Secondary Logo

Journal Logo

Pediatric Anesthesiology: Research Report

Incidence of Intraoperative Hypoxemia in Children in Relation to Age

de Graaff, Jurgen C., MD, PhD*; Bijker, Jilles B., MD, PhD; Kappen, T. H., MD; van Wolfswinkel, Leo, MD; Zuithoff, Nicolaas P. A., PhD; Kalkman, Cor J., MD, PhD

Author Information
doi: 10.1213/ANE.0b013e31829332b5

Respiratory events are the most common cause of adverse events in pediatric anesthesia, especially in younger children.1 Recently, in adults (age ≥ 16 years) the rate of hypoxemic events (oxygen saturation [SpO2] ≤ 90%) during the intraoperative period was reported to be 6.8%, and 3.5% of patients had a severe hypoxemic episode.2 Young children, and especially neonates, are perceived to be at increased risk of hypoxemia because they have smaller functional residual capacity and increased metabolic requirements compared with adults.3 Currently, there are no reliable data on the incidence of hypoxemia in children during anesthesia. Still, knowledge about the true incidence of hypoxemia in various age groups of children is important to estimate perioperative risk.

The present knowledge of the incidence of hypoxemic episodes in children is mainly based on small study populations and on self-report by anesthesiologists.4–8 However, incidence reports based on self-report may underestimate the true incidence of hypoxemia as specialized pediatric anesthesiologists are used to a frequent incidence of hypoxemia in small children and therefore are more likely to underreport the presence of hypoxemia. In contrast, an anesthesia information management system (AIMS) accurately stores all data from the anesthesia monitor and ventilator. However, as pulse oximeter readings may be inaccurate due to patient movement or sensor dislodgment, especially around induction of anesthesia and emergence in small children, simply obtaining pulse oximeter readings from the AIMS may overestimate the incidence of hypoxemia.

We hypothesized that the incidence of hypoxemia is much higher than that based on self-report and that the incidence increases in younger age groups. In the present study, we estimated the incidence of intraoperative hypoxemic episodes in children relative to age. To this aim, we first determined both the incidence of hypoxemia and the incidence of pulse oximeter reading artifacts (“false hypoxemia”) in a prospective cohort. Subsequently, we applied these results to a large retrospective cohort, derived from the AIMS database to estimate the incidence of hypoxemia relative to age.



The patients study protocol was approved by the local hospital ethics committee, which waived the need for informed consent, as patients were not subjected to investigational actions. Patient confidentiality was guaranteed according to the Dutch law on personal data protection.

The incidence of hypoxemic episodes and the incidence of such episodes based on pulse oximeter artifacts during anesthesia were estimated prospectively in a cohort of 575 anesthetic procedures (patients aged younger than 16 years) anesthetized at a tertiary pediatric university hospital (Wilhelmina Children’s Hospital, University Medical Center Utrecht, the Netherlands) between June 26 and August 26, 2005. In addition, the incidence of hypoxemic episodes during anesthesia was assessed in a retrospective cohort obtained from an AIMS database. The retrospective cohort included 8277 cases of children who underwent elective, noncardiac surgery under general anesthesia between November 1999 and March 2005 at the same hospital. Because a child could be included in the analyses more than once, the results are presented in relation to cases. A case is defined as the registration of an anesthetic procedure from induction to end of anesthesia when the patient leaves the operating room or procedure suite.

At the Wilhelmina Children’s Hospital, all pediatric anesthetic procedures are staffed with a dedicated pediatric anesthesiologist and a pediatric anesthesia nurse. Children anesthetized outside office hours (on Saturdays, Sundays, or during weekdays between 5:00 PM and 8:00 AM), procedures with durations of <10 minutes, and children undergoing cardiac surgery were excluded from both cohorts (patients with congenital heart disease undergoing general surgery were included), because of possible preexistent hypoxemia and/or expected long periods of unreliable pulse oximeter measurements, for example, during cardiopulmonary bypass.

Data Collection

In the prospective study, SpO2 of the included patients was remotely monitored in real time by a researcher outside the operating room using an interface programmed in LabVIEW® (Release 7.1, National Instruments Corporation, Austin, TX). This interface obtained the SpO2 data from the AIMS in real time and displayed an alarm for every operating room when an SpO2 value < 90% was detected. Once a patient’s SpO2 value decreased below the 90% threshold, the depth and duration of the hypoxemic episode was registered. A few minutes after restoration of saturation, the researcher contacted the involved anesthesiologist by phone to ask whether the low SpO2 was caused by an artifact or by a genuine hypoxemic event. In addition, the cause of the artifact or hypoxemic event was verified.

The AIMS was developed by one of the authors (LvW, Vierkleurenpen®, Carepoint, Ede, The Netherlands), and a first version has been used in our hospital since 1998. The method of sampling and storage in the first commercial version (version 1.0, 2004, has not been changed. It automatically stores data from the anesthesia ventilator (Dräger Cicero, Lübeck, Germany) and patient monitor (Hewlett Packard, Merlin, M1166A, 86S ACMS, Andover, MA) with an adhesive oximeter sensor (OxiMax-P, single patient use adhesive sensor, Covidien-Nellcor, Boulder, CO). The SpO2 values were obtained from the monitor and stored as the median value of every 12 values captured per minute to prevent the AIMS from storing most of the artifacts caused by positioning or electrocautery. The hardware and software for monitoring has not changed throughout the retrospective and prospective study period (1999–2005).


In the prospective study, hypoxemia was defined as any SpO2 value ≤ 90%, for at least 1 minute and not being the result of an artifact. Artifacts were defined as any SpO2 ≤90% that was the obvious result of sensor dislodgment, electrocautery, patient motion or any other motion of the SpO2 probe (e.g., manipulation of the limb where the pulse oximeter was located), and low peripheral perfusion (e.g., during blood pressure measurement on the limb where the probe was located). Hypoxemia was categorized in mild hypoxemia defined as an SpO2 ≤ 90%, and severe hypoxemia defined as an SpO2 ≤ 80%.4,5,9,10 In addition, hypoxemia was further categorized using 2 different thresholds for the minimum consecutive duration of hypoxemic episodes (at least 1 minute or >5 minutes). To obtain an estimate of the incidence of hypoxemic episodes stratified by age, patients were grouped by month and in 4 different age groups: 0 to 28 days (neonate), 29 days to 1 year (infant), 1 to 8 years (younger child), and 8 to 16 years (older child).

In the retrospective cohort, hypoxemia was categorized in the same categories regarding severity and duration of hypoxemia. The incidence of hypoxemia was reported as the percentage of cases with hypoxemic periods and as median number of episodes per 100 cases, because >1 hypoxemic episode may occur in 1 case, and as episodes per anesthetic time per hour.


The incidence of hypoxemic episodes per 100 cases including 95% confidence intervals (95% CIs) was calculated by the Clopper-Pearson method ( The frequency of low SpO2 artifacts causing false-positive hypoxemic episodes was estimated from the incidence of artifacts in the prospective evaluation. The 95% CI of the probability was estimated considering the correlation within case by Generalized Estimating Equations (GEEs) with case as subject and the incident number within as case within-subject variable. We performed univariate and multivariate logistic regression analyses to assess the association between the presence of artifacts (outcome) and age, gender, and surgical specialty with a GEE adjustment for correlated records. Linearity was tested with restricted cubic splines; this was done in R (Wien, Austria, with the “geepack” and “rms” packages.11

To extrapolate the incidence of intraoperative hypoxemic episodes (SpO2 ≤ 90%) in the retrospective cohort, we multiplied the observed incidence in the retrospective cohort by the crude ratio of true-positive hypoxemic episodes from the prospective evaluation. This extrapolated incidence (with 95% CI) was calculated for intraoperative hypoxemic episodes (SpO2 ≤ 90%) with duration of at least 1 minute. SPSS release 20 for Windows (SPSS Inc., Chicago, IL) was used for analyses except for the GEE general linear model analysis, which was performed in R.


Our study included 575 patients in the prospective cohort and 8277 cases from the retrospective database cohort. The patient characteristics of the retrospective cohort were comparable with those of the prospective cohort, except for a higher percentage of orthopedic and ear-nose-throat procedures (Table 1).

Table 1
Table 1:
Baseline Characteristics

In the prospective cohort, at least 1 episode of SpO2 ≤90% for at least 1 minute occurred in 69 of 575 cases (12%; 95% CI, 9%–15%). Furthermore, in 35 of 575 (6%; 95% CI, 4%–8%) cases at least 1 true hypoxemic event was observed. In total, 117 episodes of SpO2 ≤ 90% were observed in the prospective study, resulting in an incidence of 20 episodes per 100 cases and 0.16 incidents per hour anesthesia (Table 2). The incidence of hypoxemia increases with younger age (Table 2). The incidence of episodes with an SpO2 value ≤ 90% increased with younger age, and varied between 142 episodes per 100 cases (0.5 incidents per hour) in neonates and 13 episodes per 100 cases (0.08 incidents per hour) in adolescents (Table 2).

Table 2
Table 2:
Results of the Prospective Cohort Study

Of these 117 episodes with an SpO2 value ≤ 90% for at least 1 minute, 67 (54%; 95% CI, 42%–65%; Table 3) episodes were classified as true hypoxemia, occurring in 35 cases (6% of total cohort; 95% CI, 4%–8%), 0.09 episodes per hour anesthesia. Of the remaining 50 episodes, 47 (46%) were classified as artifacts, whereas for 3 episodes the cause of the low SpO2 episodes was unknown (Table 3). Of the 67 episodes of hypoxemia, 4 (6%) had a duration >5 minutes (range: 6–30 minutes). Of the 67 true hypoxemic episodes, SpO2 was ≤80 % in 14 episodes (3 of which lasted >5 minutes). Nineteen of these 67 hypoxemic episodes occurred during the induction period (28%; 95% CI, 18%–41%), 31 during anesthesia maintenance (46%; 95% CI, 34–59), and 17 during emergence from anesthesia (25%; 95% CI, 16%–37%). Furthermore, 8 hypoxemic episodes were accompanied by bradycardia (varying between 21% and 89%), 3 of which received atropine (SpO2: 21%, 54%, and 1 of 88%). Of the 47 pulse oximeter artifacts with SpO2 ≤ 90%, 35 (74%) had a duration of ≤1 minute, 8 (18%) had a duration between 2 and 5 minutes, and 4 (9%) lasted for >5 minutes.

Table 3
Table 3:
Causes of Hypoxemia Episodes (Spo2 < 90%) and Artifactual Measurements in the Prospective Cohorta

Univariate and multivariate analyses did not show an association between the presence of artifacts and age, gender, and surgical specialty (Table 4). However, because the number of artifacts (47) was limited, it is not likely to find a significant relation.

Table 4
Table 4:
Covariates Defining Pulse Oximeter Artifacts in the Prospective Study

Evaluation of the retrospective cases showed 2553 episodes of hypoxemia (SpO2 ≤ 90%) in 18% of the cases and 836 episodes of severe hypoxemia (SpO2 ≤ 80%) lasting for at least 1 minute in 7% of the cases; this corresponds with 31 (95% CI, 30–32) and 10 (95% CI, 9–11) episodes per 100 cases, and 0.2 and 0.069 incidents per hour respectively (Table 4). Most episodes were short; there were only 229 episodes of hypoxemia (SpO2 ≤ 90%) lasting >5 minutes in 2% of the cases and 34 episodes of severe hypoxemia >5 minutes (SpO2 ≤ 80%) in 0.4% of the cases.

The incidence of hypoxemia increased with younger age. The incidence and severity of hypoxemia was highest (up to 60% in the retrospective database) among neonates and tapered off toward the average in children older than 2 years (Fig. 1). In the retrospective database, neonates had the highest incidence of hypoxemia: 170 episodes (0.7incidents per hour anesthesia) with SpO2 ≤ 90% for at least 1 minute were recorded per 100 cases, and 47 episodes of severe hypoxemia (SpO2 ≤ 80%) for at least a period of 1 minute per 100 cases (Table 5).

Figure 1
Figure 1:
Hypoxemia in relation to age. Percentage of cases (with numbers of cases presented in bars) with at least 1 period of hypoxemia for at least 1 minute in relation to age group in the retrospective cohort (uncorrected data are presented,n = 8277). SpO2 = oxygen saturation.
Table 5
Table 5:
Number of Episodes, Number of Cases, Number of Episodes per 100 Cases and Episodes per Anesthesia Hour with Spo2 ≤ 90% as Documented in the Anesthetic Information Management System

Assuming a true-positive hypoxemia rate of 54%, the extrapolated incidence of true hypoxemic episodes (SpO2 ≤90% for at least 1 minute) in the retrospective cohort would be 17 episodes per 100 cases (54% × 30.8 episodes) with a most conservative estimate of the CI ranging from 12 (42% × 29.6) to 21 (65% × 32.0) episodes per 100 cases. When analyzing by age category, the extrapolated incidence of “true” hypoxemia (SpO2 ≤ 90% for at least 1 minute) would vary between 11 per 100 cases (54% of 20) in children aged 8 to 16 years and 92 episodes per 100 cases (54% of 170) of the cases in neonates.


Respiratory complications are the most frequent adverse events in pediatric anesthesia. In the present study, we determined the incidence of SpO2 artifacts and used AIMS data to retrospectively estimate the incidence of intraoperative hypoxemia in >8000 children who were anesthetized in a specialized pediatric hospital. The present study showed that mild to moderate hypoxemia is relatively common; true hypoxemia (SpO2 ≤ 90% for at least 1 minute) occurs in 6% of the cases, with an incidence of 12 episodes per 100 cases in the prospective cohort. The retrospective cohort confirms that the incidence of intraoperative hypoxemic episodes increases in younger age groups, with the highest incidence in neonates (92 episodes per 100 cases), and the lowest incidence in children aged 8 to 16 years (11 episodes per 100 cases) after adjusting the data for artifacts using the prospective cohort of comparable patients.

Thus far, knowledge about the validity of data stored in AIMS is scarce. AIMS is not only used for electronic record keeping and decision support, but data are increasingly used for clinical and research purposes. Therefore, knowledge about the reliability of data from AIMS and incidence of artifacts is essential. The observed high artifact frequency in the prospective study shows that retrospective data on arterial oxygen saturation from an AIMS should be interpreted with caution. Pulse oximeters are prone to false alarms, especially in neonatal and pediatric applications. Sensor dislodgment, motion artifact, or low peripheral perfusion are the most common causes of false readings and artifacts. Low perfusion caused by inflation of a blood pressure cuff is predictable and does not interfere with clinical practice. However, these measurements should be considered as artifacts because these incorrect SpO2 measurements are stored in AIMS potentially without any comment as to their artifact status. When retrospectively reviewing a case from stored AIMS data, it is no longer possible to distinguish between SpO2 artifacts and true hypoxic periods. Two studies in the critical care unit for neonates and children found that 94% of oximeter alarms were considered clinically unimportant, and 71% were false alarms.12–14 We found similar causes of false readings, but the incidence in the present study was much lower (40%), which might be the result of using taped (disposable) sensors instead of clip-on sensors, decreasing the incidence of probe dislodgment. New generation pulse oximeter technology has considerably improved correction for artifacts due to patient motion, which may further decrease the incidence of pulse oximeter artifacts in neonates. Nonetheless, considering the high proportion of pulse oximeter artifacts, the data on SpO2 in an AIMS database should be interpreted with caution when the data are to be used for clinical, medicolegal, or research purposes.15,16

Knowledge about the incidence of hypoxemia in various age groups is important for risk stratification. The present study shows that the incidence of a mild and severe hypoxemic episode is highest in neonates per case (170 incidents per 100 cases, in at least 50% of the patients) and 0.7 incidents per anesthetic hour. The incidence of hypoxemia in relation to age group has not been published thus far and current knowledge is mainly based on small study populations and on self-report by anesthesiologists.4–8 The results of both cohorts showed that hypoxemia in children was relatively common and increased considerably in younger children. A previous database cohort study in 2 centers in adults showed that 6.8% of patients had a hypoxic event (SpO2 ≤ 90%), and 3.5% of patients had a severe hypoxic event (SpO2 ≤ 85%) of 2 consecutive minutes or longer.2 The increased risk in younger children is best explained by their smaller functional residual capacity, rapid heart rate, and increased metabolic requirements compared with older children and adults, which make them more vulnerable to episodes of hypoventilation.3 The reported incidence of hypoxemia in the pediatric anesthesia literature is highly variable (1%–54%).5,8,17,18 In these publications, the higher incidence of intraoperative hypoxemia in young children (>25% of the patients) was probably related to a more liberal definition of hypoxemia (e.g., an episode with SpO2 ≤ 90% and a minimum duration of 30 seconds).5,18 Recently, a desaturation to ≤95% in 10% of the children based on self-report in a large prospective study has been reported.6 When corrected for artifacts, the incidence of intraoperative hypoxemia (SpO2 ≤ 90%) in our study (18 episodes per 100 cases; in 9% of the cases) was much lower. In addition, this difference may also be the result of the AIMS resolution that stores (median filtered) monitor data once every minute, thus excluding transient (<30 seconds) hypoxemic episodes.

The clinical relevance of transient intraoperative hypoxemia in pediatric anesthesia is disputable. As shown in the present study, transient mild to moderate hypoxemia for a relatively short period is common. Our data do not provide information on the possible clinical consequences of hypoxemia such as hypoxic cerebral injury. A recent study in pediatric cardiac surgical patients suggested that even severe central venous desaturation (≤40%) of prolonged duration (>18 minutes) was only moderately associated with increased adverse events.19 Nonetheless, although most anesthesiologists dismiss the premise that a few minutes of mild desaturation will result in an adverse outcome, it is good clinical practice to avoid any hypoxemia.

In interpreting our findings, several aspects have to be considered. First, we cannot completely exclude interobserver bias in the prospective study. Although most observations were performed by 1 investigator, the interpretation of the cause of the SpO2 value ≤ 90% was left to the discretion of the treating anesthesiologist, and anesthesiologists may have given different explanations for similar causes of hypoxemia, including the possibility of attributing artifact status to a low SpO2 value when true hypoxemia is present. The latter situation would result in overestimation of the artifact rate and overcorrection when attempting to correct the incidence of hypoxemia in the AIMS for SpO2 artifacts. Second, the awareness of anesthesiologists for situations that are prone to hypoxemia may have been increased by the telephone calls (Hawthorne effect), which may have resulted in fewer hypoxemic episodes and consequently to an underestimation of the incidence.20 Third, at the time of study, the medical history of the children was not yet routinely documented in the AIMS. Therefore, it is possible that children with a history of cardiac disease, pulmonary disease, or other diseases with preexistent hypoxemia were included in the study. We tried to minimize inclusion of patients with preexistent cardiac disease by excluding all cardiothoracic surgery and patients with initial SpO2 values ≤ 90%. Fourth, the phase of anesthesia was not considered in the retrospective study. From the prospective study, we know that most hypoxemic incidents took place during induction (28%) and emergence (25%) from anesthesia. Fifth, from the prospective study we know both low SpO2 artifact and true hypoxemia took place in a relatively small proportion of patients. Therefore, an initial low SpO2 increases the chance of a low SpO2 later in the case. Finally, the prospective study was too small to identify risk factors for pulse oximeter artifacts and to study the true incidence of hypoxemia in relation to age. Therefore, we used a large database of AIMS to study the incidence of hypoxemia and age. However, in the retrospective study it was not possible to differentiate between low SpO2 values that were caused by artifact and true hypoxemic events. Therefore, the high incidence of hypoxemic episodes documented in the AIMS (30.8 episodes per 100 cases) is at least in part the result of storing artifactual SpO2 values. Furthermore, only 12 neonatal anesthetics were included in the prospective study.

In conclusion, the incidence of intraoperative hypoxemia in young children was rather high, even for a specialized pediatric hospital, and the incidence of intraoperative hypoxemia increased steeply in children younger than 1 year. Furthermore, any retrospective review of AIMS data for hypoxemia should be interpreted with caution, as only up to 65% of all hypoxemic episodes recorded in the AIMS during pediatric anesthesia were true hypoxic events.


Name: Jurgen C. de Graaff, MD, PhD.

Contribution: This author helped design the study, analyze the data, and write the manuscript.

Attestation: Jurgen C. de Graaff has seen the original study data, reviewed the analysis of the data, approved the final manuscript, and is the author responsible for archiving the study files.

Name: Jilles B. Bijker, MD, PhD.

Contribution: This author helped design and conduct the study, analyze the data, and write the manuscript.

Attestation: Jilles B. Bijker has seen the original study data, reviewed the analysis of the data, and approved the final manuscript.

Name: T. H. Kappen, MD.

Contribution: This author helped design and conduct the study, analyze the data, and write the manuscript.

Attestation: T. H. Kappen has seen the original study data, reviewed the analysis of the data, and approved the final manuscript.

Name: Leo van Wolfswinkel, MD.

Contribution: This author helped design the study, analyze the data, and write the manuscript.

Attestation: Leo van Wolfswinkel approved the final manuscript.

Name: Nicolaas P. A. Zuithoff, PhD.

Contribution: This author helped analyze the data and write the manuscript.

Attestation: Nicolaas P. A. Zuithoff approved the final manuscript.

Name: Cor J. Kalkman, MD, PhD.

Contribution: This author helped design and conduct the study, analyze the data, and write the manuscript.

Attestation: Cor J. Kalkman approved the final manuscript.

This manuscript was handled by: Peter J. Davis, MD.


We are grateful to Dr. David Reich, Professor and Chair, Mount Sinai Hospital, New York, NY for his suggestion to perform a prospective estimation of the incidence of pulse oximetry artifacts in children. We also thank the anesthesiologists and anesthesia nurses from the Wilhelmina Children’s Hospital, University Medical Center Utrecht for their cooperation in the prospective study.


1. Bunchungmongkol N, Somboonviboon W, Suraseranivongse S, Vasinanukorn M, Chau-in W, Hintong T. Pediatric anesthesia adverse events: the Thai Anesthesia Incidents Study (THAI Study) database of 25,098 cases. J Med Assoc Thai. 2007;90:2072–9
2. Ehrenfeld JM, Funk LM, Van Schalkwyk J, Merry AF, Sandberg WS, Gawande A. The incidence of hypoxemia during surgery: evidence from two institutions. Can J Anaesth. 2010;57:888–97
3. Morrison JE Jr. Children at increased risk of hypoxia. Anesthesiology. 2000;92:1844
4. Moller JT, Johannessen NW, Berg H, Espersen K, Larsen LE. Hypoxaemia during anaesthesia–an observer study. Br J Anaesth. 1991;66:437–44
5. Schulz C, Lenz G, Madee S, Schulze M. Frequency of hypoxic episodes during general anesthesia in children. Cah Anesthesiol. 1989;37:403–7
6. von Ungern-Sternberg BS, Boda K, Chambers NA, Rebmann C, Johnson C, Sly PD, Habre W. Risk assessment for respiratory complications in paediatric anaesthesia: a prospective cohort study. Lancet. 2010;376:773–83
7. Kakavouli A, Li G, Carson MP, Sobol J, Lin C, Ohkawa S, Huang L, Galiza C, Wood A, Sun LS. Intraoperative reported adverse events in children. Paediatr Anaesth. 2009;19:732–9
8. Fossum SR, Knowles R. Perioperative oxygen saturation levels of pediatric patients. J Post Anesth Nurs. 1995;10:313–9
9. Xue FS, Luo LK, Tong SY, Liao X, Deng XM, An G. Study of the safe threshold of apneic period in children during anesthesia induction. J Clin Anesth. 1996;8:568–74
10. Thorpe CM, Gauntlett IS. Arterial oxygen saturation during induction of anaesthesia. Anaesthesia. 1990;45:1012–5
11. Halecoh U, Højsgaard S. The R Package geepack for Generalized Estimating Equations. J Stat Softw. 2006;15:1–11
12. Salyer JW. Neonatal and pediatric pulse oximetry. Respir Care. 2003;48:386–96; discussion 397–8
13. Lawless ST. Crying wolf: false alarms in a pediatric intensive care unit. Crit Care Med. 1994;22:981–5
14. Saber R, Zmora E. Nurses’ response to alarms from monitoring in NICU. Pediatr Res. 1997;41:174
15. Vigoda MM, Lubarsky DA. Failure to recognize loss of incoming data in an anesthesia record-keeping system may have increased medical liability. Anesth Analg. 2006;102:1798–802
16. Bell G. Lessons for pediatric anesthesia from audit and incident reporting. Paediatr Anaesth. 2011;21:758–64
17. Laycock GJ, McNicol LR. Hypoxaemia during induction of anaesthesia–an audit of children who underwent general anaesthesia for routine elective surgery. Anaesthesia. 1988;43:981–4
18. Laycock GJ, McNicol LR. Hypoxaemia during recovery from anaesthesia–an audit of children after general anaesthesia for routine elective surgery. Anaesthesia. 1988;43:985–7
19. Crowley R, Sanchez E, Ho JK, Lee KJ, Schwarzenberger J, Marijic J, Sopher M, Mahajan A. Prolonged central venous desaturation measured by continuous oximetry is associated with adverse outcomes in pediatric cardiac surgery. Anesthesiology. 2011;115:1033–43
20. Rheineck-Leyssius AT, Kalkman CJ, Trouwborst A. Influence of motivation of care providers on the incidence of postoperative hypoxaemia in the recovery room. Br J Anaesth. 1996;77:453–7
© 2013 International Anesthesia Research Society