A new model for diurnal blood pressure profiling. Square wave fit compared with conventional methods.

Abstract

For the characterization of diurnal blood pressure variation, we developed a simple mathematical model that nevertheless does justice to the specific form characteristics of individual blood pressure registrations. Analysis was based on 24-hour continuous intra-arterial measurement of blood pressure obtained in 23 hospitalized patients with mild-to-moderate untreated essential hypertension (mean +/- SD, 112 +/- 13 mm Hg). The day-night difference for mean arterial pressure varied markedly (mean, 18.6 mm Hg; range, 6.8-36.0). Inspection of profiles suggested a model of blood pressure as two contiguous, complementary periods of constant pressure, a so-called square wave. Determination of the times of transience between both periods (segmentation) was performed individually using a least-square error criterion. Results were compared with those obtained by conventional methods, including analysis by Fourier modeling. The square wave fit accounted for a larger fraction (66%) of circadian variance of mean arterial pressure than modeling based on segmentation by visual inspection (59%, considerable observer bias) or by clock time (50%). Application of the Minnesota Cosinor Method resulted in the poorest description (47%). Segmentation based on harmonic modeling (61%) appeared to be cumbersome (10 harmonics needed), and the significance of additional information offered over the square wave fit is dubious. Observer bias makes segmentation by visual inspection unsuitable for assessment of the circadian variance of blood pressure. Even when daily activities are strictly regulated (hospital environment), circadian variance is not well modeled by clock time. As compared with harmonic analysis, square wave fitting is simple, and it appears to best model the circadian variance. The method can also be applied to data obtained from noninvasive ambulatory blood pressure monitoring.