An efficient method to determine normal values of signal averaged ECG in healthy children

The development of exploration methods provides physicians with many tests which can be used to settle a reliable diagnosis. To evaluate the individual diagnostic value of a new test, it is necessary to assess the normal reference values to ascertain the sensitivity and the specificity of the method. For this purpose confidence intervals (CI) of the parameter are computed from a large healthy population. However, to compute the CIs, the distribution function of the parameter must be determined. Usually, CIs are computed assuming the normality of the distribution. When this assumption is not valid, the results can be strongly biased. To compute reliable CIs, the authors propose a methodology based on the modeling of the density function. This methodology uses the following steps: 1) estimating the density function; 2) deriving a model of the density function; 3) checking the validity of the model; 4) computing CIs using the theoretical density function. Signal averaged electrocardiography (SA ECG) is a new technique which shows promise for detection of children at risk for ventricular tachycardia (VT). However, normal values have not been established in a large paediatric population for the moment. This methodology has been used to compute CIs of SA ECG parameters in 530 healthy children. These computed CIs show that SA ECG can be used as a screening test of children at risk for VT