Treatment Duration After Acute Symptomatic Seizures in Neonates: A Multicenter Cohort Study

Abstract

We aimed to define determinants of duration of treatment for acute symptomatic neonatal seizures in a contemporary multicenter observational cohort study. After adjustment for potential confounders, only study site and seizure etiology remained significantly associated with the chance of continuing antiseizure medication after discharge to home.

Despite the wide-ranging impacts of seizures in the newborn, many knowledge gaps persist, and the optimal treatment strategy is unknown. In animal models, both seizures and their treatment with phenobarbital and phenytoin can cause abnormal brain development.¹ Treating clinicians are left to make decisions regarding medication, dosage, and duration without the benefit of practice guidelines. Published reports that examine management strategies for seizures in neonates have been limited by single center study designs, reliance on clinical (vs electroencephalographic) seizure detection, limited distinction between neonatal onset epilepsies vs acute symptomatic seizures, and the use of survey data.^2–5 Survey data suggest that although the initial management of neonatal seizures is similar between centers, subsequent antiseizure medication choices and duration of treatment are extremely variable.^2–6

We aimed to evaluate contemporary treatment practices related to prescription of antiseizure medications at the time of discharge to home for newborns with acute symptomatic seizures. We hypothesized that there would be substantial treatment variability across tertiary care centers because there is little evidence to guide prescribing practices for seizure medications in this patient population.

Methods

This was a prospective, observational cohort study of consecutive newborns with seizures treated at the 7 sites of the Neonatal Seizure Registry. Each site has a level IV neonatal intensive care unit and follows the American Clinical Neurophysiology Society guidelines for continuous video-electroencephalogram ([EEG] cEEG) monitoring.⁷ The local institutional review board for every site approved the study and granted a waiver of informed consent.

All newborns with seizures diagnosed clinically or with EEG confirmation were enrolled from January 2013 through November 2015. Neonates with events that were determined by EEG not to be seizures were not enrolled. The demographic and etiologic data for a subset of 427 newborns in this cohort were reported separately⁸; the present analyses include the management details of the full Registry cohort (n = 611). Indications for cEEG monitoring included differential diagnosis of abnormal paroxysmal events, screening for seizures in high-risk patients (eg, hypoxic ischemic encephalopathy), and the assessment of background abnormalities or seizures in newborns with acute encephalopathy.

Details regarding seizure etiology, medical management, and treatment were abstracted from medical records and recorded prospectively. Treatment for neonatal seizures, including medication selection and duration of therapy, was at the discretion of each neonate’s clinical team. No specific treatment algorithm or guideline was provided to the study sites, and sites did not all have an institutional standard neonatal seizure treatment pathway.

Potential confounders and covariables for treatment duration included study site, seizure etiology, electrographic confirmation of neonatal seizures, presence of status epilepticus, seizures that were refractory to the initial loading dose of antiseizure medication, and abnormal neurologic examination at the time of discharge (defined as documented abnormality in consciousness, tone, and/or reflexes). Analyses of treatment duration excluded the following infants: those who did not survive the neonatal admission, who were discharged from the neonatal intensive care unit to palliative care or hospice, or who were transferred from a study site to another hospital.

Descriptive statistics and results of ANOVA and χ² tests are presented. Variables that were significant at a level of P ≤ 0.1 in univariable analyses were included in the multivariable models. Backward stepwise regression was employed to reach the final multivariable model. Analyses were completed using Stata 12 (StataCorp, College Station, Texas).

Results

From January 2013 through November 2015, the 7 study sites enrolled 611 consecutive newborns with seizures in the Neonatal Seizure Registry (male: n = 337, 55%; >37 weeks gestation, n = 519, 85%) (Table I). Among these, 458 (75%) had acute symptomatic seizures, of whom 373 (81%) survived, and 317 (69%) were discharged to home from the study center. Demographic and clinical characteristics of the study cohort are presented in Table I.

Table I.

Clinical characteristics of 611 consecutive newborns with seizures at the 7 Neonatal Seizure Registry sites

Clinical characteristics	Overall n = 611	Site 1 n = 68	Site 2 n = 113	Site 3 n = 34	Site 4 n = 80	Site 5 n = 121	Site 6 n = 65	Site 7 n = 130	P value
Male sex	337 (55%)	39 (57%)	63 (56%)	21 (62%)	47 (59%)	59 (49%)	33 (51%)	75 (58%)	.7*
Term (>37 wk gestation)	519 (85%)	58 (85%)	95 (84%)	33 (97%)	64 (80%)	103 (85%)	54 (83%)	112 (86%)	.4*
EEG monitoring, h	66 (41, 99)	55 (25, 87)	66 (41, 107)	64 (40, 96)	63 (39, 102)	64 (37, 91)	86 (56, 106)	66 (41, 96)	.03†
Primary seizure etiology									.04*
Hypoxic-ischemic encephalopathy	231 (38%)	20 (29%)	46 (41%)	10 (29%)	29 (36%)	41 (34%)	31 (48%)	54 (41%)
Ischemic stroke	101 (17%)	10 (15%)	16 (14%)	13 (38%)	14 (18%)	18 (15%)	6 (9%)	24 (18%)
Intracranial haemorrhage	78 (13%)	7 (10%)	13 (12%)	1 (3%)	10 (13%)	17 (14%)	11 (17%)	19 (15%)
Epilepsy†	80 (13%)	15 (22%)	17 (15%)	2 (6%)	11 (14%)	17 (14%)	7 (11%)	11 (8%)
Deceased	110 (18%)	19 (28%)	20 (18%)	2 (6%)	17 (21%)	22 (18%)	14 (22%)	16 (12%)	.06*
Length of stay among survivors (d)	15 (10, 30)	11 (7, 20)	20 (10, 33)	10.5 (8, 14.5)	16 (11, 41)	10 (14, 35)	21 (14, 35)	13 (9, 34)	.05†
Discharge to home on antiseizure medication
All subjects	428 (76%)	12 (27%)	76 (90%)	27 (90%)	49 (89%)	61 (71%)	39 (83%)	61 (74%)	<.0005*
Acute symptomatic etiology (n = 318)	233 (73%)	1 (4%)	57 (89%)	21 (91%)	32 (89%)	39 (65%)	33 (84%)	50 (71%)	<.0005*

Data are presented as N(%) or median (IQR).

^*χ2.

^†ANOVA.

^‡Neonatal epilepsy includes epileptic encephalopathy, brain malformation, and benign familial neonatal epilepsy.

Initial treatment strategies were similar between sites. The site at which the patient was treated was a strong predictor of medication continuation at the time of hospital discharge to home.

Antiseizure medications were continued at the time of hospital discharge for 73% of survivors of acute symptomatic seizures (range 4%–91% across sites, P < .0005, χ²). Site 1 was the most likely to discontinue antiseizure medication before discharge. Even when that site was excluded, there was significant variability across sites’ prescription of medications at discharge (P = .003, χ²).

Phenobarbital was the most commonly prescribed medication among those discharged home; 63% of survivors were receiving phenobarbital at the time of discharge (range by site 10%–88%, P < .0005, χ²). Levetiracetam was prescribed to 24% at the time of discharge (range by site 6%–44%, P < .0005, χ²). Phenytoin was prescribed to <1% of survivors at discharge (range by site 0%–2%, P = .8, χ²).

Discharge to home on medication was also strongly associated with seizure etiology (P < .0005, χ²). Among 82 survivors of hypoxic-ischemic encephalopathy treated with therapeutic hypothermia, 47 (57%) were discharged to home on medications (range by site 0%–88%, P < .0005, χ²). Among 92 survivors of ischemic stroke, 72 (78%) were discharged to home on medications (range by site 0%–100%, P < .0005, χ²).

Among 56 survivors of intracranial hemorrhage, 46 (82%) were discharged to home on medications (range by site 0%–100%, P = .01, χ²).

In univariable analyses, among survivors of acute symptomatic seizures who were discharged home, additional clinical factors associated with continuing medication upon discharge were EEG-confirmed seizures, status epilepticus, seizures refractory to the initial loading dose of medication, and abnormal neurologic examination at the time of hospital discharge (Table II). After adjustment for each of these variables, as well as seizure etiology and study site, only study site and seizure etiology remained significantly associated with the chance of continuing medication at the time of discharge to home.

Table II.

Variables associated with medications continuation at the time of discharge to home among the 317 survivors of acute symptomatic seizures

			Univariable analyses*			Multivariable analyses†
	n	Discharged with AED	RR	95% CI	P	OR	95% CI	P
EEG confirmed seizures
Yes	266	206 (77%)	1.5	(1.2–2.0)	.0001	2.3	(0.97–5.4)	.06
No	51	26 (51%)
Status epilepticus
Yes	48	42 (88%)	1.2	(1.1–1.4)	.015	2.1	(0.6–7.3)	.2
No	269	190 (71%)
Seizures refractory to initial loading dose
Yes	196	160 (82%)	1.3	(1.1–1.5)	.0003	1.6	(0.8–3.2)	.2
No/unknown	115	71 (62%)
Abnormal examination at discharge
Yes	150	123 (82%)	1.3	(1.1–1.4)	.0008	2.0	(0.99–4.1)	.053
No	167	109 (67%)

AED, antiepileptic drug; RR, relative risk.

^*χ².

^†Wald P value adjusted for each of the risk factors plus site, and etiology, which remain highly significant (P < .0005).

Discussion

In this prospective multicenter study of consecutive newborns with seizures who were monitored with cEEG according to American Clinical Neurophysiology Society guidelines, the decision regarding whether or not to send a newborn with acute symptomatic seizures home on antiseizure medications was significantly associated with the hospital to which the infant was admitted, even after adjusting for important potential confounders such as seizure burden and seizure etiology. At some centers, few to no patients with acute symptomatic neonatal seizures were prescribed antiseizure medications at the time of discharge home, whereas at other sites, almost all newborns were prescribed antiseizure medications upon hospital discharge. This finding is in keeping with physician survey results, in which respondents offered different opinions regarding appropriate duration of medication administration for the same clinical vignette.⁵

For any major health condition, understanding the correct treatment choice and the ideal length of treatment is critical. Phenobarbital, the most commonly prescribed first line antiseizure medication for neonatal seizures, is often maintained for several months because of clinicians’ and parents’ concern that early discontinuation of medicine may result in seizure recurrence.⁵

However, continued exposure to phenobarbital is sedating, which may prolong the time it takes for a newborn to establish oral feeding, and this medicine may have deleterious long-term effects on the developing brain.^9–11 Preliminary evidence suggests that early discontinuation of medication is not harmful,^5,9,12 but the optimal duration of therapy remains unknown.

Whether phenobarbital should remain the first line medication for neonatal seizures and how long to treat a newborn with antiseizure medications remain open questions, and lead to significant practice variability.⁵ This variability is clearly reflected in these data from the Neonatal Seizure Registry, in which fewer than one-half of the sites have local treatment pathways and those local guidelines differ from one another. There is a clear need for scientific evidence to serve as the foundation not only for hospital-specific treatment pathways but also for nationally recognized treatment guidelines.

The Neonatal Seizure Registry is not designed to gather long-term outcome data. Therefore, we are unable to comment on the potential implications of early vs later discontinuation of medication for acute symptomatic neonatal seizures. However, a major strength of this study is that we evaluated actual clinical practice for a very large consecutive cohort of newborns with seizures, as opposed to surveying clinicians about their preferred treatment approaches. These real-world contemporary data highlight the need for careful evaluation of the potential long-term benefits and consequences of different treatment strategies for newborns with seizures.

Whereas neonates with early onset epilepsy typically require long-term antiseizure medications, the ideal duration of antiseizure therapy for newborns with acute symptomatic seizures is not known. The rationale for continued antiseizure medication is to prevent seizure recurrence. Yet, the risk of postneonatal epilepsy in the first several years of life is less than 25% for neonates with acute symptomatic seizures and postneonatal epilepsies (particularly West Syndrome) do not necessarily respond to the medications prescribed in the neonatal period.^13–15

Although some sites in our study discontinued antiseizure medications in most or all neonates with acute symptomatic seizures before sending the infant home, the majority of neonates were discharged home on antiseizure medications, including those with a low seizure burden or even those without confirmed electrographic seizures on study center cEEG recordings. Our results highlight the pressing need for rigorous study regarding optimal treatment duration for neonates with acute symptomatic seizures.

Glossary

cEEG

Continuous video-EEG

EEG

Electroencephalogram

PERF

Pediatric Epilepsy Research Foundation

Appendix

Additional Neonatal Seizure Registry Study Group members include:

Ann Marie Bergin, MB, ScM, MRCP (UK), Boston Children’s Hospital and Harvard Medical School, Boston, MA; Dennis Dlugos, MD, MSCE, Children’s Hospital of Philadelphia and University of Pennsylvania School of Medicine, Philadelphia, PA; Donna M. Ferriero, MD, MS, University of California- San Francisco, Benioff Children’s Hospital and University of California San Francisco School of Medicine, San Francisco, CA; Faye Silverstein, MD, University of Michigan, Ann Arbor, MI; Kevin Staley, MD, Massachusetts General Hospital and Harvard Medical School, Boston, MA.