Correcting Hypertension Prevalence Estimates for Terminal Digit Preference in Blood Pressure Readings

• Activity Number: 253
• Type: Contributed
• Date/Time: Monday, July 30, 2012 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Hao Wang*+ and Leslie A McClure and Ronald Prineas and George Howard and Virginia J Howard and Daniel F. Heitjan
• Companies: The Johns Hopkins University and The University of Alabama at Birmingham and Wake Forest School of Medicine and The University of Alabama at Birmingham and The University of Alabama at Birmingham and University of Pennsylvania
• Keywords: Digit preference ; Pearson distribution ; hypertension ; hypertension prevalence ; blook pressure

Abstract

Accurate blood pressure measurement is crucial in evaluating the health of populations and individuals. Terminal digit preference distorts the distribution of blood pressure measurements, potentially biasing estimates of hypertension prevalence. We develop a model-based method to remove the effects of digit preference in estimation of the blood pressure distribution. We apply the model to systolic blood pressure data of 30, 156 subjects from REGARDS, a national population-based cohort study to evaluate stroke risk factors. Our method estimates the percentage of subjects at risk of abnormal systolic blood pressure (= 120 mmHg) to be 64.6%, compared to the face-value estimate of 70.3%. Similarly, the estimated prevalence of hypertension decreases from 21.8% with the raw data to 18.8% under the model, and the estimated prevalence of severe high blood pressure (= 160 mmHg) declines by 24%. The method yields a description of the underlying blood pressure distribution unconfounded by misrecording, and could be applied to refine estimate of trends in blood pressure and the prevalence of hypertension.