Abstract P197: Methods to Account for Antihypertensive Drugs for Estimating Underlying Blood Pressure: The Atherosclerosis Risk in Communities (ARIC) Study
Introduction: Antihypertensive drugs are widely used to lower blood pressure (BP), particularly among older adults. Thus considerations are required in epidemiological studies dealing with the “underlying” BP (BP that would have been if antihypertensive drugs were not provided), e.g. studies exploring natural history of BP.
Materials and Methods: We compared three methods proposed in literature for accounting for antihypertensive drugs in the ARIC study, which consists of 45-64 years old with ~25 years of follow-up data (1987-2013). Method 1 simply adds a constant of 10 mmHg to systolic blood pressure (SBP) and 5 mmHg to diastolic blood pressure (DBP) to the measured BP. Method 2 adds a constant derived from the expected drug class effects as reported by short-term clinical trials (e.g. angiotensin converting enzyme inhibitors, alpha blockers, beta blockers, calcium channel blockers, diuretics, miscellaneous). Method 3 uses truncated normal regression to incorporate covariates (e.g. age, sex, race, BMI, height, sex-height interaction), where the distribution of BP is assumed to be normal and requires the underlying BP to be greater than or equal to the measured BP. We examined these methods by observing the distribution of underlying BP compared to the measured BP on antihypertensive drugs.
Results: The prevalence of antihypertensive drug use in ARIC is as follows: 30% in Visit 1 (4370/14659), 33% in Visit 2 (4732/14299), 38% in Visit 3 (4874/12842), 44% in Visit 4 (5095/11610) and 76% in Visit 5 (4904/6469). Method 3 compared to methods 1 and 2 creates a distribution of underlying BP that is shifted more to the right and with less variation. This trend for SBP and DBP was true for all visits and more noticeable with each subsequent visit (Visit 5 SBP illustrated in figure).
Conclusions: Overall, truncated normal regression (method 3) yields more plausible and stable underlying BP compared to methods 1 and 2. Estimation of underlying BP can be used in epidemiological to investigate the natural history of BP, such as in genetic studies.
Author Disclosures: P. Balakrishnan: None. E. Colantuoni: None. J. Young: None. T. Beaty: None. K. Matsushita: None.
- © 2015 by American Heart Association, Inc.