A calibrated pulse waveform analysis algorithm, which determines continuous cardiac output, compared with a noncalibrated algorithm

Introduction

We compared two pulse waveform analysis algorithms for measuring continuous cardiac output; PulseCO (LiDCO UK Ltd), which requires a transpulmonary indicator dilution (TPID) calibration; and the Pressure Recording Analytical Method (PRAM) (most∼ care, FIAB, Italy), which does not require calibration but requires the user to define whether the patient is septic or nonseptic. No study has compared these two methods simultaneously in adults.

Methods

Arterial pulse waveforms were collected prospectively from 11 patients (seven male, age 36–75 years) requiring haemo-dynamic monitoring, and were stored electronically and then replayed through each device. Nine patients were deemed 'septic' on the study day by standard clinical criteria. Each patient had a (TPID) cardiac output (CO) measurement prior to pulse wave collection (lithium dilution or thermodilution) to allow prospective calibration of the PulseCO software. The duration of continuous waveform collection was between 55 and 145 min, and these were replayed through the septic and nonseptic settings of the PRAM. Thirty second CO measurements were collected with the proprietary data acquisition software. Comparisons were made between: PulseCO(c) (TPID calibrated) to PRAM(cp) (set to clinical picture, i.e. septic or nonseptic); PulseCO(c) to PRAM(bf) (set to 'retrospective best fit'); and PulseCO(c) to PRAM(c) (in line with the clinical picture and mathematically calibrated to the TPID). It was only possible to include seven patients in the PRAM(bf) analysis.

Results

One patient was in extremis at the time of data collection, requiring epinephrine boluses, and was excluded, leaving 10 patients in the analysis group. The mean flow, bias and precision (± 2SD) of PulseCO(c) to PRAM(cp), of PulseCO(c) to PRAM (bf), and of PulseCO(c) to PRAM(c) were: 6.86, -0.58 and ± 6.03 l/min; 6.24, 0.88 and ± 1.55 l/min; and 6.67, -0.21 and ± 2.23 l/min, respectively. The 95% limits of agreement for PulseCO(c) to PRAM(cp), for PulseCO(c) to PRAM (bf), and for PulseCO(c) to PRAM(c) were 88%, 24.9% and 34%.

Conclusion

The CCO algorithm in the PRAM does not trend acceptably with PulseCO. Even though the precision of PRAM(bf) was acceptable (in seven patients) the settings were inconsistent with the clinical presentation. The noncalibrated PRAM is currently not suitable for continuous cardiac output monitoring in the ICU.