Abstract

OBJECTIVES: To investigate whether the longitudinal Doppler changes in small-for-gestational age (SGA) fetuses improves the prediction of those at risk of intrauterine fetal death (IUFD). METHODS: A longitudinal study using two cohorts of singleton pregnancies (SGA and appropriate for gestational age [AGA]). The inclusion criteria for the SGA cohort were singleton pregnancies at 20 weeks' gestation or beyond and diagnosed with SGA. The AGA fetuses consisted of singleton pregnancies deemed at high-risk of developing SGA which were followed up longitudinally but remained AGA. The last two measurements prior to birth were included in the analysis. Longitudinal models for Doppler variables were developed via linear-mixed models and predictive accuracies were tested with generalized linear models. A Bayesian framework was employed to compare model accuracies. RESULTS: In total, 1549 AGA and 941 SGA babies were included in the analysis. There were 30 (3.2%) and no IUFD cases in the SGA and AGA group, respectively. Using the last measurements, the best models for the prediction of IUFD were the cerebroplacental ratio (CPR) (accuracy: 75.0%, 95% CI: 72.6-77.2%) and umbilical artery (UA) (accuracy: 71.0%, 95% CI: 68.6-73.4%) models. The posterior probability of the CPR model having a higher accuracy compared to the UA model was 97.2% (magnitude of change (MC): 3.9%, 95% credible intervals (CrI): 0.5 to 7.3%). The predictive accuracies of the UA (accuracy: 71.0% vs. 72.8% using the standard vs. longitudinal model, respectively), middle-cerebral artery (MCA) (accuracy: 64.6% vs. 63.8% using the standard vs. longitudinal model, respectively) and CPR models (accuracy: 75.0% vs. 74.9% using the standard vs. longitudinal model, respectively) were not significantly altered with longitudinal assessment. The posterior probabilities for accuracy improvement with longitudinal assessment were 50.1% (MC: <0.1%, 95% CrI: -3.3 to 3.3%), 35.2% (MC: -0.1%, 95% CrI: -4.5 to 2.8%) and 82.2% (MC: 1.9%, 95% CrI: -1.5 to 5.3%) for CPR, MCA and UA models, respectively. The change in Doppler parameters did not improve the predictive accuracy for IUFD compared to the last scan measurement. CONCLUSION: The change in MCA, UA and CPR with advancing gestation was significantly different between live born AGA and SGA fetuses with less pronounced difference with advancing gestation. Longitudinal assessment of Doppler parameters was not useful in improving the detection of IUFD compared to a single point assessment. This article is protected by copyright. All rights reserved.