. S.M. and A.S. contributed equally to this article. Abbreviations: DCCT, Diabetes Control and Complications Trial; E/I ratio, ratio of length of R-R intervals during expiration to inspiration; ETDRS, Early Treatment of Diabetic Retinopathy Study; HF, high frequency; HFnorm, normalized high frequency component; HRV, heart rate variability; LF, low frequency; LFnorm, normalized low frequency component; RMSSD, square root of the mean square of R-R interval differences; TP, total power; UAER, urinary albumin excretion rate; VLF, very low frequency.A OBJECTIVE -Frequency domain analysis of heart rate variability (HRV) is used to assess cardiovascular autonomic function. There are no prospective data on the sensitivity of its various components to glycemia or other diabetes-related risk factors compared with conventional tests and with other complications of diabetes.
RESEARCH DESIGN AND METHODS -In 1985, possible risk factors of future complications were determined in 115 children with type 1 diabetes. In 1996, the presence of complications (HRV analysis, conventional tests of autonomic function, urinary albumin excretion rate [UAER], and retinopathy) were assessed in 83 of these patients (age 32 ± 1 years, duration of diabetes 22 ± 1 years).
RESULTS -Poor glycemic control (measured as lifetime glycemic exposure or HbA 1c in 1985)was the most important independent predictor of decreases in all measures of absolute power of HRV (total power [TP] and very low frequency, low frequency [LF], and high frequency [HF] power) and square root of the mean square of R-R interval differences but not of changes of normalized measures or ratios (normalized HF and LF, LF/HF). Other significant independent predictors of autonomic dysfunction were late age of onset of diabetes, female sex, and high BMI. To examine the sensitivity of the various tests to glycemia, the patients were divided into tertiles based on lifetime glycemic exposure (A 1c months). Glycemic exposure in the tertiles averaged 194 ± 25 A 1c months (20 years of HbA 1c 0.8% above normal), 556 ± 19 A 1c months(20 years of HbA 1c 2.3% above normal), and 963 ± 30 A 1c months (20 years of HbA 1c 4% above normal). Tests of complications that were significantly abnormal in patients already in the lowest tertile and were correlated with glycemia were TP and severity of retinopathy. Of conventional tests, only the ratio of length of R-R intervals during expiration to inspiration (E/I ratio) was significantly related to glycemic exposure, but it required high glycemic exposure (20 years of HbA 1c 4% above normal) to be abnormal. UAER was significantly increased only in the highest tertile of glycemic exposure.CONCLUSIONS -TP and retinopathy score were much more sensitive to antecedent glycemia than conventional tests of autonomic function or UAER and were significantly abnor-