Predictors of visit-to-visit blood pressure variability in patients with hypertension: an analysis of trials with an amlodipine treatment arm

Abstract

We conducted a post hoc analysis of blood pressure (BP) data from long-term antihypertensive trials to identify predictors of visit-to-visit BP variability (BPV). BPV was defined as the within-subject coefficient of variation in systolic BP from week 12 onward. BP data from the Anglo-Scandinavian Cardiac Outcomes Trial-Blood Pressure Lowering Arm, Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial, Comparison of Amlodipine Versus Enalapril to Limit Occurrences of Thrombosis, NY92011, and R-0510 trials were pooled and dichotomized into top 25th and bottom 75th percentiles because of positive skew. Significant (P < .001) predictors of BPV within the top 25th percentile were identified using logistic regression. The baseline characteristics of the pooled cohort (n = 47,558) were similar between patients who received amlodipine (n = 17,499) versus other antihypertensive drugs (n = 29,491). BPV in the top 25th percentile was lower with amlodipine versus other treatments (13.7 ± 3.2 vs. 14.3 ± 3.5), with single-study analyses of Anglo-Scandinavian Cardiac Outcomes Trial-Blood Pressure Lowering Arm, Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial, and Comparison of Amlodipine Versus Enalapril to Limit Occurrences of Thrombosis all showing BPV was the lowest with amlodipine. Baseline diastolic BP, estimated glomerular filtration rate, and smoking were predictors of BPV, with significant two-way interactions between smoking and both age and body mass index and between systolic BP or diastolic BP and being randomized to treatment other than amlodipine. In conclusion, analysis of BPV required transformation of BP data. After transformation, a number of baseline variables and combinations of variables were predictors of BPV.