[A new approach to quantitative electronic foetal heart-rate analysis]

Abstract

Background: Recently it was found that a significant correlation exists between the variables of the foetal acid-base balance (ABB) and the parameters of the foetal heart rate (FHR). This dependency can be used for diagnostic purposes. Until now FHR could be evaluated off- and online by computer using scoring procedures, i. e., discontinuous methods, which need considerable computational efforts since FHR must reliably be separated into baseline (BL), decelerations and accelerations. Therefore, the question must arise whether a continuous and less cumbersome evaluation of individual FHR-parameters might lead to even better results?

Methods: The last 30 min of 465 direct recordings of foetuses all delivered by the vaginal route were stored in a computer and analysed offline using our own MATLAB programmes. Five variables were computed for every single minute: The microfluctuation, the oscillation amplitude, the mean frequency, the sum of the absolute differences of time periods and the sum of the absolute differences of frequencies. In this paper the last two variables will be abbreviated 'total sums'. All minutes of a FHR tracing were treated equally i. e. there was no separation anymore between BL and decelerations (accelerations). In order to analyse the physiological intercorrelation of the five variables mentioned above a 'pH window' was chosen between 7.275 and 7.325 [7.275< or =pH, (umbilicalartery (UA)< or =7.325]. Thereafter, the influence of hypoxia and acidosis on the five variables was evaluated. According to the results obtained two of the five variables, the 'total sums', were not further analysed. Using the remaining variables 'frequency', 'oscillation amplitude' and 'microfluctuation' a new index, the WAS index, was created. This index offers the opportunity to evaluate the FHR continuously using 'coding lines' for each of the three parameters. The WAS index was designed according to the FHR characteristics of the awake (Wach), acidotic (Azidotisch) (pH, UA=ca. 7,000) and sleeping (Schlafend) foetus. Using the last 30 WAS indices covering the last 30 min of a single FHR tracing, a WAS score and the corresponding prognostic pH value was computed. These prognostic pH values were compared with the measured pH values of UA blood using again correlation analysis.

Results: Without any hypoxia and acidosis (pH, UA window) the five FHR variables are closely correlated with each other. Accepting minutes with decelerations only, the correlation between the 'total sums' and the remaining three parameters vanishes. Accepting hypoxia and acidosis (pH, UA > or =6.960) oscillation amplitude now offers the closest correlation with actual pH, UA (r=-0.440 p<<10 (-4)) followed by microfluctuation (r=0.224, p<0.001) and frequency (r=0.056, P: n. s.). This pattern is obviously due to the retention of all decelerations in the FHR tracings. The 'total sums' show less close correlations with pH, UA and BE(oxy) when compared with microfluctuation and oscillation amplitude (SumPER vs. pH, UA r=-0.125, p<0.001 and SumFRQ vs. pH, UA r=-0.154, p<0.001). All five variables under investigation show a better correlation with pH, UA when compared with BE(oxy) UA. The WAS score computed (mean) for the last 30 min of delivery leads to close correlations (p<<10(-4)) with all variables of the foetal ABB: pH, UA r=0.608; BE(oxy) r=0.535, pCO(2) r=-0.469 and sO(2) r=0.259. The median WAS score amounts to 0.176, the mean to 0.174+/-0.023; it is normally distributed in this sample.

Conclusions: FHR characteristics in different foetal behavioural states offer the opportunity to design a new index, the WAS index, which shows close (p<<10(-4)) correlations with all variables of the foetal ABB. Thus, the pH value in cord blood can be predicted within clinically reasonable limits. The qualitative CTG analysis remains untouched.