Linear relationship between systolic and diastolic blood pressure monitored over 24 h: assessment and correlates

Objectives Systolic blood pressure (SBP) and diastolic blood pressure (DBP) frequently display a linear relationship characterized by the systolic-versus-diastolic slope (`Slope') or the `ambulatory arterial stiffness index' [AASI = 1 - (diastolic-versus-systolic Slope)] and the correlation coefficient r. We evaluated the effect of using symmetric regression on the AASI and its dependence on clinical characteristics using 24-h ambulatory monitoring. Methods Ambulatory monitoring data of 140 patients (age 56 +/- 17 years, 45% men) were retrieved from a service database. Slope and the AASI were evaluated using symmetric regression procedures, and the AASI also by standard regression. Results Correlation between SBP and DBP was r = 0.74 +/- 0.14 ( r > 0.5 in 95% of patients). Low r-values ( when SBP correlates poorly with DBP) were tightly linked with nondipping (P < 0.00001). Use of symmetric rather than standard regression eliminated the bias in slope-related parameters and unmasked their dependence on clinical characteristics. Both symmetric Slope and the AASI were independent of mean arterial pressure and r, increased with pulse pressure (P < 0.01 and P < 0.0001, respectively), with the greater effect of wider pulse pressure in older age ( P < 0.005 for both). The symmetric slope was 1.29 +/- 0.28, showing bivariate dependence (r = 0.82) on age (exponential, with P < 0.00001) and pulse pressure dipping (P < 0.00001), increased for antihypertensive drug treatment (0.07 +/- 0.03, P < 0.05) and diabetes mellitus (0.18 +/- 0.06, P < 0.005). Conclusions Application of symmetrical regression provides a more valid estimate of the systolic-on-diastolic slope and the AASI, less influenced by goodness of fit and nocturnal dipping and more sensitive to age and disease states such as hypertension and diabetes, thus providing an improved index of arterial stiffening.