Original Article

Reliability of Conventional and New Pulse Oximetry in Neonatal Patients

Department of Pediatrics, The Neonatal Clinical Research Center, University of Colorado Health Sciences Center, Denver, CO, USA

Correspondence to: William W. Hay, Jr., MD, Box B-195, University of Colorado Health Sciences Center, 4200 East Ninth Avenue, Denver, CO 80262, USA

Abstract

OBJECTIVES: Pulse oximetry is widely used in the NICU, but clinicians often distrust the displayed values during patient motion, i.e., questionable oxygen saturation (SpO₂) and pulse rate (PR) values. Masimo Corporation (Irvine, CA) has developed pulse oximetry with claims of resistance to sources of interference. To test this premise, we compared the performance of the Masimo SET pulse oximeter to a conventional device, Nellcor N-200, and then with three other new-generation pulse oximeters, Nellcor N-395, Novametrix MARS, and Philips Viridia 24C.

STUDY DESIGN: We studied 26 nonsedated NICU infants who were on supplemental oxygen and/or mechanical ventilation. ECG heart rate (HR) from a bedside monitor and SpO₂ and PR from the two pulse oximeters were captured by a PC for a total of 156 hours. The ECG HR and pulse oximeter spectral waveform were analyzed at alarms for hypoxemia (SpO₂£85%) and/or bradycardia (HR£80 bpm). We then compared the performance of the Masimo SET to three other new-generation pulse oximeters, Agilent Viridia 24C, Nellcor N-395, and Novametrix MARS, in a similar population of seven infants for a total of 28 hours. We added to the test criteria the ability of the various pulse oximeters to track acute changes in HR.

RESULTS: Compared with Nellcor, Masimo SET had 86% fewer false alarms, which also were shorter in duration, resulting in 92% less total alarm time. Masimo SET also identified nearly all bradycardias versus 14% for the Nellcor. Compared with the new-generation pulse oximeters, false desaturations, data drop-outs, and false bradycardias were lowest for Masimo SET, as was the capture of true desaturations and bradycardias. Notably, the new-generation devices differed greatly in their ability to detect changes in HR (i.e., the frequency of frozen PR during times of ECG HR change was 0, 6, 11, and 46 for Masimo, Nellcor, Philips, and Novametrix, respectively).

CONCLUSIONS: Masimo SET pulse oximetry recorded markedly fewer false SpO₂ and PR alarms and identified more true hypoxic and bradycardic events than either conventional or other new-generation pulse oximeters. Masimo SET also most closely reflected the ECG rate irrespective of accelerations or decelerations in HR.

SPECULATION: Routine use of Masimo SET pulse oximetry in the NICU could improve clinician confidence in the parameter leading to more judicious titration of oxygen with possible reductions in hypoxic (e.g., pulmonary hypertension) and hyperoxic (e.g., retinopathy of prematurity) pathology. Additionally, a more trustworthy technology should equate with fewer confirmatory arterial blood gas analyses (less blood loss), and faster weaning from the mechanical ventilation (less chronic lung disease). Journal of Perinatology (2002) 22, 360-366 doi:10.1038/sj.jp.7210740

INTRODUCTION

Since our first evaluation of the accuracy and reliability of conventional pulse oximetry to measure blood oxygen saturation and predict PaO₂ in 1989,¹ there has been little effective resolution of the principal limitation of pulse oximetry-motion artifact. Patient motion occurs frequently in newborn infants. This problem of motion artifact is inherent to conventional pulse oximeters, because motion adds another signal to the pulse waveform signal, thereby changing the apparent amounts of light transmitted to the photoreceptor. Two wavelengths of light, one red (R) and the other infrared (IR), are used to detect the relative proportions of oxyhemoglobin and deoxyhemoglobin (red/infrared transmitted light ratio, or R/IR).² Motion adds pulsatility to nonarterial blood components (e.g., venous blood) that then is averaged with the R/IR ratio generated by the pulsatile arterial blood.³ Therefore, the magnitude of error in SpO₂ measurement will be influenced by the venous blood saturation, the magnitude of the motion, and the arterial signal amplitude. When motion is marked relative to peripheral arterial blood flow, the motion-added signal can predominate over the arterial pulse causing a falsely low SpO₂ value to be displayed. As a result, when clinicians have problems obtaining reliable pulse oximeter readings, they identify motion and low perfusion as the causes.⁴ Such erroneous SpO₂ values are disturbing at best. They also can lead to unintentional neglect of the patient's oxygenation or the pulse oximetry values, or both, and can lead to inadvertent over-titration of oxygen in response to the falsely low SpO₂ values.

Masimo Corporation (Irvine, CA) has developed unique signal processing algorithms (Masimo SET) that detect and ignore sources of SpO₂ and pulse rate (PR) interference.⁵ We conducted a larger study to evaluate the effects of motion on the incidence of SpO₂ and PR alarms in a neonatal population and to compare the results obtained from a conventional pulse oximeter (N-200, Nellcor, Pleasanton, CA) to those of a Masimo SET pulse oximeter. We then followed the same protocol to differentiate the performance of the available new-generation pulse oximeters in this population.

MATERIALS AND METHODS

Twenty-six neonates weighing 900 to 2710 g were studied in the University of Colorado Hospital level III NICU, Denver, Colorado. The study was approved by the Institutional Review Board of the University of Colorado Health Sciences Center and included parental consent and primary nurse and primary physician approval. The only entry criterion was a preexisting need for pulse oximetry. A Masimo SET and a conventional pulse oximeter (N-200) were compared. The Masimo sensor, LNOP Neo, and the Nellcor sensor, Oxisensor II N-25, were attached to different feet on each infant. To reduce effects of site bias, the sensors were switched to the opposite foot after 3hours and monitored for an additional 3 hours.

The study design was blinded to provide an objective comparison of pulse oximeter performance without affecting routine care of the infant (i.e., only the alarms of the host pulse oximeter were active and the research nurse did not participate in patient care). All data were acquired using a laptop computer, which was operated by the research nurse. The ECG heart rate (HR) from a Datascope Passport monitor (Paramus, NJ) was collected to corroborate each pulse oximeter's PR. Potential sources of artifact, including infant motion, nursing care and intervention, and parental handling and care, were logged by the research nurse. The data were reviewed without knowledge of which pulse oximeter was used to determine how well each met the clinical indices for monitoring these neonates, specifically, true and false desaturation, and bradycardic events.

In follow-up to the initial study, we compared the simultaneous performance of three new-generation pulse oximeters [N-395 (Nellcor), MARS (Novametrix, Wallingford, CT), and Viridia 24C (Philips, Andover, MA)] to Masimo SET. The same enrollment criterion and test setup was followed except for the placement of additional sensors in a randomized fashion on each limb and use of the Viridia monitor as the ECG source. Seven neonates weighing 1050 to 1950 g were studied for a total of 27.7 hours (237±32 min each).

Neonatal motion can be chaotic and aperiodic consisting of kicks, jerks, and stretching. Often intermingled in the motion periods are brief segments of unperturbed heartbeat waveforms. Postprocessing calculation of SpO₂ by the conventional spectral method (R/IR) can be used on these nonmotion segments and compared to the recorded data from the pulse oximeters.⁶ In this analysis, if an SpO₂ value is found at a frequency near the ECG HR (±5 bpm) and similar SpO₂ values exist at harmonics or multiples of the ECG HR (i.e., fundamental frequency), the SpO₂ value is deemed valid. Thus, by analyzing nonmotion spectral waveform data before, in the midst of, and after a motion event, and determining that a frequency domain match is present, the true SpO₂ is verified.

The hypoxemia alarm threshold was set at 85% such that when a pulse oximeter displayed an SpO₂ £85%, R and IR waveform analysis was performed to verify the SpO₂. The PR values were compared to the ECG HR to assess each pulse oximeter's ability to accurately reflect HR. Bradycardia was defined as an ECG HR of £80 bpm. PR data were categorized as “frozen” if the displayed PR was constant (i.e., a PR change of £1 BPM for ≥20 seconds) while the ECG measured an acute change ≥25 BPM (i.e., a change of meaningful clinical consequence, which, if missed, could be problematic).

RESULTS

The upper panel of Figure 1 shows simultaneous SpO₂ recordings of the Masimo SET and the Nellcor N-200 instruments. One nonmotion epoch (circled point 1) and one motion epoch (circled point 2) are identified. The lower panel contains the simultaneous recordings of ECG HR and pulse oximetry PR. Both devices produced similar SpO₂ and PR values during the nonmotion condition at circled point 1 (92/91). However, only the Masimo SET instrument recorded a stable value for SpO₂ and captured a PR similar to the ECG HR during the motion condition at circled point 2.

Figures 2 and 3 are postprocessing plots of data corresponding to the two epochs noted in Figure 1. The red (solid line) and infrared (dashed line) amplitudes are shown in the upper panel (A) and the frequency analysis of this data in the lower panel (B). The R and IR amplitudes in Figure 2 bear a constant relationship to each other, producing a correct value for R/IR and thus SpO₂ at a PR (199 bpm), which matches the ECG HR (200 bpm). In Figure 3, the R and IR amplitudes are disparate but the correct SpO₂ is found when the PR (203 bpm) matches the ECGHR (205 bpm) according to harmonic corroboration.

Table 1 contains the reduced frequency and duration data from 156 hours of monitoring the initial 26 infants. The Nellcor N-200 had a total of 396.6 minutes of “zero outs” and false alarms or 4.2% of operation. The Masimo SET pulse oximeter exhibited 31.3 minutes of zero outs and false alarms or 0.3% of operation. Compared with Nellcor, Masimo SET had significantly fewer false alarms (by 86%), and these were shorter in duration, resulting in 92% less total alarm time. During the monitoring period, eight of the infants had a total of 14 transient bradycardic episodes. Of these, the Masimo SET pulse oximeter caught significantly more, 12 (86%) compared to 2 (14%) for the Nellcor.

Table 2 contains the reduced frequency data from 27.7 hours of monitoring seven infants (1531±349 g). Four new-generation pulse oximeters were in simultaneous use (Philips Viridia, Masimo Radical, Nellcor N-395, and Novametrix MARS). False desaturations totaled 86 (1, 10, 33, and 42 for Masimo, Philips, Novametrix, and Nellcor, respectively), while missed desaturations were few (1, 4, 6 and 12 for Masimo, Nellcor, Philips, and Novametrix, respectively). These new-generation devices differed greatly in their ability to detect changes in PR (i.e., the frequency of frozen PR during times of ECG HR change was 0, 6, 11, and 46 for Masimo, Nellcor, Philips, and Novametrix, respectively). Figure 4 illustrates the performance differences of the new-generation pulse oximeters during a true hypoxemic and associated bradycardic event.

Although the study was not designed to ascertain user preferences or features other than the defined performance indices, the research nurses had several comments: the various sensors were equally easy to apply; the operation of the devices were intuitive; no sensor site or other untoward complications were noted.

DISCUSSION

Pulse oximetry is used routinely in patients in the NICU to monitor blood oxygenation and PR. However, it has a high false-alarm rate that to date has been at best a nuisance to caregivers and parents and at worst has led to medical mismanagement. Indeed, there is evidence that the “crying wolf” syndrome caused by conventional pulse oximetry leads to decreased response to alarms.^7,8 In addition, the inaccurate underreading of SpO₂ caused by motion artifact may lead to inappropriate O₂ titration, producing inadvertent hyperoxia. Such errors in management leading to over or under oxygenation could have detrimental effects for very preterm infants at risk for ROP in the first several days of life⁹ and both term and preterm infants at risk for pulmonary hypertension.

We used an objective measure, direct verification of the R/IR,⁶ to compare the accuracy of SpO₂ and PR versus ECG HR among multiple pulse oximeters in the same infant. Our results are consistent with other reports of increased accuracy during motion and marked reduction in false alarms using Masimo SET for monitoring adults and children. In addition, this study demonstrates an increase in PR accuracy versus conventional pulse oximetry when comparing data for true and false bradycardic events. The bradycardic events missed by the Masimo SET pulse oximeter were 5 seconds or less in duration and probably are the result of differences in averaging modes between the ECG monitor and the pulse oximeter. This added finding of more accurate PR detection indicates that Masimo SET pulse oximetry may suffice where sophisticated ECG monitoring is not available, such as in home care or step-down units.

Methods to eliminate motion artifact, other than Masimo SET, have been introduced recently. However, clinical evaluations revealed the shortcomings of freezing and reporting of old values, longer averaging of data, and reporting zero values.^10,11 Such methods can mask or ignore the true SpO₂ and PR values. The inappropriate use of one of these methods can miss significant epochs of aperiodic breathing. Some authors have cautioned that it not be used in the care of infants at risk for AOP.¹¹ Figure 4 illustrates the conundrum of displaying no value versus an erroneous value; the former is far less likely to be associated with an error in patient care.

The assessment of HR variability is part of the care routine of acutely ill neonates^12,13 and those undergoing diagnostic testing.¹⁴ It is common for a conventional pulse oximeter to simultaneously display erroneously low SpO₂ and PR values. We have corroborated that some new-generation pulse oximeters “freeze” the PR during rapid changes in HR.¹⁵ The considerably better performance exhibited by Masimo SET is likely due to the robustness of this technology.

With conventional pulse oximetry, attentive medical personnel carefully wait until motion stops, verify that the pulse waveform (if available) “looks good,” and then confirm that the PR matches the ECG HR. This added vigilance consumes caregiver time and emphasizes that considerable periods of absent or inaccurate SpO₂ values are inherent with conventional pulse oximetry. During conditions of no-motion and suspect SpO₂ values, a caregiver may unnecessarily assess the infant or obtain a confirmatory arterial blood specimen. Acceptance of such a condition runs counter to usual attempts in medicine to perfect diagnostics and minimize unnecessary interventions.

Preterm infants in a NICU may be exposed to as many as six iatrogenic changes in behavioral state per hour, with at least 60 per day lasting 10 minutes or more.¹⁶ Care interventions and noise independently result in significant changes in both the behavioral and physiological response of infants.^17,18 The combination of noise and touch is even worse. The physical assessment of the sensor site in response to a suspiciously alarming pulse oximeter is common practice. However, such practice is not without adverse consequences as this frequently disturbs the infant leading to unstable conditions. Use of Masimo SET should reduce these problems. Motion artifact can adversely affect up to 50% of conventional pulse oximetry recordings even during neonatal sleep.^19,20 Motion and need for assessing oxygenation is part and parcel of care in the delivery room,²¹ during resuscitation or transport,²² and for the discharge of preterm infants in car seats.²³ It was noted that during circumcision of newborns, the SpO₂ routinely dropped for a conventional pulse oximeter from an average of 96 to 80%, while the Masimo SET device remained stable and there were no clinical signs of hypoxemia to substantiate the low conventional values.²⁴ These are settings where more accurate and continuous pulse oximetry would benefit diagnostics, oxygen therapy, and other necessary medical and surgical management.

Based on the results of this and other studies, it is appropriate now to develop clinical trials to test whether this more accurate and reliable pulse oximeter technology is more valuable. Such trials should include evaluations of nursing time and effort, impact on neonatal behavior, and effectiveness of medical care. Greater device reliability should translate into improved clinician confidence and responsiveness to pulse oximetry, thereby helping in the treatment of prethreshold ROP or perhaps preventing such conditions as pulmonary hypertension, chronic lung disease, or the vasoconstrictive phase of ROP.^9,25 Clinical trials also could address the impact on the improved detection of apnea and bradycardia at home and in the hospital.

CONCLUSIONS

In our NICU patient population, Masimo SET pulse oximetry recorded markedly fewer false SpO₂ and PR alarms and identified more true hypoxic and bradycardic events than either a conventional or three other new-generation pulse oximeters. Decreased false alarming should benefit the infant's behavioral state and improve the caregiver's response to monitoring a vital parameter. It also has the potential to relieve parental anxiety and improve the accuracy of detecting true bradycardic and/or hypoxic events. More expansive clinical trials are indicated to test whether such improved oximetry technology and the changes in medical management it affords will improve patient outcomes.

Acknowledgements

This research was supported by National Institutes of Health GCRC Grant RR00069 to the University of Colorado School of Medicine, Department of Pediatrics and the Masimo Corporation, Irvine, California. Preliminary aspects of this study were presented at the 1999 PAS annual meeting: Hay WW, Rodden DJ, Collins SM, Melara DL, Hale KA, Fashaw LM. Pulse oximetry in the NICU: conventional versus Masimo SET. Pediatr Res 1999;45(4):304A.

References

1 Hay WW Jr, Brockway J, Eyzaguirre M. Neonatal pulse oximetry: accuracy and reliability. Pediatrics 1989; 83: 717–22. MEDLINE

2 Pologe JA. Pulse oximetry: technical aspects of machine design. Int Anesthesiol Clin 1987; 25: 137–53. MEDLINE

3 Swedlow DB, Irving SM. Monitoring and patient safety. In: Blitt CD, editor Monitoring in Anesthesia and Critical Care Medicine 2nd ed New York: Churchill Livingstone, 1990, p 45–6.

4 Industry News. Survey indicates problems in oximetry readings. RT\J Respir Care Pract 1995; 3: (3) 14.

5 Goldman JM, Petterson MT, Kopotic RJ, Barker SJ. Masimo signal extraction pulse oximetry. J Clin Monit 2000; 16: (7) 475–83.

6 Petterson MT, Novak JJ, Kopotic RJ, Goldman JM. A silver standard for corroboration of SpO₂ values during motion. Anesth Analg 2002; 94: (Suppl 1) S106, A12.

7 Lawless ST. Crying wolf: false alarms in a pediatric intensive care unit. Crit Care Med 1994; 22: 981–5. MEDLINE

8 Sabar R, Zmora E. Nurse's response to alarms from NICU monitoring systems. Pediatr Res 1997; 41: 174A.

9 Goldstein MR, Barnum PT, Vogt J, Gangitano ES, Stephenson CG, Liberman RL. Conventional pulse oximetry can give spurious data in a neonatal population at risk for retinopathy of prematurity (ROP). Pediatr Res 1998; 43: 216A.

10 Rheineck-Leyssius AT, Kalkman CJ. Influence of pulse oximeter settings on the frequency of alarms and detection of hypoxemia: theoretical effects of artifact rejection, alarm delay, averaging, median filtering or a lower setting of the alarm limit. J Clin Monit Comput 1998; 14: 151–6. MEDLINE

11 Goldstein MR, Furman GI, Sindel BD et al. SatSeconds alarm management misses short desaturations common to periodic breathing and infantile apnea. Pediatr Res 2001; 49: (4) 400A, 2296.

12 Cabal LA, Zanini B, Yeh SY, Hon EH. Neonatal heart rate variability as a prognostic tool in the management of RDS. Pediatr Res 1975; 9: 394–7.

13 Poets CF, Stebbens VA, Samuels MP, Southall DP. The relationship between bradycardia, apnea, and hypoxemia in preterm infants. Pediatr Res 1993; 34: (2) 144–7. MEDLINE

14 Galland BC, Hayman RM, Taylor BJ, Bolton DPG, Sayers RM, Williams SM. Factors affecting heart rate variability and heart rate responses to tilting in infants aged 1 and 3 months. Pediatr Res 2000; 48: (3) 360–8. MEDLINE

15 Goldstein MR, Furman GI, Pernia ML et al. Performance of motion-resistant pulse oximeters in tracking neonatal heart rate variability. Anesth Analg 2002; 94: (Suppl 1) A5.

16 Zahr LK, Balian S. Responses of premature infants to routine nursing interventions and noise in the NICU. Nurs Res 1995; 44: (3) 179–84. MEDLINE

17 Elander G, Hellstrom G. Reduction of noise levels in intensive care units for infants: evaluation of an intervention. Heart and Lung 1995; 24: 376–9. MEDLINE

18 Long JG, Lucey JF, Philip AGS. Noise and hypoxemia in the intensive care nursery. Pediatrics 1980; 65: 143–5. MEDLINE

19 Fletcher J, Page M, Jeffrey HE. Sleep states and neonatal pulse oximetry. Sleep 1998; 21: 305–10. MEDLINE

20 Brouillette RT, Lavergne J, Leimanis A, Nixon GM, Laden S, McGregor CD. Differences in pulse oximetry technology can affectdetection of sleep disorders in children. Anesth Analg 2002; 94: (Suppl 1) S47–53. MEDLINE

21 Kopotic RJ, Lindner W. Assessing high-risk infants in the delivery room with pulse oximetry. Anesth Analg 2002; 94: (Suppl 1) S31–6. MEDLINE

22 Goldstein MR, Liberman RL, Taschuk RD, Thomas A, Vogt JF. Pulse oximetry in transport of poorly perfused babies. Pediatrics 1998; 2: 818.

23 AAP Committee on Injury and Poison Prevention. Safe transportation of newborns at hospital discharge. Pediatrics 1999; 144: 986–7.

24 Sahni R, Gupta A, Rosen TS. Motion-resistant pulse oximetry during circumcision in neonates. Pediatr Res 2000; 47: (4) 430A, 2544.

25 STOP-ROP: The STOP-ROP Multicenter Study Group. Supplemental therapeutic oxygen for prethreshold retinopathy of prematurity (STOP-ROP), a randomized, controlled trial. Pediatr 2000; 105: 295–310.

Figures

Figure 1 A 20-minute plot of SpO₂ (upper panel) and PR and ECG HR (lower panel) tracings. Points 1 and 2 show data analyzed in Figures 2 and 3, respectively. Point 1 highlights a typical nonmotion state and point 2 a motion condition.

Figure 2 Postprocessed nonmotion data from the Masimo SET instrument at point 1 in Figure 1. The upper panel (A) displays a 2-second window of red and infrared light detection occurring at minute 13.7 in Figure 1. The lower panel (B) is the frequency domain analysis of the above data marked by the solid upright lines. Note the single harmonic of the ECG HR (darkened in the table) with equivalent SpO₂ values.

Figure 3 Postprocessed motion data from the Masimo SET instrument at point 2 in Figure 1. The upper panel (A) displays a 2-second window of red and infrared light detection occurring at minute 11.9 in Figure 1. The lower panel (B) is the frequency domain analysis of above data marked by the solid upright lines. Note the single harmonic of the ECG HR (darkened in the table) with equivalent SpO₂ values.

Figure 4 Saturation (upper panel) and PR and ECG HR (lower panel) tracings during simultaneous monitoring with four new-generation pulse oximeters. Points 1 and 2 shows frozen SpO₂ and PR as well as a data drop-out during a true hypoxemic and associated bradycardic event.

Tables

Table 1 Performance of the Masimo SET (New-generation) and Nellcor N-200 (Conventional) Pulse Oximeters

Table 2 Incidence of Performance Problems of Four New-generation Pulse Oximeters

July/August 2002, Volume 22, Number 5, Pages 360-366

Table of contents    Previous  Article  Next    [PDF]