Session MC.2

Non-Invasive Potassium Measurements from ECG Analysis during Hemodialysis Sessions

S Severi, C Corsi*, M Haigney, J DeBie, D Mortara

Universidad de Bologna
Bologna, Italy

It is well known that several ECG alterations are related to electrolyte changes. In particular, blood potassium concentration ([K+]) has a strong influence on T-wave morphology. However, no quantitative relations between parameters derived from ECG analysis and potassium levels in the blood have been established. We developed a new method to quantify [K+] from T-wave analysis in real-time and tested it on data from dialysis patients, since they undergo large potassium variations in a relatively short time (about four hours). Holter ECG recordings (H12+, Mortara Instrument Inc.) were acquired on 13 patients during dialysis sessions (39 sessions) as well as electrolytes values at the following times: 0, 30, 60, 120, 180, 240 min. from the start of dialysis. ECG data were exported and averaged over 5 minute windows (SuperECG, Mortara Instrument Inc.). The ratio between the T-wave descending slope and the T-wave amplitude (TS/A) was computed on each averaged beat. A significant correlation (r=0.63, p<0.0001) was found between TS/A and [K+]. Based on these results an ECG-based potassium estimator (KECG) was defined as KECG=1.21* TS/A-0.69 and compared with the reference potassium measurements. The agreement was good (absolute error: 0.49±0.16 mM) for most of the sessions (30/39 sessions) except for 9 sessions (absolute error: 1.17±0.36 mM) in which the presence of a systematic error (bias) all along the session did not allow reliable estimates. Bland-Altman analysis showed that the overall systematic (mean) error was almost null (-0.03 mM) whereas the standard deviation (sd) was 0.83 mM. The manual correction of the bias over each dialysis session resulted in excellent results for all patients (mean=0.001 mM, sd=0.45 mM). Session dependent bias seems the only major limitation of the method. Possible causes for this limitation could be session dependent factors such as calcium, initial hydration status, pH, HR, and serum magnesium. We propose a new method for non-invasive potassium concentration measurements in real-time from the ECG. Preliminary results are promising although further investigation is required to understand the reason for session-dependent bias in some patients. Following a comprehensive validation, this method could be effectively applied to monitor patients at risk for hyper- and hypokalemia which are among the main risk factors for cardiac arrhythmias as well as being indicators for worsening heart or kidney failure.

(Abstract Control Number: 60)