Anticonvulsants for neonates with seizures

Skip sharing on social media links
Share this:

Authors

Booth D, Evans DJ

Background - Methods - Results - Characetristics of Included Studies - References - Data Tables & Graphs


Dates

Date edited: 24/08/2004
Date of last substantive update: 30/03/2004
Date of last minor update: / /
Date next stage expected / /
Protocol first published: Issue 2, 2003
Review first published: Issue 4, 2004

Contact reviewer

Dr David Booth

Specialist Registrar
Peter Congdon Neonatal Unit
Clarendon Wing, Leeds General Infirmary
Leeds
West Yorkshire UK
LS2 9NS
Telephone 1: +44 113 392 7166
Telephone 2: +44 113 243 2799
Facsimile: +44 113 392 6068

E-mail: drdbooth@btopenworld.com

Contribution of reviewers

  • None noted.

Sources of Support

Internal sources of support

  • None noted.

External sources of support

  • None noted.

What's new

Date / Event Description

History

Date / Event Description

Synopsis

  • Synopsis pending

Abstract

Background

Neonatal seizures are a common problem and most neonates with seizures are treated with anticonvulsants. There is wide variation in clinical practice in both diagnosis and treatment of such seizures and this reflects the lack of clear evidence of the relative benefit and harm of the anticonvulsants used. The routine use of anticonvulsants to treat seizures in neonates needs to be evaluated.

Objectives

To assess and compare (with respect to benefits and harm) different anticonvulsants administered to neonates for the treatmentof established seizures.

Search strategy

Relevant randomised controlled trials were identified using a combination of electronic database searches (MEDLINE 1966 - March 2004, EMBASE 1980 - March 2004), the Cochrane Central Register of Controlled Trials (CENTRAL, The Cochrane Library, Issue 1, 2004) and hand searches. Identification of ongoing or unpublished trials was attempted by contacting prominent authors in the field and searching electronic registers of ongoing trials.

Selection criteria

All randomised or quasi-randomised controlled clinical trials with reported data comparing the following outcomes: mortality, neurodevelopmental disability, need for additional anticonvulsants, need for maintenance anticonvulsants at discharge and adverse events (hypotension requiring volume or inotropic support, arrhythmia, respiratory depression, hepatotoxicity) in neonates treated for seizures with systemic anticonvulsants compared to placebo, no drug or alternative anticonvulsants.

Data collection & analysis

Methodological quality and validity were assessed without consideration of the results. The first reviewer screened the title and abstracts of studies identified by the above search strategy. Full text versions of studies of potential relevance were re-screened by both reviewers. Studies meeting the pre-specified inclusion criteria were included. Relevant data were extracted and analysed separately and any disagreements were resolved by discussion.

Main results

Only two randomised controlled trials published in full could be identified. Painter 1999 showed that both of the two most commonly used anticonvulsants (phenobarbital and phenytoin) were similarly effective (RR 1.03 95% CI 0.96 to 1.62), controlling seizures in less than fifty percent of infants. Painter 1999 did not report mortality or neurodevelopmental outcome. Boylan 2004 randomised infants who failed to respond to phenobarbital to receive either lidocaine or midazolam as second-line agents. There was a trend for lidocaine to be more effective in reducing seizure burden (RR 0.40 95% CI 0.14 to 1.17) but both groups had similarly poor long term outcomes assessed at one year.

Reviewers' conclusions

At present there is little evidence from randomised controlled trials to support the use of any of the anticonvulsants currently used in the neonatal period. In the literature, there remains a body of opinion that seizures should be treated because of the concern that seizures in themselves may be harmful, although this is only supported by relatively low grade evidence (Levene 2002; Massingale 1993).

Development of safe and effective treatment strategies relies on future studies of high quality (randomised controlled trials with methodology that assures validity) and of sufficient size to have the power to detect clinically important reductions in mortality and severe neurodevelopmental disability in addition to any short term reduction in seizure burden.

[top]

Background

Neonatal seizures are common: estimates of the incidence of clinical seizures in term infants range from 0.7 to 2.7 per 1000 live births and from 57.5 to 132 per 1000 live births in premature infants (Evans 1998). The incidence of clinically silent (electrographic) seizures is unknown. Seizures represent a manifestation of neuronal compromise from a wide variety of etiologies. Seizures can substantially increase CNS metabolic demand (Younkin 1986), cause the release of excitatory neurotransmitters such as glutamate (McDonald 1990), lead to fluctuations in systemic arterial pressure (Clozel 1985) and result in hypoxia and hypercapnia. Thus seizures might themselves cause further neuronal injury although the effects of seizures on brain development are difficult to differentiate from the effects of the brain lesions causing the seizures. The prognosis following neonatal seizures is therefore varied. Mortality rates of 21% - 58% have been reported following neonatal seizures (Andre 1988; Scher 1993; Watkins 1988). There is also evidence that neonatal seizures have an adverse effect on neurodevelopmental progression and may predispose to cognitive, behavioural or epileptic complications later in life (Temple 1995; Levene 2002). The natural history of neonatal seizures is unknown but observations suggest that they may abate regardless of intervention (Painter 1999b). However, there is general consensus that neonatal seizures should be treated, particularly if they are frequent or prolonged.

The treatment of neonatal seizures has changed little over the last 50 years. Most neonates who have seizures are treated and the most common treatments are either phenobarbital or phenytoin. The efficacy of phenobarbital is unproven and is often ineffective as a first line anticonvulsant in neonates (Boylan 2002). There is no evidence that administering anticonvulsants to prevent seizures following hypoxic-ischaemic injury in term newborns is beneficial (Evans 2004).

Anticonvulsants may produce side effects (Wallace 1996). Most anticonvulsants, in particular benzodiazepines, will cause sedation and if given in sufficient dose, they will reduce respiratory effort and the infant may require ventilatory support (Ng 2002). Phenytoin, lidocaine and, to some extent, phenobarbital impair myocardial function and can cause hypotension or arrhythmia. A well recognised side effect of sodium valproate is its hepatotoxicity (Painter 2001).

When undertaking a review on the treatment of seizures, two major problems exist. The first is the definition of what constitutes a seizure requiring treatment and the second is the heterogeneity of etiology. There is poor concordance between clinical and electrical (EEG) evidence of seizure activity (Mizrahi 1987). Many studies use different methods of seizure recognition (e.g. clinical surveillance, EEG detection following initial clinical suspicion, continuous EEG monitoring). This leads to heterogeneity of seizure definition and a difference in severity of treated seizures between studies. Secondly, the etiology of the seizure, rather than anticonvulsant therapy, may have the greater bearing upon the outcome.

In this review we have adopted an inclusive approach to the definition and etiology of seizures. We will include neonates with either clinical or electrographic seizures, or both. We will attempt, where possible, to categorise the results according to seizure aetiology.

Objectives

Primary objectives:

In neonates suffering from seizures, does anticonvulsant therapy (compared to placebo or no anticonvulsant therapy) reduce neonatal mortality and long-term neurodevelopmental disability?

In neonates suffering from seizures, does one particular anticonvulsant (compared to an alternative anticonvulsant) reduce neonatal mortality and long-term neurodevelopmental disability?

Secondary objectives include need for additional anticonvulsants to gain seizure control during the neonatal period, proportion of infants discharged home on maintenance anticonvulsants, and an assessment of the adverse effects of anticonvulsants (see outcome criteria below).

Subgroup analyses are planned on the basis of:

  1. Gestation (preterm < 37 weeks, term less than/or equal to37 weeks)
  2. Method of seizure detection (clinical alone, EEG alone, clinical and EEG)
  3. Etiology of seizure (hypoxia-ischaemia, infective, metabolic, malformation)
  4. Class of anticonvulsant (barbiturate, benzodiazepines, other).

Criteria for considering studies for this review

Types of studies

Published and unpublished controlled trials utilizing either random or quasi-random patient allocation.

Types of participants

Neonates with clinical and/or electrographic seizures, commencing within the first 28 days of life.

Types of interventions

Trials comparing systemic anticonvulsant with placebo, no drug or alternative anticonvulsants.
Anticonvulsants to be considered: barbiturates (phenobarbital, thiopental), benzodiazepines (midazolam, clonazepam, diazepam, lorazepam), phenytoin (including fosphenytoin), lidocaine, primidone, paraldehyde, sodium valproate, carbamazepine, lamotrigine.

Types of outcome measures

Studies were considered if they reported one or more of the following outcomes:

Primary outcomes:
  1. death within first 28 days and within the first year of life.
  2. significant neurodevelopmental impairment (any one or combination of: cerebral palsy, developmental delay DQ > 2SD, blindness) assessed at 1-2 years of age.
  3. death or significant neurodevelopmental impairment (any one or combination of: cerebral palsy, developmental delay DQ > 2SD, blindness) assessed at 1-2 years of age.
Secondary outcomes:
  1. proportion of infants who required additional anticonvulsants during the neonatal period.
  2. proportion of infants discharged home on maintenance anticonvulsants.
  3. adverse effects of anticonvulsant therapy:
    1. hypotension requiring volume or inotropic support,
    2. arrhythmia causing circulatory disturbance,
    3. respiratory depression requiring ventilatory support,
    4. (iv) hepatotoxicity resulting in discontinuation of therapy.

Search strategy for identification of studies

The standard search strategy of the Cochrane Neonatal Review Group was used. This included electronic searches of the Cochrane Central Register of Controlled Trials (CENTRAL, The Cochrane Library, Issue 1, 2004), MEDLINE (1966-March 2004) and EMBASE (1980-March 2004). No language restriction was applied.

The electronic search strategy involved the following keywords, using the search fields of abstract, MeSH subject heading, exploded subject heading, subject heading word, text word and title: explode "Infant-Newborn/ all subheadings" OR Neonat* AND explode "Seizures"/ all subheadings OR Convuls* OR Seizur* AND explode "Anticonvulsants"/ all subheadings OR Anticonvul* OR [MeSH term for the particular anticonvulsant (exploded)] OR [textword terms for particular anticonvulsant]. The MEDLINE and EMBASE searches were assessed with and without combination with the search filter for randomised controlled trials (see Appendix 5c of the Cochrane Handbook).

1. MEDLINE Search:

Dates: 1966 - March 2004
Strategy: example phenobarbital

  1. infant, newborn
  2. neonat$.mp.
  3. #1 or #2
  4. exp SEIZURES/ or seizures.mp
  5. convuls$.mp.
  6. seizur$.mp
  7. #4 or #5 or #6
  8. #3 and #7
  9. anticonvulsants.mp. or exp ANTICONVULSANTS
  10. anticonvul$.mp
  11. barbiturate.mp or exp BARBITURATES
  12. phenorbarbitone.mp or exp phenobarbital
  13. #9 or #10 or #11 or #12
  14. #8 and #13

2. EMBASE Search

Dates: 1980 to March 2004
Strategy: example phenobarbital

  1. exp newborn
  2. neonat$.mp.
  3. #1 or #2
  4. seizures.mp. or exp Seizure
  5. convuls$.mp.
  6. seizur$.mp.
  7. #4 or #5 or #6
  8. anticonvulsants.mp or exp Anticonvulsive agent
  9. anticonvul$.mp
  10. barbiturates.mp or exp Barbituric Acid Derivative
  11. phenobarbitone.mp.
  12. #8 or #9 or #10 or #11
  13. #3 and #7 and #12
  14. limit 13 to infant < to one year>

3. Cochrane Controlled Trials Registry Search

Date: Issue 1 2004
Strategy: example phenobarbital

  1. (infant next newborn)
  2. neonat*
  3. (#1 or #2)
  4. seizures
  5. convuls*
  6. seizur*
  7. ((#4 or #5) or #6)
  8. anticonvulsants
  9. anticonvuls*
  10. barbiturates
  11. phenobarbit*
  12. (((#8 or #9) or #10) or #11)
  13. ((#3 and #7) and #12)

4. International registers of ongoing clinical trials: internet-based trials registers were searched for ongoing trials. The registers searched were:
Trials Central (www.trialscentral.org) which is a database of clinical trials registers that gives access to Current Controlled Trials, made up of two registers, IsrcTN (a database of randomised controlled trials with an international randomised controlled trial number) and mRCT (metaRegister of controlled trials), a database combining registers of ongoing randomised controlled trials in all areas of healthcare. Trials Central also links to the Australasian Perinatal Trials Registry.
These registers were searched using keywords neonatal, perinatal, seizure and anticonvulsant.

5. Personal communication with authors of identified trials was undertaken in order to identify any trials, ongoing or unpublished that may not have been identified by the outlined search strategies.

[top]

Methods of the review

The first reviewer screened the title and abstract of studies identified by the above search strategy. Both reviewers re-screened the full text of the report of each study identified as of potential relevance. Studies meeting any of the pre-specified inclusion criteria were included.

The criteria and standard methods of the Cochrane Neonatal Review Group were used to assess the methodological quality of the included trials. Quality of the trials included was evaluated in terms of allocation concealment, blinding of intervention, completeness of assessment in all randomised individuals and blinding of outcome assessment. The methodological quality of these components of the study design were scored as follows:

Selection bias (Allocation concealment - Section 6 of the Cochrane Handbook):
A = adequate, B = can't tell / unclear, C = not used / inadequate.

Performance bias (blinding of caretakers to the intervention):
Adequate; Can't tell / unclear; not used / inadequate.

Attrition bias (post-randomisation exclusions and loss to follow-up):
Where possible, the number of losses will be expressed as a percentage of the number assigned to each group

Detection bias (blinding of outcome assessment):
Adequate (where the outcome assessor was unaware, or was unlikely to correctly identify, treatment allocation); Can't tell / unclear; Not used / inadequate (includes studies where effects, or side effects, of treatment meant that it was likely that the treatment allocation could be correctly identified in a significant proportion (>20%) of participants).

A data collection form was used to aid extraction of the relevant information and data. Each reviewer extracted the data separately, compared data and resolved differences by consensus after discussion.

The treatment effects of individual trials were examined by comparing groups allocated to the treatment under study (anticonvulsant therapy versus placebo or no anticonvulsant or therapy with alternative anticonvulsant(s)). Data relating to the primary and secondary outcomes (described above) were compared. Where relevant and if possible, relative risk (RR) and risk difference (RD) were calculated for dichotomous data and weighted mean difference (WMD) for continuous data, with 95% confidence intervals (CI) for all analyses. The number needed to treat (NNT) and associated 95% CI were determined for a statistically significant reduction in the RD. Analysis of outcome data was by intention to treat. Where relevant, meta-analyses was performed using the fixed effects model. Where relevant and if possible, heterogeneity between trial results was examined using the I 2test for dichotomous outcomes and ANOVA for continuous outcomes.

Description of studies

The search strategies as outlined yielded a large number of studies, the actual number varying with each anticonvulsant.
Following screening of the titles and abstracts, five studies of potential relevance were identified and were considered for further assessment.

Two of these studies were excluded. Hall 1998 randomized term infants with severe asphyxia to receive 40mg/kg phenobarbital or placebo to assess the effect on the incidence of seizures and adverse neurological outcome at three years of age. This was a study of seizure prevention and not treatment. Ichiba 2002 randomized term infants with severe asphyxia to receive 250mg/kg/day magnesium sulphate or nothing for three days following birth. Although the presence of seizures was an inclusion criterion, the group with seizures was not randomized or reported separately and data extraction relating solely to this group was not possible.

  1. Anticonvulsant compared to (no anticonvulsant or placebo)
    1. No eligible studies were identified
  2. Anticonvulsant compared to alternative anticonvulsant
  3. Three eligible studies were identified.

Painter 1999b compared phenobarbital with phenytoin for the treatment of neonatal seizures. Infants on the neonatal unit with one or more risk factors for seizures were enrolled and subsequently screened by electroencephalography. If screening met specific criteria for epileptiform activity, the infants were randomly assigned to receive either phenobarbital or phenytoin. Infants with clinical seizures only (without electroencephalographic evidence of epileptiform activity) were not randomised. 30 infants were assigned to phenobarbital and 29 to phenytoin. In both treatment groups, one third were premature infants and two thirds were term infants. The primary endpoint was complete control of seizures, determined electroencephalographically, during treatment with one drug or after addition of the second drug. Treatment failure was defined as the occurrence of any electrical seizure after target plasma concentrations of free drug had been achieved. Crossover to the second drug was initiated if continuous seizure activity occurred for greater than 2.5 minutes or if there was 2.5 minutes of seizure activity within any 5 minute period. No adverse events were noted with either anticonvulsant.

Boylan 2004 compared benzodiazepines and lidocaine for second-line treatment of neonatal seizures in a video-EEG monitoring study. Subjects were randomly assigned to receive either midazolam (3 infants) or lidocaine (5 infants) after treatment failure with phenobarbitone. Failure to control seizures within 12 hours of second line treatment was followed by an increase in the dose of the allocated drug. The primary endpoint was control of electroencephalographic seizures (either complete absence or 80% reduction of pre-treatment burden). Neurodevelopmental outcome was assessed at 1 year. No adverse events were reported during treatment with any anticonvulsant.

Wilkinson 1989 compared four anticonvulsants in a randomised control trial. Electroencephalographically confirmed seizures of any aetiology in a total of 40 term and premature neonates were treated with either phenobarbitone, phenytoin, clonazepam or sodium valproate. The study concentrates on short term neonatal outcomes and adverse events. This study has been published in abstract form only and full methodological assessment awaits further correspondence from the authors. When the requested additional information is provided, the data from this trial will be considered for inclusion in a future update of this review.

Methodological quality of included studies

Selection

The study by Painter 1999b ensured balanced treatment assignment over time by stratifying subjects with respect to race and gestational age prior to randomisation within strata. Physicians, hospital staff and EEG technicians were aware of treatment assignments but it is unclear whether allocation concealment occurred.

The study by Boylan 2004 treated all infants with EEG confirmed seizures with phenobarbitone; if this failed to abolish seizures within 12 hours, the infants were randomised to receive either midazolam or lidocaine as a second line anticonvulsant. Consent was requested before randomisation to second-line anticonvulsant. Concealment of allocation to treatment is unclear.

If parents were not willing to allow treatment based on a random choice of drug, clonazepam was used instead of either of the trial drugs and these infants were not included in the analysis.

Performance

The study by Painter 1999b was single blind (caretaker blinding did not occur). Similarly, in the Boylan 2004 study, caretaker blinding did not occur.

Attrition

There was no post-randomisation loss in Painter 1999b. In Boylan 2004 consent was obtained prior to randomisation; the parents of 3 of 11 infants eligible for randomisation to second-line anticonvulsants withdrew consent prior to randomisation to trial drug. There were no subsequent (post-randomization) losses.

Detection

The Painter 1999b study does not state who assessed outcomes, therefore it is unclear whether the outcome assessment was blinded. In the Boylan 2004 study, it is unclear whether blinding of outcome assessment (either for seizure control or neurodevelopmental assessment at 1 year) took place.

[top]

Results

Anticonvulsant compared to (no anticonvulsant or placebo)

No eligible studies were identified

Anticonvulsant compared to alternative anticonvulsant

Phenobarbital vs Phenytoin

Primary Outcomes:
  1. Death within 1st 28 days and within 1st year of life (01.01)
    This is not assessed in Painter 1999b.
  2. Significant neurodevelopmental impairment assessed at 1 -2 years of age (01.02)
    This is not assessed in Painter 1999b.
  3. Death or significant neurodevelopmental impairment at 1 - 2 years of age (01.03)
    This is not assessed in Painter 1999b
Secondary Outcomes:
  1. Proportion of infants who require additional anticonvulsants during the neonatal period. (01.04)
    In Painter 1999b, treatment failure with either phenobarbital or phenytoin as first-line anticonvulsant led to use of the other anticonvulsant (as add-on therapy) as part of the study protocol. As such, treatment failure occurred in 17 of the 30 infants initially treated with phenobarbital and 15 of these 17 went on to add-on treatment with phenytoin. Of the 29 randomly assigned to phenytoin, treatment failure occurred in 16 and 13 of these 16 had add-on treatment with phenobarbital. Relative risk for use of additional anticonvulsant (phenobarbital v phenytoin) 1.12 (95% CI 0.65, -1.91).
  2. Proportion of infants discharged home on maintenance anticonvulsants (01.05)
    Painter 1999b does not report this outcome
  3. Adverse effects of anticonvulsant therapy (01.06)
    Painter 1999b reported no hypotension, arrhythmia or respiratory depression attributable to anticonvulsant therapy. Hepatotoxicity is not reported.
  4. Failure of seizure control (01.07)
    Although not an a priori outcome sought in this review, it was a major outcome measure in the included studies.
    Painter 1999b defined seizure control as complete control of seizures as determined electroencephalographically. An 80% reduction in the severity of seizures (calculated as mean severity per hour) as compared to seizure severity in the period prior to initial drug administration, was arbitrarily defined as 'successful treatment'. Phenobarbital controlled seizures in 13 of 30 neonates (43%). Phenytoin controlled seizures in 13 of 29 neonates (45%). RR for failure of seizure control (phenobarbital v phenytoin) 1.03 (95% CI 0.65-1.62).

Subsequent administration of phenytoin to the phenobarbital non-responders resulted in seizure control in an additional 4 subjects. Administration of phenobarbital to the phenytoin non-responders resulted in seizure control in an additional 5 subjects.

In addition to complete seizure control, the authors describe successful treatment (80% reduction in seizure activity) in a further 7 infants initially assigned to phenobarbital and in a further 3 infants initially assigned to phenytoin.

The authors report that the electroencephalographically assessed severity of seizure activity prior to treatment was strongly inversely related to the successful control of seizures. Severity was measured in channel-seconds per hour; 3000 or less was defined as mild, 20000 or more as severe and between these two as moderate. 10% of infants with a severe seizure burden responded to treatment, whereas 88% with mild burden responded. The initial drug treatment did not affect this association.

Midazolam vs Lidocaine after treatment failure with phenobarbital

Primary Outcomes:
  1. Death within 1st 28 days and within 1st year of life (02.01)
    Boylan 2004 reports 3 deaths in the 8 infants randomised to second-line anticonvulsants (1 death in the 3 receiving midazolam and 2 deaths in the 5 receiving lidocaine). Relative risk for death for midazolam v lidocaine 0.83 (95% CI 0.12-5.72)
  2. Significant neurodevelopmental impairment assessed at 1 -2 years of age (02.02)
    Boylan 2004 assessed all surviving neonates using Amiel-Tison and Griffiths neurodevelopmental assessments at 1 year of age. Of the 5 surviving neonates, 3 had severe, 1 had moderate and 1 had mild impairment. Definitions of mild, moderate and severe impairment are not stated. Relative risk of neurological impairment - all grades - (midazolam v lidocaine) 1.11 (95% CI 0.38-3.25).
  3. Death or significant neurodevelopmental impairment at 1 - 2 years of age (02.03)
    Results as stated above
Secondary Outcomes:
  1. Proportion of infants who require additional anticonvulsants during the neonatal period. (02.04)
    In Boylan 2002 all infants were treated with phenobarbital if they had electroencephalographic evidence of seizures. Treatment failure was followed by randomisation to receive either lidocaine or midazolam as second-line anticonvulsant therapy. Of the 22 neonates enrolled, 11 responded to phenobarbital. Of the 11 non-responders, 8 were randomised to receive second-line anticonvulsant (5 to lidocaine and 3 to midazolam). 3 of the 5 responded to lidocaine. 0 of 3 responded to midazolam. RR for treatment failure (midazolam v lidocaine) 2.50 (95% CI 0.85-7.31).
  2. Proportion of infants discharged home on maintenance anticonvulsants (02.05)
    Boylan 2004 does not report this outcome
  3. Adverse effects of anticonvulsant therapy (02.06)
    Boylan 2004 does not report assessment of adverse events.
  4. Failure of seizure control (02.07)
    Although not an a priori outcome sought in this review, it was a major outcome measure in the included studies.
    Boylan 2004 defined seizure control as complete cessation of electroencephalographic seizure activity and 'major improvement' as 80% or more decrease in pre-treatment seizure burden. Of the 8 infants randomised to second-line anticonvulsants, 5 received lidocaine; 2 of these became seizure-free and 1 showed major improvement. Of the 3 receiving midazolam, none responded. RR for failure of seizure control (midazolam v lidocaine) 1.67 (95% CI 0.81-3.41).

Discussion

The heterogeneity of seizure definition, the differences in severity of treated seizures between studies and the fact that the etiology of the seizure, rather than anticonvulsant therapy, may have a greater bearing upon the eventual outcome, all contribute to the difficulties in assessing studies of the treatment of neonatal seizures.

Despite an inclusive approach to definitions and etiology, we found no data available from randomised controlled trials comparing anticonvulsant to placebo or no treatment that address the important question of whether attempting to control neonatal seizures with any anticonvulsant has any measurable impact on the most important outcomes, namely mortality or neurodevelopmental impairment.

From studies comparing anticonvulsants, there are insufficient data available to support the use of one anticonvulsant over another.

In the literature, there remains a body of opinion that seizures should be treated because of the concern that seizures in themselves may be harmful, although this is only supported by relatively low-grade evidence (Levene 2002; Massingale 1993).

Reviewers' conclusions

Implications for practice

There is no evidence from randomised controlled trials that either supports or fails to support the use of anticonvulsant therapy for the treatment of seizures in neonates. Data from randomised controlled trials to support the choice of anticonvulsant are limited, and no definite recommendations based on available data can be made.

Implications for research

The majority of neonatal seizures in neonates occur against a background of pre-existing brain damage. The question of whether seizures per se are harmful to the developing brain is an extremely important question that remains largely unanswered.

It is possible that, against the background of pre-existing brain damage, the additional adverse effects of seizures on the important outcomes of death and neurodevelopmental impairment will be small. Therefore, the question of whether treatment of such seizures confers benefit will require further randomised controlled trials of sufficient sizeto have the power to detect changes in the rates of clinically important outcomes, such as mortality and severe neurodevelopmental disability.

The two studies assessed here use specific electroencephalographic (EEG) criteria in defining seizures; future studies must use EEG or at the very least amplitude-integrated EEG criteria and should not rely on clinical seizure detection alone. The heterogeneity of seizure aetiology in this population should be addressed in study design since treatment effects for one aetiological group cannot be assumed to be applicable to other groups.

It would be preferable to stratify randomization by seizure etiology (although this is difficult, as many etiologies are not apparent at the time of randomization). Nevertheless, we suggest that infants should be stratified according to gestation (preterm versus term) and where an hypoxic-ischaemic etiology is suspected.

We feel that the lack of evidence with regard to seizure treatment justifies future study using placebo comparison. If investigators do not find this approach ethically acceptable, alternatives may be to use high treatment threshold versus low treatment thresholds or high dose versus low dose anticonvulsant strategies.

The issue of adverse effects of drug treatments should be given prominence in future studies, with adverse effects clearly defined at the outset and clearly stated in the published reports.

Acknowledgements

  • None noted.

Potential conflict of interest

  • None noted.

[top]

Characteristics of studies

Characteristics of Included Studies

Study Methods Participants Interventions Outcomes Notes Allocation concealment
Boylan 2004 A randomized trial comparing second-line anticonvulsants.
Blinding of allocation: can't tell
Blinding of intervention: no
Post-randomization losses: 3 of the 11 infants requiring second-line treatment were not assigned to either midazolam or lidocaine because of parental objection to random drug assignment. Of the 8 infants randomized, there were no post-randomization losses.
Blinding of outcome: no
8 term and preterm infants with electrographically confirmed seizures not adequately controlled with phenobarbital within 12 hours of enrolment. Seizures were defined as sudden, evolving stereotyped forms with a definite beginning, middle and end lasting at least 10 seconds. All infants with electrographically confirmed seizures were treated with phenobarbital in a dose of 40mg/kg. If this failed to abolish or reduce seizure burden by 80% the infants were randomized to treatment with either midazolam (N=3) (bolus 60 micrograms per kg followed by infusion of 150 micrograms per kg per hour, increased to 300 micrograms per kilogram per hour if no response in 12 hours) or lidocaine (N=5) (bolus 4mg/kg followed by infusion of 2mg per kg per hour, increased to 4mg per kg per hour if no response in 12 hours). Complete control of seizures, determined electroencephalographically (defined as abolition of seizure activity or reduction by 80% of pre-treatment seizure burden).
Neurodevelopmental outcome assessed at 1 year using the Amiel-Tison Test and the Griffiths developmental scale for babies, plus a neurological examination.
Both study centres routinely used midazolam as a sedative at a dose of 30 to 60 micrograms per kilogram per hour. B
Painter 1999a Randomised controlled trial.
Allocation concealment: can't tell
Blinding of intervention: no
Blinding of outcome: no
Post-randomization losses: none
59 term and preterm infants with electrographically diagnosed seizures. A seizure was defined as an episode lasting at least 10 seconds and consisting of abnormal repetitive electrical discharges with demonstrable onset, waveform and amplitude. Participants were randomized to receive either phenobarbital (N =30) or phenytoin (N=29) once daily at a dose calculated to achieve plasma concentrations of free drug of 25 and 3 micrograms per millilitre respectively. Dose adjustments were made to achieve trough levels of at least 22.5 micrograms per millilitre for phenobarbital and 2.5 micrograms per millilitre for phenytoin. If treatment with one drug failed, the second drug was added. Complete control of seizures, determined electroencephalographically, plus an arbitrarily defined success end point of 80% reduction in severity of seizures. Short term (neonatal) outcomes B

Characteristics of excluded studies

Study Reason for exclusion
Hall 1998 This trial studied the effect of phenobarbitone on the prevention (rather than treatment) of seizures and the neurologic outcome in term infants with severe asphyxia. The presence of seizures was not a pre-requisite for randomisation.
Ichiba 2002 The presence of seizures was one of the eligibility criteria for this study of term infants with severe birth asphyxia. However, the group of infants with seizures were not randomised or reported separately from the group of subjects as a whole. As such, extraction of data relating only to infants with seizures was not possible. In addition, the treatment (magnesium sulphate) was not part of the a priori list of anticonvulsants studied in this review.

[top]

References to studies

Included studies

Boylan 2004

{published data only}

* Boylan GB, Rennie JM, Chorley G, Pressler RM, Fox GF, Farrer K, Morton M, Binnie CD. Second line anticonvulsant treatment of neonatal seizures: a video-EEG monitoring study. Neurology 2004;62:486-8.

Rennie JM, Boylan GB, Chorley G, Pressler R, Fox GF, Farrer K, Morton M, Binnie C. Second-line anticonvulsant treatment of neonatal seizures: an open comparative study using video-EEG monitoring. Early Human Development 2003;71:92.

Painter 1999a

{published data only}

* Painter MJ, Scher MS, Stein AD, Armatti S, Wang Z, Gardiner JC, Paneth N, Minnigh B, Alvin J. Phenobarbital compared with phenytoin for the treatment of neonatal seizures. New England Journal of Medicine 1999;341:485-9.

Scher MS, Alvin J, Gaus L, Minnigh B, Painter MJ. Uncoupling of EEG-clinical neonatal seizures after antiepileptic drug use. Pediatric Neurology 2003;28:277-80.

Excluded studies

Hall 1998

{published data only}

Hall RT, Hall FK, Daily DK. High dose phenobarbital therapy in term newborn infants with severe perinatal asphyxia: a randomized, prospective study with three-year follow-up. Journal of Pediatrics 1998;132:345-8.

Ichiba 2002

{published data only}

Ichiba H, Tamai H, Negishi H, Ueda T, Kim TJ, Sumida Y, Takahashi Y, Fujinaga H, Minami H. Randomized controlled trial of magnesium sulfate infusion for severe birth asphyxia. Pediatrics International 2002;44:505-9.

References to studies awaiting assessment

Wilkinson 1989

{published data only}

Wilkinson AR, Rochefort MJ. Phenytoin reduces frequency and duration of neonatal seizures in the newborn: a randomised trial of four anticonvulsants. Pediatric Research 1989;26:522.

* indicates the primary reference for the study

Other references

Additional references

Andre 1988

Andre M, Matisse N, Vert P, Debruille C. Neonatal seizures - recent aspects. Neuropediatrics 1988;19:201-7.

Boylan 2002

Boylan BG, Rennie JM, Pressler RM, Wilson G, Morton M, Binnie CD. Phenobarbitone, neonatal seizures and video-EEG. Archives of Disease in Childhood Fetal and Neonatal Edition 2002;86:F165-70.

Clozel 1985

Clozel M, Daval JL, Monin P, Debruc C, Morselli PL, Vert P. Regional cerebral blood flow during bicuculline-induced seizures in the newborn piglet: effect of phenobarbital. Developmental Pharmacology and Therapeutics 1985;8:189-99.

Evans 1998

Evans D, Levene M. Neonatal Seizures [Neonatal Seizures]. Archives of Disease in Childhood Fetal and Neonatal Edition 1998;78:F70-5.

Evans 2004

Evans DJ, Levene MI. Anticonvulsants for preventing mortality and morbidity in full term newborns with perinatal asphyxia (Cochrane Review). In: The Cochrane Library, Issue 3, 2004. Chichester, UK: John Wiley & Sons, Ltd.

Levene 2002

Levene M. The clinical conundrum of neonatal seizures. Archives of Disease in Childhood Fetal and Neonatal Edition 2002;86:F75-7.

Massingale 1993

Massingale TW, Buttross S. Survey of treatment practices for neonatal seizures. Journal of Perinatology 1993;13:107-10.

McDonald 1990

McDonald JW, Johnston MV. Physiological and pathophysiological roles of excitatory amino acids during central nervous system development. Brain research. Brain research reviews 1990;15:41-70.

Mizrahi 1987

Mizrahi EM, Kellaway P. Characterization and classification of neonatal seizures. Neurology 1987;37:1837-44.

Ng 2002

Ng E, Klinger G, Shah V, Taddio A. Safety of benzodiazepines in newborns. Annals of Pharamacotherapy 2002;36:1150-5.

Painter 1999b

Painter MJ, Scher MS, Stein AD, Armatti S, Wang Z, Gardiner JC, et al. Phenobarbital compared with phenytoin for the treatment of neonatal seizures. New England Journal of Medicine 1999;341:485-9.

Painter 2001

Painter MJ, Alvin J. Neonatal Seizures. Current Treatment Options in Neurology 2001;3:237-48.

Scher 1993

Scher MS, Aso K, Beggarly ME, Hamid MY, Steppe DA, Painter MJ. Electrographic seizures in preterm and full-term neonates: clinical correlates, associated brain lesions, and risk for neurologic sequelae. Pediatrics 1993;91:128-34.

Temple 1995

Temple CM, Dennis J, Carney R, Sharich J. Neonatal seizures: long-term outcome and cognitive development among "normal" survivors. Developmental Medicine and Child Neurology 1995;37:109-18.

Wallace 1996

Wallace SJ. A comparative review of the adverse effects of anticonvulsants in children with epilepsy. Drug Safety 1996;15:378-93.

Watkins 1988

Watkins A, Szymonowicz W, Jin X, Yu VV. Significance of seizures in very low-birthweight infants. Developmental Medicine and Child Neurology 1988;30:162-9.

Younkin 1986

Younkin DP, Delivoria-Papadopoulos M, Maris J, Donlon E, Clancy R, Chance B. Cerebral metabolic effects of neonatal seizures measured with in vivo 31P NMR spectroscopy. Annals of Neurology 1986;20:513-9.

[top]

Comparisons and data

01 Phenobarbital versus phenytoin

Comparison or outcome Studies Participants Statistical method Effect size
01.01 Death within 1st 28 days and within 1st year of life 0 0 RR (fixed), 95% CI No numeric data
01.02 Neurodevelopmental impairment assessed at 1-2 years of age 0 0 RR (fixed), 95% CI No numeric data
01.03 Death or neurodevelopmental impairment at 1-2 years of age 0 0 RR (fixed), 95% CI No numeric data
01.04 Proportion of infants who require additional anticonvulsants during neonatal period 1 59 RR (fixed), 95% CI 1.12 [0.65, 1.91]
01.05 Proportion of infants discharged home on maintenance anticonvulsants 0 0 RR (fixed), 95% CI No numeric data
01.06 Adverse effects of anticonvulsant therapy 0 0 RR (fixed), 95% CI No numeric data
01.07 Failure of seizure control 1 59 RR (fixed), 95% CI 1.03 [0.65, 1.62]

02 Addition of midazolam vs lidocaine after treatment failure with phenobarbitone

Comparison or outcome Studies Participants Statistical method Effect size
02.01 Death within first 28 days of life or death within 1st year of life 1 8 RR (fixed), 95% CI 0.83 [0.12, 5.72]
02.02 Neurodevelopmental impairment assessed at 1-2 years of age 1 8 RR (fixed), 95% CI 1.11 [0.38, 3.25]
02.03 Death or significant neurodevelopmental impairment at 1-2 years of age 0 0 RR (fixed), 95% CI No numeric data
02.04 Proportion of infants who require additional anticonvulsants during the neonatal period 1 8 RR (fixed), 95% CI 2.50 [0.85, 7.31]
02.05 Proportion of infants discharged home on maintenance anticonvulsants 0 0 RR (fixed), 95% CI No numeric data
02.06 Adverse effects of anticonvulsant therapy 0 0 RR (fixed), 95% CI No numeric data
02.07 Failure of seizure control 1 8 RR (fixed), 95% CI 1.67 [0.81, 3.41]

Additional tables

  • None noted.

Amended sections

Cover sheet
Synopsis
Abstract
Background
Objectives
Criteria for considering studies for this review
Search strategy for identification of studies
Methods of the review
Description of studies
Methodological quality of included studies
Results
Discussion
Reviewers' conclusions
References to studies
Other references
Characteristics of Included Studies
Characteristics of excluded studies
Characteristics of ongoing studies
Comparisons, data or analyses

Contact details for co-reviewers

Dr David J Evans

Consultant Neonatologist
Neonatal Intensive Care Unit
Southmead Hospital
Neonatal Intensive Care Unit
Southmead Hospital
Bristol
UK
BS10 5NB
Telephone 1: +44 117 959 5326
Telephone 2: +44 117 959 5325
Facsimile: +44 117 959 5324

E-mail: dj-evans@doctors.org.uk

Secondary contact person's name: Prof M.I. Levene


This review is published as a Cochrane review in The Cochrane Library, Issue 4, 2004 (see http://www.thecochranelibrary.com External Web Site Policy for information). Cochrane reviews are regularly updated as new evidence emerges and in response to comments and criticisms, and The Cochrane Library should be consulted for the most recent version of the review.