Objective: To assess the validity of a computerized methodology for ca
rdiotocogram analysis based on a recently described reproducible visua
l estimation of the baseline. Methods: Forty-two antepartum and 43 int
rapartum cardiotocograms (CTGs) acquired by a personal computer were s
elected. Antepartum tracings were performed in the 48 h that preceded
an elective cesarean section, and intrapartum tracings were performed
until delivery. FHR baselines were estimated by an expert, according t
o an objective and reproducible methodology. Using these baselines, au
tomated detection of accelerations and decelerations and estimation of
variability was performed by the personal computer. A quantitative ad
aptation of the FIGO guidelines for fetal monitoring was used to class
ify tracings. Perinatal outcome was classified according to the Apgar
score and umbilical arterial pH. Validity was then assessed by the pro
portions of agreement (PA), kappa statistic (kappa), sensitivity and s
pecificity, with 95% confidence intervals (95% CI). Cases showing a di
sagreement between CTG and perinatal classification were reviewed and
an adjustment in baseline definition was tested. Results: The initial
overall PA and kappa between CTG and perinatal classification were, re
spectively, 0.79 (95% CI: 0.69-0.87) and 0.62 (95% CI: 0.41-0.83). The
overall PA and kappa, after baseline adjustment were, respectively, 0
.89 (95% CI: 0.81-0.95)and 0.78 (95% CI: 0.58-0.98). Sensitivities and
specificities ranged between 79% (95% CH: 60-92%) and 100%(95% CI: 95
-100%). Conclusions: Good clinical prediction may be possible with an
objective methodology for cardiotocogram analysis based on a recently
described reproducible baseline estimation. (C) 1998 International Fed
eration of Gynecology and Obstetrics.