The cardiac output from blood pressure algorithms trial

Area of Science:

• Critical Care Medicine
• Biomedical Engineering
• Cardiovascular Physiology

Background:

• Estimating cardiac output (CO) from arterial blood pressure (ABP) waveforms is crucial in intensive care units (ICUs).
• The accuracy of various algorithms for CO estimation from peripheral ABP remains uncertain.
• A standardized dataset for comparing these algorithms is lacking.

Purpose of the Study:

• To evaluate the performance of investigational algorithms for estimating CO from radial ABP waveforms.
• To develop a public dataset for the comparative analysis of CO-from-ABP algorithms.
• To identify reliable methods for CO estimation in ICU patients.

Main Methods:

• Retrospective comparative analysis of eight CO-from-ABP algorithms.
• Utilized the Multiparameter Intelligent Monitoring in Intensive Care II (MIMIC-II) database.
• Data included radial ABP waveform segments and reference CO measurements by thermodilution (CO(TD)) from 120 ICU cases.

Main Results:

• All tested algorithms demonstrated superiority over mean arterial pressure (MAP) in estimating directional changes in CO(TD).
• The Liljestrand and Zander (1928) method showed significantly better quantitative CO(TD) estimation than MAP (p < 0.001).
• Accuracy of the Liljestrand method improved with cleaner ABP waveforms; other algorithms were not significantly better than MAP.

Conclusions:

• The Liljestrand and Zander method provides a more accurate estimation of CO(TD) than MAP alone, based on ICU ABP data.
• Commercial CO-from-ABP algorithms could not be evaluated due to replication issues.
• A publicly available dataset is offered for the development and validation of CO-from-ABP algorithms.