Authors: Sarah Spenard, Carlos Ivan Salazar Cerda, Mehmet N. Cizmeci
Categories: Review, Low- and middle-income countries, neonate, seizure
Source: Turkish Archives of Pediatrics
Neonatal seizures are a common cause of neonatal intensive care unit (NICU) admission and a significant source of morbidity and mortality worldwide. Over the recent decades, there have been significant improvements in perinatal and neonatal medicine and electroencephalographic monitoring that have enhanced the diagnosis and treatment of neonatal seizures in high-income countries. However, the management of neonatal seizures remains a major challenge in low- to middle-income countries, where the availabilityof resources is limited. The purpose of this article is to present a comprehensive review of the current evidence on the etiology, pathophysiology, diagnosis, and treatment of neonatal seizures and to offer practical management recommendations that could be implemented in resource-limited settings.
Keywords: Low- and middle-income countries, neonate, seizure
Neonatal seizures (NS) are defined as seizures presenting in the first 4 weeks after birth in term neonates or within 44 weeks of corrected gestational age in preterm infants.^1^ A seizure is the manifestation of excessive synchronous electrical discharge within the neurons of the central nervous system.^2-4^ This imbalance is thought to be secondary to an excessive excitation over an insufficient inhibition and impaired influx–efflux balance of major ions (i.e., sodium, potassium, and chloride) in the developing brain. Many pathophysiological mechanisms may explain the occurrence of seizures in newborns, including a failure in the adenosine triphosphate-dependent sodium–potassium pump to maintain stable neuronal membrane potential and a membrane alteration with increased sodium permeability. Specific neonatal conditions, such as hypoxic–ischemic injury and hypoglycemia, can also decrease cellular energy production and lead to the release of excessive extracellular glutamate, an excitatory neurotransmitter.^4^
Neonatal seizures can be provoked or unprovoked.^5^ A provoked seizure results from an acute illness or brain insult caused by a suspected or documented etiology, such as hypoxic–ischemic encephalopathy or central nervous system infection. On the other hand, an unprovoked seizure is not associated with a specific etiology and can meet the diagnosis of epilepsy in the presence of specific criteria. By definition, epilepsy is diagnosed if any of these conditions are (1) at least 2 unprovoked seizures separated by >24 hours, (2) one unprovoked seizure and a probability of additional seizures that is similar to the recurrence risk after 2 unprovoked seizures (at least 60%), and (3) a diagnosis of an epilepsy syndrome.^5,6^
Status epilepticus is defined as any continuous clinical seizure lasting more than 5 minutes or 2 or more discrete seizures without recovery of consciousness between seizures.^7^ Traditionally, 30 minutes was accepted as the cutoff, but it was later understood that seizures that last for more than 5 minutes have a high potential to turn into status and are operationally defined as status epilepticus.^8^ This definition is difficult to apply to neonates because NS require electroencephalographic (EEG) confirmation, and it is difficult to evaluate for interictal return to baseline mental status in infants. Therefore, the term “status epilepticus” should be used with caution in neonates.^9^
The immature neonatal brain exists in a state of excitation/inhibition imbalance; its increased excitation facilitates a number of activity-dependent developmental processes, such as neurogenesis and neural circuit development.^10^ This increased excitability makes the neonatal brain more susceptible to seizures. Hence, seizures happen more commonly in the neonatal period than at any other time over the human lifespan. The incidence rate of seizures in term neonates is estimated to be 1-5 per 1000 live births.^6,9,11,12^ Preterm infants and small-for-gestational-age infants are at increased risk of seizures, with an incidence of 11.1 per 1000 and 13.5 per 1000, respectively.^11,13^ It is important to note that these incidence rates are based on studies done in high-income countries (HIC). The incidence of reported NS in low- and middle-income countries (LMIC) is higher, ranging from 4 per 1000 live births (in Iran)^14^ to 40 per 1000 live births (in Kenya and Nigeria).^15,16^ The high variability in the reported incidence of NS in LMIC can be explained by limited access to neurocritical monitoring and care resources. Most seizures are diagnosed clinically and are not based on EEG studies. Some seizure-like events may be misdiagnosed as seizures, while subclinical seizures may be missed.^17^ Furthermore, a higher rate of home deliveries in LMIC may lead to an underestimation of NS, as some seizures may go undiagnosed in the early neonatal course. Some neonates may also die in the community from complications of seizures before reaching the hospital.^15^ Hence, the incidence of NS in LMIC remains unknown but is likely superior to HIC due to resource limitations and the high prevalence of risk factors for neonatal brain insult.
Various etiologies can increase the excitability of the neonatal brain or cause injury to neurons, which can both increase the risk for seizures. Table 1 contains the most common causes of seizures in newborns.^4,18-20^ The most common etiologies of seizures in late preterm and term infants are hypoxic–ischemic encephalopathy, followed by perinatal arterial ischemic stroke.^1,21,22^ On the other hand, intracranial hemorrhage is the most common seizure etiology in preterm infants, followed by central nervous system infections.^10,22^ The proportion of unidentified etiologies is similar throughout all gestational groups.^22^
Birth asphyxia, perinatal infection, and hypoglycemia are the most frequently identifiable etiologies of NS in LMIC, based on studies from Pakistan, Kenya, Nigeria, and Iran.^15,16,23,24^ Undiagnosed acute bilirubin encephalopathy may add to the problem in LMIC. The underrepresentation of intracranial hemorrhage in neonates living in LMIC could be explained by limited access to neuroimaging, especially brain magnetic resonance imaging (MRI). The burden of preventable neonatal morbidities remains high in LMIC. Perinatal asphyxia is a frequent reason for hospital admission, encompassing 1 out of 6 NICU admissions in resource-limited settings.^25^ The case fatality rate of early-onset group B Streptococcus (GBS) neonatal meningoencephalitis is around 10%, ranging from 5% in developed countries to up to 27% in sub-Saharan Africa.^26^ Higher rate of NS may also be partly due to the higher rate of consanguinity in LMIC countries, which can increase the risk of brain malformations and neurometabolic disorders that can cause NS. Improving antenatal and perinatal care in LMIC could likely prevent the occurrence of these neonatal morbidities and decrease the associated risk of seizures. Interventions, including neonatal resuscitation training of local birth attendants and the provision of basic resuscitation equipment to local health-care centers, could potentially decrease the burden of perinatal asphyxia.^25^ Furthermore, improving antenatal maternal serology screening, routine maternal immunization, neonatal skin care, and most importantly improved breastfeeding rates are other strategies that could be used to decrease the burden of neonatal bacterial infection in LMIC.^27^
According to the International League Against Epilepsy (ILAE) Task Force on NS, the presentation of NS can be divided into motor, nonmotor, sequential, and unclassified categories.^5^ Motor seizures can further be classified into automatism, clonic, myoclonic, tonic seizures, and epileptic spasms. Nonmotor seizures can be subdivided into autonomic and behavioral arrest. It should be remembered that the most common seizure semiology in newborns is EEG-only (electrographic, subclinical) seizures. Table 2 summarizes the different types of clinical presentation of seizures.
Nonmotor seizures can be quite subtle. Neonatologists and pediatric neurologists can find it quite challenging to distinguish apneas caused by autonomic seizures and apneas caused by other neurologic and nonneurologic disorders. Most ictal apneas are associated with the limbic/paralimbic mesial temporal cortex involvement, and they should be distinguished from nonictal apneas.^28,29^ Continuous video-integrated EEG monitoring is required for diagnosing ictal apneas since it measures cortical electrical activity and vital sign variability simultaneously, but access to continuous video-integrated EEG monitoring can be a major challenge in resource-limited settings. The evidence behind clinical characteristics specific to ictal apneas is scarce. However, some clinical clues can be helpful to distinguish ictal apneas from nonictal events. First, apneas are more frequently associated with electrical seizure activity in term compared to preterm neonates.^4^ Second, in preterm infants with recurrent apneas that persist despite the introduction of caffeine treatment, an ictal process should be kept in the differential.^30^ Of note, while tachycardia was thought to be the hallmark of apneic seizures, it should be remembered that bradycardia may also accompany apneic seizures in the newborn population.^30,31^ Neonates with 1 or more of these clinical features should be triaged for transfer to tertiary care facilities to obtain an EEG evaluation.
Various atypical neonatal movements can mimic NS, including jitteriness, benign neonatal sleep myoclonus, and tremors. Some bedside maneuvers can be helpful in distinguishing seizures from nonictal behaviors. If similar behaviors can be provoked by stimulation of the infant and can be interrupted by restraining the affected limbs, then they are unlikely to be seizures.^5^ Hyperekplexia is another seizure mimicker that should not be missed by healthcare practitioners because its diagnosis is clinical and it can lead to developmental delay.^32-34^ Hyperekplexia is commonly described as an exaggerated startle response to external stimuli. Its hereditary form involves genes that affect glycine neurotransmission and presents in the newborn period.^34^ Neonates with hereditary hyperekplexia exhibit an exaggerated startle response, usually followed by prolonged stiffening. A useful bedside clinical examination is to elicit a nonfatigable glabella reflex by tapping on the glabella region or head retraction by tapping repeatedly on the trigeminal area of the face.^33^ Hyperekplexia movements can be aborted by the Vigevano maneuver, which consists of a forced flexion of the head and legs on the trunk.^32^
Neonatal seizures have a wide range of presentations, making it challenging for medical providers to diagnose a seizure solely based on the clinical assessment. An observational study published by Malone et al^35^ evaluated the accuracy of medical providers in distinguishing true seizures from seizure mimickers caught on video recordings. Only around half of the recordings were identified correctly, and the interobserver agreement was suboptimal. Furthermore, the majority of NS are subclinical, meaning that they are electrographic only and are not associated with clinical manifestations.^36-38^ This holds especially true for neonates at risk for seizures, such as neonates with hypoxic–ischemic encephalopathy, perinatal stroke, or central nervous system infections. Murray et al^37^ showed that only around a third of electrographic seizures in term infants at risk of seizures had associated clinical manifestations caught on simultaneous video recording, and that only 9% of electrographic seizures had clinical features that were identified by physicians. A more recent study published by Chen et al^38^ showed that 80% of electrographic seizures captured on continuous video EEG monitoring in neonates with encephalopathy did not have a clinical correlate. Therefore, in order to avoid the risk of over- or undertreating seizures, EEG monitoring has now become the standard of care for diagnosing seizures in newborns.
The ILAE Task Force on NS defines a seizure as “an electrographic event with a pattern characterized by sudden, repetitive, evolving stereotyped waveforms with a beginning and end.” The group does not specify a duration in their most recent definition. They rather explain that the event “has to be sufficient to demonstrate evolution in frequency and morphology of the discharges and needs to be long enough to allow recognition of onset, evolution, and resolution of an abnormal discharge.”^5^ The omission of a specific duration contrasts with the previous definition published in 2013 by the American Clinical Neurophysiology Society critical care monitoring committee,^39^ which included that a seizure needs to be at least 10 seconds long.
Premature neonates are at higher risk of developing seizures than their term counterparts. Although the definition of seizures does not vary based on the degree of prematurity, gestational age does affect the electrographic background of EEG.^40^ A multichannel, continuous EEG (cEEG) is the method of choice to monitor seizures in premature infants because preterm infants have shorter seizures and can have a background that is challenging to differentiate from true ictal activity.^41^ Given that access to a cEEG is challenging in resource-limited settings, certain risk factors that can increase the likelihood of seizures in preterm infants can be used to prioritize the infants for further EEG assessments. These risk factors include a lower gestational age with low Apgar scores at birth, higher Clinical Risk Index for Babies Score II, evidence of hemorrhagic or ischemic brain injury, or major cerebral malformations on cranial ultrasound scans.^41,42^ Monitoring preterm infants at high risk for seizures for a maximum of 24 hours is sufficient and cost-effective.^40,42,43^
While cEEG remains the gold standard for seizure detection in neonates, it remains a resource-intensive and costly investigation tool with limited accessibility across the globe, especially in LMIC.^44-46^ Amplitude-integrated EEG (aEEG) is a reliable alternative to screen for seizures. Amplitude-integrated EEG is a simplified EEG that uses a montage of 2-4 electrodes, which captures the activity of the cortex close to each electrode, resulting in a raw EEG recording. It also compresses and rectifies the EEG tracing over time to provide a rough overview of the brain background activity and its evolution over time. Amplitude-integrated EEG can also give prognostic information based on the evolution of the background findings and the emergence of sleep–wake cycling. However, it should be remembered that aEEG has major limitations compared to cEEG. Seizures with low amplitude or brief duration or seizures occurring remotely from the aEEG leads can be easily missed with the use of aEEG.^47^ Although aEEG does not replace cEEG, they both can be used complementarily, as aEEG can help clinicians at the bedside identify neonates who need cEEG.^48^
Different strategies can be used to increase the accessibility of EEG studies in LMIC.
**Improving pediatric EEG training **
Increasing use of digital technologies * : The rapid growth of digital technology in medicine can also facilitate remote EEG specialist review and reporting. The ongoing Prevention of Epilepsy by Reducing Neonatal Encephalopathy^52^ trial aims to implement a care bundle to reduce birth injury-related epilepsy at 18 months of age in India. To improve the accessibility to EEG interpretation, video EEG will be uploaded onto a secure cloud-based server for central reporting, while aEEG will be locally read in real time to assist clinical decision-making.^53^ Smartphones can also be used to facilitate remote EEG reading. Williams et al^54^ evaluated the use of a smartphone EEG and remote online interpretation for children with epilepsy in the Republic of Guinea. The Smartphone Brain Scanner-2 allowed for EEG data acquisition, filtering, artifact removal, and recording. Data obtained from the application was then converted on a secure web-based platform, and EEGs were read by HIC board-certified neurologists. The use of this platform was found to have moderate sensitivity and high specificity for the detection of epileptiform abnormalities in children in low-income countries.
The use of EEG in the NICU may also be hindered by the lack of trained personnel who can perform EEG recordings. One potential solution to this problem is to train NICU nurses to attach the EEG electrodes to newborns, which can save time and reduce errors. This can potentially optimize the use of EEG in the NICU by increasing the availability of EEG data, facilitating timely diagnosis and treatment of NS, and improving the outcome of affected infants.
The ILAE Task Force on NS published a diagnostic framework to guide clinicians in the classification of seizures.^5^ According to this framework, neonates who present with abnormal movements require electrographic recording to confirm seizures, and those with an electrographic component can then be classified as electro-clinical or electrographic only based on the presence or absence of a clinical component. A recent study published by Yozawitz et al^56^ showed that this classification system can be used by all health-care providers to identify a predominant seizure type. There was also an average accuracy of 85% across the 5 seizure types. Electroencephalographic studies are central to this classification system as today the most common semiology of NS is electrographic only. The ILAE task force excluded “clinical-only seizures” from their classification system, as studies have shown that most clinical-only events do not truly have an epileptic origin.^37,57^ Furthermore, a distinction between focal and generalized seizures was not included in the neonatal classification because all seizures in the neonatal period have a focal onset^.5^
The first step in NS management is to assess and secure the airway, breathing, and circulation of the patient and to intervene promptly if necessary. Identifying and managing the underlying cause of a seizure is crucial for achieving adequate seizure control.^1,17,58^ One important point is to remember that the treatment should not be deferred while waiting for preliminary and further testing.
Access to neuroimaging, especially brain MRI, is far from being universal in LMIC.^44,59^ Cranial ultrasonography can be used as an initial neuroimaging modality to assess for intraventricular hemorrhage, parenchymal hemorrhage, and posthemorrhagic ventricular dilatation.^60^ However, ultrasonography is not a substitute for brain MRI and is less sensitive in detecting ischemic injury. Brain MRI can be particularly useful in diagnosing hypoxic–ischemic encephalopathy, neonatal arterial ischemic stroke, central nervous system infections, cerebral venous sinus thrombosis, some metabolic disorders, and cerebral malformations.^21^ Hence, every neonate with seizures of unknown etiology should undergo a brain MRI if available. Further investigations can be done when seizures are recurrent and resist first-line treatment or when their underlying etiology remains unknown. Metabolic testing for organic acidemias, urea cycle defects and fatty oxidation defects, screening for intrauterine infections, genetic testing, and pathological assessment of the placenta can be considered based on available resources.^1^ A pediatric neurologist should be involved early in the management whenever possible.
Despite earlier recommendations stating that antiepileptic treatment should be initiated once reaching a cumulative electrographic seizure burden of at least 30 seconds/hour, seizures should ideally be treated as soon as possible to prevent further injury.^61^ However, in resource-limited settings without access to EEG monitoring, clinicians may need to rely on their clinical assessment. The levels of diagnostic certainty published by the Brighton Collaboration Neonatal Seizures Working Group can guide clinicians in their decision to start antiepileptic treatment (Table 3).^62^ When the suspected event has a focal clonic or focal tonic presentation, with or without associated ictal EEG corroboration, administration of an anti-seizure medication (ASM) is recommended. Treatment may be considered but is not necessarily indicated when the suspected abnormal movement is not associated with focal clonic or focal tonic movements and when EEG evaluation is unavailable. Further confirmatory testing should also be considered. The timing of ASM administration may also be critical, as treating a seizure within the first hour of its onset may lower the seizure burden over the next 24 hours.^63^ It is recommended to continue electrographic monitoring for 24 hours after the last ASM dose to watch for uncoupling. Uncoupling refers to the conversion of electroclinical seizures into electrographic-only seizures after the administration of ASM.^64^
The ILAE Task Force on NS recently published treatment guidelines that include the following consensus-based ^65^
Today, phenobarbital is considered the preferred first-line ASM to treat NS.^66^ However, until recently, a growing body of studies suggested that the use of intravenous levetiracetam may be a promising first-line treatment option for NS compared to phenobarbital.^67-71^ A study from Egypt showed that levetiracetam can be an effective and safe option as a first-line ASM in NS compared to phenobarbital.^67^ After administering 1 or 2 doses of levetiracetam or phenobarbital to infants, seizure cessation was achieved in 79% compared to 65%, respectively (P = .01). Patients exposed to levetiracetam did not develop adverse effects, compared to patients treated with phenobarbital, who developed hypotension, bradycardia, and respiratory depression.
However, in a recent phase IIb randomized controlled trial, 80% of patients randomly assigned to phenobarbital remained seizure-free for 24 hours, compared with 28% of patients randomly assigned to levetiracetam [P < .001; relative risk 0.35 (95% CI, 0.22-0.56)]. Of note, more adverse effects were seen in subjects randomly assigned to phenobarbital, but this was not statistically significant.^66^ With the current evidence, intravenous phenobarbital should be the first-line ASM in neonates presenting with seizures. Administering phenobarbital preparations enterally via a nasogastric tube should not be preferred in settings where intravenous phenobarbital is not accessible. If intravenous phenobarbital is not readily available, then a trial of benzodiazepine such as lorazepam or midazolam (both of which have a shorter half-life than phenobarbital) can be administered to abort seizures.
The question of continuing antiepileptic medications after discharge from the hospital remained a subject of controversy for decades. However, several animal studies showed that antiepileptic medications can be neurotoxic to the developing brain.^72-74^ A recent study from Glass et al. also showed that the neurodevelopmental outcomes and epilepsy risk at 24 months were similar among children whose ASM was discontinued or maintained at hospital discharge after the resolution of acute symptomatic NS. Hence, the updated ILAE Task Force on NS recommends discontinuation of ASM before discharge home in neonates hospitalized for acute-provoked seizures who do not have a diagnosis of neonatal epilepsy. This recommendation holds true regardless of neuroimaging or EEG findings.^65^
As noted earlier, access to ASMs can be challenging in LMIC due to several factors. In Laos, the national production of phenobarbital does not match the population’s needs. This mismatch is caused by delays in the overpriced local production of phenobarbital and the centralization of delivery sites in urban centers.^75^ Traditional beliefs also limit children with seizures from having access to appropriate medical services. Children living with epilepsy tend to be isolated from villages and referred to traditional care or religious sacrifices. These misconceptions about epilepsy are also prevalent among Lao health-care providers, as 1/3 of physicians treating children had never prescribed ASM.^76^ Similar challenges, including population misbeliefs regarding epilepsy, financial barriers, and disparities in health infrastructures in remote districts, were also present in Madagascar^77^ and Cambodia.^78^
Compelling evidence suggests that NS significantly increase the risk of cognitive and motor delays and neurodevelopmental impairment in multiple developmental domains.^79-82^ Neonatal seizures also increase the risk of childhood epilepsy, especially in neonates with extensive perinatal brain injury to their deep gray matter and cortical regions.^83,84^ The underlying etiology of NS remains an important modulator of neurodevelopmental prognosis.
It remains unclear whether seizures are an epiphenomenon of brain injury versus an independent effect modifier further contributing to brain damage.^85^ Different mechanisms have been raised to explain how seizures can cause brain injury. Greater seizure burden has been associated with smaller brain volumes on term-equivalent age MRI in preterm infants less than 30 weeks’ gestation.^86^ A rise in cerebral blood velocity may also lead to increased intracranial pressure and, consequently, bleeding from the fragile germinal matrix.^87^ Another theory suggests that the increased cerebral metabolic demand during an ictal episode^88^ may “steal” glucose and oxygen supply from other brain regions, especially in neonates with previous brain insults.^89^
Longer duration of seizures may also worsen outcomes, especially in neonates with perinatal brain injury.^90-92^ Pisani et al^90^ found that an ictal fraction (defined as the total duration of seizure/duration of the EEG recording × hour) longer than 10 minutes was significantly associated with adverse neurodevelopmental outcomes, including cerebral palsy, intellectual disability, and postnatal epilepsy. Different studies have looked for possible risk factors that may independently influence the outcomes of neonates with seizures.^93-96^ Garfinkle and Shevell^96^ designed a 5-point scoring system to help clinicians in the prognostication of neurodevelopmental outcomes in neonates who experienced seizures.^95^ Others also reported that neonates with seizures who were born prematurely or had very low birth weight, abnormal cranial ultrasound findings, or suffered from birth asphyxia, meningitis, or septicemia, were at increased risk for death, had abnormal examination at discharge, and were phenobarbital nonresponders.
Seizures in the neonatal period remain a serious neurological emergency that necessitates admission to neonatal intensive care units across the globe, and NS is associated with increased risks of neurodevelopmental impairment and mortality. In recent decades, there have been significant improvements in perinatal and neonatal medicine and EEG monitoring that have enhanced the diagnosis and treatment of NS in HIC. However, the management of NS remains a major challenge in LMICs, where the availability and accessibility of resources are limited. Several initiatives have been launched to provide optimal neurological investigations and neuroprotective care to neonates living in LMIC. However, more efforts are needed to close the gap and ensure that all neonates with seizures receive the best possible care.^53,97,98^ We hope that this review article will provide clinicians with practical and evidence-based considerations for the diagnosis and treatment of NS and stimulate further research in this field.
This study received no funding. Mehmet N. Cizmeci received support from Dr. Karen Pape Program in Neuroplasticity for neonatal neurodevelopmental research.