Validation of Noninvasive Hemoglobin Measurement by Pulse Co-Oximeter in Newborn Infants

ORIGINAL ARTICLE Validation of noninvasive hemoglobin measurement by pulse co-oximeter in newborn infants

C Nicholas1,2, R George2, S Sardesai1, M Durand1, R Ramanathan1,2 and R Cayabyab1

OBJECTIVE: To describe the accuracy of noninvasive hemoglobin (Hb) obtained with pulse co-oximeter (SpHb) compared with total Hb (tHb) from laboratory co-oximeter in neonates. STUDY DESIGN: Neonates with birth weight (BW) o3000 g admitted to LAC+USC Medical Center neonatal intensive care unit were included. SpHb was recorded using Masimo Radical-7 and compared with tHb. A total of three data sets were obtained for each patient. Regression analysis and Bland–Altman analysis were performed. RESULT: Sixty-one patients (mean ± s.d., BW 1177 ± 610 g and gestational age 28.7 ± 3.9 weeks) were enrolled. The mean tHb value was 13.9 ± 2.0 g dl − 1 and the mean SpHb was 14.0 ± 2.0 g dl − 1. There was a moderately positive correlation between SpHb and tHb (r = 0.66, Po0.001) with a bias and precision of − 0.09 ± 1.67 g dl − 1. Data from a subgroup of infants with gestational age ⩽ 32 weeks (52/61 patients) were analyzed, and the correlation coefﬁcient was moderately positive (r = 0.69, Po0.001) with a bias and precision of − 0.23 ± 1.60 g dl − 1. CONCLUSION: Our results suggest that noninvasive SpHb may be considered as an adjunct to invasive tHb measurements in newborn infants o3000 g especially in preterm infants ⩽ 32 weeks of gestation. Journal of Perinatology (2015) 35, 617–620; doi:10.1038/jp.2015.12; published online 5 March 2015

INTRODUCTION accuracy and reliability of Hb obtained by pulse co-oximeter in Total hemoglobin (tHb) measurement is one of the most com- newborns. Previous studies have suggested that noninvasive Hb monly performed laboratory tests in patients treated in both acute measurements with pulse co-oximetry may provide clinically 7,8 and outpatient settings. This test is performed 4400 million times acceptable accuracy in the neonatal population. The aim of this per year in the United States.1 It is the most frequently ordered study was to compare the accuracy of noninvasive Hb obtained laboratory test in critical care settings such as the Neonatal with pulse co-oximeter (SpHb) with tHb from laboratory co- Intensive Care Unit (NICU) to diagnose anemia. oximeter during routine blood sampling of preterm and term The Masimo Rainbow co-oximeter technology uses spectro- neonates with a birth weight (BW) o3000 g. photometry to noninvasively measure tHb (SpHb), carboxyhemo- globin, methemoglobin, oxyhemoglobin and heart rate.2 The system emits light in both the red and infrared spectrum using METHODS multiple (7+) wavelengths of light in a pulsatile system to distin- This is a prospective observational study of infants with BW o3000 g guish between and enumerate the different types of hemoglobin admitted to the NICU at LAC+USC Medical Center. The Institutional Review (Hb). The extinction coefficient, defining the proportion of light Board at our institution approved the study and informed consent was ’ absorbed at a given wavelength, varies not only among the obtained from the subjects parents. Inclusion criteria included: (1) infants o different types of Hb but also at different wavelengths along the admitted to our NICU with BW 3000 g; (2) receiving some form of respiratory support requiring blood gases; and (3) parental consent spectrum. The algorithm uses these known extinction coefficients ⩾ fi obtained. Those infants with a BW 3000 g, major congenital anomalies at each speci ed wavelength to quantify each type of Hb based or skin disorders were excluded. 2 on the light absorbed at the sensor. Although the extinction As part of clinical care in these critically ill infants, serial blood gas coefficients of fetal Hb are shifted only slightly from that of adult analyses were obtained to determine pulmonary gas exchange. Blood Hb at most wavelengths, it is currently unknown whether this shift gases were collected per NICU standard as an arterial sample if arterial line will significantly affect the accuracy of the device when used on was in place and a capillary sample in those patients without arterial line. infants at an early age at a time when fetal Hb is the predominant Part of the blood gas analysis was determination of tHb with the use of form of Hb. laboratory co-oximeter (Gem Premier 4000 Instrumentation Laboratory, Bedford, MA, USA). At the time of the blood draw for blood gas analysis, Studies performed to compare Hb values obtained with the the R1 20 L pulse-oximeter sensor was attached to one of the extremities use of noninvasive Hb monitor and laboratory co-oximeter have of the patient and connected to the Masimo Radical-7 pulse-oximeter shown a clinically acceptable accuracy of noninvasive Hb mea- (Rainbow Pulse Co-Oximeter; Masimo, Irvine, CA, USA) to measure oxygen 3,4 surements in healthy adults, and those admitted to the saturation, perfusion index and noninvasive tHb (SpHb). The tHb measured intensive care unit;5,6 however, limited data are available on the by a laboratory co-oximeter was used as the gold standard to compare the

1Division of Neonatal Medicine, Department of Pediatrics, LAC+USC Medical Center, University of Southern California, Los Angeles, CA, USA and 2Division of Neonatal Medicine, Department of Pediatrics, Children’s Hospital Los Angeles, Keck School of Medicine, University of Southern California, Los Angeles, CA, USA. Correspondence: Dr C Nicholas, Division of Neonatal Medicine, Department of Pediatrics, LAC+USC Medical Center, Keck School of Medicine, University of Southern California, 1200N State Street, IRD Building 820, Los Angeles 90033, CA, USA. E-mail: [email protected] This work was presented in part at the Pediatric Academic Societies annual meeting, Washington, DC, May 2013. Received 1 October 2014; revised 9 January 2015; accepted 13 January 2015; published online 5 March 2015 Noninvasive hemoglobin measurements in newborns C Nicholas et al 618 accuracy of SpHb measured by the pulse co-oximeter. Laboratory co- 44 ± 10 mm Hg. Measured pH was 7.32 ± 0.06 with 41/52 patients oximeters are considered to be the most accurate and have the highest receiving endotracheal or nasal ventilation (Table 1). The majority degree of interdevice reliability of all the methods available to measure 9 of the participants were hemodynamically stable at the time of the tHb. A total of three data sets (tHb and SpHb) at three time points were study with only 3/52 patients receiving dopamine or dobutamine obtained for each patient enrolled. Both tHb and SpHb were recorded at support for hypotension. each time point. Demographic data including BW, gestational age, postnatal age, When focusing on this subgroup of 52 premature infants, 144 paired samples (68 arterial and 76 capillary samples) showed a tHb respiratory support and vital signs including body temperature were also − 1 − 1 recorded. Arterial blood pressure was recorded in patients with arterial of 13.9 ± 2.0 g dl (range 9.7 to 19.0 g dl ) and a SpHb value of − 1 − 1 lines in place for clinical care and by cuff in those patients without 14.1 ± 2.1 g dl (range 8.3 to 8.9 g dl , P = 0.09; Table 2). There arterial lines. was a moderately positive correlation between SpHb and tHb measurements (r = 0.69, Po0.001; Figure 1). Bland–Altman Statistical analysis analysis showed a good agreement of paired Hb differences. ’ The bias and precision for the tHb and SpHb values were Two-tailed, Student s t-tests were used for analysis of continuous, normally − − 1 distributed variables and the Wilcoxon-signed ranks test where appro- 0.23 ± 1.60 g dl (Figure 2). priate. Regression analysis was performed to determine the relationship between paired Hb measurements (tHb and SpHb). The mean and s.d. of the paired Hb differences (bias and precision) were calculated using Bland– DISCUSSION Altman analysis.10 Values are presented as mean ± s.d., unless otherwise The availability of accurate, noninvasive Hb monitoring would indicated. Statistical significance was set at Po0.05 for all tests. greatly impact the care of newborns in the intensive care setting, especially the very preterm infants whose blood volume is quite RESULTS limited. Currently Hb levels in the NICU are typically obtained through blood draws that are invasive, time-consuming and can A total of 61 infants with a BW of 1177 ± 610 g (mean ± s.d.), be painful. Laboratory blood testing provides data at one time gestational age of 28.7 ± 3.9 weeks and median postnatal age of point rather than a continuous measurement, and the cumulative 4 days (interquartile range 2 to 10 days) were studied between blood loss with each of these tests in the premature population February 2012 and March 2014. There was a slight male may contribute to clinically significant anemia requiring blood predominance at 54%. Seventy-two percent of those enrolled transfusion. The potential to reduce blood transfusions has been had a BW of o1500 g (44/61) and 48% of subjects had BW of described by Ehrenfeld et al11 in a study of 327 adult patients o1000 g (29/61). Eighty-five percent of infants enrolled had a undergoing orthopedic surgery. They observed that when SpHb gestational age of ⩽ 32 completed weeks (52/61). Demographic monitoring was added to standard care, there was a decrease in and clinical data of our total study population are shown in Table 1. One hundred sixty-seven paired samples from 61 patients were − 1 analyzed. The mean tHb value was 13.9 ± 2.0 g dl (range 9.7 to Table 2. Analysis of the differences between tHb and SpHb values for − 1 − 1 19.0 g dl ) and the SpHb value was 14.0 ± 2.0 g dl (range 8.3 to total population and preterm infants ⩽ 32 weeks 18.9 g dl − 1, P = 0.48; Table 2). There was a moderately positive correlation between SpHb and tHb measurements (r = 0.66, Hemoglobin Correlation P- value Bias ± o (g dl − 1) coefficient precision P 0.001). There was also a good agreement of paired Hb differ- − 1 ences. The bias (mean difference) and precision (s.d. of paired (r) (g dl ) − − 1 differences) for the tHb and SpHb values were 0.09 ± 1.67 g dl Total (n = 61) paired values = 167 (Table 2). a ± tHb 13.9 2.0 o − ± From the total study population, 52/61 infants were premature SpHba 14.0 ± 2.0 0.66 0.001 0.09 1.67 with a gestational age of ⩽ 32 completed weeks. The BW of this subgroup was 1020 ± 439 g and gestational age was 27.6 ± Preterm (n = 52) paired values = 144 2.9 weeks. This subgroup of patients was also studied early in tHba 13.9 ± 2.0 a 0.69 o0.001 − 0.23 ± 1.60 life at a median postnatal age of 4 days (interquartile range 3 to SpHb 14.1 ±2.1 10 days). At the time of the study, these infants had a tem- Abbreviations: Hb, hemoglobin; tHb, total hemoglobin; SpHb, noninvasive perature of 36.6 ± 0.3 °C, heart rate of 155 ± 16 beats per minute, hemoglobin. aMean ± s.d. systolic blood pressure 60 ± 12 mm Hg and mean blood pressure

Table 1. Demographic and clinical data for total population and infants ⩽ 32 weeks of gestation

Total population (n = 61) Infants ⩽ 32 weeks (n = 52)

Mean ± s.d. Range Mean ± s.d. Range

Gestational age (weeks) 28.7 ± 3.9 23–41 27.6 ± 2.9 23–32 Birth weight (g) 1177 ± 610 330–2795 1020 ± 439 330–2150 Postnatal age (days)a 4(2–10)a 1–66 4 (3-10)a 1–66 Temperature (°C) 36.6 ± 0.3 35.6–37.3 36.6 ± 0.3 35.6–37.3 Heart rate (b.p.m.) 154 ± 17 116–209 155 ± 16 116–197 Systolic BP (mmHg) 60 ± 12 32–92 60 ± 12 32–85 Mean BP (mmHg) 44 ± 922–76 44 ± 10 22–76 pH 7.33 ± 0.06 7.12–7.48 7.32 ± 0.06 7.12–7.48 FiO2 0.25 ± 0.07 0.21–0.50 0.25 ± 0.06 0.21–0.50 Abbreviations: BP, blood pressure; b.p.m., beats per minute. aMedian (interquartile range)

Journal of Perinatology (2015), 617 – 620 © 2015 Nature America, Inc. Noninvasive hemoglobin measurements in newborns C Nicholas et al 619 SpHb monitoring with the Radical-7 co-oximeter was noted to give lower values when compared with laboratory measurements in adult patients during hemorrhagic surgery.17 It also showed variable bias (influenced by tHb concentration) in children undergoing neurosurgery18 and over the course of elective cesarean sections, as well as the post-operative time period.19 The data remain conflicting as other studies have shown SpHb to be accurate within 1.0 g dl − 1 (1 s.d.) compared with laboratory co- oximeter tHb measurements in adult subjects undergoing hemodilution.20 In addition, several other studies have reported a clinically acceptable accuracy of noninvasive SpHb in healthy adults3,4 and those admitted to the intensive care unit.5,6 Our study of neonates and specifically premature infants shows results similar to previous studies reporting a good agreement between the paired SpHb and tHb samples in newborn infants.7 Bland–Altman analysis shows a clinically acceptable agreement with a bias of − 0.09 g dl − 1 in our total population and − 0.23 g dl − 1 in the preterm cohort. The agreement in the preterm cohort is further strengthened as the study results shows 95.8% of paired Figure 1. Relationship between total hemoglobin (tHb) and non- ’ – invasive hemoglobin (SpHb) in 52 patients ⩽ 32 weeks of gestation data falling within two s.d. s of the mean difference on the Bland (144 paired samples). The correlation coefficient (r) of tHb versus Altman plot. This is similar to that previously reported in the 7 SpHb was moderately positive (r = 0.69, P o0.001). neonatal population by Jung et al. at 94.8% of paired data falling within two s.d.’s of the mean difference. There was a moderately positive correlation of paired Hb values in our study with a correlation coefficient (r) = 0.66 in the entire population and r = 0.69 in the preterm infants ⩽ 32 weeks of gestation. This value is within the wide range of that described in the pediatric and neonatal population (r = 0.53 to 0.99),7,8,18 however it does fall somewhat below the values described in studies specific to the neonatal population.7,8 One study used the Radical-7 pulse co-oximeter to measure SpHb as was done in this study, and compared those measurements with venous Hb values (r = 0.76) in 56 neonates (gestational age (GA) 31 weeks, BW 1440 g, postnatal age 20 days, median values).7 Another neonatal study compared Hb levels obtained using a transcutaneous spectroscopic device (Mediscan 2000, MBR Optical Systems, Wuppertal, Germany) with venous or capillary blood samples (r = 0.99) in 80 infants (GA 29.5 weeks, BW1350 g, corrected GA at study 33.1 weeks, mean values).8 However, our patients had a lower mean BW and GA (total population: 1177 g, 28.7 weeks; infants ⩽ 32 weeks’ gestation: 1020 g, 27.6 weeks), and paired samples in our study were obtained earlier in life than previously described with a median age of 4 days. Previous investigators reported significant differences in fetal Hb with BW with smaller – Figure 2. Bland Altman analysis of the differences between total infants having higher fetal Hb concentrations.21 Thus, the hemoglobin (tHb) and noninvasive hemoglobin (SpHb) in 52 ⩽ predominant presence of fetal Hb may have played a part in patients 32 weeks of gestation (144 paired samples). There was fi a good agreement between paired hemoglobin values with a mean the slightly lower correlation coef cient in our study; however, difference (bias) and s.d. of the differences (precision) of despite this, the results do show a moderately positive correlation. − 0.23 ± 1.60 g dl − 1. Use of SpHb in the clinical setting, possibly as an adjunct to direct laboratory Hb monitoring, would be of considerable value as it is continuous and noninvasive. the number of patients receiving blood transfusions from 4.5 to Limitations of our study include the investigation of a rather 0.6%.11 A prospective cohort study in 106 adult patients at risk for homogeneous patient population including only those infants on high blood loss during neurosurgery showed that the addition of respiratory support requiring blood gases. This tended to select noninvasive SpHb monitoring to standard care resulted in a 47% for the smaller preterm population ⩽ 32 weeks of gestation. reduction in the average number of red blood cell transfusions per We studied a convenient sample size of 61 patients with 167 paired patient. In this study, clinicians using SpHb monitoring were able Hb measurements (144 paired measurements in 52 patients to initiate transfusions faster compared with physicians not using ⩽ 32 weeks of gestation) to include a representative group of the technology, because they did not have to wait for a laboratory patients o3000 g at birth admitted to our NICU. However, Hb value.12 Monitoring SpHb in the NICU would make immediate additional studies with a larger sample of late preterm and term and clinically relevant information available if found to be accurate infants are warranted. A very small proportion of infants in this in the neonatal population. This continuous stream of data would study required blood pressure support, and healthy infants on also call attention to trending values and prompt more rapid room air were not enrolled. We also looked only at point medical intervention as has been described for pregnant patients measurements, whereas continuous SpHb monitoring may allow at high risk of obstetric hemorrhage.13 for decreasing variability and more accurate trending of Hb values Some investigators have reported a hesitance to make over time. Further studies will need to be done to determine the clinical decisions regarding blood transfusions based on the clinical application in a diverse patient population over a number measurements obtained with noninvasive Hb monitoring.14–16 of different clinical scenarios including patients with anemia and

© 2015 Nature America, Inc. Journal of Perinatology (2015), 617 – 620 Noninvasive hemoglobin measurements in newborns C Nicholas et al 620 polycythemia, as it has been suggested that peripheral perfusion 7 Jung YH, Lee J, Kim HS, Shin SH, Sohn JA, Kim EK et al. The efficacy of noninvasive may have a role in the accuracy of the SpHb monitor.22 hemoglobin measurement by pulse co-oximetry in neonates. Pediatr Crit Care Med 2013; 14:70–73. 8 Rabe H, Alvarez RF, Whitfield T, Lawson F, Jungmann H. Spectroscopic non- CONCLUSION invasive measurement of hemoglobin compared with capillary and venous values in neonates. Neonatology 2010; 98:1–5. Our results show that noninvasive SpHb may be considered as o 9 Shapiro BA, Peruzzi WT, Templin R. Clinical Application of Blood Gases. 5th edn, an adjunct to invasive tHb in newborn infants 3000 g, and that CV Mosby: Chicago, IL, USA, 1993; 334–336. these results hold true for the preterm population ⩽ 32 weeks of 10 Bland JM, Altman DG. Statistical methods for assessing agreement between two gestation where this noninvasive form of monitoring may provide methods of clinical measurement. Lancet 1986; 1(8476): 307–310. the largest benefit in prevention of iatrogenic anemia. However, 11 Ehrenfeld JM, Henneman JP, Sandberg WS. Impact of continuous and noninvasive additional studies with a larger sample size are needed to determine hemoglobin monitoring on intraoperative blood transfusions. Proceedings of the whether the correlation and agreement persists between these 2010 American Society of Anesthesiologists Annual Meeting. San Diego, CA, 2010. Abstract LB05. two measurements. 12 Aweda WNMF, Maher F. Reduction in red blood cell transfusions during neurosurgery with noninvasive and continuous hemoglobin monitoring. Anesth Analg 2013; 117(4); 50–51 Abstract. CONFLICT OF INTEREST 13 Butwick AJ, Hilton G, Riley ET, Carvalho B. Non-invasive measurement of hemo- The authors declare no conflict of interest. globin during cesarean hysterectomy: a case series. Int J Obstet Anesth 2011; 20: 240–245. 14 Dewhirst E, Naguib A, Winch P, Rice J, Galantowicz M, McConnell P et al. Accuracy ACKNOWLEDGEMENTS of noninvasive and continuous hemoglobin measurement by pulse co-oximetry 29 – We thank Masimo Corporation for providing the probes used in this study; however, during preoperative phlebotomy. J Intensive Care Med 2014; : 238 242. 15 Rice MJ, Gravenstein N, Morey TE. Noninvasive hemoglobin monitoring: how they did not participate in the study design, analysis and interpretation of the data, accurate is enough? Anesth Analg 2013; 117:902–907. or the preparation or approval of the manuscript. We also thank the respiratory 16 Gayat E, Bodin A, Sportiello C, Boisson M, Dreyfus JF, Mathieu E et al. Performance therapists, neonatologists, neonatal fellows and the staff of LAC+USC Neonatal evaluation of a noninvasive hemoglobin monitoring device. Ann Emerg Med 2011; Intensive Care Unit for their cooperation with the study. 57:330–333. 17 Lamhaut L, Apriotesei R, Combes X, Lejay M, Carli P, Vivien B. Comparison of the REFERENCES accuracy of noninvasive hemoglobin monitoring by spectrophotometry (SpHb) and HemoCue with automated laboratory hemoglobin measurement. Anesthe- 1 DeFrances CJ, Lucas CA, Buie VC, Golosinskiy A. 2006 National Hospital siology 2011; 115:548–554. Discharge Survey. Natl Health Stat Report 2008; 30(5): 1–20. 18 Park YH, Lee JH, Song HG, Byon HJ, Kim HS, Kim JT. The accuracy of noninvasive 2 Radical-7 Signal Extraction Pulse CO-Oximeter Operator’s Manual 2010; 1–6. hemoglobin monitoring using the Radical-7 pulse co-oximeter in children 3 Noiri E, Kobayashi N, Takamura Y, Lijima T, Takagi T, Doi K et al. Pulse total- undergoing neurosurgery. Anesth Analg 2012; 115: 1302–1307. hemoglobinometer provides accurate noninvasive monitoring. Crit Care Med 19 Butwick A, Hilton G, Carvalho B. Non-invasive haemoglobin measurement in 2005; 33: 2831–2835. patients undergoing elective caesarean section. Br J Anaesth 2012; 108: 271–277. 4 Raikhel M. Accuracy of noninvasive and invasive point-of-care total blood 20 Macknet MR, Allard M, Applegate RL, Rook J. The accuracy of noninvasive and hemoglobin measurement in an outpatient setting. Postgrad Med 2012; 124: continuous total hemoglobin measurement by pulse co-oximetry in human 250–255. subjects undergoing hemodilution. Anesth Analg 2010; 111: 1424–1426. 5 Causey MW, Miller S, Foster A, Beekley A, Zenger D, Martin M. Validation of 21 Delivoria-Papadopoulos M, Roncevic NP, Oski FA. Postnatal changes in oxygen noninvasive hemoglobin measurements using the Masimo Radical-7 SpHb Sta- transport of term, premature, and sick infants: the role of red cell 2, 3-diphospho- tion. Am J Surg 2011; 201:592–598. glycerate and adult hemoglobin. Pediatr Res 1971; 5:235–245. 6 Frasca D, Dahyot-Fizelier C, Catherine K, Levrat Q, Debaene B, Mimoz O. Accuracy 22 Isosu T, Obara S, Hosono A, Ohashi S, Nakano Y, Imaizumi T et al. Validation of of a continuous noninvasive hemoglobin monitor in intensive care unit patients. continuous and noninvasive hemoglobin monitoring by pulse CO-oximetry in Crit Care Med 2011; 39: 2277–2282. Japanese surgical patients. J Clin Monit Comput 2013; 27:55–60.