Biomed Tech 2006; 51:163–166  2006 by Walter de Gruyter • Berlin • New York. DOI 10.1515/BMT.2006.028 2006/514006 Article in press - uncorrected proof Multivariate and multidimensional analysis of cardiovascular oscillations in patients with heart failure Andreas Voss 1, *, Rico Schroeder 1 , Sandra Truebner 1 , Mathias Baumert 2 , Matthias Goernig 3 , Andreas Hagenow 4 and Hans-Reiner Figulla 3 1 Department of Medical Engineering, University of Applied Sciences Jena, Jena, Germany 2 Department of Medical Engineering, University of Applied Sciences Jena, Jena, Germany 3 Clinic of Internal Medicine I, Friedrich Schiller University Jena, Jena, Germany 4 Centre of Internal Medicine Elsterwerda, Elsterwerda, Germany Abstract Within 5 years of first diagnosis, nearly 60% of patients with heart failure (HF) suffer from cardiac death. Early diagnosis of HF and reliable risk prediction are still required. Therefore, the objective of this study was to develop a parameter set for enhanced risk stratification in HF patients. In 43 patients suffering from HF (NYHA class GII, ejection fraction -45%) and 10 healthy sub- jects (REF), heart rate and blood pressure variability (HRV and BPV), interactions between heart rate and blood pressure (joint symbolic dynamics, JSD) and blood pres- sure morphology (BPM) were analysed. BPV, BPM and JSD measures revealed high significance (p-0.0001) in discriminating REF and HF. A set of three parameters from BPV, JSD and BPM was developed for risk stratifi- cation (sensitivity 76.5%, specificity 84.2%, area under the receiver operating characteristic curve 81.4%) in patients with HF. Keywords: blood pressure variability; cardiovascular dis- ease; heart rate variability. Introduction In Europe approximately 14 million people suffer from heart failure (HF). Statistical analyses show that the inci- dence of HF will dramatically increase in the future to approximately 30 million by the year 2020. Furthermore, 60% of HF patients suffer from cardiac death within 5 years of diagnosis and HF is the most common reason for hospitalisation in the over 65 age group w 9x . HF is a complex cardiovascular disease resulting from functional or structural cardiac disorder, mostly caused *Corresponding author: Andreas Voss, Prof. Dr.-Ing., Department of Medical Engineering, University of Applied Sciences Jena, Carl-Zeiss-Promenade 2, 07745 Jena, Germany Phone: q49-3641-205601 Fax: q49-3641-205626 E-mail: voss@fh-jena.de by coronary artery disease, hypertension and cardiomyo- pathy and characterised by impaired ventricular filling or reduced ventricular ejection fraction (EF) w 6x . Physicians often assess the extent of HF according to the New York Heart Association (NYHA) functional classification sys- tem, which places patients in NYHA classes I–IV based on how much they are limited during physical activity. However, currently applied strategies for early diagnosis of HF are far from being optimal. Improved risk stratifi- cation for HF is necessary to assess prognosis and to identify individual drug treatment strategies or optimal timing for pacemaker or cardioverter defibrillator implan- tation or heart transplantation. The aim of this study was to develop a multivariate parameter set for enhanced risk stratification in HF patients. Materials and methods For this study, 43 patients suffering from HF characteri- sed by NYHA G2 and EF -45% and 10 healthy subjects as reference (REF) were enrolled. All patients were treat- ed with ACE inhibitors (85% of all patients), beta-block- ers (85%), digitalis (24%) and diuretics (80%). ECG (30 min) and non-invasive blood pressure (NIBP) were recorded under standardised resting conditions (supine position, quiet environment, same time of day and location) using the Portapres M2 blood pressure monitor (TNO-TPD, Amsterdam, Netherlands). Using a commercially available amplifier system (Twente Medical Systems, Enschede, Netherlands), ECG and NIBP were sampled at f s s1600 Hz. After discretisation, the signals were stored in a database together with the patient data. The diagnosis of all HF patients was confirmed by an experienced cardiologist using short- and long-term ECG, as well as stress ECG, echocardiography and heart catheter examination. From the 30-min data records, time series of heart rate (tachogram) consisting of beat-to-beat intervals (BBI) and of blood pressure values (systogram/diastogram) were extracted. Ectopic beats and other disturbances were removed. To quantify the heart rate variability (HRV) and blood pressure variability (BPV), several parameters of the time domain, frequency domain and non-linear dynamics were calculated from each tachogram, systo- gram and diastogram. From the time domain, the following parameters w 10x were calculated (dia_, diastolic; sys_, systolic): • meanNN: mean value of BBI time series w msx ; • sdNN: standard deviation of BBI series w msx ; • rmssd: square root of the mean squared differences of successive beat-to-beat intervals w msx ;