The Impact of Skin Pigment on Measurement of Noninvasive Hemoglobin

Background: Anemia is common in critically ill patients, with 90% of patients becoming anemic by ICU Day three. Noninvasive hemoglobin (SpHb) monitoring provides continuous, hemoglobin (Hb) measurement. Currently available devices have limitations: low sensitivity for detecting serial changes in Hb, inaccuracy at the lower range (< 9 g/dL) and overestimation of Hb in the presence of poor peripheral perfusion. Recently, a number of studies have demonstrated greater inaccuracy of oxygen saturation (SpO2) in dark skinned patients. We hypothesized that skin pigment might also impact the accuracy of SpHb. Methods: We compared measured Hb from an arterial blood sample to SpHb measured by an oximeter (Masimo, Irving, CA). The study was approved by the IRB and informed consent was obtained from subjects or their LAR. Data collected included pulse rate, SpO2, SpHb, perfusion index and complete blood count (CBC). Demographics including age, sex, and primary diagnosis were recorded. Patients self-identified as Black or white, we used Bland-Altman analysis to assess the bias, precision, and limits of agreement (LOA) of SpHb monitoring in the group as a whole and in the two groups based on skin pigment. Results: We enrolled 123 subjects, 47 (38%) were Black and 76 (62%) were white. 83% of Black subjects and 68% of white subjects were male. The mean age for Black subjects was 59 ± 15 y and 60 ± 12 y for white subjects. Groups did not differ significantly by age, gender, or Hb measured by CBC 10 ± 2.0 g/dL for Black subjects, and 9.9 ± 2.7 g/dL for white subjects. The SpHb bias for Black subjects was 0.27 g/dL with precision 1.39 and LOA was 3.0 g/dL, – 2.5 g/dL. The SpHb bias for white subjects was 0.32 g/dL with a precision of 1.3 and LOA of 3.1 g/dL, – 2.0 g/dL. The SpHb bias for all subjects was 0.43, precision was 1.33 and LOA was 3.0, -2.17. See figure. Proportional bias was present for both racial groups, with the difference between SpHb and CBC tending to decrease as the Hb increased. Spearman’s rho was -0.30 (P = .04) for Black patients, and -0.44 (P < .001) for white patients. Conclusions: SpHb monitoring can display continuous Hb without repeated blood draws. However, SpHb monitoring has significant limitations. Our study demonstrates that SpHb tends to overestimate the true Hb in patients with low Hb (<9 g/dL), which could lead to patients not receiving necessary blood transfusions. Skin pigment did not impact SpHb accuracy.