No Pulse, No Problem! Measuring Blood Pressure and Estimating Pulse Pressure in Centrifugal Flow Left Ventricular Assist Devices

Blood pressure control is critical to mitigate the risk of adverse events. However, in patients with an HVAD, the relationship between the Doppler opening pressure with mean arterial pressure (MAP) vs. systolic blood pressure (SBP) remains uncertain. Also, the association between waveform pulsatility and actual pulse pressure (PP) remains unknown. The goals of this study were to: 1) assess if Doppler measures SBP or MAP, 2) determine if HVAD pulsatility and PP correlate, and 3) develop a prediction rule to estimate PP based on HVAD pulsatility. Doppler measurements were made in triplicate and arterial catheter measurements were made in duplicate among 14 consecutive HVAD patients. Scatter plots and Bland-Altman plots assessed BP agreement between Doppler and arterial measurements. The relationship between PP and pulsatility were analyzed with scatter plots and a linear mixed model in order to estimate PP. There were 14 patients with a total of 25 longitudinal observations. Scatterplots demonstrated a perfectly linear relationship between Doppler and SBP along the line of equality (slope = 1). Bland-Altman analysis revealed good agreement between Doppler and SBP (mean difference: 0.4 mmHg; 95% limits of agreement (LoA): -12.6, 13.4 mmHg) but poor agreement between Doppler and MAP (mean difference: 5.4 mmHg; 95% LoA: -7.6 to 18.4 mmHg). There was a positive linear relationship between PP and pulsatility (Fig). Using pulsatility as the predictor and PP as the outcome, a linear mixed model was fit to give: Pulse pressure (mmHg) = (Pulsatility * 5) - 5.4. Measurement error was 3.1 mmHg and between patient variability was 1.9 mmHg. (Example) Given a pulsatility of 2 L/min and 4 L/min, the prediction equation would estimate a PP of 5 mmHg and 15 mmHg respectively. Doppler BP more closely reflects SBP than MAP. Device pulsatility correlates closely with PP in an HVAD population. A simple prediction model may be used to estimate PP based on pulsatility.