared with other non-invasive blood pressure methods.12 13 In addition, case reports of patients with Takayasu’s arteritis concluded the method to be a useful alternative for blood pressure measurement in this clinical situation, for which the oscillometric device did not work.16

To date, the only other paediatric study was performed on 50 primarily premature neonates by Langbaum and Eyal.14 Comparison of simultaneous measurements of the intra-arterial, oscillometric, and oximeter systolic blood pressure demonstrated that, in addition to correlating well with intra-arterial measurements, oximeter measurements had significantly better accuracy than oscillometric measurements. Our study differed from that of Langbaum and Eyal in that waveform return of the oximeter tracing provided the most consistent data. This is not surprising because it has been shown that less blood flow is required to restart the pulse oximeter tracing during blood pressure cuff deflation than is needed to maintain the tracing during blood pressure cuff inflation.17 The neonatal study, however, found no significant difference in blood pressure measurements between those recorded at waveform disappearance, return, or their average.14 Like the study by Langbaum and Eyal, we also found that the oximeter measurements remained reliable at lower blood pressures, whereas the accuracy of the oscillometric measurements deteriorated further. Our study also demonstrated that the oximeter blood pressure measurements were more accurate in the smallest patients, who tended to have the lowest blood pressures.

There are several important limitations to using the pulse oximeter to measure blood pressure indirectly. First, only systolic blood pressure can be measured. Also, there is an inherent error in the measurements, based on the 2–5 mm Hg pressure increments used to obtain them. An inherent error of similar magnitude exists for the oscillometric device as well.4 In addition, oximeter measurements are subject to observer variation. Location of blood pressure measurement constitutes another potential source of error in this study. There is evidence that blood pressure might differ between the upper and lower extremities and the cuff measurements were always taken on a different type of extremity from the direct measurements.18 However, because our primary goal was to determine the absolute accuracy of the oscillometric method and the pulse oximeter method compared with intra-arterial measurements, the two non-invasive measurements were always carried out on the same type of extremity. Thus, we could still make a meaningful comparison between the pulse oximeter and oscillometric methods and the direct intra-arterial measurement. Finally, most arterial catheters in this study were placed in peripheral arteries. Although considered the gold standard, intra-arterial measurement of blood pressure may vary with arterial catheter location. This variability can be secondary to the catheter itself (that is “ringing”) as well as differences in reflection of the arterial wave at more peripheral locations.19 Many of our patients were on vasoactive medications, which may influence arterial wave reflection through changes in vasomotor tone.19 However, this should have affected both non-invasive methods equally.

Direct arterial and indirect oscillometric measurements are the standard methods of recording systemic blood pressure in our paediatric intensive care unit. Intra-arterial catheterisation is the gold standard for monitoring blood pressure, but it is invasive and carries potential morbidity.20 Unfortunately, the non-invasive oscillometric method of measurement is less accurate, especially at the low blood pressures common in children. As such, this non-invasive measurement of blood pressure is used primarily for patients without haemodynamic compromise or in whom intra-arterial access has been unsuccessful. A more accurate and precise non-invasive measurement of blood pressure in the paediatric intensive care population would be beneficial.

We conclude that using the reappearance of the pulse oximeter plethysmographic waveform is an accurate and reliable way to measure systolic blood pressure non-invasively in children after cardiac surgery. This method of non-invasive blood pressure measurement is superior to the oscillometric method when compared with direct intra-arterial measurements. Although it is not a replacement for direct arterial measurement in haemodynamically compromised patients, the method offers another alternative in patients for whom intra-arterial access was unsuccessful. It might also be useful in patients for whom oscillometric measurements are unobtainable, such as those with very low blood pressure. Finally, it has the additional advantage of being easy to perform, without requiring any new or costly equipment.

View Abstract

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.