[Comparison of variability (STV and LTV) indices by computer-simulated data]

Many investigators have proposed their own formula for quantitative evaluation of STV (short term variability) and LTV (long term variability). There has even been confusion over the definition itself of STV. We compared those indices by using 'beat to beat interval' computer-simulated series and found that there is little difference between the STV indices. According to the definition of STV by the Japan Society of Obstetrics and Gynecology, it is the variance in beat to beat difference, which is determined by the variance in beat to beat intervals and the autocorrelation function of the same intervals. For all the 2,500 combinations of the variability and the autocorrelation, normal random interval series were generated by the computer and five indices (Dalton, Cabal, Yeh, Heilbron and de-Haan) were compared with calculated index values using these interval series. Assigning the variability and the autocorrelation to the X-axis and the Y-axis respectively, the Z-axis was used to indicate the value of the STV. Five curved surfaces thus formed have a similar shape, which shows us that these STV indices behave the same way in accordance with the change in STV. Further, the study of LTV indices with the same simulated data showed that Yeh's LTV index indicates only the variability but has no relation-ship with the autocorrelation and both de-Haan and Heilbron's LTV indices represent the variability and the autocorrelation. However, they increase as the autocorrelation increases unlike the STV indices.