Article Text

G488 Pulse oximetry is an unreliable measure of haemoglobin oxygen saturation as calculated by earlobe blood gas co-oximetry in ambulatory paediatric sickle cell disease patients
  1. S Zheng1,
  2. F Chan2,
  3. G Ruiz1,
  4. D Rees3,
  5. A Gupta1
  1. 1Paediatric Respiratory, King’s College Hospital, London, UK
  2. 2Eastbourne District General Hospital, East Sussex Healthcare NHS Trust, Eastbourne, UK
  3. 3Paediatric Haematology, King’s College Hospital, London, UK


Aim Accurate measurement of Hb saturation is essential in the care of children with sickle cell disease (SCD). The gold standard of Hb saturation measurement is by ABG with co-oximetry. Minute-by-minute changes are detected with pulse oximetry (SpO2). SpO2 shows good correlation to ABG co-oximetry in well patients with normal Hb. SCD patients represent a different population: previous small studies have shown discrepancies in calculated ABG and SpO2 Hb saturations. SpO2 may be falsely low due to high carboxyhaemoglobin and metahaemaglobin levels following heme catabolism, or high due to hyperbilirubinaemia. Both occur in acute and chronic haemolysis. With treatment decisions in acute and ambulatory settings driven by bedside SpO2, inaccuracies can result in mismanagement. Earlobe blood gas (EBG) co-oximetry is an alternative better tolerated method of measuring Hb O2 saturation in arterialised capillary blood (SaO2), with good correlation to ABG measurements. We sought to investigate the correlation between SpO2 and SaO2 measured by EBG co-oximetry in ambulatory SCD children.

Methods We retrospectively reviewed paediatric SCD patients attending a UK tertiary sickle-respiratory clinic from February 2012–August 2015, and who had simultaneous EBG and SpO2 measurements. Positive difference results from SpO2 value greater than EBG SaO2.

Results We identified 39 simultaneous paired SaO2 and SpO2 readings from 33 patients (52% male, median age 10years, range: 5–17). Mean difference between readings was –0.7% (Figure 1 and 2). In 23% of cases, difference was >±2%. SaO2 was overestimated in one-third and underestimated in two-thirds. Of 17 hypoxaemic measurements (SpO2 <94%), only 76% had SaO2 <94% (means: SpO2 90.3%, SaO2 92.1%; difference –1.8%; CI –3.3 to –0.4, p = 0.02) (Table 1). In terms of determining hypoxia based on a cut-off of 94%, SpO2 had sensitivity 85%, specificity 100%, PPV 100% and NPV 76%.

Abstract G488 Figure 1

Bland-Altman plot showing mean simultaneous O2 saturations by EBG and pulse oximerty and difference between results. Unbroken line indicates mean difference (-0.7%) and broken lines indicates limits of agreement (-5.6 to +4.3%) (mean ± 2SD). Shaded region represents a difference of ±2%, the accepted error range

Abstract G488 Figure 2

Simultaneous SpO2 and SaO2 measurements. Straight line indicates SpO2 = SaO2

Conclusion Pulse oximetry as a measure of Hb oxygen saturation was inaccurate in a significant proportion of ambulatory SCD children. SpO2 tended to be lower than actual SaO2 as measured with EBG, with SpO2≥94% more likely to predict true result than SpO2 <94%. In patients with SpO2 <94%, there was a significant difference in calculated Hb saturation. EBG should be used more widely as an accurate means of detecting changes in arterial Hb oxygen saturation.

Statistics from

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.