Electrophysiology Including patients with diabetes mellitus or coronary artery bypass grafting decreases the association between heart rate variability and mortality after myocardial infarction Phyllis K. Stein, PhD, a Peter P. Domitrovich, PhD, b and Robert E. Kleiger, MD, FACC, a for the CAST Investigators St Louis, Mo Background Decreased heart rate variability (HRV) is often assumed to be associated with mortality in all patients after myocardial infarction (MI), independent of clinical factors or time after MI. Method HRV was determined from Holter tapes in the Cardiac Arrhythmia Suppression Trial (CAST). Patients were 71  120 days after MI. A total of 735 pre-therapy tapes were analyzed in patients who had ventricular premature con- tractions (VPCs) suppressed on the first treatment. The period of follow-up was 362  243 days (69 deaths). The associa- tion of clinical and demographic factors and 24-hour, daytime, and nighttime HRV to mortality in all patients, patients without coronary artery bypass graft (CABG) surgery between the qualifying MI and the Holter monitoring, and patients with neither CABG nor diabetes mellitus was determined with univariate Cox regression analysis. Results For the entire group and the subgroup without CABG, the strongest association was with increased daytime normalized high frequency power (NHF day). Further excluding patients with diabetes mellitus strengthened the associa- tion of HRV with mortality rate. Decreased natural logarithm (ln) 24-hour total and ultra low frequency (ULF) power were the strongest predictors of mortality. The best cutoff point for ln ULF for separating survivors and non-survivors was deter- mined. After including a history of MI, congestive heart failure, or both as co-factors, ln ULF 7.85 identified patients at approximately 4-times the relative risk of mortality, but did not risk-stratify patients without prior MI or history of congestive heart failure. Conclusions HRV predicts mortality rate in a broad range of times after MI. Excluding patients with CABG after MI or with diabetes mellitus significantly strengthens the association of HRV with mortality. HRV measures beyond the peri- infarction period, with clinical factors, can identify subgroups at an elevated risk of mortality. (Am Heart J 2004;147: 309 –16.) The usefulness of 24-hour time and frequency do- main heart rate variability (HRV) measured in the peri- infarction period for risk stratification of patients after myocardial infarction (MI) has been validated in nu- merous investigations, from the Multi-Center Post-In- farction Project (MPIP) in the late 1980s to the recent Autonomic Tone and Reflexes After Myocardial Infarc- tion (ATRAMI ) study. 1–13 It is often assumed that de- creased HRV is associated with an increased risk in virtually all patients after MI. Recent subgroup analyses have suggested that the same HRV risk factors may have different prognostic value in different groups. For example, in ATRAMI, the predictive value of low HRV in the subgroup with low left ventricular ejection frac- tion was far greater for patients aged 65 years. Stud- ies combining HRV with other risk factors affirm that there are subgroups in which decreased HRV has espe- cially high prognostic value. 6,13–15 The association of decreased HRV and mortality rate has generally been validated for the peri-infarction pe- riod. The Cardiac Arrhythmia Pilot Study (CAPS), a pi- From the a Cardiovascular Division, Department of Medicine, Washington University School of Medicine, and b Barnes-Jewish Hospital, St. Louis, Mo. Supported by a grant from the National Heart, Lung, and Blood Institute (RO-3 Grant HL 53776). Submitted August 20, 2002; accepted July 3, 2003. Reprint requests: Phyllis K. Stein, PhD, Washington University School of Medicine Heart Rate Variability Laboratory, 4625 Lindell Blvd, Suite 402, St. Louis, MO 63108. E-mail: pstein@im.wustl.edu 0002-8703/$ - see front matter © 2004, Elsevier Inc. All rights reserved. doi:10.1016/S0002-8703(03)00520-9