388 DIABETES CARE, VOLUME 22, NUMBER 3, MARCH 1999 N oninvasive exploration of the auto- nomic nervous system is important for scientists and physicians because cardiac autonomic nervous system dys- function is a precocious complication in diabetes in which diabetic neuropathy pre- dicts an increased risk in cardiovascular morbidity and mortality (1,2). In auto- nomic neuropathy, blood pressure (BP) reg- ulation is impaired by both vascular sympathetic and cardiac baroreflex alter- ations. The current diagnosis of cardiovas- cular neuropathy is based on a battery of well-established tests stimulating the car- diovascular autonomic system (3). These tests, first described by Wheeler and Watkins (4) and later by Ewing et al. (5), must be rigorously standardized and imply a close cooperation of the subjects. More simple and reliable methods of detecting and quantifying cardiac dysau- tonomy are of clinical interest. Autonomic control of BP may be reflected by sponta- neous beat-to-beat BP and heart rate (HR) variabilities obtained by a Finapres record- ing. Various mathematical analysis cus- tomized to BP and HR beat-to-beat variations quantify the strength of their relations. Thus, the spectral analysis was reported to provide useful information on autonomic nervous function (6–8). Car- diac baroreflex sensitivity (BRS) was reported to be reliably estimated by cross- spectral analysis (including Fourier analy- sis, as well as autoregressive modeling), by the sequence method (analysis of linear covariations of systolic blood pressure [sBP] and HR), and by the Z method (analysis of the statistical dependence of BP and HR) (9–13). Furthermore, the nonlin- ear dynamic of the cardiovascular system can be analyzed by the fractal and chaos analysis that quantify its degree of com- plexity (14–16). All these methods were applied to diabetic subjects to quantify their cardiac dysautonomy, but their rela- tive interest has never been assessed. The purpose of this work, then, was to compare indices of autonomic nervous system func- tion obtained from diabetic subjects exhibiting nonsymptomatic orthostatic hypotension. From the Département de Néphrologie et Hypertension Artérielle (M.D., M.L., J.P .F.) and the Département d’Endocrinologie, Diabète, Nutrition (S.A.-A., C.T.), Hopital E. Herriot; and the Département de Physiolo- gie et Pharmacologie Clinique (M.D., C.C., M.P.G., C.Z.P.), Faculté de Pharmacie, Lyon, France. Address correspondence and reprint requests to J.P. Fauvel, Département de Néphrologie et Hypertension Artérielle, Hopital E. Herriot, 69437 Lyon, France. Received for publication 11 May 1998 and accepted in revised form 17 November 1998. Abbreviations: BP, blood pressure; BRS, baroreflex sensitivity; dBP, diastolic blood pressure; HF, high fre- quency; HR, heart rate; LF, low frequency; RD, relative dispersion; sBP, systolic blood pressure; VLF, very low frequency. A table elsewhere in this issue shows conventional and Système International (SI) units and conversion factors for many substances. Noninvasive Exploration of Cardiac Autonomic Neuropathy Four reliable methods for diabetes? O R I G I N A L A R T I C L E OBJECTIVE — The purpose of this work was to assess relevant information that could be provided by various mathematical analyses of spontaneous blood pressure (BP) and heart rate (HR) variabilities in diabetic cardiovascular neuropathy. RESEARCH DESIGN AND METHODS — There were 10 healthy volunteers and 11 diabetic subjects included in the study. Diabetic patients were selected for nonsymptomatic orthostatic hypotension in an assessment of their cardiovascular autonomic impairment. Car- diac autonomic function was scored according to Ewing’s methodology adapted to the use of a Finapres device. The spontaneous beat-to-beat BP and HR variabilities were then analyzed on a 1-h recording in supine subjects. The global variabilities were assessed by standard devi- ation, fractal dimension, and spectral power. The cardiac baroreflex function was estimated by cross-spectral sequences and Z analyses. RESULTS — In diabetic patients, Ewing’s scores ranged from 1 to 4.5, confirming cardio- vascular autonomic dysfunction. In these diabetic patients, global indices of variabilities were consistently lower than in healthy subjects. Furthermore, some of them (standard deviation and fractal dimension of HR, spectral power of systolic blood pressure and HR) were signifi- cantly correlated with the Ewing’s scores. The Z methods and the spectral analysis found that the cardiac baroreflex was less effective in diabetic subjects. However, the baroreflex sensitiv- ity could not be reliably assessed in all the patients. The sequence method pointed out a decreased number of baro re flex sequences in diabetic subjects that was correlated to the Ewing’s score. CONCLUSIONS — Indices of HR spontaneous beat-to-beat variability are consistently related to the degree of cardiac autonomic dysfunction, according to Ewing’s methodology. The Z method and spectral analysis confirmed that the cardiac baroreflex was impaired in diabetic patients. These methods might be clinically relevant for use in detecting incipient neuropathy in diabetic patients. Diabetes Care 22:388–393, 1999 MICHEL DUCHER, PHD CATHERINE CERUTTI, PHD MARIE PAULE GUSTIN, PHD SAWSAN ABOU-AMARA, MD CHARLES THIVOLET , MD, PHD MAURICE LAVILLE, MD CHRISTIAN ZACHARIE PAULTRE, PHD JEAN PIERRE F AUVEL, MD, PHD Clinical Care/Education/Nutrition