Confusion matrix, sensitivity, specificity, positive and negative predictive value (PPV and NPV) of EEG at three different time points and the relationship with neurodevelopmental outcome in all available infants
| EEG | Outcome | P value* | AUC (95% CI) | Sensitivity (95% CI) | Specificity (95% CI) | PPV (95% CI) | NPV (95% CI) | |||
| Normal | Abnormal | |||||||||
| EEG-1 | All infants (n=57) | Normal (n=18) | 17 | 1 | 0.011 | 0.68 (0.55 to 0.80) | 94 (71 to 100) | 43 (27 to 59) | 41 (26 to 58) | 94 (73 to 100) |
| Abnormal (n=39) | 23 | 16 | ||||||||
| EEG-32 | All infants (n=53) | Normal (n=27) | 27 | 0 | <0.001 | 0.84 (0.70 to 0.92) | 100 (75 to 100) | 68 (51 to 81) | 50 (30 to 70) | 100 (87 to 100) |
| Abnormal (n=26) | 13 | 13 | ||||||||
| EEG-35 | All infants (n=45) | Normal (n=29) | 29 | 0 | <0.001 | 0.91 (0.83 to 1.00) | 100 (69 to 100) | 83 (66 to 93) | 63 (35 to 85) | 100 (88 to 100) |
| Abnormal (n=16) | 6 | 10 | ||||||||
Sensitivity calculates the ability of the EEG to correctly classify infants with an abnormal outcome (ie, true positives among total true positives and false negatives), while specificity calculates the ability of the EEG to correctly classify the infants with a normal outcome (ie, true negatives among total true negatives and false positives). PPV calculates the EEG’s ability to predict infants with an abnormal outcome (ie, true positives among total true positives and false positives), while NPV calculates the EEG’s ability to predict infants with a normal outcome (ie, true negatives among total true negatives and false negatives).
*Fisher’s exact test.
AUC, area under the receiver operating characteristic curve; EEG, electroencephalography.