Article Text
Abstract
Objective To determine reference values for pulse oximeter saturation (SpO2) variables and desaturation event indices in healthy term neonates during their first 5 days of life, and to compare two definitions for the identification of desaturation events.
Design Observational study (case series).
Setting Maternity ward, Department of Neonatology, University Children's Hospital, Tuebingen, Germany.
Patients 209 healthy term neonates (50% boys), median (minimum–maximum) age 2 (0–5) days.
Main outcome measures SpO2 variables (eg, median SpO2) and desaturation events obtained by motion-resistant pulse oximetry (VitaGuard 310; Getemed, incorporating Masimo SET). Desaturation events were identified based either on a good signal quality (SIQ) provided by the device or on the combination of a good SIQ and an undisturbed pulse waveform (SIQ+PW). Desaturation event indices were calculated as desaturation events divided by hour of artefact-free recording time.
Results The mean (SD) of the obtained median SpO2 was 97.3% (1.4%). There were 36 (17%) subjects with desaturation events to <80% SpO2 based on SIQ, and 26 (12%) based on SIQ+PW. Median desaturation event rate to <80% SpO2/h (75th centile; 95th centile; maximum) was 0 (0; 0.6; 2.3) based on SIQ, and 0 (0; 0.4; 1.7) based on SIQ+PW.
Conclusions Desaturation events to <80% SpO2 were rare in our sample of healthy term neonates during their first 5 days of life. Analysis of SIQ alone could be a quick and simple alternative to traditional analysis of PW. The presented reference values may be used for clinical decision making.
Statistics from Altmetric.com
Introduction
Indications for supplemental oxygen for neonates remain a frequent concern. Such treatment decisions are often based on oxygen saturation measured by pulse oximetry (SpO2). To interpret such data, reference values for SpO2 in healthy term neonates are needed.
Reference data on SpO2 in neonates are limited, as is their interpretation. Most studies involved only a small number of subjects,1,–,3 were performed at high altitude4 5 or included only spot measurements of SpO2.6 Some studies did not exclude spuriously low values caused by motion.6 7 In addition, the introduction of motion-resistant oximetry may require the establishment of a new set of reference data.8
What is already known on this topic
▶ Desaturation events to <80% SpO2 may occur in healthy term neonates.
▶ SpO2 tends to become more stable with increasing postnatal age.
What this study adds
▶ Reference values for SpO2 data and desaturation events using a motion-resistant technology in a large cohort of healthy newborns.
▶ Analysis of desaturation events based on a device-specific signal quality indicator yields results that are somewhat higher than those obtained by operator-based analysis.
Regarding desaturation events, previous studies used time-consuming and operator-dependant visual assessment of the pulse waveform (PW) associated with these events.9 Identification of such events based only on signal quality (SIQ; provided by the device) may offer a simple, valid and time-saving alternative. No study, however, has yet compared these two analytical methods.
Embedded in a larger study on SpO2 in term neonates,10 we aimed to determine reference values for SpO2 data and desaturation events using a motion-resistant pulse oximeter in a large sample of healthy neonates during their first 5 days of life, and compare PW to SIQ analysis of desaturation events.
Methods
Subjects
The recruitment strategy and basic characteristics of the original study sample are described elsewhere.10 In summary, pulse oximetry recordings were performed in 476 term neonates (ie, born at ≥37 weeks gestational age) aged 0–5 days, born at Tuebingen University Hospital. Of these neonates, 242 (51%) were admitted to the neonatal unit, and 234 (49%) stayed in the maternity ward. Of the neonates staying in the maternity ward, all those suspected of having an underlying condition or illness were excluded (n=25). Hence, only data from the remaining 209 (44% of the original sample) neonates staying in the maternity ward and considered clinically healthy at the time of enrolment were included in this study. Written informed parental consent was obtained for each neonate. The study protocol was approved by the hospital's ethics committee.
Study protocol
Subjects were studied for 12 h overnight and were placed to sleep either in a horizontal supine or side position using a random number allocation procedure.10 After 6 h, they were changed to the alternate position. Sleep position order was recorded. Otherwise, the neonates received routine care including demand feeds. Caregivers were asked to keep a note of feeding periods. For the present study, only data recorded while infants were in a horizontal supine position were included.
Pulse oximetry
A home monitor (VitaGuard 310; Getemed, Teltow, Germany) with a motion-resistant pulse oximeter module (Masimo, Irvine, California, USA; 2–4 s averaging mode) capable of storing trend and event data was used.11 Monitor alarms sounded if SpO2 was <70%, pulse rate was <60 or there were >260 beats per minute. SpO2, perfusion index and SIQ were recorded into trend data memory at a sampling rate of 1 Hz. SIQ, derived from the raw red-to-infrared absorption signal, indicates the signal-to-noise ratio and, thereby, the confidence of the measured SpO2.12
In addition to these data, desaturation events, defined as SpO2 <90%, were recorded separately over a 120 s period prior to and during the event into event data memory. These event data comprised the PW signal, sampled at 32 Hz, in addition to the above-mentioned data. The oximeter updated its memory every second, yielding a minimum event duration of 1 s. This minimum duration and threshold levels were extrapolated from previous work on desaturation events to <80% SpO2 using an oximeter in the beat-to-beat mode, and were considered suitable to trace maturational and disease-related changes in respiratory control during infancy.2 9
Analysis
Recordings were analysed for SpO2 values using data analysis software (Matlab; MathSoft, Cambridge, Massachusetts, USA). SpO2 values from both baseline and desaturation events were included. First, total recording time was registered and artefact-free recording time (AFRT) calculated. For calculation of AFRT, feeding periods and episodes with poor signal identified by the data analysis software (ie, SIQ ≤0.3) were subtracted from total recording time. Then, the following values were calculated for each recording using the above-mentioned analysis software: median SpO2 and the SpO2 below which the child spent 10% and 5% of AFRT.
Desaturation events to <90%, <85% and <80% SpO2 were visually analysed by an investigator using software provided by the device manufacturer (VitaWin, v 2.3.2; Getemed) using two different definitions: (1) SIQ >0.3 and no low perfusion tag within 7 s prior to the desaturation event (these 7 s correspond to the signal processing time of the monitor; definition ‘SIQ’); and (2) SIQ >0.3, no low perfusion tag and an undisturbed PW within 7 s prior to the event (definition ‘SIQ+PW’). Desaturation events with SIQ ≤0.3, a low perfusion tag or a distorted PW within 7 s prior to the event (the latter only for definition 2), as well as abrupt falls in SpO2 by more than 7% SpO2 were considered artefactual and excluded. Figures 1 and 2 show examples of desaturation events. Based on both definitions, the following desaturation event indices (ie, events per hour of AFRT) were calculated by summing all events identified by the respective method and dividing them by AFRT: desaturation events to <90%, <85% and <80% SpO2 per hour. In addition, the minimum SpO2 value associated with desaturation events identified by SIQ+PW was registered.
Statistics
Descriptive statistics as numbers and percentages, mean and SD, and median/minimum/maximum were used to summarise demographic, clinical and oximetry characteristics. In addition, the 5th and 25th centiles were calculated for SpO2 values, and the 75th and 95th centiles for desaturation event indices. Associations with potentially influencing factors (ie, age and gender) were investigated using linear regression analysis for SpO2 values and logistic regression analysis for desaturation event indices. For the latter analysis, indices were dichotomised into ‘0’ or ‘>0’. Comparison of SpO2 values and desaturation event indices in unassisted versus assisted deliveries was performed using the Mann–Whitney U test. A p value <0.05 was considered statistically significant. All analyses were performed with statistical software (PASW Statistics 18.0 for Windows; IBM, Chicago, Illinois, USA).
Results
Recordings were obtained for 209 healthy subjects (104 boys). Of these, 145 (69%) were born by unassisted vaginal delivery, 54 (26%) by caesarean section and 10 (5%) by assisted vaginal delivery. Median (minimum–maximum) postnatal age at the onset of recording was 2 (0–5) days. One recording was started at 8 h of age, and the rest at more than 12 h after birth. Race was predominantly Caucasian (ie, >95%). Demographic and clinical characteristics are given in table 1. There were no relevant gender-specific differences regarding gestational or postnatal age, type of delivery, birth weight, Apgar scores at 1, 5 and 10 min, or umbilical cord arterial pH.
The mean±SD (minimum–maximum) duration of artefact-free recording time was 4.9±0.8 (3.0–7.1) h. The mean±SD of the obtained median and minimum SpO2 (pulse oximeter saturation) was 97.3±1.4% and 82.9±5.6%, respectively. Table 2 gives descriptive statistics for SpO2 variables. Based on SIQ, there were 193 (92%) infants with desaturation events to <90% SpO2, 102 (49%) with desaturation events to <85% SpO2 and 36 (17%) with desaturation events to <80% SpO2. Based on SIQ+PW, the corresponding figures were 166 (79%), 74 (35%) and 26 (12%).
Descriptive statistics for all desaturation event indices under study are given in table 3. Desaturation event index to <80% SpO2 (based on SIQ+PW) was <0.5 in 202 (97%) subjects, between 0.5 and 1.5 in 5 (2%), and >1.5 in only 2 (1%) infants. Maximum desaturation event index to <80% SpO2 was 2.3 based on SIQ and 1.7 based on SIQ+PW.
Compared to girls, boys showed lower SpO2 values throughout. In boys, median SpO2 and SpO2 below which the child spent 10% and 5% of AFRT were (mean±SD) 97.1±1.5%, 94.6±1.9% and 93.6±2.3%, respectively. In girls, the corresponding figures were 97.5±1.3%, 95.4±1.7% and 94.6±1.8% (all p values <0.05). However, we considered the magnitude of the differences between boys and girls (mostly <1% SpO2) as clinically not relevant and opted against presenting gender-specific reference values. Of all desaturation event indices, only the desaturation event index to <90% SpO2 (based on SIQ+PW) was associated with gender. The odds ratio of having this index >0 was 1.5 (p=0.04) for boys compared to girls.
There were no associations between SpO2 values or desaturation event indices and age.
Subjects born by unassisted delivery had significantly higher median SpO2 (98% vs 97%; p<0.001), but only slightly higher minimum SpO2 values (84% vs 83.5%; p=0.08) than those born by caesarean section. Desaturation event indices were not significantly different between these groups (p>0.05).
Discussion
We established reference values for SpO2 variables and desaturation event indices in 0–5-day-old neonates using motion-resistant pulse oximetry. While desaturation events to <90% SpO2 occurred in the majority of infants, SpO2 rarely fell to <80%. Desaturation event indices were higher if analysis was based on the manufacturer's signal quality indicator compared to additional operator dependant visual analysis based on an undisturbed PW.
To our knowledge, this is the largest study on SpO2 reference values for overnight recordings from healthy 0–5-day-old neonates. In addition to the already known association between age and SpO2,9 the present study is also the first to describe gender-specific differences in SpO2 in this population. As boys showed no clinical or demographic differences to girls, the above-mentioned association cannot be explained. A possibly more immature respiratory pattern in boys could be hypothesised as a physiological explanation. In any case, differences were too small to be considered clinically relevant.
The use of motion-resistant pulse oximetry in this study likely increased the accuracy of our measurements compared to studies involving conventional pulse oximetry. This increased accuracy is supported by Sahni et al, who investigated 15 quietly sleeping infants for 10 min with a motion-resistant pulse oximeter, obtaining significantly higher SpO2 values than with a non-motion-resistant oximeter.13
In the present study, median SpO2 was slightly lower than the so-called ‘baseline’ SpO2 reported in some previous studies.2 9 This is likely related to the fact that baseline SpO2 in those studies was only analysed during regular breathing, at least 10 s away from apnoeas and predominantly in a sleep state with low oxygen consumption. In contrast to that, the median SpO2 includes data from baseline periods and desaturation events, which will decrease the actual value. Automatic software-based determination of SpO2 variables as used here has the advantage of not requiring an additional recording of breathing movements and time-consuming expert analysis, but does require a new set of reference values.
In order to provide comprehensive information on oxygenation in this age group, we determined reference values for median SpO2, minimum SpO2, SpO2 values below which the child spent 10% and 5% of AFRT and desaturation event indices to various threshold levels. In contrast to the high frequency of desaturation events to <90% seen here, desaturation events to <80% SpO2 were much rarer. Previously, one study using beat-to-beat pulse oximetry, but a minimum event duration of 4 s, found that 35% of full-term neonates had desaturation events to <80% SpO2 in recordings obtained during their first week of life.2 These differences between studies underscore the need for standardised recording and analysis techniques for SpO2 data. As almost all included subjects (97%) showed a desaturation event index to <80% (based on SIQ+PW) of <0.5 and only 1% a value >1.5, we propose the following cut-offs for the visual analysis of SpO2 recordings obtained in neonates between 0 and 5 days of age with Masimo SET in 2–4 s averaging mode: a desaturation event index to <80% SpO2 of <0.5 as normal, between 0.5 and 1.5 as borderline, and >1.5 as exceeding the normal range.
We found different results for both desaturation event definitions used here. According to SIQ and SIQ+PW, 36 and 26 subjects, respectively, showed desaturation events to <80% SpO2. This means that in <5% of the total sample, SIQ based analysis identified such events that were not confirmed by additional operator-based assessment of PW. These results seem promising, as an analysis based solely on SIQ may be a relatively simple way to assess desaturation events. The manufacturer claims that its proprietary technology can read through motion, and although we have no non-invasive method to substantiate this claim, there are indications to suggest that SpO2 data obtained with motion-resistant instruments can generally be considered valid, particularly if used in combination with information on signal quality.8 14 Identification of desaturation events based on this method is likely cheaper and more readily available than a visual analysis of PW. Depending on the number of desaturation events, it takes between minutes and a couple of hours to visually analyse a 6 h recording, whereas analysis based exclusively on SIQ is software-based and thus potentially almost immediate. Whichever method is used, we suggest comparing results for desaturation event indices against a reference that has been obtained with the same analytical method, which is why we present data for both analytical methods in this study.
A possible limitation of the present study is that we did not record electroencephalographic or other cardiorespiratory signals. Desaturation event indices were therefore based on total AFRT and not on total sleep time. Nevertheless, we did not intend to determine the effects of different sleep states on SpO2. In addition, we aimed to establish reference values based on a method that is simple, cheap and easy to use. Hence, we feel that unattended SpO2 recordings correspond more closely to routine neonatal care conditions than a full sleep assessment using multiple electrodes. Although we tried to identify and separate periods with feeding and poor signal, it is possible that some low SpO2 values and desaturation events may have been artefactual. Also, desaturation event indices were obtained using 2–4 s signal averaging and may therefore be different from those obtained with longer averaging times often used during routine SpO2 monitoring in the neonatal intensive care unit. Results may thus still be difficult to translate into the routine clinical setting, which should be explored further.
Conclusion
We defined reference values for SpO2 variables and desaturation event indices in a large sample of healthy neonates during their first 5 days of life using motion-resistant oximetry. Our data showed an average median SpO2 of 97.3% and an SpO2 >94% for 95% of the time. Gender was significantly associated with median SpO2 and desaturation event indices to <90% SpO2. Desaturation events to <80% SpO2 were seen in only a minority. These data may help clinicians in deciding which neonates may have an excessive number of desaturation events, or other abnormal SpO2 values, and require further investigation. Future studies should try to identify the consequences of desaturation events.
Acknowledgments
The authors thank the parents and newborns for participating in this study. They also thank Lena Froehling, MD, Gisela Trapp, MD, Florian Nowotny and Viviane Ishak-Lesnik for their help with performing these recordings.
References
Footnotes
PEB and AP contributed equally to this work.
-
Competing interests CFP has served as an adviser to Masimo Inc.
-
Ethics approval This study was conducted with the approval of the ethics committee of the University Children's Hospital, Tuebingen, Germany.
-
Provenance and peer review Not commissioned; externally peer reviewed.