A seizure is the most frequent sign of neurologic dysfunction in the neonate. Since seizures may be the only sign of a central nervous system disorder, their recognition is important. Seizures are not only more frequent during the neonatal period than at any other age, they are also more difficult to diagnose because of the subtle nature of behavioral and EEG manifestations. The absence of precise defini-tions has hampered epidemiological studies resulting in reduced accuracy of analyzing therapeutic effects and long term outcome in these infants.

Clinicians and nurses caring for neonates with seizures are inevitably faced with questions regarding diagnosis, evaluation, management and outcome. Although, controversies exist in the diagnosis of neonatal seizures, once the diagnosis is made, these babies clearly have significant mortality and risk of major neuro-developmental disability.

Reports of the incidence of neonatal seizures have suffered from both over and under estimations. The incidence of clinical neonatal seizures ranges from 0.5% of term infants(1) to 20.2% of preterm infants(2). Lanska et al. reported the overall risk of neonatal seizures as 2.84 per 1000 live births in a National Hospital Discharge survey (1980-1991). In a large population-based study from Newfoundland, Ronen et al. found the incidence of clinical neonatal seizures to be 2 per 1000 live term births to 11.1 per 1000 live preterm births(3). With such discrepant incidence reports, subjectivity of seizures reporting clearly hinders defining the true incidence.

The classification of neonatal seizures has been based on the clinical recognition of repetitive and stereotypic motor activity or behavioral phenomena that are often difficult to recognize in neonates, even by experienced observers. This is especially true in premature infants, whose abnormal motor or behavioral activities may be indistinguishable from normal activity. Subtle seizures are the most common type of neonatal seizure recorded in clinical studies, followed by the multifocal clonic variety. Two other conditions, jitteriness and benign neonatal sleep myoclonus, should be differentiated from neonatal seizures.

The landmark study of Mizrahi and Kellaway provided the first scientific basis for understanding the true nature of many subtle seizures(4). Their study used time synchronized video-EEG monitoring to characterize a variety of clinical paroxysmal behaviors. A clinical seizure was considered to arise from a specific epileptic basis if it occurred simultaneously and was consistently synchro-nized to an electrographic seizure displayed on the coincident EEG. For example, in their classification of clinical seizures, focal clonic seizures arise coincidentally with an electro-graphic seizure; therefore, the specific clinical event represents an epileptic seizure. Converse-ly, most of the clinical activity and behaviors that fall within the definition of subtle or minimal seizures were shown to occur without cortical electrographic seizure activity. Conse-quently, most clinically subtle seizures are nonepileptic, since they display no consistent relationship to the EEG. We therefore conclude that the clinical observation of a neonate with suspected seizures is often inadequate if we are to characterize and understand the nature of the epileptic or nonepileptic events that are occur-ring. Questioning the diagnostic validity of clinical observations for the appropriate classification is important both diagnostically and prognostically. It is particularly challenging to characterize stereotypic movements with inconsistent EEG correlates. Mizrahi and Kellaway suggest that certain automatisms previously believed to be seizures represent brain stem release phenomena(4).

The observation that dramatic clinical movements may or may not be associated with electrographic seizure activity raises the possibility that not all seizures are accom-panied by EEG correlates. This phenomenon is termed electro-clinical dissociation. In a retrospective evaluation, Weiner et al. evaluated 110 infants with electrographic seizures. Sixteen per cent of the infants with electrographically confirmed seizures and 19% of the 243 analyzed seizures displayed electro-clinical dissocia-tion(5). The two groups were similar with respect to perinatal factors, etiology and outcome. In another retrospective study, Sheth found that the confirmation of seizures by EEG increases as gestation increases (63% at 28 weeks vs. 77% at term)(6).

When infants with suspected seizures were studied prospectively with prolonged video-EEG recordings, O’Meara et al. reported that several infants had electrographic seizures with reduced or no clinical manifestations. This was especially evident after the administration of anticonvulsant medication(7). As many as 85% of all seizures recorded in their study had no clinical correlates. Among the infants with clear clinical correlates, clinical observation under-estimated electrographic seizures in individual neonates by a mean of 54% (range 0-95%). In view of the number of neonates that did not show clinical correlates of electrographic seizures and the difficulty associated with the clinical identification of some seizures, they suggested that EEG is essential to diagnose seizures and to monitor the effect of treatment. In a related study, the authors found evidence of reduced clinical signs after sequential anticonvulsant medication(8). Our experience at the Children’s Hospital at Strong is similar. Since August 1994, we have changed our approach to neonatal seizures from treating only clinically recognized seizures to screening for electrographic seizures. We now treat infants based on electrographic seizures, independent of clinical correlates. We have been impressed with the number, duration and intensity of clinically silent electrographic seizures, despite treatment with conventional anticonvulsants.

In humans, many parts of the brain are quite immature at birth. This immaturity implies selective vulnerability, as well as selective resistance to specific disease pro-cesses. Rapid brain growth imposes rigid constraints. Therefore, an event that interferes with the developmental cascade has the potential for long term effects.

Controversy about the impact of electrographic seizures on the developing brain is ongoing. The effects of neonatal seizures on brain development are difficult to differentiate from those of the brain lesions causing the seizures. Some animal studies suggest that the developing brain is selectively vulnerable to seizures(9), while many others have shown its selective resis-tance(10,11). A considerable body of evidence suggests that the consequences of seizures in the immature brain are considerably different from those in adults. Seizures do not have to cause cell death to result in adverse outcomes. Following kindling, immature rats, have a permanent reduction in seizure threshold. In newborn rats, seizures inhibit brain protein synthesis(12), reduce brain size, and delay developmental mile-stones(13,14). Recurrent seizures during the neonatal period also result in deficits in learning and memory when animals are studied as adults, despite lack of cell loss(15). While no data are available regarding the effect of neonatal seizures on the growth of the human brain, the extent of postnatal mitotic activity in man suggests potential vulnerability. Approximately 80% of the cells in the human brain are generated after birth(16), and mitotic activity continues in the human cerebellum for at least 1 year postnatally.

The saying "Treat the patient, not the EEG" may not be valid in the arena of neonatal seizures. First, the electrographic seizure activity itself often eludes clinical observation. Second, neonatal seizures are subtle and clinical observation cannot be depended on. It is not widely appreciated that the end point of therapy (reduction or elimination of electrographic activity) often fails. Continuous EEG monitoring of the neonates with seizures is the only objective means to judge the response to treatment.

Currently, the most common drug used to treat neonatal seizures is phenobarbital(17). Phenobarbital enhances postsynaptic GABA inhibition. Additionally, based on animal studies, GABA may be excitatory in the early developmental period, thus, phenobarbital may not be an optimal drug for neonatal seizures. In fact, it appears to have limited efficacy in treating neonatal seizures. Phenytoin, a drug that causes use-dependent inhibition of Na+ channels necessary for activation of action potentials, is used as an adjuvant to phenobarbital therapy. Other drugs used in adults that act at the Na+ channels, such as carbamazepine and lamotri-gine, have not been studied in neonates.

At the Children’s Hospital at Strong, we analyzed continuous EEG recordings of 41 consecutive neonates to quanify this "burden" of electrographic seizures. The electrographic seizures in these neonates were only partially responsive to standard anti-convulsants. Only, 25% of the infants responded to 40 mg/kg of phenobarbital. Thirty per cent of neonates continued to have electrographic seizures despite full loading doses of both phenobarbital and phenytoin(18). Similarly, Painter et al.(19), in a randomized trial, found phenobarbital and phenytoin to be equally but incompletely effective as anticonvulsants in neonates. With either drug given alone, the seizures were controlled in fewer than half the neonates. Neonates whose seizures were not controlled by the assigned drug were then treated with both drugs. Despite treatment with both phenobarbital and phenytoin, electro graphic seizure control was achieved in only 60% of neonates. The severity of the seizures was a better predictor of the success of treatment than was the assigned agent. Neonates with mild seizures or with seizures that were decreasing in severity before treatment were more likely to have their seizures end regardless of the treatment assignment. Other studies have not found benzodiazepines(20,21) to be more effective.

Based on the sequential developmental pattern of pre- and postsynaptic neurotrans-mitter receptor development, drugs that have a potential to be more beneficial in neonatal seizures are likely to be those that affect the NMDA receptor, presynaptic GABAB receptors or voltage gated channels. Unfortunately, drugs that block the NMDA receptor, such as topiramate and felbamate, have not been tested in the newborn because of the concerns that treatment with excitatory amino acid antagonists may interfere with learning, memory or brain plasticity. No drug currently available speci-fically antagonizes the GABAB receptor.

There is little question that neonatal seizures are associated with a high morbidity and mortality rate. Even with recent advances in obstetric and perinatal care, seizures continue to be a significant predictor of a poor neuro-logical outcome . Etiology of the seizures is an important factor influencing outcome. EEG background and absence of state change also significantly predict neurologic outcome(22). Using continuous EEG monitoring, we found higher mortality (p <0.01) and higher risk for cerebral palsy (p <0.01) in infants with electrographic seizures as compared with those with similar neonatal risk factors but without seizures. Infants with electrographic seizures tended to have microcephaly (p = 0.7) and failure to thrive (p = 0.14) more often than thsoe without. In addition, having more than 75 seizures recorded was highly correlated with microcephaly and severe cerebral palsy (p = 0.03)(18). Legido et al. found a moderate to severely abnormal background to be associated with an unfavorable outcome in 81% of their cases, whereas, a normal or mildly abnormal background was predictive of a favorable prognosis in 67% of cases(23). They also found that more than 10 seizures per hour were associated with the development of epilepsy in 86% of cases, whereas the rate was 45% when the seizure frequency was less than ten.

Clinical observation of the neonate has inherent and significant limitations. Because a clinical diagnosis of suspicious neonatal behaviors is limited, EEG must be accepted as a standard to establish the diagnosis of neonatal seizures. It is, therefore, important that a definition for electrographic seizure should be explicit and easily applied and be generally accepted by the EEG community.

Continuous EEG monitoring is the only real measure of effectiveness of anticonvulsant medication. However, this requires a constant attendance by technologists and the continuous observation of the EEG online to recognize and treat seizures as they arise. Few institutions have the equipment or manpower to provide this ideal level of continuous patient EEG monitoring. Much work has been done to develop computerized seizure detection algorithms that are specific to the newborn. Multichannel and single channel long term EEG recording systems have been in the experimental realm for many years. The tape-recorded EEG (Oxford Medilog 9000 systems) offers the possibility of more channels and higher reliability when diagnosing short subclinical seizures. Un-fortunately, the system does not have the capability at its current stage of development for continuous online analysis and review. The amplitude-integrated EEG (Olympic Medical) on the other hand, has the advantage that it is immediately available as a continuous single channel EEG recording. It is possible that in the years to come, neonatal intensive care nurseries will use a combination of a standard EEG for diagnosis followed by continuous EEG monitoring of at risk infants.

Studies at the Children’s Hospital at Strong, have shown that background abnormalities on the initial 30-60 minute EEG of ‘at risk’ neonates is strongly associated with electro-graphic seizures in the subsequent 18-24 hours(24). Administration of anticonvulsant medication before the start of the EEG recording does not affect this association. Thus screening infants at risk for neonatal seizures with a routine EEG allows identification of infants at highest risk for seizures; thus, conserving resources required for continuous EEG monitoring and facilitating early intervention for seizures.

Seizures in the newborn are often diffi-cult to recognize clinically. Clinical seizures may represent the tip of the iceberg as innumerable seizures may continue electrographically. The phenomenon of electro-clinical dissociation becomes even more prominent once anticonvul-sants are used. The gold standard for the diagnosis of seizures in the newborn should be an EEG. Continuous EEG monitoring must be used to evaluate the response to anticonvulsant medication, with the aim to reduce or preferably eliminate electrographic seizure activity. Indirect evidence suggests that seizures in themselves may be harmful to the brain, perhaps adding insult to the neurologic injury that caused them. This hypothesis can only be tested if we can conduct randomized clinical trials with a drug capable of eliminating electrographic seizure activity.

The author wishes to thank Ronnie Guillet for her assistance in the preparation of this manuscript.

Nirupama Laroia,
Assistant Professor,
Department of Pediatrics/ Neonatology,
Children’s Hospital at Strong,
University of Rochester,
Rochester, NY.
Address for correspondence:
601 Elmwood Avenue, Box 651.
Rochester, NY 14642, USA
E-mail: nirupama_laroia@ urmc.rochester.edu

1. Eriksson M, Zetterstrom R. Neonatal convulsions: Incidence and causes in the Stockholm area. Acta Paediatr Scand 1979; 68: 807-811.

2. Seay AR BP. Significance of seizures in infants weighing less than 2500 g. Arch Neurol 1977; 34: 381-382.

3. Ronen GM, Penney S, Andrews W. The epidemiology of clinical neonatal seizures in Newfoundland: A population-based study. J Pediatr 1999; 134: 71-75.

4. Mizrahi EM, Kellaway P. Characterization and classification of neonatal seizures. Neurology 1987; 37: 1837-1844.

5. Weiner SP, Painter MJ, Geva D, Guthrie RD, Scher MS. Neonatal seizures: electroclinical dissociation. Pediatric Neurol 1991; 7: 363-368.

6. Sheth RD. Electroencephalogram confirmatory rate in neonatal seizures. Pediatr Neurol 1999; 20: 27-30.

7. O’Meara MW, Bye AM, Flanagan D. Clinical features of neonatal seizures. J Paediatr Child Health 1995; 31: 237-240.

8. Bye AM, Flanagan D. Spatial and temporal characteristics of neonatal seizures. Epilepsia 1995; 36: 1009-1016.

9. Moshe SL, Albala BJ, Ackermann RF, Engel J. Increased seizure susceptibility of the immature brain. Brain Res 1983; 283: 81-85.

10. Moshe SL, Sharpless NS, Kaplan J. Kindling in developing rats: Variability of after discharge thresholds with age. Brain Res 1981; 211: 190-195.

11. Michelson HB, Lothman EW. An ontogenetic study of kindling using rapidly recurring hippocampal seizures. Brain Res Dev Brain Res 1991; 61: 79-85.

12. Dwyer BE, Wasterlain CG, Fujiwaka DG, Yamada L. Brain protein metabolism in epilepsy. In: Basic Mechanisms of the Epilepsies: Mole-cular and Cellular Approaches. Eds. Delgado WJA Escueta AV, Woodbury DM, Porter RJ., New York, Raven Press, 1986; pp 903-918.

13. Wasterlain CG, Fujiwaka DG, Dwyer BE, Vanucci RC, Schwartz PH, Morin AM. Brain damage in the neonate: Multiple periods of selective vulnerability reflect discrete molecular events resulting from normal development. In: Neonatal Seizures. Ed. Wasterlain CG New York, Raven Press, 1990; pp 69-82.

14. Holmes GL, Gairsa JL, Chevassus-Au-Louis N, Ben-Ari Y. Consequences of neonatal seizures in the rat: Morphological and behavioral effects. Ann Neurol 1998; 44: 845-857.

15. Neill JC, Liu Z, Sarkisian M, Tandon P, Yang Y, Stabstrom CE, et al. Recurrent seizures in immature rats: Effect on auditory and visual discrimination. Brain Res Dev Brain Res 1996; 95: 283-292.

16. Dobbing J, Sands J. Quantitative growth and development of human brain. Arch Dis Child 1973; 48: 757-767.

17. Massingale TW, Buttross S. Survey of treatment practices for neonatal seizures. J Perinatol 1993; 13: 107-110.

18. Laroia N, McBride M, Guillet R. Neonatal electrographic seizure burden correlates with neurologic outcome. Pediatr Res 1999; 45: 344A.

19. Painter MJ, Cher MS, Stein AD, Armatti S, Wang Z, Gardiner JC, et al. Phenobarbital compared with phenytoin for the treatment of neonatal seizures. New Engl J Med 1999; 341: 485-489.

20. Deshmukh A, Wittert W, Schnitzler E, Mangurten HH. Lorazepam in the treatment of refractory neonatal seizures. A pilot study. Am J Dis Child 1986; 140: 1042-1044.

21. Andre M, Boutroy MJ, Dubruc C, Thenot JP, Bianchett G, Sola L, et al. Clonazepam pharmacokinetics and therapeutic efficacy in neonatal seizures. Eur J Clin Pharmacol 1986; 30: 585-589.

22. McBride M. 1999 (personal communication).

23. Legido A, Clancy RR, Berman PH. Neurologic outcome after electroencephalographically proven neonatal seizures. Pediatrics 1991; 88: 583-596.

24. Laroia N, Guillet R, Burchfiel J, McBride MC. EEG background as predictor of electrographic seizures in high-risk neonates. Epilepsia 1998; 39: 545-551.