Neonatal treatment decisions are based on prognosis, which depends on the severity of HIE. Therapeutic hypothermia is currently applied to neonates with moderate-to-severe HIE but is not considered effective in those with severe HIE.[2] As such, it is crucial to develop tools to predict the probability of severe adverse outcomes. Predictive models must be developed using data obtainable within 6 h of birth, because the efficacy of initiated therapeutic hypothermia beyond this period is unclear. Accurate predictive models or scoring systems that are established after 6 h are not clinically useful as evidence supporting the efficacy of hypothermia treatment beyond 6 h after birth remains unreliable.[10] In one study of 96 asphyxiated infants, Perlman and Risser identified that an Apgar score < 5 at five min, intubation in the delivery room, and an umbilical artery blood pH < 7.0 were both predictors of seizures within 24 h. However, these factors only accounted for approximately half of all infants who developed seizures within 24 h due to asphyxia.[11] Carter et al. previously demonstrated that persistent fetal bradycardia, umbilical artery blood pH < 7.0, umbilical artery blood base deficit ≥ 10, and an Apgar score < 3 at five min were all significant predictors of multiple organ failure in neonates with asphyxia. However, these factors were not indicative of neurological outcomes.[12]
Imaging modalities such as MRI, magnetic resonance spectroscopy, and diffusion-weighted imaging are recommended between days two and eight after birth; however, their prognostic utility on day one remains unestablished.[13–16] This study showed that flat trace or continuous low voltage on an aEEG, a Thompson score ≥ 12, and chest compressions during resuscitation were associated with an increased risk of death, cerebral palsy, or a DQ < 70 at 18 months in neonates with HIE. Shankaran et al. also found that neurological examinations at 12 months could predict severe disability at five-years-old.[17] As such, our findings are likely to also reflect outcomes beyond 18 months of age.
An aEEG is useful in predicting outcomes in neonates with HIE. In one study, approximately 58% of infants with good outcomes exhibited a sleep-wake cycle due to rewarming, with at least 5% showing an early cycle.[18] Failure to establish a cycle by day seven has previously been found to indicate a higher risk of death or severe brain injury.[19] Multiple studies conducted prior to the introduction of therapeutic hypothermia have shown that aEEG is an excellent early predictor of neurological outcomes in neonates with HIE, with a sensitivity of 93% and a specificity of 91% when conducted within 24 h of birth.[5] The positive predictive value of hypothermia for death or disability in infants with HIE who did not undergo hypothermia exceeded 80%.[20] However, therapeutic hypothermia alters the prognostic value of aEEG. The Total Body Hypothermia for Neonatal Encephalopathy Trial (TOBY) found that the predictive value of aEEG within 6 h of birth was lower, with positive predictive values of 55% and 63% in the cooled and non-cooled groups.[21] The aEEG background took longer to normalize in infants who received hypothermia, with approximately 50% showing recovery within 24–48 h. Persistent abnormal aEEG patterns (flat trace or continuous low voltage) after 48 h predicted poor outcomes.[22, 23] Meta-analysis further revealed that a flat trace, continuous low voltage, or BS on an aEEG within 6 h, before hypothermia treatment, had a sensitivity of 96% (95% CI, 91–98) for adverse outcomes; however, the specificity was only 39% (95% CI, 32–46). The positive prognostic value of aEEG at 6 h was poor, and good outcomes could occur despite an abnormal aEEG. Conversely, a normal aEEG at 6 h had a good negative predictive value, although adverse outcomes were not excluded.[24] Combining early neurological examinations and aEEG abnormalities further improved predictive accuracy.[35]
Multicenter prospective studies in HIE neonates undergoing hypothermia revealed that moderate to severe encephalopathy graded by Sarnat, within 6 h, was insufficient to accurately predict poor outcomes (sensitivity, 42%; specificity, 94%).[26] Notably, the progression of encephalopathy during the first three days of hypothermia therapy predicted better outcomes than the initial HIE stages. Severe encephalopathy persisting according to the Sarnat classification after 72 h of hypothermia was associated with death or poor neurodevelopmental outcomes at 18–22 months.[27, 28] The Thompson score has also previously been shown to be a good tool for predicting outcomes at 4–5 years in neonates with HIE, receiving hypothermia. Mendler et al. similarly found that a maximum Thompson score ≤ 10 was associated with normal cognitive function.[29] Other studies have further confirmed that the Thompson score is effective at predicting short-term prognosis, showing a correlation with death before discharge, seizure onset, and a severely abnormal aEEG.[30, 31] Clinical scoring systems, such as the Sarnat and Thompson scores, are also valuable in predicting outcomes in neonates with hypoxic-ischemic encephalopathy (HIE), treated with hypothermia. However, their original evaluations were based on assessments performed after the first week of life, making these scores inadequate for use within the first 6 h after birth.[32, 33] Nonetheless, the previous study reported that a Thompson score ≥ 12 within two h of birth was associated with increased odds for death before discharge and severe seizures, compared to a Thompson score between 7–11 (OR: 3.9, 95% CI: 1.3–11.2, p = 0.01; OR: 8.4, 95% CI: 2.5–27.8, p < 0.001).[30] Regarding resuscitation at birth, neonates requiring chest compressions for > 1 min had a higher likelihood of death or severe neurological impairment (odds ratio [OR], 3.24; 95% CI: 1.44–7.32).[34] Applying multiple predictive factors within the first 6 h can potentially predict outcomes before the initiation of therapeutic hypothermia. In the present study, the presence of flat trace or continuous low voltage on an aEEG, a Thompson score ≥ 12, and the need for chest compressions during resuscitation were all found to be associated with an increased risk of death, CP, or a DQ < 70 in neonates with moderate or severe HIE. Among the different arterial blood gas parameters, prior studies have reported the umbilical artery blood pH as the most significant, with the severity of acidosis correlated with the extent of damage.[35] However, other studies have noted that umbilical artery blood gas results alone are limited and should not be used as the sole indicator for diagnosing HIE. Williams et al. also previously reported that there was no correlation between pH and neonatal mortality in infants with HIE.[36]
This study has few limitations. First, as a single-center study, the generalizability of the findings to other settings or populations may be limited. However, as our hospital treats all patients with neonatal asphyxia in Kagoshima Prefecture, this study encompassed the entire population of this region, thereby eliminating selection bias. Second, information bias is a concern because of the reliance on historical medical records, which may contain inaccurate or incomplete data. However, although this study utilized historical data, temporal bias was not a concern, as the criteria for initiating therapeutic hypothermia and the methods of administration remained consistent. Finally, confounding factors were controlled for by the inclusion of multiple covariates in the regression models. Overall, while this study provides valuable insights, further research with larger multicenter cohorts is required to validate the findings obtained.
With regard to prognostic prediction in neonates with HIE, the use of aEEGs (flat trace or continuous low voltage), a Thompson score ≥ 12, and the presence of chest compressions during resuscitation may help to predict outcomes before the initiation of therapeutic hypothermia.