Can perioperative hemodilution be monitored with non-invasive measurement of blood hemoglobin?

Abstract

Background: Trends in non-invasive measurements of blood hemoglobin (Hb) may be useful for identifying the need for transfusion in the perioperative period.

Methods: Crystalloid fluid (5-20 mL/kg) was administered intravenously or by mouth to 30 volunteers and 33 surgical patients in five non-randomized clinical studies where Hb was measured on 915 occasions by non-invasive (Radical-7™) and invasive methodology. The hemodilution curves were compared by volume kinetic analysis and linear regression, with the slope and scattering of the data as key outcome measures.

Results: The slope was 1.0, indicating unity between the two modes of measuring Hb when crystalloid fluid was infused in volunteers; however, only 40-45% of the variability in the non-invasive Hb could be explained by the invasive Hb. Patients undergoing major surgery, who showed the most pronounced hemodilution (median 24 g/L); non-invasive Hb explained 72% of the variability but indicated only half the magnitude of the invasive Hb changes (slope 0.48, P < 0.001 versus the volunteers). Simulations based on volume kinetic parameters from the volunteers showed 25% less plasma volume expansion after infusion when based on non-invasive as compared to invasive Hb, while no difference was found during infusion.

Conclusions: In volunteers the non-invasive Hb had good accuracy (low bias) but poor precision. In surgical patients the non-invasive Hb had good precision but systematically underestimated the hemodilution. Despite severe limitations, the non-invasive technology can be used to follow Hb trends during surgery if supported by occasional invasive measurements to assure acceptable quality of the hemodilution curve.

Trial registrations: ControlledTrials.gov NCT01195025, NCT01062776, NCT01458678, NCT03848507, and NCT01360333 on September 3, 2010, February 4, 2010, October 25, 2011, February 20, 2019, and May 25, 2011, respectively.

Keywords: Hemoglobin; Non-invasive measurement; Point of care tests; Pulse oximetry.

Conflict of interest statement

RGH holds a grant from Grifols for the study of 20% albumin as infusion fluid.

The other authors declare that they have no conflict of interest.

Figures

Fig. 1

a The invasive versus non-invasive Hb concentration at the onset of the infusions. b Bland-Altman plot of the data shown in the previous subplot. c Relationship between the median “scatter” around the regression lines for each of the five groups, i.e. the coefficient of determination

Fig. 2

Three experiments with different slopes. Distances between the marks are set equal in each subplot to facilitate comparison. a Flat slope. The non-invasive Hb changes much less than the invasive Hb. b Normal slope. The rise in non-invasive Hb is similar to the rise of the invasive Hb. c Steep slope. Greater increase in non-invasive Hb than for the invasive technology

Fig. 3

Relationship between the hemodilution as calculated from the invasive versus the non-invasive Hb measurements in the five studies (a to e) and the correlation between perfusion index and the deviation between the two modes of measuring Hb in a single patient from Study 4

Fig. 4

a Hemodilution as obtained by the invasive Hb versus the non-invasive Hb for each patient. b Distribution of slope factors in the five studies

Fig. 5

The perfusion index versus the difference in hemodilution as obtained by the two modes of measuring Hb in (a) Study 1 and (b) Study 4. These were the only cohorts in which the perfusion index occasionally dropped below 1.0

Fig. 6

Volume kinetics. Curve-fitting of the Hb data from Studies 1–3 as obtained by the (a) invasive and (b) non-invasive measurements. c Schematic drawing of the kinetic model. d The model-predicted versus the measured plasma dilution as obtained with the (d) invasive and (e) non-invasive Hb. f Simulation of the volume expansion of the central body fluid space (the plasma) when 1 L of Ringer’s acetate is infused over 30 min