Enhancing central blood pressure accuracy through statistical modeling: A proof-of-concept study.

Background: Non-invasive estimation of central blood pressure (BP) may have better prognostic value than brachial BP. The accuracy of central BP is limited in certain populations, such as in females and the elderly. This study aims to examine whether statistical modeling of central BP for clinical and hemodynamic parameters results in enhanced accuracy.

Methods: This study is a cross-sectional analysis of 500 patients who underwent cardiac catheterization. Non-invasive brachial cuff and central BP were measured simultaneously to invasive aortic systolic BP (AoSBP). Central BP was calibrated for brachial systolic (SBP) and diastolic BP (Type I calibration; C1SBP) or brachial mean and diastolic BP (Type II calibration; C2SBP). Differences between central SBP and the corresponding AoSBP were assessed with linear regression models using clinical and hemodynamic parameters. These parameters were then added to C1SBP and C2SBP in adjusted models to predict AoSBP. Accuracy and precision were computed in the overall population and per age or sex strata.

Results: C1SBP underestimated AoSBP by 11.2 mmHg (±13.5) and C2SBP overestimated it by 6.2 mmHg (±14.8). Estimated SBP amplification and heart rate were the greatest predictors of C1- and C2-AoSBP accuracies, respectively. Statistical modeling improved both accuracy (0.0 mmHg) and precision (±11.4) but more importantly, eliminated the differences of accuracy seen in different sex and age groups.

Conclusion: Statistical modeling greatly enhances the accuracy of central BP measurements and abolishes sex- and age-based differences. Such factors could easily be implemented in central BP devices to improve their accuracy.