RESUMEN
AIM: Infants born preterm are at risk of cerebral palsy (CP) and motor or cognitive developmental delay. For clinicians, it is essential to know the relative predictive accuracy of the most commonly used neuroimaging and neurophysiological tests for the early prediction of adverse neurodevelopmental outcome. The aim of this study was to compare the accuracy of these tests in survivors of a population of infants born very preterm. METHOD: A retrospective cohort study was performed in 163 children born before 32 weeks gestational age. We compared the accuracy in predicting adverse neurodevelopmental outcome at the age of 2 years 6 months of early and late cranial ultrasound (CUS), magnetic resonance imaging, somatosensory evoked potentials after stimulation of the posterior tibial nerve, and electroencephalography by calculating positive and negative likelihood ratios. RESULTS: An abnormal early CUS is the best predictor of the presence of CP (positive likelihood ratio 6.09), motor developmental delay (positive likelihood ratio 3.11), and cognitive developmental delay (positive likelihood ratio 5.66). Overall, negative likelihood ratios were poor, ranging between 0.49 and 0.98, meaning that a normal test result had only minimal influence on the probability of adverse neurological outcome. INTERPRETATION: None of the diagnostic tests had a good performance in predicting future neurodevelopmental problems in infants born preterm. A normal test result provided very little clinically useful information. WHAT THIS PAPER ADDS: An abnormal early cranial ultrasound (positive test result) is the best predictor of adverse neurodevelopmental outcome. All negative results have poor predictive value of future neurodevelopmental problems.