The incidence and significance of impaired heart rate variability (HRV) after acute myocardial infarction (AMI) have not yet been evaluated in cohorts of patients in whom early reperfusion was systematically attempted. Therefore, HRV was evaluated in 412 unselected patients with AMI (311 men, mean age: 60±12 years, anterior AMI in 172 patients) treated with direct coronary angioplasty (PTCA) within 12 hours of symptom onset (mean 3.5±2.0 h). Standard deviation of normal RR intervals (SDNN), square root of the mean of the sum of the squares of differences between adjacent RR intervals (RMSSD) and left ventricular ejection fraction (LVEF, mean: 55±15%) were measured 11±9 days after AMI before discharge. Angiotensin-converting enzyme (ACE) inhibitors and beta-blockers were prescribed at discharge in 81.1% and 70.1% of patients, respectively.

Mean SDNN was 94±30 ms (range 14–155). SDNN was <50 ms in 7% of patients. Mean RMSSD was 34±32 ms (range 2–234). RMSSD was <15 ms in 21% of patients. Low SDNN (<50 ms) was unrelated to gender, age, infarct location or extension of CHD but was related to low LVEF (p<0.001, logistic regression analysis). During mean follow-up of 4.3±3 years, there were 31 deaths; 24 were cardiac. SDNN was higher in long-term survivors (102±39 ms) as compared to nonsurvivors (81±33 ms, p=0.02) but RMSSD was unrelated to the long-term vital status. Four-year survival of patients with a SDNN <50 ms vs >50 ms was 80% vs 92%, respectively (p<0.001, Kaplan Meier analysis). Low SDNN (odds ratio OR=2.0, p<0.05) but not RMSSD was an independent denominator for long-term mortality as were low LVEF (OR=1.0 decrease in LVEF, p<0.01, proportional hazards model) and age (OR=1.1, p<0.001). Only 3/31 fatalities and 1/24 cardiac deaths were predicted by a SDNN <50 ms and only 5/31 fatalities by a RMSSD <15 ms.

The incidence of severely depressed HRV in patients after AMI is low (<10%) in the era of early reperfusion of the infarct vessel using direct PTCA. Mortality in patients with a very low HRV when assessed by SDNN is substantial but the positive predictive value of this parameter is low.