How to avoid overestimating the burden of hypertension in epidemiological studies? A new methodology to account for within-person variability of blood pressure.

Abstract

Hypertension (HT) is a major modifiable risk factor for cardiovascular and renal diseases. HT is diagnosed as blood pressure (BP) exceeding a certain threshold. Because of the high within-person variability of BP, the estimation of HT in epidemiological studies based on single visit measurements tends to overestimate HT prevalence. Our study aimed to determine the correction factors to correct the bias in the estimation of HT prevalence in epidemiological studies. We used data from the National Health and Nutrition Examination Survey study in which BP was measured at three visits. A Bayesian hierarchical model was used to estimate the different components of BP variability (between individuals, between visits or between measures) by age and sex. These components allowed us to calculate the correction factors necessary to correct HT prevalence in epidemiological studies with single BP measurement. The method was then applied to data from the French Esteban study in which three standardized BP measurements were performed at a single clinical examination. The components of BP variability varied with age and sex, with different patterns observed for systolic and diastolic BP. Between-visit and between-individual variations drove BP variability, with between-measure variability being much lower. The uncorrected prevalence of HT in the Esteban study overestimated the burden of HT by 12.6% in the French population, with more than one in five untreated individuals being misclassified. The overestimation was higher in younger adults and women. Taking into account within-person BP variability in epidemiological studies could avoid a substantial overestimation of the burden of HT.