Abstract
For evaluation of reproducibility of ventricular premature beat (VPB) by ambulatory electrocardiogram recording (AEG), we analysed individual property of VPB frequency using both the distribution and statistical standard deviation to explain long-term and short-term variance on AEG. In 22 outpatients with VPBs we compared the difference of VPB frequency on two succesive days' AEG (experiment I) with that on two days' AEG separated by 2 weeks (experiment II). The frequency distribution was classified into two types by plotting pattern of cumulative number of VPBs on the vertical axis with the number of VPBs per minute on the horizontal axis. An ‘unimodal’ type was defined that the distribution was almost normal and ‘multimodal’ type was the other. The ratio of changeless type was significantly higher in experiment I (95.4%) than that in experiment II (59.1%) (p<0.05). Each correlation coefficient in experiment I was significantly higher than that in experiment II (p<0.05) as to various standard deviations defined as follows: a average of 24 standard deviations per hour (SDa), standard deviation of mean VPB frequency per minute as a short term variance (SD) and of mean VPB frequency per hour as a long term variance (SDb)·√SDa2+SDb2/SD as an index of variability of VPB frequency was significantly small in experiment I than that in experiment II (p<0.05). As reproducibility of VPB frequency on AEG was better in experiment I, an estimation formula of individual ranges of the number of VPB on the following day might be possible. The difference between numbers of VPBs in experiment I was below some value obtained by √SD2+SDb2. We calculated statstically new reduction rate of upper confidence limit with each patient by the approximate formula. We concluded that we may get more accurate individual VPB estimation on the following day but that the estimation after more days is more difficult.
