Point of Care Measurement of Hemoglobin and Hematocrit in Newborn Infants Using a Blood Gas Analyzer
Celine Chedid1, Aiman Rahmani MD2, Naveed Durrani MD2, Moghis Ur Rehman MD2, Iram Musharraf MD2, Ahmad Imran MD2
1University of Balamand, Koura, Lebanon
2Tawam Hospital, Al Ain, United Arab Emirates
Ahmad Ali Imran, MD
PO Box 15258
Abu Dhabi, United Arab Emirates
Author Disclosure: The authors have nothing to disclose.
Objective: To assess the reliability of measuring hemoglobin (Hb) and hematocrit (Hct) in neonates using a point of care blood gas analyzer.
Methods: Hb and Hct results were compared between the point of care blood gas analyzer (OMNI-S, Roche) and laboratory values on 131 paired blood samples, collected from 67 newborns. The bias and limits of agreement were assessed with the Bland Altman method. Paired results were classified in four therapeutic groups according to the Canadian Pediatric Society (CPS) guidelines for blood transfusion; the percent of agreement and weighed kappa were calculated.
Results: The mean difference between paired samples was – 0.14 g/dl (± 1.4) and -0.11% (± 8.8) for Hb and Hct respectively; 97% of Hb and 77.4% of Hct values fell within the acceptable limits of the Clinical Laboratories Improvement Amendment. The correlation coefficient between the paired results was 0.95 (P 0.0005) and 0.84 (P 0.0005) respectively. The data distribution in the four CPS groups showed a high level of agreement for Hb (86.3%) and less for Hct (65%) corresponding to a weighted kappa of 0.75 and 0.45 respectively. The disagreement between the Hb and Hct pairs modified the clinical indication of blood transfusion in 2.3% and 3.2% respectively.
Conclusion: Our data showed a high level of agreement between paired Hb values measured by the blood gas analyzer and the laboratory permitting its routine use in the point of care setting. The wide limits of agreement between Hct values hinder its clinical use.
Key Words: point of care, hemoglobin, hematocrit, newborn
The availability of a quick and reliable bedside measurement of neonatal hemoglobin (Hb) and hematocrit (Hct) using the blood gas analyzer is important in a modern neonatal intensive care unit (NICU). The tests can be done concomitantly with blood gases, alleviating the need for extra blood drawing and therefore decreasing the risk of iatrogenic anemia.1-3 In a study by Madan et al in 2005, point of care testing (POCT) was associated with a 46% reduction of blood transfusions in extreme low birth weight infants in the first two weeks of life.4 In addition the POCT provides immediate results allowing timely intervention and require a much smaller blood sample (50 µL) than the central laboratory testing (500 µL).
Previous studies have validated the use of near-patient testing of Hb concentration in neonates using the ABL725 Radiometer blood gas analyzer5 and an in line monitor, directly connected to the patient’s intravascular catheter (VIA LVM).6 Both studies measured the bias of the POCT and its limits of agreement with laboratory results but they did not address clinical agreement. We aimed to compare hemoglobin and Hct results obtained on OMNI-S series blood gas analyzers with those from the laboratory and assess the level of clinical agreement with regard to blood transfusion.
This study was conducted retrospectively in a tertiary neonatal intensive care unit (NICU) at Al Ain, United Arab Emirates. Our practice is to measure Hb and Hct in the hospital laboratory using a Cell-Dyn Sapphire analyzer (Cell-Dyn Sapphire, Abbott, IL, USA) which measures Hb by spectrophotometry and calculates Hct using the equation Hct = (red blood cell count x mean corpuscular volume)/10. Blood gas specimens were processed in the neonatal unit using the OMNI-S blood gas analyzer (Roche Diagnostics, Indianapolis, IN, USA) which measures Hb and Hct by spectrophotometry and conductivity respectively. Both analyzers are calibrated daily and are subject to rigorous quality control. The laboratory analyzer needs a minimum of 500 µl of heparinized blood in an EDTA bottle while the blood gas analyzer needs only about 50 µl blood per sample usually collected in a pre-heparinized capillary tube.
The data were retrospectively collected from the computerized hospital database; 131 paired results of Hb and Hct were gathered from 67 newborn infants. The same blood sample was simultaneously tested for Hb and Hct by the Roche OMNI-S blood gas analyzer (POCT) and the hospital laboratory (reference method). All the blood tests were requested by the attending physician as clinically indicated. All patients were included without applying any exclusion criteria.
The paired Hb and Hct values were classified in a four by four table according to the Canadian Pediatric Society recommendations for blood transfusion in the newborn.7 Initially we measured the agreement between the paired values of Hb and Hct without considering the patients’ clinical condition then we reviewed the medical files and assessed the severity of the clinical condition in all patients who were classified differently by the POCT and reference method. Our aim was to find if the laboratory disagreement had an impact on the decision to administer a blood transfusion (clinical disagreement).
We analyzed the data using the SPSS (version 18.0). The correlation between the POCT and the laboratory results was measured with the Pearson correlation coefficient. A scatter plot with line of best fit was generated and simple linear regression was used to measure the slope and intercept of the Hb and Hct lines. The percentage of agreement and the Cohen’s Kappa were used to assess if the POCT concurred with the reference laboratory in assigning the patients to the different therapeutic groups dictated by the standard transfusion guidelines. The difference between the POCT and reference laboratory results for the Hb and Hct were plotted against their respective means following the Bland and Altman method.8 The bias (an indicator of accuracy) was calculated as the mean difference between paired Hb and Hct values, the limits of agreement (an indicator of inter-sample reproducibility) were defined as two standard deviation of that difference. The precision was assessed using the guidelines of the Clinical Laboratories Improvement Act (CLIA), which requires at least 80% of the POCT results to fall within acceptable tolerance limits of 7% for Hb and 6% for Hct.9 The precision was also assessed against the European criteria which allow 4.1% error for Hb and Hct.10
The Hospital Ethics Committee granted permission to conduct this retrospective study.
One hundred and thirty one blood samples were collected from 67 study participants. Seventy eight percent of patients were preterm; their mean birth weight was 1390 g (SD 845) and mean gestational age 30.6 weeks (SD 5.6, range 22 to 40). The age at the time of blood extraction was 12 days (SD 10).
Scatter plot diagrams (Figure 1 and Figure 2) and the Pearson correlation coefficient showed a significant correlation between the blood gas analyzer and the laboratory results for Hb (r = 0.96, P 0.0005) and Hct (r = 0.84, P 0.0005). The regression line for Hb had a slope of 1.02 and a constant of 0.39, that of Hct had a slope on 0.79 and a constant of 7.66.
Bland-Altman plots revealed clinically non-significant biases of -0.14 (SE 0.08) and 0.11 (SE 0.4) for Hb and Hct respectively demonstrating a high accuracy of the POCT (Figure 3 and Figure 4). The limits of agreement were narrow for the Hb (-1.54 to +1.26) and relatively wide for Hct (-8.71 to +8.93). The vertical scatter of the paired data points around the bias showed that 97% of Hb values and 77.4% of Hct values fell within the acceptable limits of the CLIA. The more stringent European requirements were fulfilled by 89.2% and 58.1% of Hb and Hct values respectively.
A cross tabulation of Hb and Hct pairs, according to the classification of the Canadian Pediatric Society for blood transfusion is shown in Table 1. Eighty-six percent of the 131 Hb and 65% of the 124 Hct pairs were classified in the same group corresponding to a Cohen’s Kappa of 0.79 (P < 0.001) and 0.45 (P < 0.001) respectively. The analysis of the patient’s clinical condition revealed that 15 out of the 18 discordant Hb pairs and 39 out of the 43 discordant Hct pairs did not have any impact on the indication of blood transfusion. The percentage of agreement between the POCT and reference method in selecting the patients who require blood transfusion rose to 97.7% and 96.8% for Hb and Hct respectively after combining the laboratory values and clinical criteria.
The measurement of Hb and Hct using the blood gas analyzer gives immediate results at the point of care allowing early intervention at less cost. The blood gas analyzer measures Hb, Hct, electrolytes, ionized calcium, glucose, bilirubin and lactate concomitantly with blood gas reducing the need for extra blood sampling and allowing frequent measurements at the bedside. The POCT must be submitted to stringent quality control measures to ensure the accuracy and reproducibility of the results. Previous studies have validated the use of near-patient testing of hemoglobin concentration in neonates using a haemoglobino-meter (HemoCue, Sheffield, UK).11 Similar studies in adults validated the POCT with the ABL725 Radiometer blood gas analyzer (Radiometer, Copenhagen, Denmark) and the Chiron 855 ABG analyzer (Chiron Diagnostics, Medfield, MA, USA) but none of these studies addressed the use of the OMNI-S blood gas analyzer to measure Hb in newborns.
The high product-moment correlation coefficient (r 0.95, P 0.0005 for Hb and 0.84, P 0.0005 for Hct) between the POCT and the laboratory values denotes that the results are spread close to a straight line, but should not be considered as a sign of equality between the paired values unless the slope of the best fit line is close to one and the intercept is close to zero. The slope of the regression equation for Hb (1.02) is close to 1 and the intercept (- 0.39) is close to zero suggesting equality and minimal bias. The slope for the Hct (0.79) was significantly different from one and the intercept (7.66) is far from zero indicating an important difference between the instruments.
The mean differences between the paired values of Hb (0.14, SE 0.08) and Hct (0.11 SE 0.4) were close to zero suggesting accurate results and minimal bias. The narrow limits of agreement for Hb (-1.54 to +1.26), assessed with the bland Altman method suggest highly reproducible results. The same was not demonstrated for Hct, in which the limits of agreement (-8.71 to +8.93) were too wide for it to be an acceptable transfusion trigger in clinical practice. Ninety-seven percent of Hb values were < 7% different from of their laboratory counterparts, thus satisfying the CLIA which requires at least 80% of values to fall within the 7% acceptable limits. The more stringent European requirements, which allows only 4% difference limit were also fulfilled by 89% of Hb values. On the other hand, Hct measurement with the POCT was not satisfactory as 23% and 45% of values exceeded the 6% and 4% permitted by the CLIA and European guidelines respectively. Our laboratory testing calculates the Hct using the equation Hct = (RBC count x MCV)/10 while the OMNI-S blood gas analyzer measures it directly by radiometry; this explains the lack of agreement between the two instruments. Hb is measured by the same method (Co-oximetry) in the laboratory and by the POCT; this explains the high precision and narrow limits of agreement for Hb.
Misclassified Hb (13.8%) and Hct (34.6%) values affected the indication for blood transfusion in 2.3% and 3.2% respectively. The use of a restrictive blood transfusion policy based on the Canadian Pediatric Society Guidelines and the inclusion of clinical criteria limited the need for blood transfusion and increased the rate of clinical agreement regarding blood transfusion.
The number of patients with Hb values below 10 g/dl (N = 30) or Hct less than 35% (N = 64) is relatively small, which may be considered as a limitation of the study since these values are clinically significant. All our patients had Hct values > 30% and only one patient had a Hb less than 6 g/dl limiting the ability of our study to assess the reliability of the OMNI-S blood gas analyzer at very low Hb and Hct values.
Point of care testing should meet the same accreditation and regulatory requirements as central laboratory testing. The measurement of Hb by the OMNI-S blood gas analyzer in our study was highly satisfactory over a broad array of values, allowing its clinical use at the bed side. In contrast, the variations of Hct values exceeded the limits allowed by the CLIA limiting its discriminative ability as POCT. Further studies recruiting a larger number of patients with Hb < 6 g/dl and/or Hct <30% are required to assess the reliability of the OMNI-S blood gas analyzer in case of severe anemia.
- Ray J, Post J, Hamielec C. Use of a rapid arterial blood gas analyzer to estimate blood haemoglobin concentration among critically ill adults. Crit Care 2002; 6:72-5.
- Lin J, Strauss R, Kulhavy J, Johnson KJ, Zimmerman MB, Cress GA, et al. Phlebotomy overdraw in the neonatal intensive care nursery. Pediatrics 2000; 106:e19-24.
- Widness J, Madan A, Grindeanu L, Zimmerman MB, Wong DK, Stevenson DK. Reduction in red blood cell transfusions among preterm infants: results of a randomized trial with an in-line blood gas and chemistry monitor. Pediatrics 2005; 115:1299-306.
- Madan A, Kumar R, Adams MM, Benitz WE, Geaghan SM, Widness JA. Reduction in red blood cell transfusions using a bedside analyzer in extremely low birth weight infants. J Perinatol 2005; 25:21-5.
- Hinds LE, Brown CL, Clark SJ. Point of care estimation of hemoglobin in neonates. Arch Dis Child Fetal Neonatal Ed 2007; 92:F378-80.
- Widness JA, Kulhavy JC, Johnson KJ, Cress GA, Kromer IJ, Acarregui MJ, et al. Clinical performance of an in-line point-of-care monitor in neonates. Pediatrics 2000; 106:497-504.
- Fetus and Newborn Committee, Canadian Paediatric Society. Red blood cell transfusion in newborn infants: revised guidelines. Paediatr Child Health 2002; 7:553-8.
- Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 8476:307-10.
- Clinical laboratories improvement amendment (CLIA). Federal Register February 28, 1992; 57:70002-186.
- Ricos C, Alvarez V, Cava F, Garcia-Lario JV, Hernandez A, Jimenez CV, et al. Current databases on biological variation: pros, cons and progress. J Clin Lab Invest 2000; 60:733-5.
- Rechner I, Twigg A, Davies A, Imong S. Evaluation of the HemoCue compared with the Coulter STKS for measurement of neonatal haemoglobin. Arch Dis Child Fetal Neonatal Ed 2002; 86:F188-9.