ABSTRACT Volume status of a patient is difficult to assess clinically. Methods to measure plasma volume as well as changes in plasma volume in connection with fluid therapy are therefore valuable for research purposes and could potentially be used bedside to guide fluid therapy in individual patients. The objective of the present study was to evaluate hematocrit as a marker of changes in plasma volume and an anthropometric formula to estimate absolute plasma volume against the reference method, plasma volume measurements using radiolabeled albumin. A total of 64 postoperative patients received 10 mL/kg of albumin. The plasma volumes were measured with the reference method (radiolabeled albumin) at baseline, after 30 and 180 min. Plasma volumes were compared with plasma volumes derived from either the baseline measurements and subsequent changes in hematocrit (calibrated method) or from an anthropometric formula and subsequent hematocrit changes (anthropometric method). Bland–Altman plots were used to test agreements between methods. The mean difference in plasma volume between the anthropometric and the reference method at baseline was −0.1 mL/kg (95% CI –2.1 to 1.8) with lower and upper LOA of −18.0 and 17.0 mL/kg. The agreement remained essentially unchanged after fluid therapy (mean difference −0.1 mL/kg (95% CI –1.8 to 1.5) with LOAs of −18.0 and 17.0 mL/kg). The mean difference between the calibrated and reference methods was −0.1 mL/kg (95% CI –0.9 to 0.7) with LOAs of −9.4 and 9.2 mL/kg. The anthropometric formula yields low precision in predicting plasma volumes in postsurgical patients. Changes in hematocrit during fluid therapy cannot replace the reference method to assess changes in plasma volume in the individual patient because of the imprecision. Editorial Comment Clinical appreciation of plasma volume is relevant for assessing treatment where intravenous fluid resuscitation is involved. This analysis, using a reference method for plasma volume assessment, and comparing to simpler methods to estimate the same, demonstrates that there are important limitations with some simpler and readily acceptable methods to perform this estimation.
Grubb et al. (Thu,) studied this question.