Arterial oxygen saturation in preterm infants at discharge from the hospital and six weeks later

FETAL AND NEONATAL MEDICINE

Arterial oxygen saturation in preterm infants at discharge from the hospital and six weeks later Christian F. Poets, MD,* V a l e r i e A. S t e b b e n s , BSc, John R. A l e x a n d e r , MSc, FSS,FIS, William A. Arrowsmith, FRCP, DCH, S t e p h e n A. W. Salfield, MB,BS,MRCP, DCH, a n d D a v i d P. Southall, MD, FRCP From the National Heart and Lung Institute, Brompton Hospital, London, Syntex Research Maidenhead, Doncaster Royal Infirmary, Doncaster, and Rotherham District General Hospital, Rotherham, England To obtain normal data on arterial o x y g e n saturation (SaO2) in preterm infants and to study early d e v e l o p m e n t a l c h a n g e s in SaO2, we o b t a i n e d overnight t a p e recordings of $a02 and breathing movements in 160 preterm infants at their disc h a r g e from three special care b a b y units (mean gestational a g e at birth 33 weeks; at time of study, 37 weeks). One hundred ten infants (69%) underwent a s e c o n d recording 6 weeks later. Median baseline SaO2 during regular breathing was 99.5% (range 88.7% to 100%) at discharge, and 100% (range 95.3% to 100%) at follow-up (p <0.001). The number of episodes of desaturation, d e f i n e d as a fall in SaG2 to __<80%for at least 4 seconds, c o r r e c t e d to the mean duration of recording (12.2 hours), d e c r e a s e d from a m e d i a n of 3 (0 to 355) to 0 (0 to 17) (p <0.001). The median duration of e a c h e p i s o d e of desaturation remained unc h a n g e d (5.2 [4.0 to 22.7) vs 5.5 [4.2 to 24.0] seconds). At discharge, a small minority of infants had a clinically unrecognized low baseline SaG2 (lowest, 88.7%; 5th percentile, 95.7%) or a high number of desaturation episodes (the highest was six times the 95th percentile value). At follow-up, all outlying values had normalized. Follow-up recordings m a d e b e t w e e n 42 and 47 weeks of gestational a g e (n = 53) were c o m p a r e d with similar recordings from 67 term infants at the same gestational age. The preterm infants had a significantly higher baseline $aO2 and no more desaturation than the infants born at term. Knowle d g e of normal ranges of o x y g e n a t i o n and their c h a n g e s with a g e may b e of value in identifying clinically u n d e t e c t e d h y p o x e m i a in preterm infants at disc h a r g e from the hospital. The potential influence of such h y p o x e m i a on clinical o u t c o m e remains to be determined. (J PEDIATR1992;120:447-54)

Supported in part by the Deutsche Forschungsgemeinschaft, Bonn, Germany (Dr. Poets), by the Moorgate Trust (Mrs. Stebbens), by Hazelton, U.K. Ltd. (Mr. Alexander), and by Nellcor Inc., Hayward, Calif., and the Foundation for the Study of Infant Death (Dr. Southall). Submitted for publication Aug. 2, 1991; accepted Sept. 26, 1991. Reprint requests: D. P. Southa!l, MD, FRCP, Department of Paediatrics, National Heart and Lung Institute, Dovehouse St., London SW3 6HP, United Kingdom. 9 9Now at the Department of Pediatrics, Medizinische Hochschule Hannover, Hannover, Germany.

9/23/34027

N o r m a l data concerning arterial oxygen saturation in infants are sparse. In two recent studies we analyzed baseline SaO2, the frequency of falls in SaO2 to 80% or less, and GA SaO2 SIDS

I

Gestational age Arterial oxygen saturation Sudden infant death syndrome

[

I

their association with apneic pauses in preterm and term infants 1,2by means of a validated pulse oximeter in the beat-to-beat mode. 3, 4 The preterm infants, all of whom

447

448

Poets et al.

The Journal of Pediatrics March 1992

Table I. Clinical data of all infants and of the subgroup with follow-up recording Discharge, all (n No. of male/female infants GA at birth (wk) Birth weight (gm) GA at recording (wk) Postnatal age at recording (days) Weight at recording (gin) No. (%) of infants having had RDS No. (%) of infants having had IPPV No. (%) of infants having had Fio2 >0.21 for >28 days No. (%) of infants with history of AOP

90/70 32.8 • 2.5 1844 • 481 36.6 • 2.1 26.3 • 24.0 2228 • 359 84 (53) 40 (25) 11 (7) 41 (26)

=

160)

Follow-up (n

=

tt0)

58/52 32.8 • 2.5 1843 • 494 42.9 • 3.3 70.3 • 29.4 3658 • 787 58 (53) 29 (26) 7 (6) 24 (22)

Results are mean _+SD except as indicated;percentagegivenin parentheses. RDS, Respiratorydistress syndrome;IPPV, intermittentpositive-pressureventilation;Fio2,inspiredoxygenfraction;AOP, apneaof prematurity.

were studied shortly after term (>__37 weeks of gestational age), had a significantly lower baseline SaO2 and a higher frequency and duration of episodic falls in SaO2 than did the infants born at term. Data comparability was limited, however, because the preterm infants were of lower GA at recording. In addition, these studies did not address (1) how baseline SaO2 and the frequency of desaturation change with age, (2) whether premature birth is an independent risk factor for a low baseline SaO2 or frequent desaturation, and (3) how long an outlying result will persist in an infant. To answer these questions and to provide data representative for most preterm infants at discharge from the hospital, the previously described study2 was enlarged by including (1) recordings from infants who had not reached term at the time of the study, (2) recordings from those who had been studied within 3 days after discharge, and (3) follow-up recordings, which were obtained from two thirds of the infants at about 6 weeks after discharge. METHODS Patients. During a 1-year period (August 1986 to July 1987), the parents of all preterm infants born at three maternity hospitals and admitted to special-care baby units immediately after birth were approached for permission to obtain an overnight tape recording of physiologic variables on their baby at discharge. Parents were approached only if the infant had a good prognosis for survival. A total of 305 infants were available, and in 261 cases (86%) parental consent was given for this investigation. For reasons of homogeneity, only those 175 infants (67%) from whom recordings were obtained within the 3 days before or after discharge were included in this study. After parental permission was given, 113 of the 175 infants also had a follow-up recording at home. Eighteen recordings (15 discharge, 3 follow-up) were of poor signal quality throughout and were removed from the study. Thus the study involved analysis of 160 recordings obtained at discharge from hos-

pital and 110 follow-up recordings obtained at home about 6 weeks later. In addition, a comparison was made between those 53 of the 110 follow-up recordings obtained at a GA of 42 to 47 weeks and the recordings from 67 term infants of similar GA studied with the same method.1 The project, including the decision not to use the data collected to influence the treatment of the infants, was approved by the hospitals' ethics committees. Clinical data were obtained from the discharge summaries. Mean GAs and birth weights of all 160 infants studied at discharge were virtually identical to those of the 110 from whom a follow-up recording was obtained (Table I). The 160 infants studied at discharge, however, had lower GAs at birth and birth weights than did those (n = 145) who were not enrolled in the study (mean GA 32.8 _+ 2.5 weeks vs 33.1 _+ 2.4 weeks, p = 0 . 0 4 ; mean birth weight 1844 + 481 gm vs 1930 _+ 544 gm, p <0.01). These differences were caused by difficulties in obtaining recordings within 3 days before or after discharge and the higher refusal rate among parents of infants who were born at higher GAs and who therefore spent only a brief period in special care. Thirty-five infants were born at less than 32 weeks of gestation, 49 were born at 32 to 33 weeks, 35 were born at 34 weeks, and 41 were born at 35 to 36 weeks. Prematurity-related problems were found predominantly in the group of infants born at less than 32 weeks of gestation, which contained 10 of the 11 infants who had received supplemental oxygen for more than 28 days, 26 of the 40 who had received mechanical ventilation, and 21 of the 41 who had had apnea of prematurity (defined as episodes of apnea [>20 seconds], bradycardia [heart rate <100 beats/min], color change as determined by nursing observations and standard ICU monitors, or a combination of these signs). When the recordings were made at discharge from the hospital, the following symptoms were reported: nasal obstruction (n = 2), skin infection (thrush or staphylococcal

Volume 120 Number 3

Sao2 in preterm infants

Baseline

SaO2

(%) []

100

0

[]

98

~

~

L

IT~J~-A-- -5O .~oo& $

[]

O

[]

[]

Do 0

n

oa .o

0

& ~

[]

0 0

O

96

~ v-

-

[]

449

D0

[]

<32 w e e k s GA

*

32-33weeks GA

A

34 w e e k s GA

o

35-36 w e e k s G A

A O&

94

92

90 n

88

. 32

34

.

. 36

. 38

.

.

40

. 42

. 44

.

GA

46

48

(weeks)

50

Fig. t. Baseline SaO2 in recordings made at discharge (n = 160), plotted against GA at recording (r = -0.1; p = 0.3). Horizontal line represents 5th percentile for baseline SaO2 in all 160 recordings (95.7%).

T a b l e II. Results regarding oxygenation for both entire group and follow-up group at discharge and for subgroup at follow-up Discharge, all (n = t60)

Baseline Sao2 (%) SD of individual baseline measurements No of recordings with episodes of desaturationt Episodes of desaturation per recording Median duration of desaturation episodes (sec) Desaturation episodes (%) with apneic pause Desaturation episodes (%) in periodic apnea

99.6 0.5 114 3.0 5.3 83 67

(88.7-100) (0.0-3.2) (71) (0-355) (4.0-22.7) (0-100) (0-100)

Discharge, follow-up (n = tt0)

99.5 0.6 82 3.0 5.2 83 60

(88.7-100) (0.0-3.2) (75) (0-355) (4.0-22.7) (0-100) (0-100)

Follow-up (n = tt0)

100 (95.3-100)* 0.1 (0.0-1.8)* 34 (31)* 0.0 (0-17)* 5.5 (4.2-24.0) 100 (0-100) 0 (0-100):~

Results are medians with ranges in parentheses. *p <0.0001, differencebetween discharge and follow-upby McNemar chi-square or Wilcoxonmatched-pairs tests, as appropriate. tNumber in parentheses represents percentage of recordings. :~p <0.05, differencebetween discharge and follow-upby Wilcoxonmatched-pairs test.

infection, n = 4), and episodes of bradycardia (n -- 2). For these symptoms, four infants were receiving systemic antibiotics and one was receiving theophylline. Five infants (3%) had congenital heart lesions: one had a ventricular septal defect, one had an aortopulmonary window, and three had a patent ductus arteriosus; one of these last was receiving furosemide. A t the time of follow-up recordings, 14 infants had a nasal obstruction or a cold, five had a skin infection, and three had colic; six of these were receiving systemic antibiotics and two were receiving nasal decongestants. Data concerning the infants' survival were available from the National Health Service Register at the Office of Population Censuses and Surveys. Four infants have subse-

quently died. Two deaths were certified as resulting from sudden infant death syndrome (both at 7 months of age; neither patient had used a home monitor). One patient died at 5 months of age in the postoperative period after closure of a ventricular septal defect, and one died at 11 months Of age in status asthmaticus. Measurement of data. We obtained overnight tape recordings of SaO2 (Nellcor N 100, with new software equivalent to N200 and modified to provide beat-to-beat data; Nellcor Inc., Hayward, Calif.), photoplethysmographic waveforms from the oximeter for the validation of the SaO2 signal, and breathing movements through a volume-expansion capsule (Graseby M e d i c a l / R i g e l Medical Electronic, Millersville, Md.) or inductance plethysmography (Studley

450

Poets et al.

The Journal o f Pediatrics March 1992

N Desaturati0ns 1000

@ []

1,1

100

A @

Ln

[] . 10

ae [] f o & ~ao & O~

O

Lo~

O

[]

o o% In ~

oo~

l~m q []

0 0

@

r

A

AmOr

0 ~ 00, a

(~

0 0

0

[]n

I

0 ! 32

,-" ~ 34 36

38

40

. 42

.

. 44

.

. 46

GA (weeks) 48

50

Fig. 2. Numberofepisodesofdesaturationperrecordingatdischarge(n= 160), corrected to a mean duration of recording of l 2.2 hours and plotted against GA at recording (r = -0.2; p <0.05). Note logarithmic scale on y-axis. Symbols as in Fig. 1. Horizontal line represents 95th percentile for all recordings (61 episodes per 12.2 hours of recording).

Data Systems). One hundred fifty-fiVe recordings (97% of those made at discharge) were obtained on the ward and 115 (including all 110 follow-up recordings) were obtained at the patients' homes. The recordings were stored on tape (Racat FM4; Racal Recorders Inc., lrvine, Calif.) and subsequently printed onto graph paper (at 3.2 mm/sec) by means of an ink-jet recorder. The study was prospective and noninterventional.This approach was considered legitimate because it was not routine on these wards to obtain recordings of oxygenation before discharge; thus no information that would otherwise have been available was withheld from the clinicians. The printouts were analyzed by two experienced workers without knowledge of the infants' history or the timing of the recordings. Episodes in which SaO2 fell to 80% or less for at least 4 seconds were counted and their durations were measured. These episodes were classified according to their relationship with an apneic pause of 4 seconds or longer. An association with an apneic pause was considered to be present if the desaturation began within 2 to 12 seconds after the onset of the pause. The minimum apneic pause duration (4 seconds) was arbitrarily defined to conform with previous work from our group.l' 2 The minimum duration of desaturation (4 seconds) was chosen to concentrate analysis only on relatively prolonged falls in SaO2; in term infants, 95% of all episodic falls in SaO2 to --<80% are less than 4 seconds in duration, z Periodic apnea was defined as a sequence of three or more apneic pauses separated by fewer than 20 breaths. ] Baseline SaO2 was determined

during the pattern of regular breathing as described previouslyl; in summary, the oxygen saturation values were measured at end inspiration for each of five successive breaths in the middle of each episode of regular breathing, at least 10 seconds away from sighs or apneic pauses. The number of desaturation episodes, corrected to the mean duration of recordings (see Results), and their median durations were calculated for each recording. Results are presented as means + SD for the clinical data and as medians and ranges for the data from the recordings, because the latter all showed a nonnormal distribution. Statistical significancewas assessed by means of the paired or unpaired t test for the clinical data, the Wilcoxon rank-sum test or McNemar chi-square test for the comparison between unpaired data, and the Wilcoxon matched-pairs test for the longitudinal study. Correlations were tested by the Spearman rank correlation coefficient. RESULTS The mean duration of all recordings (discharge and follow-up) was 12.2 _+ 1.3 hours. Mean duration was similar for discharge and follow-up recordings. Infants with a history of positive-pressure ventilation, prolonged oxygen requirements, or apnea of prematurity had higher postnatal and postconceptional ages at the time of recording than infants of similar GA at birth and birth weight who had no such history, but there were no significant differences for any of the variables regarding oxygenation between these matched groups (data not shown). None of the infants with

Volume 120 Number 3

Sao2 in preterm infants

45 1

Fig. 3. Section from recording of an infant with relatively frequent desaturation episodes (n = 61) who was born at 31 weeks of GA, had not required mechanical ventilatory support and was studied at 16 days of age. There are frequent falls in SaO2, associated with periodic apnea.

Table Ill. Comparison between recordings in preterm and term infants with comparable GAs at the time of the recording

No. of recordings GA at birth (wk) Age at recording (days) GA at recording (wk) Baseline Sao2 (%) SD of individual baseline measurements No. of (%) of recordings with episodes of desaturation Episodes of desaturation per recording Desaturation episodes (%) with apneic pause Desaturation episodes (%) in periodic apnea

Preterm

Term

53 33 + 2.6 76 + 22.1 44 _+ 1.7 100 (97.4-100) 0.0 (0.0-1.2) 13 (25) 0.0 (O-9) 1oo (0-100) 17 (0-100)

67 40 _+ 1.0 40 • 6.1 45 • 1.4 99.8 (97.0-100)* 0.2 (O.O-1.4)t 11 (16) 0.0 (O-9) 100 (0-1oo) 50 (0-100)

Results are mean + SD for clinicaldata, and, for recordingresults, medians with ranges in parentheses or numbers with percentages in parentheses, as indicated. Significance assessed by Wilcoxonrank sum test. *p <0.05. tp <0.01.

a congenital heart lesion or minor symptoms had outlying results for any of the variables studied. Results are therefore presented without reference to any of these clinical features. Discharge recordings (n = 160, Table II) Baseline Sa02. Baseline SaO2 was greater than 97%, the lowest value recorded in a study of infants born at term, 1 in 145 recordings (91%). The 5th percentile was 95.7% (Fig. 1); eight infants had values less than this value. Of the four infants with the lowest values (88.7%, 92.1%, 92.7%, and 93.2%), three had received additional oxygen for longer than 28 days but none had been considered to have clinical hypoxemia at the time of the recording. The infant with the lowest baseline value (88.7%) had an SaO2 of 80% or less

during substantial parts of her recording, which was therefore excluded from analysis of episodic desaturation. Analysis of the variability between baseline measurements obtained during individual periods of regular breathing in each infant showed that on average two thirds of the values lay within 0.7% of the infant's mean baseline SaO2. This variability was inversely correlated with baseline SaO2 (r = -0.9; p <0.001). Episodes ofdesaturation. A total of 2096 episodes of desaturation were found in 114 of the 160 discharge recordings, 164 of which (8% in 48 recordings) lasted for longer than 10 seconds. Most episodes were associated with an apneie pause and most of these pauses occurred during episodes of periodic apnea (Table II); only 5% (113/2096)

452

Poets el al.

of all desaturation episodes, however, were associated with a pause of 12 seconds or longer. Desaturation was uncommon during regular breathing (25 episodes [1%] in 10 infants). There was a weak but significant inverse correlation between the frequency of desaturation and baseline SaO2 (r = - 0 . 2 ; p <0.01); the infants with a lower baseline SaO2 had more frequent desaturation. The 95th percentile for the number of desaturation episodes per recording was 61 (Fig. 2). Of the eight infants with values greater than the 95th percentile, only one had a low baseline SaO2 (95.7%); three had history of apnea of prematurity (considered clinically to have ceased at the time of the recording) and six had not reached term (GA >--37 weeks) at the time of the study. An example of a recording from one of these relatively young infants is shown in Fig. 3. Longitudinal study (n = 110, Table II). Comparison of the recordings made at discharge with those obtained about 6 weeks later (median age difference 37 days; range 19 to 110 days; interquartile range 32 to 49 days) showed highly significant differences for almost all measurements; at follow-up the infants had a higher baseline SaO2 (5th percentile 97.9%) and fewer episodes of desaturation (95th percentile three episodes per recording). The total number of episodes of desaturation in all follow-up recordings was 78, eight of which (10%) were longer than 10 seconds and three of which (4%) were associated with an apneic pause of 12 seconds or longer. Only one episode occurred during regular breathing. As in the discharge recordings, there was a negative correlation between baseline SaO2 and the frequency of desaturation (r = -0.3; p <0.001). For six of the eight infants with a baseline SaO2 less than the 5th percentile at discharge, including the one with a baseline value of 88.7%, a follow-up recording was made. The lowest value found at follow-up in these six infants was 98.4%. For five of the eight infants with frequency of desaturation greater than the 95th percentile, a follow-up recording was made. The infant with the highest frequency of desaturation at discharge (n = 355) had the second highest number in the follow-up recordings (n = 9). The remaining four infants had a maximum of two episodes at follow-up. The correlation between discharge data and follow-up recordings was poor (r = 0.1 and p >0.05 for both baseline SaO2 and number of desaturation episodes). Comparison with term infants at comparable GA (Table III). Comparison of the subgroup of 53 infants whose G A was 42 to 47 weeks at the time the follow-up recording was made with the term infants studied previously1 showed a significantly higher baseline SaO 2 in the preterm infants (who were chronologically older) but no significant difference in the number of desaturation episodes. Recordings of infants who died. Two recordings were obtained from one of the two infants in the study who subse-

The Journal o f Pediatrics March 1992

quently died of SIDS; one was obtained from the other infant. The latter infant was born at 25 weeks of GA (birth weight 750 gm) and had received ventilatory assistance for 13 days. She had had recurrent episodes of cyanosis and bradycardia of unknown cause for some time before discharge but these had resolved by the time of study (they recurred, however, at some time after discharge and continued until her death). She was studied at 41 weeks of GA (114 days of age), had a baseline SaO2 of 100%, and had one episode of desaturation in her recording. The other infant, born at 33 weeks of GA (birth weight 2000 gm) and studied at 9 and 50 days of age, had baseline values of 97.9% and 100% in her recordings at discharge and follow-up, and one desaturation episode in her first recording but none in her second recording. No complications were reported during or after her hospital stay. The two infants who subsequently died of causes other than SIDS also did not have outlying results in their recordings. DISCUSSION These overnight tape recordings of preterm infants who were broadly representative of a population of infants admitted to special-care baby units showed that (1) a minority of infants clinically considered to be ready for discharge from the hospital had an unrecognized low baseline SaO2 or a high number of desaturation episodes (the highest being six times the 95th percentile value); (2) baseline SaO2 had risen and the frequency of desaturation had decreased when recordings were repeated about 6 weeks later, with particularly marked changes in those infants who had initially had outlying results; and (3) the preterm infants studied at about 5 to 10 weeks after term did not have less stable oxygenation than a group of term infants studied during their second month of life. In a minority of preterm infants at discharge from the hospital, clinically unrecognized baseline or intermittent hypoxemia has been reported. 2 It has also been shown that hypoxemia has potentially deleterious effects on development and may be a cause of sudden deathl s-7 We did not collect information that would have allowed us to assess the potential effects of hypoxemia on growth and development~ However, we do know that the two infants who later died of SIDS did not have significant abnormalities in their recordings. The study demonstrates an inverse correlation between baseline SaO2 and episodic desaturation. This relationship could be determined by an increase in the frequency of apneic pauses in response to airway hypoxia, 8 which would lead to an increase in the number of desaturation episodes because there is a strong relationship between apneic pauses and desaturation, t' 2 It could also be due to an increase in the proportion of apneic pauses associated with desaturation in the infants with a lower baseline SaO2. Whatever the

Volume 120 Number 3

mechanism, we speculate that our data on the relationship between baseline SaO2 and episodic desaturation indicate that raising the baseline SaO2 by providing additional oxygen may also reduce the frequency of episodic desaturation. It was surprising to find that baseline SaO2 was not lower in the group of infants born very prematurely (<32 weeks of GA), despite the fact that 31% of the infants in this group had a history of prolonged oxygen requirement and 74% had received mechanical ventilation. This observation may be explained by the fact that these infants generally had greater postnatal ages and GAs at the time of study than those born at a greater GA (see below). Age effects may also be responsible for the finding that infants who had received ventilatory support did not differ significantly from infants who had not received ventilatory support after the groups were matched for GA at birth. Clinically this finding is reassuring, because it shows that whatever their history almost all infants had a normal baseline SaO2 at discharge. An examination of the group of infants with outlying results, however, showed overrepresentation of subjects with prolonged oxygen requirements. The influence of age on baseline SaO2 and on the frequency of desaturation episodes was demonstrated by the significant increase in the former and decrease in the latter between initial study and follow-up. We cannot say whether an increase in postnatal age, in postconceptional age, or in a combination of the two caused these changes, but they may be related to improvements in the matching of ventilation to perfusion during the first weeks of life.9 The decrease in the frequency of desaturation episodes with increasing age may also be due to a reduced frequency of apneic pauses. Frequency of apneic events in preterm infants decreases by more than 80% between 40 and 52 weeks of GA, by which time preterm infants have fewer apneic pauses than do term infants of similar GA. l~ This decrease has been interpreted as being due to a compensatory mechanism for dealing with chronic hypoxia, l~ a hypothesis that is not supported by our data. The reduction with age in the frequency of desaturation observed in our study, however, exceeded that reported for apneic pauses, 1~ 11 and additional factors, possibly including improvements in chest wall stability12 and in ventilation-perfusion matching, 13 may be involved in this developmental process. Analysis of episodic falls in SaO2, because it provides direct information about the effects of respiratory phenomena on blood gas homeostasis, differs from measurements of the duration and frequency of apneic pauses. We previously showed that the proportion of apneic pauses leading to desaturation is extremely variable and depends not only on the duration of a pause but also on the breathing pattern in which it occurs. 1 Moreover, desaturation can also occur without a cessation of breathing movements,1 airflow,2, 14or

Sao2 in preterm infants

453

both. The regulation of breathing during sleep is predominantly influenced by chemical control mechanisms, and oxygen levels change far more rapidly than carbon dioxide levelslS; thus multichannel recordings including continuous measurements of SaO2 may provide a more direct approach to the study of respiratory control than recordings of breathing movements alone. Our findings regarding the developmental changes in the occurrence of desaturation suggest that the maturation of respiratory control in most of these preterm infants had caught up with that in term infants within 1 to two months after term. The study has a number of limitations. Recordings were performed in different environments (home and hospital), but this difference would not be expected to affect data comparability. 16 We have not collected data on sleepingawake periods or on sleep states, so information regarding the possible effects of these factors on the frequency of desaturation is not available. However, baseline SaO2 measurements are likely to have been performed only during quiet sleep, because regular breathing is strongly associated with this state. 17 The frequency of apneic pauses was not analyzed because it was our aim to provide normal data for arterial oxygenation rather than for breathing patterns. In contrast to our previous studies, l, 2 we have corrected our data only to the mean duration of recordings and not to the duration of an artifact-free SaO2 signal. In our previous studies, the variability in the frequency of desaturation was more than 100 times greater than that in the proportion of uninterpretable SaO2 signal, and there was no consistent relationship between this proportion and GA at birth, GA at recording, or postnatal age. We therefore believe that the potential influence of differences in artifact-free time can be ignored. Finally, we stress that the values for oxygenation presented in this study might have been different had a longer averaging mode or another pulse oximeter been used. Although different modes of averaging do not affect baseline measurements from the same instrument (our unpublished observations), they do affect the measured depth and duration of short-lived desaturation. Differences in SaO2 measurements between oximeters from different manufacturers, and even between different software versions for the same instrument,is are considerable. This study may provide a basis for the identification of infants with an abnormally low baseline SaO2 or an increased frequency of desaturation. It may also help to decide whether an infant will require supplemental oxygen. Further studies of the relationship between arterial oxygenation and clinical outcome in preterm infants are required. We thank the parents who allowed us to perform this study, the medical and nursing staff who helped us with the recordings, and J. Kelly and P. Mills, who analyzed the recordings.

454

Poets et al.

The Journal o f Pediatrics March 1992

REFERENCES

1. Stebbens VA, Poets CF, Alexander JA, Arrowsmith WA, Southall DP. Oxygen saturations and breathing patterns in infancy. I. Fullterm infants in the second month of life. Arch Dis Child 1991;66:569-73. 2. Poets CF, Stebbens VA, Alexander JR, Arrowsmith WA, Salfield SAW, Southall DP. Oxygen saturations and breathing patterns in infancy. II. Preterm infants at discharge from special care. Arch Dis Child 1991;66:574-8. 3. Fanconi S, Doherty P, Edmonds JF, Barker GA, Bohn DJ. Pulse oximetry in pediatric intensive care: comparison with measured saturations and transcutaneous oxygen tension. J PEDIATR 1985;107:362-6. 4. Southall DP, Bignall S, Stebbens VA, Alexander JR, Rivers RPA, Lissauer T. Pulse oximeter and transcutaneous arterial oxygen measurements in neonatal and paediatric intensive care. Arch Dis Child 1987;62:882-8. 5. Groothuis JR, Rosenberg AA. Home oxygen promotes weight gain in infants with bronchopulmonary dysplasia. Am J Dis Child 1987;141:992-5. 6. Teitel D, Sidi D, Bernstein D, Heyman MA, Rudolph AM. Chronic hypoxemia in the newborn lamb: cardiovascular, hematopoietic, and growth adaptations. Pediatr Res 1985; 19:1004-10. 7. Rognum TO, Saugstad OD, (byasoeter S, Olaisen B. Elevated levels of hypoxanthine in vitreous humor indicate prolonged cerebral hypoxia in victims of sudden infant death syndrome. Pediatrics 1988;82:615-7. 8. Rigatto H, Brady JB. Periodic breathing and apnea in preterm infants. II. Hypoxia as a primary event. Pediatrics 1972;50:21927.

9. Krauss AN, Auld PAM. Ventilation-perfusion abnormalities in the premature infant: triple gradient. Pediatr Res 1969; 3:255-64. 10. Albani M, Bentele KHP, Budde C, Schulte FJ. Infant sleep apnea profile: preterm vs term infants. Eur J Pediatr 1985; 143:261-8. 11. McCulloch K, Kelly DH. Prospective pneumogram recordings in preterm infants with and without clinical apnea and bradycardia. Pediatr Pulmonol 1990;8:33-9. 12. Heldt GP. Development of stability of the respiratory system in preterm infants. J Appl Physiol 1988;65:441-4. 13. Woodrum DE, Oliver TK, Hodson WA. The effect of prematurity and hyaline membrane disease on oxygen exchange in the lung. Pediatrics 1972;50:380-6. 14. Poets CF, Stebbens VA, Samuels MP, Southall DP. Patterns of breathing during hypoxaemia in preterm infants at around term [Abstract]. Early Hum Dev 1990;22:107-8. 15. Sullivan CE, Kozar LF, Murphy E, Phillipson EA. Primary role of respiratory afferents in sustaining breathing rhythm. J Appl Physiol 1978;45:11-7. 16. Keens TG, Davidson Ward SL, et al. A comparison of pneumogram recordings in infants in the hospital and at home. Pediatr Pulmonol 1986;2:373-7. 17. Anders T, Erode R, Parmelee A. A manual of standardized terminology techniques and criteria for scoring of states of sleep and wakefulness in newborn infants. Los Angeles: UCLA Brain Information Service, 1971. 18. Severinghaus JW, Naifeh KH, Koh SO. Errors in 14 pulse oximeters during profound hypoxia. J Clin Monit 1989;5:7281.

BOUND VOLUMES AVAILABLE TO SUBSCRIBERS Bound volumes of the 1992 issues of THE JOURNAL OF PEDIATRICS are available to subscribers (only) from the Publisher, at a cost of $58.00 for domestic, $78.06 for Canadian, and $74.00 for international subscribers, for Vol. 120 (January-June) and Vol. 121 (July-December), shipping charges included. Each bound volume contains subject and author indexes, and all advertising is removed. Copies are shipped within 60 days after publication of the last issue in the volume. The binding is durable buckram, with the Journal name, volume number, and year stamped in gold on the spine. P a y m e n t m u s t accompany all-orders. Contact M o s b y - Y e a r Book, Inc., Subscription Services, 11830 Westline Industrial Dr., St. Louis, M O 63146-3318, USA/800-325-4177, ext. 4351, or 314-453-4351. Subscriptions must be in force to qualify. Bound volumes are not available in place of a regular Journal subscription.