ESTIMATING PREDICTIVE VALUES FOR BLOOD-PRESSURE MEASUREMENTS FROM MULTIVARIATE REGRESSION-MODELS WITH COVARIATES

Predictive values are useful in estimating the probability distributio n of a 'true' or underlying measurement, that is, without measurement error or within-person variability. They have been applied to blood pr essure data to estimate the true probability that a person is hyperten sive currently, or that he/she will become hypertensive based on previ ous data from childhood. The current work extends these results to sit uations where covariates are of interest. One can use multivariate reg ression models to model predictive values for future levels as functio ns of covariates as well as current measured levels. I compare predict ive value estimates obtained from these models to those obtained from ordinary linear regression and from logistic regression with use of da ta on childhood blood pressure from East Boston, MA. Estimates obtaine d using the multivariate model are preferable either in terms of bias in the estimates themselves or in terms of their variability. This is particularly true with covariates included in the model. The differenc e between the multivariate and ordinary regression estimates depends o n the conditional reliability of future levels given current blood pre ssure levels and covariates. I also discuss predictive value estimates for true current level given observed level as well as covariates. Th ese also depend on the reliability of the current measure given values of covariates.