Measurement accuracy of non-invasively obtained central blood pressure: a systematic review and meta-analysis

Abstract

<h4>Background</h4>Blood pressures determined at different sites vary considerably. Non-invasive methods are available to estimate central aortic blood pressure, the blood pressure at the origin of all arterial pulses. These methods obtain estimated central blood pressure by calibration and/or mathematical calculations for peripheral pulse waveforms. However, the accuracy of these methods has not been systematically examined.<h4>Objectives</h4>The review aimed to synthesise the best evidence on the accuracy of non-invasive measurement methods for central blood pressure.<h4>Inclusion criteria</h4>Types of participantsStudies with adult patients receiving invasive and non-invasive measurements of central blood pressure were considered.<h4>Phenomena of interest</h4>Studies were considered for inclusion if the focus was accuracy of non-invasive central BP estimating methods compared to invasively obtained corresponding values.<h4>Types of studies</h4>Studies examining agreement between measurements using non-invasive central blood pressure estimating methods compared to invasive corresponding values were considered.<h4>Types of outcomes</h4>This review included the means and standard deviation of differences between estimated and invasively measured central blood pressure.<h4>Search strategy</h4>The search sought to identify any relevant published or unpublished studies with a three-step search strategy.<h4>Methodological quality</h4>Two independent reviewers assessed methodological quality of studies by a critical appraisal tool modified from Cochrane Diagnostic Test Accuracy Working Group.<h4>Data collection</h4>We used an original form to extract from included studies all study characteristics possibly related to agreement.<h4>Data synthesis</h4>Inverse variance weighted approach and DerSimonian-Laird weights for the random effects model, which incorporates a between-study variance, were used to obtain pooled estimates of systematic and random error from individual study estimates of the mean and standard deviation of differences between the paired measurements. Heterogeneity was assessed using Cochran Q. All analyses were performed in Microsoft Excel 2003.<h4>Results</h4>Twenty eight studies were eligible for inclusion and critically appraised in this review. Appropriate data for agreement were extracted from papers or authors in 20 studies, which were further included in meta-analysis. Acquired peripheral waveforms in these studies were directly measured, calibrated to match invasively obtained aortic mean blood pressure and diastolic blood pressure, or calibrated using brachial blood pressure measured by sphygomomanometer, the cuff blood pressure. Estimated central blood pressure of the studies using the last totally non-invasively methods (real world practices) were subject to meta-analysis separately from studies with the former two invasive methods (theoretical practice). Of the invasive methods, mean difference of the estimated central blood pressure was small (-1.2 ± 4.2mmHg for central systolic blood pressure, -0.6 ± 2.1mmHg for central diastolic blood pressure, and -1.1 ± 5.3 mmHg for central pulse pressure). However, the errors of the non-invasive method inflated considerably (-8.1 ± 10.7mmHg for central systolic blood pressure, 8.8 ± 9.5mmHg for central diastolic blood pressure, and -11.8 ± 13.3 mmHg for central pulse pressure). The findings were similar in subgroup analysis by different central blood pressure methods and by validated cuff monitors.<h4>Conclusions</h4>Current central blood pressure estimating methods are acceptable in theory with small systematic and random error. However, the error of these methods was evident when cuff blood pressure was used for calibration and probably made them clinically inapplicable.