The Norsjö-Cooperstown healthy heart project: A case study combining data from different studies without the use of meta-analysis

Abstract

This paper aims to develop and describe a method for combining. comparing, and maximizing the statistical power of two longitudinal studies of risk factors for cardiovascular disease that did not have identical data collection methodologies. Subjects from a 1986 cross-sectional study (n = 180) were pair-matched with subjects of corresponding gender and age (+5 years) from a 1990 cross-sectional study. The methodology is described and results are calculated for various measures of cardiovascular risk or risk factors (e.g. cholesterol. Finnish Risk Score). Box's test of equality and symmetry of covariance matrices gave chi-square values of 223.8 and 710.0 for two cardiovascular risk factors (cholesterol and cardiac risk score, respectively); these values were highly significant (p=0.0001) For the North Karelia Risk Score, repeated measures ANOVA revealed a borderline significant interaction for treatment by time (p=0.054) and a significant interaction for treatment by time by country (p=0.035). These probabilities compared favorably with a randomized blocks model. Creation of a synthetic longitudinal control group resulted in a statistically valid ANOVA model that increased the statistical power of the study.