Validity of non-invasive hemoglobin measured by pulse co-oximeter in neonates - An observational study

Background: Hemoglobin is a frequently ordered investigation in neonatal intensive care units. There is a need for hemoglobin estimation by point of care methods to reduce iatrogenic anemia and to alleviate pain associated with repeated venous sampling. Pulse co-oximeters have been developed to measure hemoglobin non-invasively based on spectrophotometric method. We compared hemoglobin measured by pulse co-oximeter with reference venous hemoglobin in neonates admitted to a tertiary care newborn unit. Design: This was an observational study. Duration: The study duration was from November 2016 to December 2016. Setting: Department of Neonatology, Institute of Child Health and Hospital for Children, Egmore. Methods: Neonates admitted in nursery who warranted hemoglobin estimation underwent both invasive venous hemoglobin estimation by automated hematology analyzer and non-invasive hemoglobin estimation by pulse co-oximeter (Masimo SET radical 7). Results: Of 158 newborns enrolled, the device failure rate was 12.5%. The bias between transcutaneous and venous hemoglobin was 1.66±2.26 g/dl (mean ± standard deviation). Transcutaneous and venous hemoglobin showed moderate agreement on Bland Altman plot with intraclass correlation coefficient of 0.56. At lower levels of hemoglobin, we noted higher bias. It was 2.69±1.87 g/dl at hemoglobin <13 g/dl and 3.29±1.86 at hemoglobin ?10 g/dl. On regression analysis, only the level of hemoglobin influenced bias and device failure rate. Conclusion: Non-invasive hemoglobin measured by pulse co-oximeter shows only a moderate agreement with reference venous hemoglobin in neonates admitted to nursery. We report a high device failure rate of 12.6%. Level of hemoglobin is the single most determinant of device failure and degree of agreement. With high device failure rates and poor agreement at low hemoglobin levels, the clinical utility appears negligible.