We tested an innovative approach for estimating baroreflex sensitivity (BRS) from the gain function between spontaneous oscillations of systolic arterial pressure (SAP) and heart period (HP). The major goal was to assess the practical implications of abandoning the classical coherence criterion (⩾0.5) as regards measurability of BRS, and agreement with values of BRS obtained using the phenylephrine test (Phe-BRS). We studied 19 normal subjects, 44 patients with a history of previous myocardial infarction (MI) and 45 patients with chronic heart failure (CHF). The experimental protocol included recording of SAP and HP for 10min of supine rest, and evaluation of Phe-BRS. From resting SAP and HP, the gain and coherence functions were computed. The new BRS index was obtained in all subjects by averaging the gain function over the whole low-frequency band (0.04-0.15Hz) (whole-band average BRS, WBA-BRS). WBA-BRS was 7.4 (5.8-10.8)ms/mmHg [median (25th-75th percentile)] in normal controls, 3.1 (1.4-5.4)ms/mmHg in MI patients (P<0.001 compared with normals) and 5.0 (3.2-6.9)ms/mmHg in CHF patients (P<0.01 compared with normals). Using the coherence criterion, BRS could be measured in only 43% and 49% of MI and CHF patients respectively, and the proportion of the low-frequency band contributing to the measurement was 21% (14-47%) and 29% (16-35%) respectively. The correlation between WBA-BRS and Phe-BRS was 0.47, 0.63 and 0.36 in the normal, MI and CHF groups respectively (all P<0.001). The relative bias of WBA-BRS was -5.2ms/mmHg (P<0.001) in normals, -1.4ms/mmHg (P = 0.004) in MI patients and -1.0ms/mmHg (P = 0.11) in CHF patients. The limits of agreement were -13 to 2.6, -7.4 to 4.6 and -9.3 to 7.3ms/mmHg in the normal, MI and CHF groups respectively. Thus the WBA-BRS method standardizes the computation of BRS among subjects, and dramatically increases its measurability in subjects with pathology compared with the classical spectral technique based on the coherence criterion. Compared with Phe-BRS, WBA-BRS tends to give negatively biased results. The correlation and the magnitude of the limits of agreement between the two methods are similar to those observed previously using coherence-based spectral methods.

This content is only available as a PDF.
You do not currently have access to this content.