Relationships between metrics of visit-to-visit variability of blood pressure

This paper examines relationships between metrics of visit-to-visit variability (VVV) of blood pressure (BP) to determine which metrics should be calculated in studies of the association of VVV with health outcomes. We examined correlation and agreement between quintiles for standard deviation (s.d.), standard deviation independent of the mean (SDIM), coefficient of variation (CV), successive variation (SV), average real variability (ARV), range, maximum, peak size and trough size of systolic BP in the Trial of Preventing Hypertension placebo arm (n = 288). The average age of participants was 48 years. Mean systolic BP was 133.5 mm Hg. VVV metrics were all significantly correlated (P < 0.001). Correlations between s.d., SDIM, CV and range and between ARV and SV were >= 0.90. Kappa statistics between quintiles of SD, SDIM, CV and range and between ARV and SV were >= 0.80. In studies of the relationship of VVV with health outcomes, we recommend reporting results for one of the metrics of overall variability (s.d., SDIM, CV), one of the metrics of variability between consecutive visits (SV, ARV), and one or more of the metrics of extreme values at a single visit (maximum, peak size, trough size).