| Based on the first 10 biennial examinations of the Framingham Heart Study, bivariate survival methods are used to investigate the relationship between the age at first detection of hypertension and the age at onset of cardiovascular disease. Four groups of people are defined: 317 hypertensives who experienced cardiovascular disease, 410 normotensives who experienced cardiovascular disease, 875 hypertensives free of cardiovascular disease by examination 10, and 2929 normotensives free of cardiovascular disease by examination 10. These four groups provide motivation for the construction of a likelihood function using bivariate survival functions derived from Weibull distributions. The following risk factors for cardiovascular disease are introduced into the Weibull scale parameters as dichotomous covariates: self-reported use of anti-hypertensive medication, self-reported smoking, gender, elevated total serum cholesterol, and large body mass. FORTRAN programs are presented for analysis of data using iterative maximum likelihood methods to obtain estimates of the parameters and standard errors.;Conditional expectations, which account for the effects of risk factors, calculated for the expected age at onset of cardiovascular disease conditional on knowing the age at first detection of hypertension. Conclusions drawn from the model with the largest log likelihood demonstrate statistically significant effects of the risk factors smoking, male and large body mass. These risk factors (and the not statistically significant risk factor elevated total serum cholesterol) lead to an earlier expected age of onset of cardiovascular disease. Self-reported anti-hypertensive medication use is statistically significant and results in delayed expected age of onset of cardiovascular disease. |
