Pediatric Pulmonology 37:76–80 (2004)

Diagnostic and Therapeutic Methods

Sleep Desaturation: Comparison of Two Oximeters

Ha Trang, MD, PhD,1,2* Souham Boureghda, MD,1 and Vivian Leske, MD1

Summary. Oxygen saturation is measured by pulse oximetry during sleep studies. Body

movements and peripheral vasoconstriction related to respiratory events may interfere with

measurements byconventional oximeters. Our objective was to compare the detection rate of sleep

desaturations by two oximeters, one of which used new motion-resistant technology. We studied

34 children (median age, 13 years; range, 3–18) with suspected sleep-disordered breathing.

During polysomnography, oxygen saturation was measured by two oximeters set on fast mode: the

motion-resistant Radical oximeter (2-sec averaging), and the conventional Nellcor N-200 oximeter

(2–3-sec averaging). Respiratory events were identified based on airflow signal. The numbers of

respiratory event-related desaturations�3% or�5% detected by each oximeter were determined.

Valid desaturations were defined using the Nellcor plethysmographic waveform and the Radical

signal-quality data. Hypoxemic respiratory events were those with associated valid desaturation. In

total, 1,278 respiratory events were identified and pooled. Basal oxygen saturation measured just

before event onset was not different between oximeters (Radical: 98%; range, 84–100; Nellcor:

97%; range, 86–100; P¼ns). However, the Radical detected a greater number of valid desa-

turations than did the Nellcor for any level of desaturation (respectively, N¼ 651 and 476

desaturations�3%,P< 0.001; and N¼232 and 146 desaturations�5%,P¼ 0.01). Consequently,

for each patient, the number of hypoxemic respiratory events per hour of sleep was greater using

the Radical than using the Nellcor (P¼0.002, and P¼0.021, for desaturation �3% and �5%,

respectively). In conclusion, standardized oximeter settings are required to achieve more accurate

assessments of hypoxemia in children with sleep-disordered breathing. Pediatr Pulmonol. 2004;

37:76–80. � 2004 Wiley-Liss, Inc.

Key words: desaturation; oximetry; sleep apnea; child.

INTRODUCTION

Arterial oxygenation is commonly measured by non-invasive pulse oximetry (SpO2). The technology used inconventional pulse oximeters rests on the assumption thatarterial blood is the only light-absorbing pulsatile com-ponent in the optical path. The Nellcor-200 (Nellcor,Pleasanton, CA) is a conventional oximeter used in manypediatric sleep centres. However, motion imparts pulsa-tions to nonarterial blood components, and may thereforeaffect arterial saturation measurement by conventionaloximeters.1 Recent advances in signal processing underlow signal-to-noise conditions were used to develop a newgeneration of pulse oximeters.2 One of these is the Radicaloximeter (Masimo Corp., Irvine, CA), which was shown toaccurately measure arterial oxygenation during exercise-induced motion3 and to reduce the number of false alarmswhen used in intensive and postanesthesia care units.4–7

This new oximeter was used recently to obtain referencevalues for nocturnal pulse oximetry in young children.8

Sleep-disordered breathing (SDB) in children is char-acterized by brief respiratory events (REs) which mayinduce transient hypoxemia.9–11 Although there is no re-ference technique to which pulse oximeters can be com-pared, all sleep centers use data derived from pulseoximetry, both in clinical and research settings. We hypo-thesized that new pulse oximetry technology may improvethe detection of brief hypoxic episodes in children with

1Service de Physiologie, Hopital Robert Debre, Paris, France.

2INSERM E-9935, Universite Paris VII, Paris, France.

*Correspondence to: Ha Trang, M.D., Ph.D., Service de Physiologie,

Hopital Robert Debre, 48 Boulevard Serurier, 75019 Paris, France.

E-mail: ha.trang@rdb.ap-hop-paris.fr

Received 1 March 2003; Revised 18 July 2003; Accepted 20 July 2003.

DOI 10.1002/ppul.10411

Published online in Wiley InterScience (www.interscience.wiley.com).

� 2004 Wiley-Liss, Inc.

suspected SDB. The aim of this study was to evaluate theextent to which differences in oximeter technology mightaffect detection of sleep desaturation. Although chronicepisodic hypoxemia was shown to potentially impaircognitive and cardiac functions,9 it is not clear whether acertain degree of desaturation from a normal baselinewithout associated respiratory event or arousal or changein sympathetic tone may be of any clinical importance,and whether detecting these desaturations is an advantage.For this reason, we chose in this study to analyze RE-related desaturations only. SpO2 was measured simulta-neously using Radical and Nellcor oximeters during sleepstudies.

PATIENTS AND METHODS

Patients

Thirty-four consecutive children referred to our depart-ment for polysomnography to evaluate a clinical suspicionof SDB were studied. There were 20 boys and 14 girls,with a median age of 13 years (range, 3–18). Underlyingdisorders included enlarged tonsils (n¼ 14), obesity(n¼ 16), laryngomalacia (n¼ 2), and craniofacial mal-formations (n¼ 2). The study protocol was approved bythe appropriate ethics committee, and informed consentwas obtained from the parents.

Polysomnography (PSG)

Attended overnight PSGs were performed usingstandard techniques, as recommended by consensus.10

Furthermore, airflow was measured by a nasal cannula-pressure transducer (Protech, Mukilteo, WA), as describedin a previous study.12 All data were recorded using acomputerized PSG system (Alice, Respironics).

In addition, SpO2 was recorded by two pulse oximeterssimultaneously. Both were set on fast mode: the Radicaloximeter (2-sec averaging time) and the conventionalNellcor-200 oximeter (2–3-sec averaging time). Eachoximeter probe was chosen to match finger size, and wasattached to a finger selected at random. A Masimo probe(LNOP) was connected to the Radical oximeter, and aNellcor probe (Oxisensor II) was connected to the Nellcoroximeter. Probes were wrapped in dark disposable foamto eliminate sensor-to-sensor optical cross-talk and toprovide shielding from extraneous light. SpO2 values ofboth oximeters, as well as the Nellcor plethysmographicsignal, were recorded directly onto the PSG system. In

addition, data for Radical signal quality (perfusion andenvironment noise) were collected using DPLOG soft-ware (Masimo, Inc.)8 onto a second computer, which wassynchronized with the PSG system (�1 sec).

Scoring

On-screen scoring was performed using different passesof PSG tracings. Sleep was staged on the first pass.10

On the second pass, REs were determined based on theairflow amplitude measured by the nasal cannula-pressuretransducer:12 apnea was determined as an absence of air-flow signal;10 since no consensus definition exists forhypopnea, we scored hypopnea in this study as a 50% orgreater decrease in peak airflow compared to baseline,associated with changes in any other respiratory signaland/or associated with movement/arousal, as described inprevious studies.12,13 No attempt was made to differentiatecentral from obstructive REs. We considered only REslasting longer than 5 sec, which is the duration of 2 or 3breaths in children.14 SpO2 were deleted from the com-puter screen during this phase of scoring.

On the third pass, recordings were reviewed to matchSpO2 data with REs. SpO2 was analyzed during the RE andfirst 30 sec that followed, for determination of basal SpO2

(mean of 3 values before onset of RE), lowest SpO2, anddepth of desaturation (difference between basal and lowestSpO2 values). Desaturation events were not analyzedwhen SpO2 values were below 50%. Hypoxemia wasdefined as desaturation �3% and �5%, the level used inmost pediatric studies of sleep.10,11 Valid desaturationswere identified for the Nellcor as a plethysmographicwaveform signal regular in amplitude and rate,15 and forthe Radical, based on signal-quality data (perfusion >1%and environment noise >10,000). Artifactual desatura-tions were those that did not meet these criteria.

Analysis

First, all REs were pooled and analyzed as a whole. Wedetermined the numbers: 1) of RE-related detected de-saturations, and 2) valid desaturations �3% or �5% byeach oximeter. Second, for each patient, the number ofhypoxemic REs per hour of sleep was calculated. Hypo-xemic REs were those REs with concomitant validhypoxemia.

Statistical Analysis

Data are shown as medians and range, unless otherwisespecified. Differences between oximeters were assessedusing Wilcoxon signed-rank tests. All analyses wereperformed with SPSS software (SPSS, Chicago, IL).P< 0.05 was considered statistically significant.

ABBREVIATIONS

PSG Polysomnography

RE Respiratory event

SDB Sleep-disordered breathing

SpO2 Arterial oxygenation measured by pulse oximetry

Desaturation Detection During Sleep 77

RESULTS

Detection of Desaturation for All REs Pooled

In total, 1,278 REs were identified and pooled, 43% ofwhich occurred during rapid-eye movement sleep. Medianduration of REs was 11 sec (range, 5–41). Many REsoccurred as a succession of events with repetitive phasesof desaturation and reoxygenation; thus the basal SpO2

value determined arbitraly at the onset of many REs waslow. However, no significant differences were foundbetween oximeters for basal SpO2 measured just beforeRE onset (Radical: 98%; range, 84–100; Nellcor: 97%;range, 86–100; P¼ ns) or for the lowest SpO2 values(Radical: 95%; range, 72–98; Nellcor: 94%; range, 74–98; P¼ ns).

The numbers of detected desaturations �3% were 664by the Radical and 483 by the Nellcor (P< 0.001, Fig. 1).Among these, 349 desaturations were detected by bothoximeters, all 349 being valid desaturations. There were127 artifactual desaturations by the Nellcor and only 13 bythe Radical (P< 0.001, Figs. 1 and 2). Moreover, manydesaturations were detected by the Radical oximeter only,and not by the Nellcor (Fig. 3). Consequently, the numberof RE-related valid desaturations�3% by the Radical wassignificantly greater than that by the Nellcor (Radical: 651REs; Nellcor: 476 REs; P< 0.001).

Similar results were found for desaturation �5%. Thetwo oximeters detected a comparable number of desatura-tions (N¼ 252 and 243, for the Radical and the Nellcorrespectively, P¼ ns), among which 112 were valid de-saturations detected by both oximeters. Artifactualdesaturation occurred more frequently with the Nellcorthan the Radical (respectively, N¼ 115 and 10, P¼ 0.01).Thus, the Radical detected more valid desaturations �5%than the Nellcor (N¼ 232 and 146, respectively,P¼ 0.01).

Determination of Number of HypoxemicREs for Each Patient

The study patients presented with a number of REsranging from 0.1–19 per hour of sleep (median value, 4),and the longest REs lasting from 13–41 sec (median,26 sec). Median total sleep time was 7 hr (range, 6.6–9.5),with 18% of rapid-eye movement sleep (range 12–23%).SpO2 values during wakefulness were not different

Fig. 1. Number of valid and artifactual desaturations �3%

detected by Radical only, by Nellcor only, and by both oximeters,

among 1,278 REs analyzed.

Fig. 3. PSG tracing showing obstructive hypopnea, associated

with desaturation of 4% detected by Radical and not Nellcor

oximeter. For abbreviations, see Figure 2; @, validated respira-

tory event.

Fig. 2. PSG tracing showing obstructive apnea followed by

prolonged period of awakening and body movements. Valid

desaturation was detected by Radical, whereas artifactual

desaturation occurred with Nellcor (due to its highly irregular

plethysmographic signal waveform). C4A2, O2A2, electroence-

phalogram; REOG, right electrooculogram; Chin, submental

electromyogram; ECG, electrocardiogram; NBF, nasobuccal

airflow using thermistor; NF, nasal airflow using a nasal

cannula-pressure transducer; THO, ABD, respiratory move-

ments of thorax and abdomen; RAD, Radical SpO2; N200F,

Nellcor-200 SpO2; O2g, Nellcor plethysmographic signal; CO2g,

end-tidal CO2 waveform; time in seconds; @, validated respira-

tory event.

78 Trang et al.

between oximeters (Radical: 98%; range, 97–100;Nellcor: 97%; range, 96–100; P¼ ns).

For each patient, the number of hypoxemic REs perhour of sleep (i.e., REs associated with valid desatura-tions) was greater with the Radical than with N-200oximeters, for both levels of desaturation (P¼ 0.002 fordesaturation �3% (Fig. 4) and P¼ 0.021 for desaturation�5%).

DISCUSSION

This study shows that, in children with suspected SDB,the Radical oximeter detected a greater number of RE-related valid desaturations than did the conventionalNellcor-200. Artifactual desaturations were more frequentwith the Nellcor than with the Radical. Consequently, foreach patient, the number of hypoxemic REs per hour ofsleep was greater with the Radical than with the Nellcor,for both levels of desaturation �3% or �5%.

Indeed, oximetry signals were taken from probes placedat different sites, and this may have influenced the agree-ment between devices. However, we previously showed,in a preliminary study, that no differences existed in pulseoximetry with changing sensor site positions.17

In the present study, both oximeters were set on fastmode, as short averaging time on conventional pulseoximeters was shown to maximize detection of transientdesaturation during sleep.16 Although there was a minordifference in averaging between the two tested oximeters,with the Radical averaging over 2 sec and the Nellcor over2–3 sec, it is unlikely that this could account for the higherrates of detection by the Radical.

The most likely explanation for differences betweenoximeters is that the Radical, with its new signalprocessing technology, is more likely to detect moderate,transient desaturation than is the Nellcor, under challen-ging conditions.2 Children with SDB may present withfrequent body movements and arousals, associated with orterminating REs, with a pattern of fragmented and restlesssleep.18 The present study provides further evidence ofmotion artifact resistance of the Radical oximeter, inaccordance with previous investigations.3–8 In contrast,losses of oximetry signals are likely to occur with theNellcor at critical moments, caused by body movementsthat follow REs during sleep. This may produce eitherzero-out signals, or false-negative desaturations, or on thecontrary, false-positive desaturations. In any case, thisleads to missing and erroneous oximetry results for sleepstudies. Furthermore, it was shown that obstructive apneasinduce profound, transient vasoconstriction at the fingers,where pulse oximeter probes are usually placed,19,20 andthat vasoconstriction is further accentuated by hypoxia.20

We found in this study that many desaturations weredetected by the Radical but not by the Nellcor oximeter.Therefore, RE-induced vasoconstriction combined withbody movements likely alter pulse oximetry signal-to-noise ratios, and thereby interfere with oxygen saturationmeasurement by conventional oximeters during sleepstudies. These factors likely account for both the higherrate of artifactual desaturations by the Nellcor and thelower rate of valid desaturations, as compared to the newRadical oximeter.

In the clinical setting, the history of hypoxemia is amajor criterion upon which many management decisionsare predicated. Various desaturation levels and indices areused to assess the hypoxemia burden.8,10 Determining thedesaturation index values that impact the clinical assess-ment and care of children with SDB is beyond the scope ofthis study. As the technical characteristics of oximetersused during sleep studies may influence desaturation indexvalues, standardized device settings and scoring criteriaare required.

Useful reference values for nocturnal oximetry wererecently established, using the Radical oximeter in youngchildren.8 However, many differences in techniques ofrecordings (home-setting, no sleep recording) and analysis(no respiratory event scoring) precluded any comparisonbetween these data and ours.

Early studies of pediatric sleep counted all REs lastingat least 5 sec, which is the duration of 2 or 3 breaths inchildren.14 However, of the two studies that reportednormative data in children and adolescents, one count-ed apneas of any duration,21 whereas the other countedapneas and hypopneas longer than 10 sec only.22 We foundin the present study that a substantial proportion of briefREs was associated with valid desaturation. The clinicalrelevance of scoring brief REs in pediatric SDB should

Fig. 4. Comparison of numbers of REs with valid desaturation

�3% per hour of sleep between two oximeters (N¼34 patients).

Desaturation Detection During Sleep 79

be assessed in multicenter trials to determine the optimalcutoff values for these respiratory variables.

In summary, we show in this study that the Radical pulseoximeter detected more valid desaturations than did theconventional N-200 oximeter during polysomnographies.Standardization of oximeter settings is required to im-prove the comparability of sleep studies done for bothclinical and research purposes.

REFERENCES

1. Wukitsch MW, Petterson MT, Tobler DR, Pologe JA. Pulse

oximetry: analysis of theory, technology, and practice. J Clin

Monit 1988;4:290–301.

2. Goldman JM, Petterson MT, Kopotic RJ, Barker SJ. Masimo

signal extraction pulse oximetry. J Clin Monit 2000;16:475–

483.

3. Yamaha Y, Bogaard HJ, Wagner PD, Niizeki K, Hopkins SR.

Validity of pulse oximetry during maximal exercise in normoxia,

hypoxia, and hyperoxia. J Appl Physiol 2002;92:162–168.

4. Dumas C, Wahr JA, Tremper KK. Clinical evaluation of a proto-

type motion artifact resistant pulse oximeter in the recovery room.

Anesth Analg 1996;83:269–272.

5. Barker SJ, Shah NK. The effects of motion on the performance of

pulse oximeters in volunteers. Anesthesiology 1997;86:101–108.

6. Bohnhorst B, Peter CS, Poets CF. Pulse oximeters’ reliability in

detecting hypoxemia and bradycardia: comparison between

Nellcor N-200, N-3000 and Masimo SET. Crit Care Med 2000;

28:1565–1568.

7. Malviya S, Reynolds PI, Voepel-Lewis T, Siewert M, Watson D,

Tait AR, Tremper K. False alarms and sensitivity of conventional

pulse oximeters versus the Masimo SET technology in the

pediatric postanesthesia care unit. Anesth Analg 2000;90:1336–

1340.

8. Urschitz MS, Wolff J, von Einem V, Urschitz-Duprat PM, Schlaud

M, Poets C. Reference values for nocturnal home pulse oximetry

during sleep in primary school children. Chest 2003;123:96–101.

9. Marcus CL. Sleep-disordered breathing in children. Am J Respir

Crit Care Med 2001;164:16–30.

10. American Thoracic Society. Standards and indications for

cardiopulmonary sleep studies in children. Am J Respir Crit Care

Med 1996;153:866–878.

11. American Thoracic Society. Cardiorespiratory sleep studies in

children. Establishment of normative data and polysomnographic

predictors of morbidity. Am J Respir Crit Care Med 1999;160:

1381–1387.

12. Trang H, Leske V, Gaultier C. Use of nasal cannula for detecting

apneas and hypopneas in infants and children. Am J Respir Crit

Care Med 2002;166:464–468.

13. Mograss MA, Ducharme FM, Brouillette RT. Movement-arousal.

Description, classification, and relationship to sleep apnea in

children. Am J Respir Crit Care Med 1994;150:1690–1696.

14. Carskadon MA, Harvey K, Dement WC, Guilleminault C,

Simmons B, Anders TF. Respiration during sleep in children.

West J Med 1978;478:477–481.

15. Lafontaine VM, Ducharme FM, Brouillette RT. Pulse oximetry:

accuracy of methods of interpreting graphic summaries. Pediatr

Pulmonol 1996;21:121–131.

16. Farre R, Montserrat JM, Ballester E, Hernandez L, Rotger M,

Navajas D. Importance of the pulse oximeter averaging time when

measuring oxygen desaturation in sleep apnea. Sleep 1998;21:

386–390.

17. Trang H, Leske V, Boureghda S, Gaultier C. Masimo SET pulse

oximeter improves detection of sleep apnea-related hypoxemia.

Am J Respir Crit Care Med 2001;163, A 298.

18. Scholle S, Zwacka G. Arousals and obstructive sleep apnea

syndrome in children. Clin Neurophysiol 2001;112:984–991.

19. Imadojemu VA, Gleeson G, Gray KS, Sinoway LI, Leuenberger

UA. Obstructive apnea during sleep is associated with peripheral

vasoconstriction. Am J Respir Crit Care Med 2002;165: 61–66.

20. Leuenberger UA, Hardy JC, Herr MD, Gray KS, Sinoway LI.

Hypoxia augments apnea-induced peripheral vasoconstriction in

humans. J Appl Physiol 2001;90:1516–1522.

21. Marcus CL, Omlin KJ, Basinki DJ, Bailey SL, Rachal AB, Von

Pechmann WS, Keens TG, Ward SL. Normal polysomnographic

values for children and adolescents. Am Rev Respir Dis 1992;

146:1235–1239.

22. Acebo C, Millman RP, Rosenberg C, Cavallo A, Carskadon MA.

Sleep, breathing, and cephalometrics in older children and young

adults. Part I—normative values. Chest 1996;109:664–672.

80 Trang et al.