Using noninvasive hemoglobin measurements to estimate measured hemoglobin in a pediatric hemodialysis unit

Maintaining a dialysis patient's hemoglobin (Hgb) within a very narrow range can be challenging. Relying on Hgb measurements only once or twice a month can cause large fluctuations in their measurements. Utilizing the Hgb measurement from noninvasive modalities has been studied in adult populations. Our study focused on a pediatric hemodialysis population where blood volumes are much smaller to see if these measurements would adequately work for adjusting erythropoietin dosages. We reviewed our patients' data over a 6-month time period and collected simultaneous measurements of Hgb performed in the laboratory, as well as the initial Crit-Line measurement. We then analyzed the correlation of the two estimates of the patient's Hgb using linear regression as well as Bland-Altman plot and ROCs. There were 407 simultaneous measurements of Hgb in our 32 pediatric hemodialysis patients during this time. Linear regression showed good correlation with an R value of 0.85 (P value < 0.0001). The Bland-Altman plot showed excellent agreement between the two methods. The ROC analysis showed that the performance of the noninvasive Hgb measurement was very good at predicting low measurements. Predicting Hgb less than 10 g/dL had an area under the curve of 0.94. Predicting Hgb greater than 12 g/dL had an area under the curve of 0.91. There were 100 simultaneous measurements of hematocrit. The analysis revealed similar results as the hemoglobin. Noninvasive in-line monitoring of Hgb can be a very useful way of assessing the patient's response to erythropoietin on a day-to-day time frame. Utilizing this methodology should help reduce the variability in the pediatric patients' Hgb measurements.