Therapeutic hypothermia (TH) is the recommended treatment for neonates with signs and symptoms of neonatal encephalopathy (NE) occurring after a significant perinatal event, presumed to be caused by hypoxic ischemia. The clinical presentation of NE includes depressed level of consciousness, decreased spontaneous movement, low muscle tone, poor sucking reflex, and seizures. This commonly occurs in the context of a perinatal sentinel event (e.g., cord prolapse, placental abruption causing hypoxic-ischemic encephalopathy) and can be associated with end-organ involvement and long-term neurodevelopmental sequela. NE is traditionally categorized as mild, moderate, and severe using modified Sarnat scoring and other assessment tools, with TH considered standard treatment for moderate and severe NE (1, 2). Published data demonstrate the benefits of initiating TH in neonates prior to 6 h of life and continuing for 72 h (2–4).

In many centers around the country, agitation, discomfort, and pain during TH are routinely treated with opioids such as morphine and fentanyl (continuous infusion and/or intermittent bolus administration) (1). Morphine has anti-nociceptive properties and reduces the release of glutamate and the sensitivity of glutamate receptors. Glutamate is an excitatory neurotransmitter and is responsible for triggering apoptosis during secondary energy failure characteristic of neonatal NE. This is contrary to what would be expected since morphine inhibits neighboring gamma-aminobutyric acid (GABA) inhibitory neurons, which leads to excitation of ventral tegmental area dopamine (VTA-DA) neurons (5). Secondary energy failure typically occurs approximately 6–15 h after the injury and is characterized by cytotoxic edema, excitotoxicity, and eventually complete failure of mitochondria. Glutamate release activates glutamate and N-methyl-D-aspartate (NMDA) receptors which results in an influx of sodium, chloride and calcium which results in apoptosis and necrosis (6). Inhibition of glutamate release and reduction of glutamate receptor sensitivity is thought to result in neuroprotection (7). Angeles et al. (7) found that neonates who were given morphine during TH had less evidence of hypoxic-ischemic injury on magnetic resonance imaging compared to those who did not receive opioids.

Repeated neonatal pain may be associated with adverse neurodevelopmental outcomes. Williams and Lascelles (8) suggested that long-term consequences of untreated neonatal pain are associated with adverse sensorimotor/cognitive development and subsequent future responses to pain. Walker (9) also reviewed the long-term effects of neonatal pain on neurodevelopment and suggested that there are distinct differences when morphine is administered as a sedative in the absence of pain compared to when used as an analgesic for true pain. Gunderson et al. (10) reported a prospectively collected cohort of 282 neonates with NE treated with TH and exposed to opioids and found no adverse effects on neurodevelopmental outcomes at 18–24 months of age.

Natarajan et al. (11) demonstrated that sedation during TH does not impact neurologic outcomes but suggested that further studies should be studied in neonates exposed to TH. Liow et al. (12) performed a secondary analysis of a large multinational prospective observational study examining the association between morphine infusion during TH on brain injury and neurodevelopmental outcomes. They concluded that morphine does not offer any neuroprotective benefits and may be associated with a prolonged hospital stay (12). Additionally, Frymoyer et al. (13) conducted a pharmacokinetic study in 20 neonates with NE receiving morphine during TH and found clearance to be significantly slower compared to term controls. Similarly, another study found that serum morphine concentrations during the 24–72 h after birth were higher in neonates who underwent TH compared to their normothermic controls despite similar morphine infusion rates and doses (14). The potential adverse effects of opioids (e.g., hypotension, respiratory depression) are especially concerning given that clearance is decreased during TH (13, 14).

The goal of the present study was to determine whether opioid use during TH for moderate and severe NE affects short-term clinical outcomes including length of hospitalization, duration and mode of respiratory support, and need for vasopressor medications for blood pressure support. Our hypothesis was that morphine or fentanyl administration during TH for NE does impact these short-term clinical outcomes. Current research is lacking data regarding these specific short-term clinical outcomes. The goal of our study was to provide further evidence for the potential risks of routine opioid use during TH.

The benefits and risks of opioid use during TH remain controversial. While some pre-clinical models suggest that morphine may enhance the neuroprotective effects of TH by reducing oxidative stress, other animal studies have suggested that morphine induces neuronal apoptosis which could be harmful (4, 18, 19). The aim of this study was to examine short-term outcomes in neonates requiring TH and exposed to no or minimal doses of opioids compared to those receiving higher MME at three academic medical centers. The results suggest that higher cumulative doses of MME during TH for NE (often via continuous infusions) are associated with a greater need and longer duration of mechanical ventilatory support, vasopressors to support blood pressure, and a longer hospital stay. This is likely due to the direct effects of opioids on respiratory control centers of the brain and on vasomotor tone of blood vessels; effects that may be accentuated when combined with TH (which may delay opioid clearance). The TH protocols at the three centers are similar except for the approach to sedation. Intermittent morphine doses (0.05–0.1 mg/kg) are used only as needed at Center 1, routine administration of morphine by continuous infusion occurs at Center 2, and routine administration of fentanyl by continuous infusion occurs at Center 3. Our analysis demonstrated that the cumulative MME dose was significantly different among the three sites which was somewhat unexpected since the general perception is that continuous infusion decreases the cumulative MME amount required for pain control and comfort.

Our analysis also demonstrated that the cumulative MME dose was significantly different among the three sites with the centers using continuous infusions having the highest cumulative MME dose exposure. This was somewhat unexpected since the general perception is that continuous infusion decreases the cumulative MME amount required for control of pain and discomfort. The substantial difference in the cumulative MME dose at one center compared to the other two centers is related to the preferential use of fentanyl, with significantly higher analgesic potency (100 times) compared to morphine.

Although patient demographics at the three centers were similar, some differences were identified. Most neonates receiving TH at all three centers were outborn which is consistent with current evidence (17). There was a higher number of mothers with substance use disorder in the Center 2 cohort and anxiety and depression at Center 1 (20, 21). The data also showed minor differences in practice patterns among sites with respect to modes of ventilation, fluid resuscitation, laboratory data, and initiation of feeds (Table 3A). Neonates with EEG confirmed seizures and subsequent exposure to antiepileptic medications in addition to opioids required a longer hospitalization, a greater need for vasopressors, and greater need and duration of both invasive and non-invasive ventilation. Duration of vasopressor use though was not found to be statistically significant.

This is a pragmatic study, which enabled the comparison of effect of opioids on short-term outcomes at three centers with similar TH management, despite some differences in clinical practice (e.g., opioid administration). The difference in pressor use could be partially related to variations in the approach to blood pressure measurement at the three institutions. Variations in the use of intra-arterial lines for blood pressure monitoring may influence vasopressor use and contribute to the duration of central line use (22). Higher use of vasopressors may also result in slower advancement of feeds and may partially explain why the delay in initiation trophic feeds. Likewise, differences in ventilation may also be due to the individual center practice or provider preferences regarding the timing of extubation and use of non-invasive ventilation.

The main limitations of our study include the relatively small sample size, retrospective data collection, and clinical practice variations between the three institutions despite very similar TH protocols. In addition, we did not analyze standardized pain scores (due to their subjective nature) among sites, which could potentially have influenced opioid use. Since the concomitant use of anti-seizure medication may be a confounding variable influencing our primary and secondary outcomes, EEG confirmed seizures was used to adjust analyses. The criteria for EEG confirmed seizures was chosen over exposure to antiepileptic medication criteria because many neonates treated empirically with antiepileptic medications did not have EEG proven seizure activity and the medications were stopped (Table 3A, B). At all three centers, only a small percentage of neonates that received anti-epileptics also had EEG confirmed seizures. Another limitation was the lack of an objective tool to more accurately compare the severity of NE and any pain/discomfort associated with TH. Information used as a surrogate biomarker included APGAR scores, blood gases, the presence of multi-organ involvement, and Sarnat scoring. Data collected in this study showed that neonates with a diagnosis of severe NE did not consistently have evidence of multi-organ involvement or the most severe MRI findings.

Although the use of MRI as a biomarker of underlying brain injury is still being debated, MRI findings do correlate with future neurodevelopmental outcomes associated with NE (23). The discrepancy between unfavorable short-term outcomes and MRI results as a long-term prognostic tool emphasizes the need for long-term outcome data to definitively determine benefit or harm of opioid use in the management of neonatal NE during TH. For this study, each center's MRIs were interpreted by their own pediatric neuroradiologist. We were not able to have a single blinded radiologist interpret the results for all the MRIs and this prevented us from formulating statistical conclusions regarding the MRI findings.

The cumulative impact of opioids on long-term neurodevelopmental outcomes after TH could not be analyzed given the limited amounts of follow-up data available from the three centers. Long-term outcomes are critical to establish safety of these medication practices. The optimal therapeutic approach to neonatal pain management using opioids is still unresolved and a standardized approach is lacking. Newer sedatives such as dexmedetomidine are emerging as alternative options and their use in neonates is becoming much more widespread without sufficient compelling data on safety and dosing, especially during TH when the metabolism of many drugs change (24).

This study is one of the few involving real-world data, an important approach considering the paucity of data on the safety and efficacy of opioid use in neonates (especially those with underlying brain injury). While the current results raise concerns regarding the use of opioids in neonates undergoing TH, the findings indicate the complexity of a pragmatic approach to study opioid therapy in TH. A larger multicenter trial with standardized approach to opioid therapy and a detailed pharmacokinetic analyses and comprehensive neurodevelopmental follow-up will ultimately be needed to further investigate the short- and long-term safety and efficacy of opioids during TH.