Abstract

INTRODUCTION: One of the limitations reported with cardiotocography (CTG) is the modest interobserver agreement observed in tracing interpretation. This study compared agreement, reliability and accuracy of CTG interpretation using the FIGO, ACOG and NICE guidelines. MATERIAL AND METHODS: A total of 151 tracings was evaluated by 27 clinicians from three centers where FIGO, ACOG and NICE guidelines were routinely used. Interobserver agreement was evaluated using the proportions of agreement (PA) and reliability with the kappa (k) statistic. The accuracy of tracings classified as "pathological/category III" was assessed for prediction of newborn acidemia. For all measures, 95% confidence intervals (95%CI) were calculated RESULTS: CTG classifications were more distributed with FIGO (9%, 52%, 39%) and NICE (30%, 33%, 37%) than with ACOG (13%, 81%, 6%). The category with the highest agreement was ACOG category II (PA=0.73 95%CI 0.70-76), and the ones with the lowest agreement were ACOG categories I and III. Reliability was significantly higher with FIGO (k=0.37, 95%CI 0.31-0.43), and NICE (k=0.33, 95%CI 0.28-0.39) than with ACOG (k= 0.15, 95%CI 0.10-0.21), however all represent only slight/fair reliability. FIGO and NICE showed a trend towards higher sensitivities in prediction of newborn acidemia (89% and 97% respectively) than ACOG (32%,), but the latter achieved a significantly higher specificity (95%) CONCLUSIONS: With ACOG guidelines there is high agreement in category II, low reliability, low sensitivity and high specificity in prediction of acidemia. With FIGO and NICE guidelines there is higher reliability, a trend towards higher sensitivity, and lower specificity in prediction of acidemia. This article is protected by copyright. All rights reserved.