Modeling blood pressure: Comparative study of seemingly unrelated regression and ordinary least squares estimators

Abstract

Most authors have focused on Systolic Blood Pressure(SBP) and Diastolic Blood Pressure(DBP) separately. The effect of some identified risk factors on SBP and DBP can be estimated separately since they are affected by different factors.This study is aimed at developing a model that can appropriately capture the relationship between SBP and DBP rather than estimating the two separately. Also, to compare the efficiency of joint estimator; Seemingly Unrelated Regression(SUR) over the separate estimator; Ordinary Least Squares(OLS). The SUR model which is a special case of multivariate regression model was used to  simultaneously capture the effect ofSBP and DBP. Data collected on age, sex, weight, waist, profession, history,  Triglycerides(TRS) and Body Mass Index (BMI) from 100 patients of Olabisi Onabanjo University Teaching Hospital were used for the analysis. The results showed that there is a positive correlation between the SBP and DBP.The standard errors of SUR estimator were consistently lower than that of OLS estimator. The Correlation between SBP and DBP is |ρ| = 0.4434 which confirmed the report of Zellner, 1962, Dieliman1989 Aebayo, 2003;and shows that SUR is an appropriate estimator for this study.The simultaneous estimation of SBP and DBP gave a higher precision in this study

Key words: systolic and diastolic blood pressures, joint and separate estimators