Evaluation of interobserver agreement of cardiotocograms

Abstract

Objective: To evaluate interobserver agreement in visual analysis of each cardiotocographic event.

Methods: Three experts independently divided 16 antepartum and 17 intrapartum cardiotocograms into baseline segments, accelerations and decelerations, according to the FIGO guidelines. Baseline segments were further classified as having normal, reduced or increased variability and decelerations as early, late and variable. Uterine activity was divided into tonus and contractions. Agreement was assessed by the proportions of agreement (pa) with 95% confidence intervals.

Results: Reproducibility in assessment of baseline segments with normal variability, accelerations and uterine activity was acceptable (pa = 0.56-0.71) whereas that of other segments was not (pa = 0.14-0.45).

Conclusions: Analysis of most cardiotocographic events is poorly reproducible, even when experts use the FIGO guidelines. This may be explained by some still ambiguous guidelines, by eyeball limitations in evaluation of subtle events, and by the incapacity of busy clinicians to assess complex and multiple cardiotocographic events in a systematic and disciplined fashion.