Performance evaluation of computerized antepartum fetal heart rate monitoring: Dawes-Redman algorithm at term.

OBJECTIVES

To assess the effectiveness of the Dawes-Redman algorithm in identifying fetal wellbeing at term by analyzing 30 years of retrospective clinical data, comparing normal and adverse pregnancy outcomes, evaluating key metrics and testing its performance when used 0-48 h before delivery.

METHODS

Antepartum fetal heart rate (FHR) traces from term singleton pregnancies at 37 + 0 to 41 + 6 weeks' gestation obtained between 1991 and 2024 were extracted from the Oxford University Hospitals database. Traces with > 30% of their signal information missing or with incomplete Dawes-Redman analyses were excluded. Only traces performed within 48 h prior to delivery were considered. A cohort of pregnancies with subsequent normal pregnancy outcome (NPO) was established using rigorous inclusion and exclusion criteria. Another cohort of pregnancies with adverse pregnancy outcome (APO) was developed if the neonate experienced at least one of seven APOs after delivery. Propensity score matching (PSM) facilitated a balanced comparison between NPO and APO cohorts using six factors: gestational age at FHR monitoring, fetal sex, maternal body mass index at presentation, maternal age at delivery, parity and time interval between FHR trace and delivery. FHR traces were categorized as either 'criteria met' (indicating fetal wellbeing) or 'criteria not met' (indicating a need for further evaluation) according to the Dawes-Redman algorithm, which informed the evaluation of predictive performance metrics. Performance was assessed using accuracy, sensitivity, specificity, positive predictive value, and negative predictive value (NPV) adjusted for various population risk prevalences of APO.

RESULTS

A balanced dataset of 3316 antepartum FHR traces was developed with PSM (standardized mean difference < 0.10). The Dawes-Redman algorithm showed a high specificity of 90.7% (95% CI, 89.2-92.0%) for ruling out APO. Sensitivity was 18.2% (95% CI, 16.3-20.0%). The NPV varied with the population prevalence of APO and was high in very-low-risk settings (NPV, 99.1% (95% CI, 98.9-99.3%) at 1% APO prevalence) and decreased with increasing risk of APO (NPV, 72.1% (95% CI, 67.7-76.1%) at 30% APO prevalence). Temporal proximity of FHR assessment to delivery indicated robust specificity, which was similar for assessments performed at 0-24 h and 24-48 h prior to delivery (specificity at 0-24 h, 90.8% (95% CI, 88.8-92.7%); specificity at 24-48 h, 90.3% (95% CI, 88.2-92.3%); P = 0.898). Across the different adverse outcomes comprising the APO cohort, the performance of the Dawes-Redman algorithm remained consistent, with high specificity (ranging from 87.7% to 94.7%) and NPVs (ranging from 95.4% to 96.0%), confirming its utility in identifying fetal wellbeing.

CONCLUSION

These findings indicate that the Dawes-Redman algorithm is effective for its intended purpose: identifying a state of fetal wellbeing. This is evidenced by its high specificity. However, its low sensitivity suggests limitations in its ability to identify fetuses at risk of APO. The predictive accuracy of the algorithm is affected significantly by the prevalence of healthy pregnancies within the population. Clinical interpretation of FHR traces that do not satisfy the 10 Dawes-Redman criteria warrant further expert clinical evaluation. While the algorithm proves reliable for its primary objective, the development of an algorithm optimized for high-risk pregnancy scenarios remains an area of interest for future study. © 2025 The Author(s). Ultrasound in Obstetrics & Gynecology published by John Wiley & Sons Ltd on behalf of International Society of Ultrasound in Obstetrics and Gynecology.