A comparison of the incidence and reproducibility of ventricular late potentials (LP) in the time and frequency domain in patients with cardiovascular (CV) disease

DE Balderson, DWG Harron, OJ Escalona, RH Mitchell

Research output: Contribution to journalArticlepeer-review


IntroductionHigh resolution signal averaging (HR-SA) of the surface ECG has demonstrated LP which are regarded as prognostic indicators for identifying those subjects at risk from a major or fatal arrhythmic event (Vatterott et al., 1988). However, there is controversy with regard to the methods of analysis for detection of LP. The present study compared a number of methods in the time and frequency domain for the detection of LP. MethodsHR-SA ECG were recorded (Predictor-Corazonix) from 82 subjects using the orthogonal XYZ lead system; averages of 600 beats or until noise was <0.3,uV were taken (those > 1,u Vwere ignored). Two systems for evaluation of LP were investigated. The first system used published parameters for the 25- and 40-250 Hz bandwidths. Five analyses were actually performed using variations of the three parameters: the QRS duration (QRSd), the Root MeanSquare of the terminal 40 ms of the QRS (RMS40) and the Low Amplitude Signal duration (LASd) (<40 ,uV). For the first 2 analyses A and B (25-250 Hz) LP parametersfor A were QRSd > 120 ms, LASd > 35 ms and RMS40 <25 ,uV, for B LASd > 40 ms. Methods C, D and E (40-250 Hz); LP parameters for C were QRSd > 110 ms, LASd > 35 ms, RMS40 <20 ,uV, D LASd > 40 ms and E QRSd > 120 ms and LASd > 40 ms. At least two of the parameters had to be abnormal for LP definition. Fourier transform (FT) (method F) was performed using a Blackman-Harris window on a 120 ms length starting at the 40 ,uV level (25-250 Hz) and extending out towards the T wave. The area for 60-120/0-120 Hz was calculated for each lead then meaned (%) (Pierce et al., 1989): a value of > 3% was used for LP detection. The second system of analysis involved setting parameterswhich were outside the normal distribution (methods G, H and I); G (25-250 Hz) QRSd > 114 ms,RMS40 <18 ,uV, LASd > 40 ms, H (40-250 Hz) QRSd > 113 ms, RMS40 <11.5 ,uV, LASd > 40 ms. For FT (I), a value of > 3.6% was used. Five groups were studied; Group 1 (G1) Healthy (n = 13), G2 Hypertensive (HP) (n = 10), G3 HP + CV disease (n = 23), G4 = G2 + G3 (n = 33), G5 CV (n = 24) and G6 'Others' (n = 12).ResultsThe results indicate the incidence of LP in each group using each method of analysis. A Pearson correlation compared the methods and the subjects with LP for the first system of analysis (methods A to F). This demonstrated a correlation (P <0.001) between methods AB,CD,CE and DE in the time domain but none between time and frequency. The results of thesecond system of analysis showed no correlation between the groups. The number of subjects detected in the first system as having late potentials by every method (A-F) was four compared with two for methods G-I. ConclusionIn conclusion the present study confirms a correlation between methods A to E in the time domain for evaluation of LP. However, further studies need to be undertaken with regard to the frequency domain, for example, setting area calculations at different bandwidths.
Original languageEnglish
Pages (from-to)244P-244P
JournalBritish Journal of Clinical Pharmacology
Issue number2
Publication statusPublished - 15 Jan 1991


  • cardiovascular disease
  • conference paper
  • heart arrhythmia
  • heart muscle fiber membrane potential
  • heart ventricle function
  • signal averaged ECG
  • late potentials.

Fingerprint Dive into the research topics of 'A comparison of the incidence and reproducibility of ventricular late potentials (LP) in the time and frequency domain in patients with cardiovascular (CV) disease'. Together they form a unique fingerprint.

Cite this