Objective: To define the accuracy of the umbilical artery Doppler flow velocity waveforms, according to different cut-off values, in predicting adverse perinatal outcomes among fetuses at high risk of hypoxic complications.
Subjects: Two hundred and sixty-five pregnant women with diagnosis of small for gestational age fetus and/or pregnancy induced hypertension studied in four Italian ultrasound units.
Methods: Prospective study. Serial Doppler ultrasound measurements of the umbilical artery were performed. Results were not available for clinical management. Cut-off curves, corresponding to different age-specific centiles of the pulsatility index distribution among pregnancies resulting in healthy newborns, regardless of birthweight, were computed by regression methods. Sensitivity, specificity, positive predictive value and negative predictive value of such cut-off curves, and of absent/reverse end-diastolic flow, in predicting different adverse outcomes were estimated. The adverse outcomes were: perinatal or neonatal death (OUTCOME 1). Death or Apgar<7 at 5' or need for admission to intensive care unit or other hypoxic related abnormalities (OUTCOME 2). Either OUTCOME 2 or birthweight<l0th centile (OUTCOME 3). Both OUTCOME 2 and birthweight<10th centile (OUTCOME 4).
Results: The best accuracy was in predicting OUTCOMES 1 and 4. Positive predictive value increased strongly with higher cut-off curves while negative predictive value only decreased slightly. For absent/reverse end-diastolic flow, negative predictive value and positive predictive value were respectively 94% and 39% for death, and 81% and 72% respectively for OUTCOME 2. For the 95th centile curve the corresponding figures were 96%, 33% and 84%, 67%. The 60th centile curve had a 85% and 74% sensitivity value for death and OUTCOME 2 respectively, but the corresponding positive predictive values were 18% and 40% only.
Conclusions: The findings of an absent end diastolic flow or of pulsatility index values above the 95th centile curve strongly suggest it is time to deliver the fetus. The 60th centile curve is the most suitable to recognize fetuses at risk for abnormal outcome, but early delivery should be avoided because of its low positive predictive value.