The point correlation dimension of R-R intervals predicts sudden cardiac death among high-risk patients

An algorithm for the correlation dimension, the PD2, accurately tracks dimension in nonstationary data. Low-dimensional PD2s predict ventricular fibrillation (VF) in experimental myocardial infarction in animals. The authors now demonstrate in 61 electrocardiographic recordings from ambulatory high-risk cardiac patients that low-dimensional chaos in the heartbeat intervals, as represented by a reduction in PD2 below 1.2 dimensions, precedes lethal VF by a mean of 12 hours. This PD2-reduction occurs only in those high-risk patients who actually experience PD2-reduction had a high sensitivity (100%) and specificity (83%) in predicting VF among the high-risk patients. In contrast, the standard deviations of the same heartbeat intervals provided no such discrimination. In symptomatic extremely-high-risk patients, each having a history of syncope and episodic ventricular tachycardia, but who did not experience VF that day, PD2s<1.2 were observed in 9 of 10 subjects. This latter result suggests that VF-prediction and VF-occurrence may be separated by days not just hours. It is concluded that the PD2 of heartbeat intervals may provide an important diagnostic test and early warning signal for sudden cardiac death