Original Paper Received: March 7, 2014 Accepted after revision: September 4, 2014 Published online: December 3, 2014
Neonatology 2015;107:113–119 DOI: 10.1159/000368178
Reliability of Pulse Oximetry during Cardiopulmonary Resuscitation in a Piglet Model of Neonatal Cardiac Arrest Mohammad Ahmad Hassan a, b Marc Mendler a Miriam Maurer a Markus Waitz a Li Huang a Helmut D. Hummler a
a
Division of Neonatology and Pediatric Critical Care, Department of Pediatrics and Adolescent Medicine, Ulm University, Ulm, Germany; b Department of Pediatrics, Sohag University, Sohag, Egypt
Abstract Background: Pulse oximetry is widely used in intensive care and emergency conditions to monitor arterial oxygenation and to guide oxygen therapy. Objective: To study the reliability of pulse oximetry in comparison with CO-oximetry in newborn piglets during cardiopulmonary resuscitation (CPR). Methodology: In a prospective cohort study in 30 healthy newborn piglets, cardiac arrest was induced, and thereafter each piglet received CPR for 20 min. Arterial oxygen saturation was monitored continuously by pulse oximetry (SpO2). Arterial blood was analyzed for functional oxygenation (SaO2) every 2 min. SpO2 was compared with coinciding SaO2 values and bias considered whenever the difference (SpO2 – SaO2) was beyond ±5%. Results: Bias values were decreased at the baseline measurements (mean: 2.5 ± 4.6%) with higher precision and accuracy compared with values across the experiment. Two minutes after cardiac arrest, there was a marked decrease in precision and accuracy as well as an increase in bias up to 13 ± 34%, reaching
© 2014 S. Karger AG, Basel 1661–7800/14/1072–0113$39.50/0 E-Mail [email protected] www.karger.com/neo
a maximum of 45.6 ± 28.3% after 10 min over a mean SaO2 range of 29–58%. Conclusion: Pulse oximetry showed increased bias and decreased accuracy and precision during CPR in a model of neonatal cardiac arrest. We recommend further studies to clarify the exact mechanisms of these false readings to improve reliability of pulse oximetry during the marked desaturation and hypoperfusion found during CPR. © 2014 S. Karger AG, Basel
Introduction
Pulse oximetry was widely introduced in the early 1980s, first in perioperative care and then in neonatal, pediatric, and adult intensive care to monitor oxygen levels and to guide FiO2 adjustments [1, 2]. Motion, ambient light, hypoxemia, and hypoperfusion can affect the reliability of pulse oximetry [3]. The reliability of pulse oximetry during hypoxemia has been studied during desaturation episodes, breathing of hypoxic gas mixtures by healthy volunteers, and in children with cyanotic congenital heart disease [4–6]. The effect of low perfusion on pulse oximetry has been studied by using vasoconstrictive therapy, experimental occlusion, during hemorrhagic Mohammad Ahmad Hassan Division of Neonatology and Pediatric Critical Care Department of Pediatrics and Adolescent Medicine, Ulm University Eythstrasse 24, DE–89075 Ulm (Germany) E-Mail dmohamed81 @ gmail.com
Downloaded by: UCSF Library & CKM 169.230.243.252 - 12/11/2014 11:22:59 PM
Key Words Cardiopulmonary resuscitation · Neonate · Oximetry · Pulse oximetry · Hypoxemia
Materials and Methods All animals were cared for according to German law concerning animal protection and the protocol was approved by the appropriate German authorities. Anesthetized (continuous propofol infusion) newborn piglets were intubated and placed underneath an overhead warmer to maintain a core temperature of 39.0– 39.5 ° C, and on pressure-controlled ventilation using a Stephanie Ventilator (Stephan Medizintechnik GmbH, Gackenbach, Germany) with the following setting: FiO2: 0.3, PIP: 20 cm H2O, PEEP: 5 cm H2O, inspiratory time: 0.4 s, respiratory rate: 30/min. The rate was adjusted to maintain a PaCO2 within 35–45 mm Hg. Dextrose 2.5% with Na 94.5 mmol/l, K 9 mmol/l, and 1 U heparin/ml were administered at 8 ml/kg/h into a peripheral vein. A 3.5-Fr arterial femoral line was inserted for arterial blood sampling and for continuous blood pressure monitoring. A 3.5-Fr double-lumen catheter was introduced via the right femoral vein and placed in the inferior vena cava. A pulse oximetry sensor (LNOP NeoPt-L, Masimo SET; Masimo, Irvine, Calif., USA) was applied to the proximal tail, shielded, and connected to the pulse oximeter (Radical, Masimo SET V4.6.0.2; Masimo) to measure pulse rate and SpO2. Furthermore, 3 ECG electrodes were placed on the shaved chest. All signals were digitized and recorded simultaneously using a data recording system (DATAQ Instruments Inc., Akron, Ohio, USA). Approximately 15 min after instrumentation, baseline measurements were recorded and an arterial blood sample was drawn to measure SaO2, PaO2, PCO2, pH, and lactate by a blood gas analyzer (ABL 700 series; Radiometer, Copenhagen, Denmark). Thereafter, a bolus of 2 mmol/kg KCl was given and flushed with 2 ml normal saline via the central venous catheter to induce diastolic cardiac arrest [17]. Furthermore, 3 mmol KCl/kg/h were given continuously to maintain hyperkalemia. Asystole was defined as a loss of pulsatility in the arterial blood pressure trace and absence of regular ECG activity. Respiratory support was stopped for 30 s and started thereafter. External cardiac compressions (ECC) with the 2 thumb technique [18] were started after 30 s of respira
114
Neonatology 2015;107:113–119 DOI: 10.1159/000368178
tory support with the animals randomized to three different types of respiratory support: mechanical ventilation, ventilation using a T-piece, or ventilation using a self-inflating bag to assess the effects of these different techniques on gas exchange. For the purpose of this study, all animals were analyzed as one group. Respiratory support was initiated with an FiO2 of 0.21, as suggested by current clinical evidence [19] and continued for 10 min. FiO2 was then increased to 1.0 for another 10 min. Whereas current guidelines suggest increasing the FiO2 to 1.0 earlier, this approach allowed us to collect several blood samples to study gas exchange and SpO2 response during CPR. The ECC rate (3 compressions + 1 breath in 2-second cycles [13, 20]) was guided using a metronome (rate of 120/min). ECC was standardized to generate a systolic arterial blood pressure of 50 mm Hg. Resuscitative efforts were continued for a total of 20 min after asystole. Whenever the resuscitator sensed fatigue, another person took over ECC to maintain the CPR quality. Arterial blood samples were taken every 2 min and processed immediately. SpO2 measurements were analyzed taking the average of the 10 s that coincided with the arterial blood sampling every 2 min. The primary outcome was the difference between oxygen saturation as measured by pulse oximetry and CO-oximetry throughout the experimental time. Bias is the total systematic error and was calculated as the average difference between SpO2 and SaO2. An acceptable bias was defined to be within ±5%. Accuracy was defined as the closeness of agreement between test results and reference values. It was calculated as the accuracy root-mean-square (Arms) difference (SpO2 – SaO2). A higher Arms value reflects lower accuracy. Precision was defined as the closeness of agreement among independent test results obtained under stipulated conditions. It is recognized as the scatter of data points about the best fitting curve of agreement and depends on the random error. It was calculated as the standard deviation of residuals (Sres) from a linear regression model where higher values reflect low precision [21]. A modified Bland-Altman plot was used to compare time-dependent changes of the bias across the experimental time. The original Bland-Altman plot was used to assess the agreement between SpO2 and SaO2 using 95% limits of agreement (±2 SD of the bias) [22]. Paired t and Wilcoxon signed-ranks tests were used where appropriate to compare SpO2 and SaO2. Statistical analyses were performed using Excel and SPSS, version 19.0 (SPSS Inc., Chicago, Ill., USA).
Results
Thirty healthy newborn piglets were included in the study with a median age of 5 days (range: 2–11) and median weight of 1,875 g (range: 1,200–2,980). Table 1 shows potassium, sodium, glucose, and lactate levels at baseline and across the experiment. Potassium levels were maintained >10 mEq/l across the experiment. Lactate levels increased throughout the experiment, suggesting the presence of poor tissue perfusion during CPR. Table 2 shows the pulse oximetry signal dropout time for both SpO2 and pulse rate. Every subject had at least Hassan /Mendler /Maurer /Waitz /Huang / Hummler
Downloaded by: UCSF Library & CKM 169.230.243.252 - 12/11/2014 11:22:59 PM
hypotension, and in sepsis-associated hypoperfusion [7– 10]. During the transitional period and up to the second week of life, there is a poor correlation between skin color and arterial oxygen saturation [11, 12]. Therefore, pulse oximetry is recommended during neonatal resuscitation [13, 14]. Pulse oximetry performance during cardiopulmonary resuscitation (CPR) has been evaluated in a few adult studies. Moorthy et al. [15] reported that errors of pulse oximetry can lead to incorrect judgment of the indication for and adequacy of CPR, but the use of pulse oximetry during CPR in 20 adult patients significantly changed the management of 7 patients, 5 of whom survived [16]. The objective of this investigation was to study the reliability of pulse oximetry in comparison with CO-oximetry as the gold standard during CPR in newborn piglets.
Color version available online
FiO2 0.21–1.0
Oxygen saturation (%)
100 80
SpO2
60 40
SaO2
20
Fig. 1. Mean ± SD of SpO2 and SaO2 across
the experimental time with a table of the number of SpO2 values available at each time point of the experimental time. Time 0 is the baseline.
0
2
4
6
8 10 12 14 Experimental time (min)
16
18
20
30
30
29
28
28 28 28 28 SpO2 values obtained (n)
28
28
28
one episode of signal dropout and most had an episode with the onset of the cardiac arrest. The pulse rate was 120 ± 10 beats/min as expected in all animals during CPR and the pulse oximeter appeared to detect the effects of the individual breath between the ECCs. Figure 1 shows the mean ± SD of SpO2 and corresponding SaO2 measurements across the experimental time. Both SpO2 and SaO2 are in proximity at baseline. Two minutes after asystole, there was already a mean difference of 13% with SpO2 (71 ± 22.7%) versus SaO2 (58 ± 20.6%) and the difference continued to increase throughout the experiment. SaO2 showed a minimal increase after 10 min that coincided with the increase of the FiO2 from 0.21 to 1.0, while SpO2 grossly overestimated SaO2. The difference between the measurements of SpO2 and SaO2 was statistically significant (p < 0.001; paired t test) for all measurements across the 20 min of the experiment including the baseline measurements (p = 0.009; Wilcoxon signed-rank tests).
Table 1. Sodium (Na), potassium (K), glucose, and lactate levels at the baseline measurements and across the experiment
Bias, Accuracy and Precision Table 3 and figure 2 show the mean bias and variability of individual bias values throughout the experiment. The average bias and variability at baseline were small (2.5 ± 4.6%), but increased dramatically to 13 ± 34% after 2 min of resuscitation and increased further across the 20 min of CPR. The individual bias values were beyond ±5% at 269/313 measurements (86%) with 37/313 (12%) underestimated values and 232/313 (74%) overestimated SaO2 values. Accuracy and precision were high at baseline
Table 2. Pulse oximetry for SpO2 and pulse rate
Reliability of Pulse Oximetry during CPR
Neonatology 2015;107:113–119 DOI: 10.1159/000368178
Baseline 2 min 4 min 6 min 8 min 10 min 12 min 14 min 16 min 18 min 20 min
Na (mmol/l)
K (mmol/l)
Glucose (mg/dl)
Lactate (mmol/l)
132 ± 4 124 ± 6 127 ± 5 128 ± 5 128 ± 6 130 ± 5 130 ± 5 131 ± 5 131 ± 5 132 ± 5 132 ± 5
4±1 20 ± 2 18 ± 3 16 ± 4 15 ± 3 15 ± 3 15 ± 2 15 ± 2 15 ± 2 14 ± 2 15 ± 2
115 ± 31 113 ± 30 145 ± 38 168 ± 43 183 ± 51 199 ± 56 213 ± 62 223 ± 71 232 ± 75 235 ± 80 239 ± 84
2.3 ± 1.2 2.8 ± 1.3 4.4 ± 1.5 5.7 ± 1.6 6.7 ± 1.6 7.5 ± 1.8 8.2 ± 2.1 8.5 ± 2.1 8.9 ± 2.2 9.3 ± 2.4 9.5 ± 2.5
Values are means ± SD.
SpO2 Total dropout time, min Episodes of signal dropout, n Duration of episodes of signal dropout, min
Pulse rate
1.44 (0.29 – 19.89) 1.48 (0.61 – 19.68) 3.5 (1 – 31)
2 (1 – 15)
0.34 (0.03 – 14.65) 0.85 (0.13 – 15.35)
Values are presented as medians (min.–max.).
115
Downloaded by: UCSF Library & CKM 169.230.243.252 - 12/11/2014 11:22:59 PM
0
Color version available online
120
mean + 2 SD
SpO2 – SaO2 (%)
80
40 mean 0
–40
mean – 2 SD
Fig. 2. Modified Bland-Altman plot of the
measured bias values (SpO2 – SaO2) across the experimental time. Each dot refers to one measurement at each time point. Mean bias along with the mean ± 2 SD values of these individual bias values are also plotted. The y-axis has been extended to include all the individual values, including the outliers. Time 0 is the baseline.
–80
–120 –2
0
2
4
6
8
10 12 Time (min)
14
16
18
20
22
Table 3. Number of matched SpO2 and SaO2 measurements available, SpO2 and SaO2 values and bias (mean ± SD), accuracy (Arms), and precision (Sres) across the experimental time
Baseline 2 min 4 min 6 min 8 min 10 min 12 min 14 min 16 min 18 min 20 min
SpO2 + SaO2 measurements, n
SpO2
SaO2
Bias
Accuracy (Arms)
Precision (Sres)
30 30 29 28 28 28 28 28 28 28 28
92.8 ± 5.3 71 ± 22.7 64 ± 25 71.2 ± 21.8 73.7 ± 20 75.6 ± 20.3 75.4 ± 16.6 74.3 ± 19 77.3 ± 15.8 74.8 ± 18.5 72.7 ± 22.5
90.3 ± 6.3 58 ± 20.6 39.2 ± 12.7 35.8 ± 14.7 32.8 ± 17.3 29.3 ± 15.4 38.9 ± 21.7 40.8 ± 20.9 39.4 ± 21.3 38.3 ± 21.5 37.7 ± 21.7
2.49 ± 4.58 12.9 ± 33.8 24.1 ± 25.3 35.6 ± 27.1 40.2 ± 28.1 45.6 ± 28.3 35.2 ± 32.5 32.4 ± 30.4 36.9 ± 28.2 36.8 ± 31.1 34.9 ± 32.3
5.2 35.7 34.6 44.4 48.8 53.4 47.5 44.1 46.2 47.8 47.7
3.8 22.6 24.8 22.2 20.6 20.2 15.5 19.3 16 18.5 22.8
(Arms 5.1% and Sres 3.8%), but then dramatically decreased at 2 min of CPR (Arms 35.7% and Sres 22.5%) and remained poor thereafter. Limits of Agreement Figure 3 represents the original Bland-Altman plots assessing the agreement between SpO2 and SaO2 at baseline, at 10 min, and at 20 min after the onset of the ex116
Neonatology 2015;107:113–119 DOI: 10.1159/000368178
periment. The limits of agreement at baseline were small and indicate that SpO2 can be 11.5% above or 6.5% below SaO2. At 10 min, SpO2 was grossly overestimating SaO2 values in most animals with wide limits of agreement indicating that SpO2 can be almost 100% above or 10% below SaO2 and at 20 min can be 99% above or 30% below SaO2. Bias levels did not systemically depend on the oxygenation status, as measured by SaO2. Hassan /Mendler /Maurer /Waitz /Huang / Hummler
Downloaded by: UCSF Library & CKM 169.230.243.252 - 12/11/2014 11:22:59 PM
Mean SpO2 was significantly higher than mean SaO2 (p < 0.01; paired t test or Wilcoxon signed-rank test where appropriate).
Discussion 100
Baseline bias
Mean + 2 SD 0
Mean Mean – 2 SD
–50
–100 75
a
80
85 90 95 SpO2 + SaO2/2 at baseline
100
100
Mean + 2 SD
50
Mean
0 Mean – 2 SD –50
–100 30
b
40
50 60 70 SpO2 + SaO2/2 at 10 min
100
80
Mean + 2 SD
50 Mean 0 Mean – 2 SD
–50
–100
c
20
40 60 SpO2 + SaO2/2 at 20 min
80
Fig. 3. Bland-Altman plots of the bias against the average of SpO2 and SaO2 at the baseline measurements (a), at 10 min of the experiment (b) and at 20 min (c). One dot refers to one animal (n = 30).
Neonatology 2015;107:113–119 DOI: 10.1159/000368178
117
Downloaded by: UCSF Library & CKM 169.230.243.252 - 12/11/2014 11:22:59 PM
Reliability of Pulse Oximetry during CPR
Bias (SpO2 – SaO2) at 10 min
50
Bias (SpO2 – SaO2) at 20 min
Pulse oximetry is considered by some authors as the fifth vital sign [23]. Several limitations of pulse oximetry represent a major challenge especially in emergency situations. We tried to overcome many of the known limitations by keeping a core temperature of 39.0–39.5 ° C, using dim light throughout the experiment, shielding of the pulse oximeter sensor, good fixation of the animal, use of continuous sedation, maintaining a systolic arterial blood pressure of 50 mm Hg with CPR by an experienced resuscitation team, and the use of one of the most recently available pulse oximetry devices to limit the effects of motion and low perfusion [24]. Additionally, major electrolyte disturbances (other than hyperkalemia) and hypoglycemia were prevented by continuous intravenous fluid infusion (table 1). In summary, the study setting was standardized to provide an environment of optimized, consistent CPR for all study subjects. We demonstrated mean bias between SpO2 and SaO2 of 2.5 ± 4.6% at baseline with mean SaO2 saturation of 90% and normal hemodynamics. Although this bias comparing SpO2 with SaO2 values was significant from the statistical point of view, it was somewhat close to values reported before with the mean bias
Neonatology 2015;107:113–119 DOI: 10.1159/000368178
Reliability of Pulse Oximetry during Cardiopulmonary Resuscitation in a Piglet Model of Neonatal Cardiac Arrest Mohammad Ahmad Hassan a, b Marc Mendler a Miriam Maurer a Markus Waitz a Li Huang a Helmut D. Hummler a
a
Division of Neonatology and Pediatric Critical Care, Department of Pediatrics and Adolescent Medicine, Ulm University, Ulm, Germany; b Department of Pediatrics, Sohag University, Sohag, Egypt
Abstract Background: Pulse oximetry is widely used in intensive care and emergency conditions to monitor arterial oxygenation and to guide oxygen therapy. Objective: To study the reliability of pulse oximetry in comparison with CO-oximetry in newborn piglets during cardiopulmonary resuscitation (CPR). Methodology: In a prospective cohort study in 30 healthy newborn piglets, cardiac arrest was induced, and thereafter each piglet received CPR for 20 min. Arterial oxygen saturation was monitored continuously by pulse oximetry (SpO2). Arterial blood was analyzed for functional oxygenation (SaO2) every 2 min. SpO2 was compared with coinciding SaO2 values and bias considered whenever the difference (SpO2 – SaO2) was beyond ±5%. Results: Bias values were decreased at the baseline measurements (mean: 2.5 ± 4.6%) with higher precision and accuracy compared with values across the experiment. Two minutes after cardiac arrest, there was a marked decrease in precision and accuracy as well as an increase in bias up to 13 ± 34%, reaching
© 2014 S. Karger AG, Basel 1661–7800/14/1072–0113$39.50/0 E-Mail [email protected] www.karger.com/neo
a maximum of 45.6 ± 28.3% after 10 min over a mean SaO2 range of 29–58%. Conclusion: Pulse oximetry showed increased bias and decreased accuracy and precision during CPR in a model of neonatal cardiac arrest. We recommend further studies to clarify the exact mechanisms of these false readings to improve reliability of pulse oximetry during the marked desaturation and hypoperfusion found during CPR. © 2014 S. Karger AG, Basel
Introduction
Pulse oximetry was widely introduced in the early 1980s, first in perioperative care and then in neonatal, pediatric, and adult intensive care to monitor oxygen levels and to guide FiO2 adjustments [1, 2]. Motion, ambient light, hypoxemia, and hypoperfusion can affect the reliability of pulse oximetry [3]. The reliability of pulse oximetry during hypoxemia has been studied during desaturation episodes, breathing of hypoxic gas mixtures by healthy volunteers, and in children with cyanotic congenital heart disease [4–6]. The effect of low perfusion on pulse oximetry has been studied by using vasoconstrictive therapy, experimental occlusion, during hemorrhagic Mohammad Ahmad Hassan Division of Neonatology and Pediatric Critical Care Department of Pediatrics and Adolescent Medicine, Ulm University Eythstrasse 24, DE–89075 Ulm (Germany) E-Mail dmohamed81 @ gmail.com
Downloaded by: UCSF Library & CKM 169.230.243.252 - 12/11/2014 11:22:59 PM
Key Words Cardiopulmonary resuscitation · Neonate · Oximetry · Pulse oximetry · Hypoxemia
Materials and Methods All animals were cared for according to German law concerning animal protection and the protocol was approved by the appropriate German authorities. Anesthetized (continuous propofol infusion) newborn piglets were intubated and placed underneath an overhead warmer to maintain a core temperature of 39.0– 39.5 ° C, and on pressure-controlled ventilation using a Stephanie Ventilator (Stephan Medizintechnik GmbH, Gackenbach, Germany) with the following setting: FiO2: 0.3, PIP: 20 cm H2O, PEEP: 5 cm H2O, inspiratory time: 0.4 s, respiratory rate: 30/min. The rate was adjusted to maintain a PaCO2 within 35–45 mm Hg. Dextrose 2.5% with Na 94.5 mmol/l, K 9 mmol/l, and 1 U heparin/ml were administered at 8 ml/kg/h into a peripheral vein. A 3.5-Fr arterial femoral line was inserted for arterial blood sampling and for continuous blood pressure monitoring. A 3.5-Fr double-lumen catheter was introduced via the right femoral vein and placed in the inferior vena cava. A pulse oximetry sensor (LNOP NeoPt-L, Masimo SET; Masimo, Irvine, Calif., USA) was applied to the proximal tail, shielded, and connected to the pulse oximeter (Radical, Masimo SET V4.6.0.2; Masimo) to measure pulse rate and SpO2. Furthermore, 3 ECG electrodes were placed on the shaved chest. All signals were digitized and recorded simultaneously using a data recording system (DATAQ Instruments Inc., Akron, Ohio, USA). Approximately 15 min after instrumentation, baseline measurements were recorded and an arterial blood sample was drawn to measure SaO2, PaO2, PCO2, pH, and lactate by a blood gas analyzer (ABL 700 series; Radiometer, Copenhagen, Denmark). Thereafter, a bolus of 2 mmol/kg KCl was given and flushed with 2 ml normal saline via the central venous catheter to induce diastolic cardiac arrest [17]. Furthermore, 3 mmol KCl/kg/h were given continuously to maintain hyperkalemia. Asystole was defined as a loss of pulsatility in the arterial blood pressure trace and absence of regular ECG activity. Respiratory support was stopped for 30 s and started thereafter. External cardiac compressions (ECC) with the 2 thumb technique [18] were started after 30 s of respira
114
Neonatology 2015;107:113–119 DOI: 10.1159/000368178
tory support with the animals randomized to three different types of respiratory support: mechanical ventilation, ventilation using a T-piece, or ventilation using a self-inflating bag to assess the effects of these different techniques on gas exchange. For the purpose of this study, all animals were analyzed as one group. Respiratory support was initiated with an FiO2 of 0.21, as suggested by current clinical evidence [19] and continued for 10 min. FiO2 was then increased to 1.0 for another 10 min. Whereas current guidelines suggest increasing the FiO2 to 1.0 earlier, this approach allowed us to collect several blood samples to study gas exchange and SpO2 response during CPR. The ECC rate (3 compressions + 1 breath in 2-second cycles [13, 20]) was guided using a metronome (rate of 120/min). ECC was standardized to generate a systolic arterial blood pressure of 50 mm Hg. Resuscitative efforts were continued for a total of 20 min after asystole. Whenever the resuscitator sensed fatigue, another person took over ECC to maintain the CPR quality. Arterial blood samples were taken every 2 min and processed immediately. SpO2 measurements were analyzed taking the average of the 10 s that coincided with the arterial blood sampling every 2 min. The primary outcome was the difference between oxygen saturation as measured by pulse oximetry and CO-oximetry throughout the experimental time. Bias is the total systematic error and was calculated as the average difference between SpO2 and SaO2. An acceptable bias was defined to be within ±5%. Accuracy was defined as the closeness of agreement between test results and reference values. It was calculated as the accuracy root-mean-square (Arms) difference (SpO2 – SaO2). A higher Arms value reflects lower accuracy. Precision was defined as the closeness of agreement among independent test results obtained under stipulated conditions. It is recognized as the scatter of data points about the best fitting curve of agreement and depends on the random error. It was calculated as the standard deviation of residuals (Sres) from a linear regression model where higher values reflect low precision [21]. A modified Bland-Altman plot was used to compare time-dependent changes of the bias across the experimental time. The original Bland-Altman plot was used to assess the agreement between SpO2 and SaO2 using 95% limits of agreement (±2 SD of the bias) [22]. Paired t and Wilcoxon signed-ranks tests were used where appropriate to compare SpO2 and SaO2. Statistical analyses were performed using Excel and SPSS, version 19.0 (SPSS Inc., Chicago, Ill., USA).
Results
Thirty healthy newborn piglets were included in the study with a median age of 5 days (range: 2–11) and median weight of 1,875 g (range: 1,200–2,980). Table 1 shows potassium, sodium, glucose, and lactate levels at baseline and across the experiment. Potassium levels were maintained >10 mEq/l across the experiment. Lactate levels increased throughout the experiment, suggesting the presence of poor tissue perfusion during CPR. Table 2 shows the pulse oximetry signal dropout time for both SpO2 and pulse rate. Every subject had at least Hassan /Mendler /Maurer /Waitz /Huang / Hummler
Downloaded by: UCSF Library & CKM 169.230.243.252 - 12/11/2014 11:22:59 PM
hypotension, and in sepsis-associated hypoperfusion [7– 10]. During the transitional period and up to the second week of life, there is a poor correlation between skin color and arterial oxygen saturation [11, 12]. Therefore, pulse oximetry is recommended during neonatal resuscitation [13, 14]. Pulse oximetry performance during cardiopulmonary resuscitation (CPR) has been evaluated in a few adult studies. Moorthy et al. [15] reported that errors of pulse oximetry can lead to incorrect judgment of the indication for and adequacy of CPR, but the use of pulse oximetry during CPR in 20 adult patients significantly changed the management of 7 patients, 5 of whom survived [16]. The objective of this investigation was to study the reliability of pulse oximetry in comparison with CO-oximetry as the gold standard during CPR in newborn piglets.
Color version available online
FiO2 0.21–1.0
Oxygen saturation (%)
100 80
SpO2
60 40
SaO2
20
Fig. 1. Mean ± SD of SpO2 and SaO2 across
the experimental time with a table of the number of SpO2 values available at each time point of the experimental time. Time 0 is the baseline.
0
2
4
6
8 10 12 14 Experimental time (min)
16
18
20
30
30
29
28
28 28 28 28 SpO2 values obtained (n)
28
28
28
one episode of signal dropout and most had an episode with the onset of the cardiac arrest. The pulse rate was 120 ± 10 beats/min as expected in all animals during CPR and the pulse oximeter appeared to detect the effects of the individual breath between the ECCs. Figure 1 shows the mean ± SD of SpO2 and corresponding SaO2 measurements across the experimental time. Both SpO2 and SaO2 are in proximity at baseline. Two minutes after asystole, there was already a mean difference of 13% with SpO2 (71 ± 22.7%) versus SaO2 (58 ± 20.6%) and the difference continued to increase throughout the experiment. SaO2 showed a minimal increase after 10 min that coincided with the increase of the FiO2 from 0.21 to 1.0, while SpO2 grossly overestimated SaO2. The difference between the measurements of SpO2 and SaO2 was statistically significant (p < 0.001; paired t test) for all measurements across the 20 min of the experiment including the baseline measurements (p = 0.009; Wilcoxon signed-rank tests).
Table 1. Sodium (Na), potassium (K), glucose, and lactate levels at the baseline measurements and across the experiment
Bias, Accuracy and Precision Table 3 and figure 2 show the mean bias and variability of individual bias values throughout the experiment. The average bias and variability at baseline were small (2.5 ± 4.6%), but increased dramatically to 13 ± 34% after 2 min of resuscitation and increased further across the 20 min of CPR. The individual bias values were beyond ±5% at 269/313 measurements (86%) with 37/313 (12%) underestimated values and 232/313 (74%) overestimated SaO2 values. Accuracy and precision were high at baseline
Table 2. Pulse oximetry for SpO2 and pulse rate
Reliability of Pulse Oximetry during CPR
Neonatology 2015;107:113–119 DOI: 10.1159/000368178
Baseline 2 min 4 min 6 min 8 min 10 min 12 min 14 min 16 min 18 min 20 min
Na (mmol/l)
K (mmol/l)
Glucose (mg/dl)
Lactate (mmol/l)
132 ± 4 124 ± 6 127 ± 5 128 ± 5 128 ± 6 130 ± 5 130 ± 5 131 ± 5 131 ± 5 132 ± 5 132 ± 5
4±1 20 ± 2 18 ± 3 16 ± 4 15 ± 3 15 ± 3 15 ± 2 15 ± 2 15 ± 2 14 ± 2 15 ± 2
115 ± 31 113 ± 30 145 ± 38 168 ± 43 183 ± 51 199 ± 56 213 ± 62 223 ± 71 232 ± 75 235 ± 80 239 ± 84
2.3 ± 1.2 2.8 ± 1.3 4.4 ± 1.5 5.7 ± 1.6 6.7 ± 1.6 7.5 ± 1.8 8.2 ± 2.1 8.5 ± 2.1 8.9 ± 2.2 9.3 ± 2.4 9.5 ± 2.5
Values are means ± SD.
SpO2 Total dropout time, min Episodes of signal dropout, n Duration of episodes of signal dropout, min
Pulse rate
1.44 (0.29 – 19.89) 1.48 (0.61 – 19.68) 3.5 (1 – 31)
2 (1 – 15)
0.34 (0.03 – 14.65) 0.85 (0.13 – 15.35)
Values are presented as medians (min.–max.).
115
Downloaded by: UCSF Library & CKM 169.230.243.252 - 12/11/2014 11:22:59 PM
0
Color version available online
120
mean + 2 SD
SpO2 – SaO2 (%)
80
40 mean 0
–40
mean – 2 SD
Fig. 2. Modified Bland-Altman plot of the
measured bias values (SpO2 – SaO2) across the experimental time. Each dot refers to one measurement at each time point. Mean bias along with the mean ± 2 SD values of these individual bias values are also plotted. The y-axis has been extended to include all the individual values, including the outliers. Time 0 is the baseline.
–80
–120 –2
0
2
4
6
8
10 12 Time (min)
14
16
18
20
22
Table 3. Number of matched SpO2 and SaO2 measurements available, SpO2 and SaO2 values and bias (mean ± SD), accuracy (Arms), and precision (Sres) across the experimental time
Baseline 2 min 4 min 6 min 8 min 10 min 12 min 14 min 16 min 18 min 20 min
SpO2 + SaO2 measurements, n
SpO2
SaO2
Bias
Accuracy (Arms)
Precision (Sres)
30 30 29 28 28 28 28 28 28 28 28
92.8 ± 5.3 71 ± 22.7 64 ± 25 71.2 ± 21.8 73.7 ± 20 75.6 ± 20.3 75.4 ± 16.6 74.3 ± 19 77.3 ± 15.8 74.8 ± 18.5 72.7 ± 22.5
90.3 ± 6.3 58 ± 20.6 39.2 ± 12.7 35.8 ± 14.7 32.8 ± 17.3 29.3 ± 15.4 38.9 ± 21.7 40.8 ± 20.9 39.4 ± 21.3 38.3 ± 21.5 37.7 ± 21.7
2.49 ± 4.58 12.9 ± 33.8 24.1 ± 25.3 35.6 ± 27.1 40.2 ± 28.1 45.6 ± 28.3 35.2 ± 32.5 32.4 ± 30.4 36.9 ± 28.2 36.8 ± 31.1 34.9 ± 32.3
5.2 35.7 34.6 44.4 48.8 53.4 47.5 44.1 46.2 47.8 47.7
3.8 22.6 24.8 22.2 20.6 20.2 15.5 19.3 16 18.5 22.8
(Arms 5.1% and Sres 3.8%), but then dramatically decreased at 2 min of CPR (Arms 35.7% and Sres 22.5%) and remained poor thereafter. Limits of Agreement Figure 3 represents the original Bland-Altman plots assessing the agreement between SpO2 and SaO2 at baseline, at 10 min, and at 20 min after the onset of the ex116
Neonatology 2015;107:113–119 DOI: 10.1159/000368178
periment. The limits of agreement at baseline were small and indicate that SpO2 can be 11.5% above or 6.5% below SaO2. At 10 min, SpO2 was grossly overestimating SaO2 values in most animals with wide limits of agreement indicating that SpO2 can be almost 100% above or 10% below SaO2 and at 20 min can be 99% above or 30% below SaO2. Bias levels did not systemically depend on the oxygenation status, as measured by SaO2. Hassan /Mendler /Maurer /Waitz /Huang / Hummler
Downloaded by: UCSF Library & CKM 169.230.243.252 - 12/11/2014 11:22:59 PM
Mean SpO2 was significantly higher than mean SaO2 (p < 0.01; paired t test or Wilcoxon signed-rank test where appropriate).
Discussion 100
Baseline bias
Mean + 2 SD 0
Mean Mean – 2 SD
–50
–100 75
a
80
85 90 95 SpO2 + SaO2/2 at baseline
100
100
Mean + 2 SD
50
Mean
0 Mean – 2 SD –50
–100 30
b
40
50 60 70 SpO2 + SaO2/2 at 10 min
100
80
Mean + 2 SD
50 Mean 0 Mean – 2 SD
–50
–100
c
20
40 60 SpO2 + SaO2/2 at 20 min
80
Fig. 3. Bland-Altman plots of the bias against the average of SpO2 and SaO2 at the baseline measurements (a), at 10 min of the experiment (b) and at 20 min (c). One dot refers to one animal (n = 30).
Neonatology 2015;107:113–119 DOI: 10.1159/000368178
117
Downloaded by: UCSF Library & CKM 169.230.243.252 - 12/11/2014 11:22:59 PM
Reliability of Pulse Oximetry during CPR
Bias (SpO2 – SaO2) at 10 min
50
Bias (SpO2 – SaO2) at 20 min
Pulse oximetry is considered by some authors as the fifth vital sign [23]. Several limitations of pulse oximetry represent a major challenge especially in emergency situations. We tried to overcome many of the known limitations by keeping a core temperature of 39.0–39.5 ° C, using dim light throughout the experiment, shielding of the pulse oximeter sensor, good fixation of the animal, use of continuous sedation, maintaining a systolic arterial blood pressure of 50 mm Hg with CPR by an experienced resuscitation team, and the use of one of the most recently available pulse oximetry devices to limit the effects of motion and low perfusion [24]. Additionally, major electrolyte disturbances (other than hyperkalemia) and hypoglycemia were prevented by continuous intravenous fluid infusion (table 1). In summary, the study setting was standardized to provide an environment of optimized, consistent CPR for all study subjects. We demonstrated mean bias between SpO2 and SaO2 of 2.5 ± 4.6% at baseline with mean SaO2 saturation of 90% and normal hemodynamics. Although this bias comparing SpO2 with SaO2 values was significant from the statistical point of view, it was somewhat close to values reported before with the mean bias
