Continuous and Noninvasive Total Blood Hemoglobin Measurement Using Near-Infrared Reflectance Spectrometry.

Abstract

Continuous measurement of total hemoglobin (tHb) can be of great clinical value during surgery, visualizing tHb trends and guiding blood transfusion practices. Devices such as those pulse oximetry-based have been available and studied over the past decade with varying results on their accuracy. We recently developed a first-of-its-kind near-infrared reflectance spectroscopy (NIRS) based algorithm to measure tHb continuously and noninvasively. This study was conducted to evaluate the performance of the NIRS based tHb algorithm using a cerebral oximetry sensor. We performed a post hoc retrospective analysis of data collected during cardiac surgery from 3 clinical sites. Both NIRS data and blood gas tHb data were collected. Then the NIRS data was postprocessed through the NIRS based tHb algorithm to generate continuous measurement of tHb. Comparison between the 2 tHb measurements was assessed using the Bland-Altman analysis, mean absolute error (MAE), root mean square error (RMSE), 4-quadrant concordance, and error-grid analysis. One hundred and eighty-nine (189) patients were included in the analysis. The bias (or mean difference) and precision (or 1 standard deviation of the difference) (95% confidence interval) are 0.08 (0.02-0.19) g/dL and 1.01 (0.93-1.09) g/dL, respectively. The limits of agreement were -1.90 and 2.06 g/dL. The MAE is 0.79 (0.69-0.91) g/dL, the RMSE is 1.12 (0.94-1.30 g/dL), and the concordance is 86.1 (79.8-92.1) %. The error-grid analysis demonstrated that the majority (84%) of the measurements are in the green zone and 0% in the red zone. The exploratory study shows that the NIRS based tHb provided an accurate measurement of tHb in cardiac surgery patients. Further research may be needed to evaluate its accuracy and implementation in other clinical settings.