Research Project
2735347
This project will use measures of heart rate variability (HRV) to provide new and clinically relevant information concerning pathological processes affecting the cardiovascular system (CV). HRV refers to changes in the time between consecutive heart beats. HRV measurements made from electrocardiograms (ECGs) taken 2 weeks following myocardial infarction (MI) will be used to predict cardiac death within 4 years. While measures such as left ventricular ejection fraction can help risk stratify patients, present risk factors do not have a good positive predictive accuracy for mortality following MI. HRV measures using power spectra (PS) based on Fourier transforms have already been shown to improve prediction of mortality following MI. However, newer HRV measures promises even better clinical utility. Compared to PS measures of HRV, these "nonlinear" measures have different theoretical bases (requiring fewer physiological assumptions), measure different aspects of HRV and are not necessarily altered in the same way as PS measures by changes in the CV. Thus, along with PS measures of HRV, we will analyze: the dimension of HRV with point dimension correlation (PD2), the regularity of HRV with approximate entropy (ApEn), and the grammar complexity (GC) of HRV. ECG data from the Multicenter Post-Infarct Project which have already been generated and analyzed with respect to risk factors following MI will be reanalyzed with the new HRV measures. Short (about 10 minute) segments from the afternoon will be selected from the 24-hour ECGs. The hypothesis that the different types of HRV measures (PS, PD2, ApEn, GC) provide nonduplicative information about the CV post-MI will be examined by the use of Spearman correlations. The hypothesis that nonlinear measures of HRV (PD2, ApEn and GC) will enhance prediction of cardiac death will be tested by comparing relative risks from Cox proportional hazards models using individual nonlinear HRV measures to models using PS HRV measures. Additionally, stepwise multivariate Cox proportional hazards procedures will be used to test the benefit of combining different types of HRV measures. Not only will this project provide important information for predicting mortality risk following myocardial infarction, but it also will provide new and valuable information concerning the benefit of using different types of HRV measures to monitor compromise of the cardiovascular system which may be generally applicable to heart diseases and their treatments.
