A comparison of three trough to peak estimators derived from ambulatory blood pressure data

Abstract

A general multivariate model is proposed to analyze ABPM data obtained from multiple subjects in which two ABPM series, typically at baseline and on randomized therapy in a clinical trial, are obtained on each subject. The model is a specific formulation of the general repeated measures multivariate regression model in which the covariance matrix associated with each of the two ABPM profiles are associated equal and with unspecified structure. Exact maximum likelihood estimates of the unknown parameters and asymptotic properties for the case of no missing data are derived.Three methods to estimate the trough to peak ratio using ABPM data are examined in this dissertation. These include a method based on between subject time of day averages, the second method using a fitted Fourier curve using the repeated measures multivariate model, and a third simple Fourier regression model. Asymptotic properties of the three estimators are derived and compared, as well as a Monte Carlo simulation to determine finite sample properties.