Non-Invasive Venous waveform Analysis (NIVA) for monitoring blood loss in human blood donors and validation in a porcine hemorrhage model

Abstract

Study objective: There is an unmet need for a non-invasive approach to diagnose hemorrhage early, before changes in vital signs occur. Non-Invasive Venous waveform Analysis (NIVA) uses a unique physiological signal (the peripheral venous waveform) to assess intravascular volume. We hypothesized changes in the venous waveform would be observed with blood loss in healthy adult blood donors and characterized hemorrhage using invasive monitoring in a porcine model.

Design: Prospective observational study.

Setting: American Red Cross donation center.

Patients: 50 human blood donors and 12 non-donating controls; 7 Yorkshire pigs.

Interventions: A venous waveform capturing prototype (NIVA device) was secured to the volar aspect of the wrist in human subjects. A central venous catheter was used to obtain hemodynamic indices and venous waveforms were obtained using the prototype NIVA device over the saphenous vein during 400 mL of graded hemorrhage in a porcine model.

Measurements: Venous waveforms were transformed from the time to the frequency domain. The ratiometric power contributions of the cardiac frequencies were used to calculate a NIVA value representative of volume status.

Main results: A significant decrease in NIVA value was observed after 500 mL of whole blood donation (p < .05). A ROC curve for the ability of the NIVA to detect 500 mL of blood loss demonstrated an area under the curve (AUC) of 0.94. In the porcine model, change in NIVA value correlated linearly with blood loss and with changes in hemodynamic indices.

Conclusions: This study provides proof-of-concept for a potential application of NIVA in detection of blood loss. NIVA represents a novel physiologic signal for detection of early blood loss that may be useful in early triage and perioperative management.

Keywords: Hemorrhage; Monitoring; Venous; Venous waveform analysis.

Figure 1:. Representation of the NIVA prototype and venous waveform example with corresponding fast Fourier transform signal.

(A) The NIVA prototype is composed of a control box containing an amplifier, analogue to digital converter, microcontroller, and SD card. Data from the control box can be transferred to a computer via USB port on the control box. The sensor is comprised of a piezo-electric crystal encased in silicone (AD Instruments) connected to the control box by a wire. (B) Venous waveform recording from human subject (top) and the corresponding fast Fourier transform of that signal (bottom). (C) Description of how the fast Fourier transform identifies cardiac frequencies that are incorporated into an algorithm to calculate a “NIVA value”.

Figure 2:. NIVA value before and after 500 mL blood loss in healthy human subjects.

NIVA values (A) before (black) and after (grey) 500 mL blood donation in 50 healthy human volunteers, (B) before (black) and after (grey) 500 mL blood donation in males (n=23) and females (n=27). *signifies p-value<0.05 (C) The area under the curve (AUC) for detecting 500 mL blood loss in donors compared to controls was 0.94 with a sensitivity of 92% and specificity of 84% for prediction of blood loss.

Figure 3:. Porcine hemorrhage model to determine correlation of NIVA with volume loss.

NIVA values were obtained before controlled hemorrhage of 0 (baseline), and at 100 mL, 200 mL, 300 mL, and 400 mL of blood loss in a porcine model. NIVA values correlated with increments of 100 mL of blood loss (r = 0.39, n = 7 pigs, A). The change in NIVA value (Δ NIVA value) correlated with each 100 mL change in blood volume (Δ blood volume) (r = 0.68, n = 7 pigs, B).

Figure 4:. NIVA values and hemodynamics obtained during controlled hemorrhage.

NIVA values were obtained concurrently with heart rate (HR, A), central venous pressure (CVP, B), cardiac output (CO, C) pulmonary artery diastolic pressure (PAD, D), pulmonary capillary wedge pressure (PCWP, E) and NIVA (F) during controlled hemorrhage. CO, PCWP, and NIVA value decreased after 200 mL hemorrhage (C, E, F). PAD, CVP, CO, PCWP and NIVA value decreased after 400 mL hemorrhage (B, C, D, E, F, n = 7 pigs); * p < 0.05. Pearson correlation coefficients for heart rate, hemodynamic indices, and NIVA compared to blood volume removed (G) and the change (Δ) in heart rate, hemodynamic indices and NIVA after incremental blood volume removed (H). Significantly non-zero slopes (p<0.05) are denoted by asterisks (*).

Figure 5:. Correlation of NIVA value and hemodynamic indices during hemorrhage in a porcine model.

NIVA value correlated with CO (A, r = 0.82, p<0.05), PAD (B, r = 0.75, p < 0.05,), and PCWP (C, r = 0.67, p < 0.05) during hemorrhage (n = 7).

Figure 6:. NIVA derived pulse rate correlates with pulse oximetry.

Pulse rate (PR) derived using from the venous waveform signal obtained with NIVA device correlated strongly with PR obtained with Nellcor™ pulse oximetry device during porcine hemorrhage (r = 0.96; n = 7).