Correcting for Measurement Error of Tracking Cardiovascular Risk Factors in Meta-analysis

Abstract

Hypertension and obesity are the acknowledged risk factors of coronary heart disease and cardiovascular diseases. To prevent hypertension or obesity, it is beneficial to identify “prehypertensives” earlier in life and to control the body mass in long term. Therefore, tracking of blood pressure (BP) and body mass index (BMI) appears to be paramount. However, the measurement errors are often not taken into account when the effects of tracking are estimated and conclusions are drawn. In our study, the classical additive measurement error model is used to analyze the measurement error in measuring blood pressure, height and weight. Moreover, reliability is a useful measure to quantify the amount of error in measurement and to correct the attenuation effect of correlation coefficients. The purpose of this study is to correct measurement errors in tracking the two cardiovascular risk factors blood pressure and BMI in meta-analysis. Due to the absence of the necessary reliability information in each primary study, the issue of corrections for measurement error in meta-analysis becomes more complicated. To reach the purpose, the reliability is estimated by gathering of data from other studies: the reliability coefficient of BP measurement is estimated by assessing two components, between-visit and within-visit variation; and the reliability of BMI is derived from its components, height and weight. The reliability of BMI is much higher than that of BP. Comparing the tracking effects before and after correction of measurement errors, the corrected tracking correlations of blood pressure from childhood to manhood rise substantially, whereas, there are no obvious improvements for tracking effect of BMI.