Fármacos antiepilépticos para la prevención primaria y secundaria de las convulsiones después del accidente cerebrovascular
Resumen
Antecedentes
Las convulsiones después de un accidente cerebrovascular son un problema clínico importante y pueden dar lugar a desenlaces desfavorables. Las indicaciones de administrar fármacos antiepilépticos (FAE) para la profilaxis de las convulsiones después de un accidente cerebrovascular siguen sin estar claras.
Esta es una versión actualizada de la revisión Cochrane publicada anteriormente en 2014.
Objetivos
Evaluar los efectos de los FAE para la prevención primaria y secundaria de las convulsiones después de un accidente cerebrovascular. En cuanto a la prevención primaria, el objetivo fue evaluar si los FAE reducen la probabilidad de convulsiones en las personas que sufren un accidente cerebrovascular pero no presentan convulsiones. En cuanto a la prevención secundaria, el objetivo fue evaluar si los FAE reducen la probabilidad de que se produzcan nuevas convulsiones en las personas que sufren un accidente cerebrovascular y presentan al menos una convulsión posterior al mismo.
Métodos de búsqueda
El 9 de marzo de 2021 se realizaron búsquedas en las siguientes bases de datos: Registro Cochrane de Estudios (Cochrane Register of Studies) (CRS Web), MEDLINE (Ovid, 1946 al 8 de marzo de 2021). El CRS Web incluye ensayos controlados aleatorizados o cuasialeatorizados de PubMed, Embase, ClinicalTrials.gov, la Plataforma de registros internacionales de ensayos clínicos (ICTRP) de la Organización Mundial de la Salud, el Registro Cochrane central de ensayos controlados (Cochrane Central Register of Controlled Trials; CENTRAL) y los Registros especializados de los Grupos Cochrane de revisión que incluyen el Grupo Cochrane de Epilepsia (Cochrane Epilepsy) y el Grupo Cochrane de Accidentes cerebrovasculares (Cochrane Stroke). También se verificaron las listas de referencias de los artículos identificados a partir de las búsquedas.
Criterios de selección
Se seleccionaron los estudios controlados aleatorizados y cuasialeatorizados que reclutaron a participantes con un diagnóstico clínico de accidente cerebrovascular, ya fuera isquémico o hemorrágico. Se excluyeron los estudios que sólo reclutaron participantes con hemorragia subaracnoidea, hemorragia subdural, hemorragia extradural u otros diagnósticos no relacionados con el accidente cerebrovascular, como el infarto o la hemorragia relacionados con un tumor o una infección. También se excluyeron los estudios que sólo reclutaron a participantes sometidos a neurocirugía. Se incluyeron participantes de todas las edades que presentaban cualquier tipo de convulsión y se asignaron a los grupos de FAE o placebo.
Obtención y análisis de los datos
De acuerdo con los procedimientos metodológicos estándar previstos por la Colaboración Cochrane, dos autores de la revisión evaluaron de forma independiente los ensayos para su inclusión antes de evaluar el riesgo de sesgo de los mismos y extraer los datos relevantes. El desenlace principal evaluado fue la proporción de participantes que presentaron convulsiones en el período de seguimiento. Los resultados se presentaron como razones de riesgos (RR) globales con intervalos de confianza (IC) del 95% para los desenlaces dicotómicos y diferencias de medias (DM) con IC del 95% para los desenlaces continuos. Cuando se dispuso de datos suficientes, se calcularon metanálisis de efectos aleatorios (Mantel‐Haenszel) para los desenlaces dicotómicos; de lo contrario, los resultados se proporcionaron de manera narrativa. Se utilizó la estadística I2 para evaluar la heterogeneidad estadística. Se planificó el uso de gráficos de embudo para evaluar el sesgo de publicación en los metanálisis en los que se incluyeron al menos diez estudios. Para evaluar la certeza de la evidencia se utilizó el método GRADE.
Resultados principales
Se incluyeron dos estudios con un total de 856 participantes. Los FAE no mostraron ser eficaces en la profilaxis primaria de las convulsiones después del accidente cerebrovascular (RR 0,65; IC del 95%: 0,34 a 1,26; dos estudios, 856 participantes; evidencia de certeza moderada).
El primer estudio fue un estudio aleatorizado doble ciego que comparó el ácido valproico con placebo para la prevención primaria de las convulsiones hasta un año después del accidente cerebrovascular. En el estudio participaron 72 adultos con hemorragia intracerebral. No hubo diferencias en el riesgo de convulsiones después del accidente cerebrovascular (RR 0,88; IC del 95%: 0,35 a 2,16) ni de muerte (RR 1,20; IC del 95%: 0,40 a 3,58).
El segundo estudio fue un subestudio sobre el uso del diazepam en el accidente cerebrovascular agudo. Se trató de un estudio aleatorizado doble ciego, que comparó un tratamiento de tres días con diazepam versus placebo para la prevención primaria de las convulsiones hasta tres meses después del accidente cerebrovascular en 784 adultos con un accidente cerebrovascular agudo. No hubo evidencia de una diferencia en el riesgo de convulsiones posteriores al accidente cerebrovascular para todos los accidentes cerebrovasculares ni subgrupos de accidentes cerebrovasculares hemorrágicos o isquémicos (RR para todos los accidentes cerebrovasculares 0,47; IC del 95%: 0,18 a 1,22). En un análisis de subgrupos de infartos corticales de la circulación anterior, la profilaxis primaria con diazepam se asoció con un menor riesgo de convulsiones después del accidente cerebrovascular (RR 0,21; IC del 95%: 0,05 a 0,95). Los riesgos de mortalidad no difirieron entre el grupo de diazepam y el grupo placebo a las dos semanas (RR 0,84; IC del 95%: 0,56 a 1,26) ni a los tres meses de seguimiento (RR 0,95; IC del 95%: 0,72 a 1,26).
Se consideró que ambos estudios tenían un bajo riesgo general de sesgo. La certeza general de la evidencia se consideró baja a moderada con el uso del método GRADE.
Conclusiones de los autores
No hay evidencia suficiente que apoye el uso sistemático de FAE en la prevención primaria y secundaria de las convulsiones después del accidente cerebrovascular. El desarrollo de estudios bien realizados sobre este importante problema clínico está justificado.
PICO
Resumen en términos sencillos
Efectos de los medicamentos antiepilépticos para la prevención primaria y secundaria de las convulsiones después del ictus
Pregunta de la revisión
¿Existe evidencia que apoye el uso habitual de medicamentos antiepilépticos para la prevención primaria y secundaria de las convulsiones después de un ictus?
Antecedentes
Las convulsiones después de un ictus son clínicamente importantes. No está claro si los medicamentos antiepilépticos son efectivos para prevenir las convulsiones después de un ictus en los adultos.
Resultados
Se encontraron dos ensayos prospectivos aleatorizados, doble ciego, controlados con placebo, que evaluaron el efecto de medicamentos antiepilépticos en la prevención primaria de las convulsiones después de un ictus. El primer estudio incluyó a 72 adultos, comparó el ácido valproico con placebo y no mostró diferencias en las convulsiones posteriores al ictus entre el grupo de intervención y el de control. El segundo estudio incluyó a 784 adultos, comparó el diazepam con placebo y no mostró diferencias en las convulsiones posteriores a un ictus entre el grupo de diazepam y el de placebo. Sin embargo, un análisis de subgrupos de los infartos corticales de la circulación anterior mostró un posible efecto beneficioso con el diazepam profiláctico en los primeros tres meses después del ictus. En general, no hay evidencia suficiente para respaldar el uso habitual de los medicamentos antiepilépticos para prevenir las convulsiones después de un ictus. Se justifica la realización de más estudios de investigación sobre si la profilaxis con medicamentos antiepilépticos está indicada en todos los ictus o en aquellos con características específicas.
Calidad de la evidencia
Se consideró que los estudios incluidos tenían un riesgo general de sesgo bajo y que la certeza de la evidencia fue de baja a moderada. Esto significa que es probable que los estudios de investigación adicionales tengan un impacto importante en la confianza en la estimación del efecto y podrían cambiar las conclusiones.
La evidencia está actualizada hasta el 9 de marzo de 2021.
Authors' conclusions
Summary of findings
| AEDs compared with placebo for post‐stroke seizure prophylaxis | ||||||
| Patient or population: patients with stroke Settings: clinical Intervention: AEDs Comparison: placebo | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect (95% CI) | No. of participants | Certainty of the evidence | Comments | |
| Assumed risk | Corresponding risk | |||||
| Primary outcomes | ||||||
| Primary seizure prophylaxis after stroke | 49 per 1000 | 31 per 1000 | RR 0.65 (0.34 to 1.26) | 856 (2 studies) | ⊕⊕⊕⊝a | In subgroup analysis of van Tuijl 2021, 408 ischaemic stroke patients with a total or partial anterior circulation infarction, the risk of post‐stroke was lower in the diazepam group (2 out of 211) than the placebo group (9 out of 197) (RR 0.21, 95% CI 0.05 to 0.95). |
| Secondary seizure prophylaxis after stroke | No evidence identified | |||||
| Secondary outcomes | ||||||
| Proportion of participants who had died or become dependent at the end of the scheduled follow‐up period | 195 per 1000 (mortality) | 188 per 1000 (mortality) | RR 0.97 (0.73 to 1.27) | 856 (2 studies) | ⊕⊕⊕⊝a | |
| NIHSS from admission for SICH to 12 months follow‐up | mean 4.4 (+/‐ 4.1) out of 42 (30 participants) | mean 8.6 (+/‐ 6.1) out of 42 (31 participants) | 72 (1 study) | ⊕⊕⊝⊝b, c | 6 patients in the VPA group and 5 patients in the placebo group died before 12 months follow‐up | |
| *The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect Moderate certainty: we are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different Low certainty: our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect Very low certainty: we have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect a Downgraded one level: impression for wide confidence interval bDowngraded one level: information from one study. c Downgraded one level: small number of participants included in this study. | ||||||
Background
Description of the condition
Stroke is a major global health problem. In industrialised countries, it is the third most common cause of death, just after ischaemic heart disease and cancers. Almost half of all stroke survivors are left with permanent disabilities (Bamford 1991). Although stroke mainly affects the elderly population, it is worth noting that about 30% of all strokes occur in people aged under 65 (Kelly‐Hayes 2010).
Cerebrovascular disease is one of the most common structural etiologies of epilepsy. Post‐stroke seizures account for 11% of all epilepsy, 22% of all cases of status epilepticus, and 55% of newly‐diagnosed seizures amongst older people (Camilo 2004; DeLorenzo 1996; Herman 2002). The reported incidences of post‐stroke seizures vary widely between epidemiological studies, ranging from 2% to 33% for early‐onset seizures and 3% to 67% for late‐onset seizures (Camilo 2004). This is mainly due to the varying methods of case ascertainment and definitions used in relation to the timing of post‐stroke seizures. According to epidemiological guidelines developed by the International League Against Epilepsy (ILAE), seizures occurring within the first week of stroke are defined as 'early post‐stroke seizures', and those occurring after the first week are defined as 'late post‐stroke seizures' (ILAE 1981). Using this definition, approximately 2% to 6% of people with stroke suffer early seizures (Lamy 2003; So 1996), and 3% to 5% suffer late seizures (Lamy 2003; So 1996). In the longer term, one community‐based study found that the cumulative actuarial risk of having a post‐stroke seizure was 4.2% at one year and 11.5% at five years (Burn 1997).
Possible risk factors for post‐stroke seizures include: 1. stroke subtype: cerebral haemorrhage (especially subarachnoid haemorrhage); 2. location of the lesion: cortical involvement, stroke occurring within the carotid artery territory; 3. stroke severity (but correlation may be weaker after adjusting for stroke subtype and location); 4. occurrence of post‐stroke bacterial infections (Burn 1997; Camilo 2004; Kwan 2007; Lamy 2003; Shinton 1988).
The pathophysiology of early‐ and late‐onset seizures after stroke is believed to be different. In the first few days following an ischaemic brain lesion, cellular biochemical dysfunction can lead to cortical excitability and seizure activity (Kessler 2002). Acute ischaemia leads to the massive release of glutamate, causing excessive activation of glutamate receptors. This process is believed to be responsible for secondary neuronal injury and epileptogenesis in ischaemic stroke (Sun 2001; Sun 2002). Within the ischaemic penumbra, a mixed population of dead, dying, and surviving neurons can become the underlying substrate for ischaemia‐induced epileptogenesis (Kessler 2002). In contrast, late‐onset seizures may be caused by the development of gliosis and meningocerebral cicatrices, with changes in membrane properties, deafferentation, selective neuronal loss, and collateral sprouting (Camilo 2004). This can result in cortical hyperexcitability and neuronal synchrony sufficient to cause seizures (Kessler 2002).
After a stroke, seizures and later development of epilepsy can have a devastating impact on patient outcome and quality of life. Predicting the risk of developing post‐stroke epilepsy is currently an inexact science, and improving the accuracy of this process could lead to more‐targeted use of prophylactic antiepileptic drugs for a defined period of time (Kwan 2010a). Several studies have found that post‐stroke seizures may predict worse functional outcome (Menon 2009), but many of these studies have not adjusted for important covariates, such as stroke severity (Camilo 2004). One large study showed that before adjusting for stroke severity, the occurrence of early post‐stroke seizures increased the risk of 30‐day mortality (odds ratio (OR) 4.3, 95% confidence interval (CI) 1.5 to 12.5; (Labovitz 2001)). However, after adjusting for stroke severity using the US National Institutes of Health Stroke Scale (NIHSS) score, this association was not statistically significant (OR 2.1, 95% CI 0.6 to 7.1). In another large community‐based study, early post‐stroke seizures actually predicted a better neurological outcome (Reith 1997). There is also some evidence that post‐stroke seizures may significantly affect health‐related quality of life (Leidy 1999), but a large prospective study did not find any adverse effect on rehabilitation outcomes, as measured by the Barthel Index or the Rivermead Mobility Index (Paolucci 1997).
Description of the intervention
AEDs are the mainstay therapy for seizure prevention in epilepsy. They work by various mechanisms including sodium channel blockade (e.g. phenytoin, carbamazepine, oxcarbazepine, lacosamide and rufinamide), gamma‐aminobutyric acid (GABA) analogue (e.g. gabapentin and pregabalin), and synaptic vesicle glycoprotein 2A (SV2A) binding (e.g. levetiracetam and brivaracetam). The antiepileptic drugs could be divided into old‐generation and new‐generation AEDs. In general, the old‐generation AEDs have more adverse effects and drug interactions. The number of new‐generation AEDs has increased exponentially in recent decades.
AEDs can result in adverse effects including drowsiness, dizziness, tremor, etc. Some AEDs (e.g. topiramate) can cause cognitive impairment, while some (e.g. levetiracetam and perampanel) may lead to psychiatric side effects. Ultimately, these potential side effects may hinder stroke rehabilitation progress and lead to significant morbidity. Drug interaction is another concern, as many of the AEDs are involved in cytochrome P450‐mediated metabolism, including many of the old‐generation AEDs and some new‐generation AEDs at high doses. This may have potential implications on many drugs that are often concomitantly prescribed for stroke patients, including anticoagulants.
How the intervention might work
More than 70% of all epilepsy patients will achieve complete seizure control with AEDs (Kwan 2000). Some evidence suggests that the number of seizures before commencing the AED may affect the chance of long‐term remission, though this has not been demonstrated in randomised controlled trial. Stroke involves neuronal damage in both acute and delayed phases resulting in epileptogenesis, AEDs may have a theoretical role in antagonise the process. Timing may be crucial in the above hypothesis. AEDs may have a role to halt or attenuate the neuronal network damage resulted from stroke. It may be beneficial to use AEDs as primary prophylaxis, that is even before the occurrence of first seizure after stroke.
Why it is important to do this review
Indications of AEDs as both primary and secondary seizure prophylaxis after stroke are debatable. The benefit of AED prophylaxis is uncertain in the post‐stroke population. Also, the medications are potentially harmful due to their adverse effects or drug interactions. Over‐treatment may therefore cause unnecessary morbidity. Hence, it is meaningful to have a systematic review on the current evidence on the indications of AEDs for seizure prophylaxis after stroke.
Objectives
To assess the effects of AEDs for the primary and secondary prevention of seizures after stroke. For primary prevention, we aimed to assess whether AEDs reduce the likelihood of seizures in people who have a stroke but do not have a seizure. For secondary prevention, we aimed to assess whether AEDs reduce the likelihood of further seizures in people who have a stroke and at least one post‐stroke seizure.
Methods
Criteria for considering studies for this review
Types of studies
We considered all randomised and quasi‐randomised controlled trials in which participants were assigned to either treatment or control groups. By control groups, we mean those in which participants received placebo or no drug. We did not consider studies with head‐to‐head drug comparisons (i.e. drug A versus drug B), or those comparing combinations of different drugs. The main reason for this decision was that the conclusion from such a comparison would not answer our objective, which was to assess whether the use of any AED (compared to no AED) was effective for the primary or secondary prevention of seizures after stroke.
Types of participants
We used the World Health Organisation's definition of stroke in this review (WHO 1989). We considered all studies that recruited participants with a new neurological deficit consistent with a clinical diagnosis of stroke. We considered studies that included participants with either ischaemic or haemorrhagic stroke, but we excluded studies that only recruited participants with subarachnoid haemorrhage, subdural haemorrhage, extradural haemorrhage, or other non‐stroke diagnoses such as tumour‐ or infection‐related infarction or haemorrhage. We also excluded studies that recruited only participants who had undergone any type of neurosurgery. The management of these excluded patient groups is likely to be substantially different from the generality of people with stroke. For studies that reported the results for a mixture of participant groups, we attempted to separate them, and identify those which were relevant to the participant groups of interest. When we found that this was not possible, we excluded the studies. We included participants of all ages suffering any seizure type.
Types of interventions
The AEDs included were: acetazolamide, barbexaclone, beclamide, brivaracetam, carbamazepine, carisbamate, chlormethiazole, clobazam, clonazepam, clorazepate, diazepam, dimethadione, divaproex, esclicarbazepine, extazolam, ethadione, ethosuximide, ethotoin, felbamate, flunarizine, fosphenytoin, gabapentin, ganaxolone, lacosamide, lamotrigine, levetiracetam, lorazepam, losigamone, magnesium sulphate, medazepam, mephenytoin, meprobamate, mesuximide, methazolamide, methylphenobarbital, midazolam, nimetazepam, nitrazepam, oxcarbazepine, paraldehyde, paramethadione, perampanel, phenacemide, pheneturide, phenobarbitone, phensuximide, phenytoin, pregabalin, primidone, progabide, propofol, remacemide, retigabine, riluzole, rufinamide, seletracetam, stiripentol, sulthiame, talampanel, temazepam, thiopental, tiagabine, tiletamine, topiramate, trimethadione, valnoctamide, valproic acid, valpromide, vigabatrin and zonisamide.
We excluded studies comparing two AEDs.
Types of outcome measures
We included the following primary and secondary outcomes.
Primary outcomes
-
Proportion of participants who experienced seizures in the scheduled follow‐up period. In cases where seizures had occurred, we noted their nature (generalised or focal) and timing, if reported.
Secondary outcomes
-
Proportion of participants who achieved remission for a predefined period of time (e.g. 12 or 24 months).
-
Proportion of participants who withdrew from the allocated treatment within the scheduled follow‐up period. This is a composite outcome, which takes into account several factors, including adverse events, compliance, and effectiveness of treatment. We were particularly interested in the occurrence of side effects for the different AEDs, which might be physical or neurobehavioural (e.g. problems with memory, attention, and performance skills).
-
Proportion of participants who had died or become dependent at the end of the scheduled follow‐up period. 'Independent' individuals were defined as those who did not require regular physical assistance from another person for activities of daily living, such as mobility, dressing, transfers, and feeding. 'Dependent' individuals were those who failed to meet one or more of these criteria.
-
NIHSS upon follow‐up.
-
Quality of life (e.g. using a recognised scoring system such as SF36 and EuroQol).
-
Duration of stay for the acute phase of stroke recovery.
-
Optimal duration of treatment (i.e. length of time that the intervention should be continued).
Search methods for identification of studies
We used the following search methods to identify studies.
Electronic searches
Searches were run for the original review in July 2009. Subsequent update searches were run in May 2011, August 2012, August 2013, June 2016, November 2017, and October 2019. For the latest update, we searched the following databases on 9 March 2021:
-
Cochrane Register of Studies (CRS Web), using the search strategy shown in Appendix 1.
-
MEDLINE (Ovid, 1946 to March 08, 2021), using the search strategy shown in Appendix 2.
CRS Web includes randomised or quasi‐randomised, controlled trials from PubMed, Embase, ClinicalTrials.gov, the World Health Organization International Clinical Trials Registry Platform (ICTRP), the Cochrane Central Register of Controlled Trials (CENTRAL), and the Specialised Registers of Cochrane Review Groups, including the Epilepsy and Stroke Review Groups.
Searching other resources
We also checked the reference lists of articles retrieved from the above searches. Where clarification of information was needed, we attempted to contact the investigators of the relevant studies.
Data collection and analysis
Selection of studies
Two review authors (JK and RC) independently screened all the titles, abstracts, and keywords of publications identified by the searches, to assess their eligibility. The review authors were blinded to the names of study authors, institutions where the work had been carried out, and the journals (i.e. by printing out the titles, abstracts and keywords without the author names, etc). Publications that clearly did not meet the inclusion criteria were excluded at this stage. We obtained a paper copy of the full publication of every study that appeared to be relevant. Both review authors independently assessed their suitability for inclusion according to prespecified selection criteria, resolving any disagreements by discussion.
Data extraction and management
Two review authors (JK and RC) independently extracted data directly from the two studies that fulfilled our inclusion criteria.
Assessment of risk of bias in included studies
Two review authors (JC and RC) independently assessed the risk of bias for each trial using the Cochrane 'Risk of bias' tool as described in the Cochrane Handbook (Higgins 2011). We rated included studies to be at high, low or unclear risk of bias on six domains applicable to RCTs: randomisation method, allocation concealment, blinding methods, incomplete outcome data, selective outcome reporting and other sources of bias.
Measures of treatment effect
We undertook data analyses according to the methods described in Chapter 9 of the Cochrane Handbook (Deeks 2017). We presented results as summary risk ratios (RRs) with 95% confidence intervals (CIs) (if available) for dichotomous outcomes. We reported mean differences (MDs) with 95% CIs (if available) for continuous outcomes.
Unit of analysis issues
The unit of analysis was the study participant. We ensured that it was the number of participants with post‐stroke seizures that was analysed rather than repeated seizure episodes within the same participant. We did not identify cross‐over trials or cluster‐randomised trials.
Dealing with missing data
We contacted study authors to obtain missing outcome data if necessary.
Assessment of heterogeneity
We visually assessed clinical heterogeneity by comparing the characteristics of the participants and interventions, and methodological heterogeneity by comparing methodological factors (such as study designs, concealment of allocation, blinding, etc.) between studies that met our inclusion criteria. We assessed statistical heterogeneity using the I2 statistic. We assessed the percentage ranges of I2 statistic as follows (Higgins 2011):
-
0% to 40%: may not be important;
-
30% to 60%: represents moderate heterogeneity;
-
50% to 90%: represents substantial heterogeneity;
-
75% to 100%: represents considerable heterogeneity.
Visual inspection of the forest plots also helped us to assess whether or not heterogeneity was present.
Assessment of reporting biases
Had we identified more than 10 studies, we would have visually inspected funnel plots for asymmetry. We would have investigated reasons for asymmetry (if any) including publication bias, outcome reporting bias, language bias, citation bias, poor methodological design, and heterogeneity.
Data synthesis
Only two studies were included; hence we performed no data analysis beyond that performed and reported in the study itself.
Subgroup analysis and investigation of heterogeneity
We did not carry out any subgroup analyses.
Sensitivity analysis
We did not perform a sensitivity analysis, as the number of studies was small.
Summary of findings and assessment of the certainty of the evidence
We used the GRADE approach to interpret findings (Schünemann 2011). We used GRADEpro GDT software (GRADEPro GDT 2020), and imported data from Review Manager 5 (Review Manager 2020), to create a 'Summary of findings' table.
The 'Summary of findings' table includes information on certainty of the evidence from the trials and information of importance for healthcare decision‐making. The GRADE approach determines the certainty of evidence for each outcome across studies on the basis of an evaluation of eight criteria (risk of bias, inconsistency, indirectness, imprecision, risk of publication bias, large magnitude of effect, the presence of plausible confounding that could have influenced the effect, and the dose‐response gradient). We used our GRADE assessments of evidence certainty to guide our conclusions.
Results
Description of studies
Two studies met the inclusion criteria. We describe them below. Please see Table 1 for a detailed description of the excluded studies, and Figure 1 for the PRISMA study flow diagram.
| Study | Participants randomised/ analysed | Experimental treatment | Control treatment | Primary outcomes | Secondary outcomes |
| 71 adults with late post‐stroke seizures | Gabapentin | Nil | Efficacy as measured by seizure recurrence average and the recurrence rate (estimated recurrence per year) of seizures after the start of medication and after completing the dose titration period | Tolerability as measured by incidence of side effects and withdrawal | |
| Meta‐analytic narrative review | |||||
| 35 patients with post‐stroke seizures | Levetiracetam | Sustained‐release carbamazepine | Compare time to second seizure | Cognitive function and quality of life, EEG changes, seizure frequency and safety of the drugs | |
| 128 adults with post‐stroke seizures | Levetiracetam | Sustained‐release carbamazepine | Efficacy as measured by seizure freedom | Time recurrence to the first seizure, EEG tracings, cognitive functions and side effects | |
| 54 patients with early post‐stroke seizures | Levetiracetam | Control (details not mentioned) | Efficacy as measured by seizure recurrence | Nil | |
| Nil (ongoing trial) | Levetiracetam | Valproic acid | Efficacy as measured by seizure freedom during the 12 months |
| |
| 64 patients with post‐stroke seizures | Lamotrigine | Sustained‐release carbamazepine | Efficacy as measured by secure freedom | Tolerability as measured by withdrawal due to side effects | |
| 97 with TBI or stroke, 32 patients in the phenytoin arm and 65 patients in the levetiracetam arm | Levetiracetam | Phenytoin | Seizure frequency | Cost, adverse drug reactions | |
| Nil (Review, not a study) | |||||
| 522 patients with TBI were included, 272 patients in study arm and 250 patients in control arm | Levetiracetam | No treatment | Incidence of early seizures | Nil | |
| 60 adults with late post‐stroke seizures after cerebral haemorrhage | Traditional Chinese medicine combined with valproic acid | Valproic acid | TCM syndrome efficacy as defined by Guiding Principles for Clinical Research of Treatment of Epilepsy Using New TCM; seizure frequency | EEG performance; Quality of life measured by Minnesota life quality scale; cognitive function as measured by Wechsler Adult Intelligence Scale revised; safety indicators as confirmed by routine laboratory blood, urine and stool tests, etc. | |
| 295 patients with intracerebral haemorrhage | AEDs | No treatment | mRS on day 90 | Nil | |
| 84 SAH patients | Extended levetiracetam | Brief levetiracetam | In‐hospital seizures | Nil | |
| Nil (ongoing trial) | Levetiracetam | No treatment | Incidence of seizure | Nil | |
| Nil (terminated trial), targeted patients with aneurysmal subarachnoid haemorrhage | Levetiracetam | No treatment | Randomisation yield | Protocol adherence yield | |
| Nil (terminated trial), targeted patients undergoing surgical repair of unruptured intracranial aneurysms | Levetiracetam | No treatment | Incidence of seizure | Return to daily activities and return to work | |
| 178 patients with traumatic or spontaneous subarachnoid haemorrhage | Levetiracetam Phenytoin | No treatment | Seizure preceding and during hospitalisation | Hospital length of stay Mortality | |
| 462 patients with acute ischaemic stroke | Fosphenytoin | Placebo | Stroke outcome assessed at months 1 and 3 by the mRS | Stroke outcome assessments at months 1 and 3 including the BI, Glasgow Outcome Scale, and NIHSS | |
| 1. Initial monotherapy trial: 61 patients with post‐stroke epilepsy | Lacosamide | Carbamazepine | Treatment‐emergent adverse events (TEAEs) during treatment | Seizure free for 6 and 12 consecutive months | |
| 2. Conversion to lacosamide monotherapy trial: 30 patients with post‐stroke epilepsy | Lacosamide | Nil | Treatment‐emergent adverse events (TEAEs) during treatment | 50% and 75% seizure reduction and seizure freedom | |
| 3. Lacosamide adjective to one baseline AED: 83 patients with post‐stroke epilepsy | Lacosamide in addition to one AED | Nil | Treatment‐emergent adverse events (TEAEs) during treatment | 50% and 75% seizure reduction and seizure freedom | |
| 593 elderly patients of age 60 or above | Lamotrigine and gabapentin | Carbamazepine | Retention in the trial for 12 months | Seizure freedom at 12 months, time to first seizure, and drug toxicity | |
| 716 patients with epilepsy | Valproate, topiramate and lamotrigine | Nil | Time to treatment failure and time to 1‐year remission | Time from randomisation to a first seizure; time to achieve a 2‐year remission; and the frequency of clinically important adverse events and side effects emerging after randomisation; quality of life outcomes as measured by Newly Diagnosed Epilepsy Quality of Life battery; and cost‐effectiveness | |
| 1721 patients with epilepsy | Carbamazepine, gabapentin, lamotrigine, oxcarbazepine, or topiramate | Nil | Time to treatment failure and time to 1‐year remission | Time from randomisation to a first seizure; time to achieve a 2‐year remission; and the frequency of clinically important adverse events and side effects emerging after randomisation; quality of life outcomes as measured by Newly Diagnosed Epilepsy Quality of Life battery; and cost‐effectiveness | |
| 89 adult patients with late onset post‐stroke epilepsy | Phenobarbital or levetiracetam | No treatment | PR and QTc intervals in ECG | Nil | |
| Nil (ongoing trial), targeted cerebrovascular disease patients | Antiepileptic drug | No treatment | Seizure | Nil | |
| Abstract or full text could not be retrieved | Unknown | Unknown | Unknown | Unknown | |
| Nil (ongoing trial), targeted 80 patients with poor‐grade subarachnoid haemorrhage | Levetiracetam | Placebo | Functional outcome using the mRS | 1. Frequency of convulsion 2. Frequency of abnormal findings on electroencephalograph 3. Frequency of delayed cerebral ischaemia and occurrence of performing endovascular therapy against cerebral vasospasm 4. Rate of brain atrophy at 6 months after the onset 5. Functional improvement using Mini‐Mental State Examination (MMSE), Frontal Assessment Battery (FAB), and Barthel Index (BI) at 1 and 6 months after onset 6. Safety | |
| 25 patients with non‐convulsive status epilepticus | Lacosamide | Nil | Time for Termination of non‐convulsive status epilepticus | Time to improvement in the EEG and in clinical observation Side effects of treatment | |
| 92 patients with epilepsy after stroke | Gastrodin in combination with folate and vitamin‐B12 in additional to regular antiepileptics | Regular antiepileptics | Frequency of epileptic seizure and Montreal cognitive assessment (MoCA) scores |
Changes in serum high‐mobility group protein B1, interleukin‐2 and interleukin‐6 | |
| BI: Barthel scale; LDH: lactate dehydrogenase; NIHSS: National Institutes of Health Stroke Scale; mRS: modified Rankin Scale; PDH: Pirovate dehydrogenase; SAH: subarachnoid haemorrhage; TBI: traumatic brain injury | |||||
Study flow diagram
Results of the search
References through database searching from 2011 to 2021 were combined with those included in the previous reviews in 2010 and 2014. After removal of duplicates, 349 references were screened and 316 were rejected. We obtained the full‐texts for 33 studies and assessed their eligibility. Twenty‐eight were excluded as they did not meet the inclusion criteria. Three were excluded as they were ongoing studies. Two studies were included in qualitative and quantitative data synthesis.
Included studies
Design and sample size
Two studies were included in this review (Gilad 2011 and van Tuijl 2021). Both were prospective, randomised, double‐blind, placebo‐controlled trials. Gilad 2011 was single‐centre while van Tuijl 2021 was multi‐centre. The van Tuijl 2021 study was a substudy of the Early GABA‐ergic Activation Study In Stroke (EGASIS) study (Lodder 2006). A total of 856 participants were included. Gilad 2011 recruited 72 participants while van Tuijl 2021 recruited 784. Gilad 2011 included spontaneous non‐traumatic and non‐aneurysmatic spontaneous intracerebral haemorrhage. All participants recruited in both trials were adults (over 18 years of age). In van Tuijl 2021, both ischaemic and haemorrhagic stroke were included. Patients with transient ischaemic attack or pre‐existing epilepsy were excluded. See 'Characteristics of included studies'.
Interventions
Gilad 2011 studied the efficacy of valproic acid versus placebo in the primary prevention of seizure. Oral valproic acid 400 mg twice daily or placebo was given for one month. In the van Tuijl 2021 study, the efficacy of primary seizure prevention within three months post‐stroke with diazepam was investigated. Rectal diazepam 10 mg or placebo was given within 12 hours after stroke onset, followed by oral 10 mg tablets twice daily for three days.
Outcomes
Gilad 2011 defined primary outcome as seizure occurrence at one year of follow‐up. Early seizures were defined as seizure attack within 14 days after stroke onset while late seizures were defined as seizure attack beyond 14 days. Secondary outcome was the mortality rate and NIHSS score at one year of follow‐up. The primary outcome measured by van Tuijl 2021 was any seizure occurred within 3 months of follow‐up period. Secondary outcome was the mortality rate in the follow‐up period. Refer to Table 2 for more details.
| Setting | Hospital, Israel | Multi‐centre, International |
| Design | Prospective randomised placebo controlled trial | Prospective randomised placebo controlled trial |
| Total number of participants | 72 | 784 |
| Treatment group: placebo group participants | 36:36 | 389:395 |
| Mean age in treatment group | 68.5 | 70.53 |
| Mean age in placebo group | 71 | 70.73 |
| Gender in treatment group (male number (%)) | 24 (66.7) | 205 (52.7) |
| Gender in placebo group (male number (%)) | 23 (63.9) | 223(56.5) |
| Primary Outcome | ||
| Total number of seizures | ||
| Treatment group Placebo group | 7 8 | 6a 13b |
| Number of early seizures | ||
| Treatment group Placebo group | 1 4 | NA NA |
| Number of late seizures | ||
| Treatment group Placebo group | 6 4 | NA NA |
| Secondary Outcome | ||
| NIHSS score | ||
| Treatment group Placebo group | 4.4 8.6 | NA NA |
| Mortality | ||
| Treatment group Placebo group | 6 5 | 74c 79c |
| a 2 out of 43 ICH, 4 out of 346 IS, 2 out of 211 PACI/TACI b 3 out of 48 ICH, 10 out of 347 IS, 9 out of 197 PACI/TACI c upon 3 months follow‐up NA: Not available ICH: intracerebral haemorrhage IS: ischaemic stroke PACI: partial anterior circulation infarction TACI: total anterior circulation infarction | ||
Excluded studies
We excluded 28 studies, and provide reasons for their exclusion (see Characteristics of excluded studies and Table 1 for details).
Ongoing studies
Hu 2014 is a published study protocol. The study is designed to investigate whether short‐duration valproate could prevent seizure onset or improve the survival and neurological outcome in adults after acute spontaneous supratentorial intracerebral haemorrhage. It is proposed to be a randomised, double‐blinded, placebo‐controlled trial. It aims to include 258 participants. In addition to standard care, participants are to be randomly assigned to receive a seven‐day prophylaxis of valproate 500 mg daily (alternatively 400 mg daily by intravenous infusion in 250 ml 0.9% normal saline) or matching placebo. Participants are aimed to be followed up at seven days, three months, six months, and twelve months. The primary outcomes include early‐ (up to seven days), and late‐onset seizures (later than seven days). Secondary outcomes include neuroimaging features (midline shift and enlargement of haematoma), all‐cause mortality, adverse effects, and functional outcome assessed by Glasgow Outcome Scale and modified Rankin Scale (mRS). No results have been reported yet.
CTRI/2018/02/011926 2018 is a randomised double‐blinded trial to study the prophylactic role of levetiracetam in prevention of post‐stroke seizures when compared to placebo. It recruits both arterial and venous stroke patients within one to two weeks of onset with a cortical syndrome. The patients are recruited into three groups (cortical ischaemic stroke, supratentorial, non‐aneurysmal, nontraumatic cortical intracerebral haemorrhage (ICH) and cortical venous thrombosis with a cortical syndrome). In each subgroup, the patients are randomised to two arms (three months levetiracetam versus three months placebo). In each group, it is supposed to have a sample size of 200 patients in each arm. Hence, the total sample size is 1200 patients. All patients satisfying the inclusion criteria are block‐randomised using a computer‐generated randomisation sequence into active and control arms. The primary outcome is the occurrence of first seizure. Diagnosis of seizure is established by a neurologist based on direct observation of seizures by hospital’s medical staff or on reliable description by patients, family members, caregivers, or other eyewitnesses. The secondary outcomes are time from stroke to occurrence of a late epileptic seizure, occurrence of early epileptic seizures after stroke, seizure severity, neurological function, midline shift, enlargement of haematoma, death (all cause), functional outcome and the occurrence of side effects of the trial medication. The characteristics of the recruited cohort have been recently published (Agarwal 2021), but the results of the trial are pending to be reported.
PEACH 2015 is a randomised controlled trial aiming at evaluating the impact of prophylactic antiepileptic treatment with levetiracetam versus placebo in acute supratentorial spontaneous intracerebral haemorrhage. The aim is to recruit 104 patients with acute supratentorial spontaneous intracerebral haemorrhage over two years. The intervention is levetiracetam. The primary endpoint is the occurrence of at least one clinical or electrical epileptic seizure recorded on continuous 48 hour Holter electroencephalogram (EEG). Secondary outcomes include the number of EEG seizures, the total duration of epileptic seizures continuously recorded on EEG, the occurrence of some paroxysmal EEG patterns, the number of clinical seizures occurred during 72 hours of diagnosis, the occurrence of early (day 0 to day 30) and late (from day 30 to 12 months) clinical seizures, and the functional prognosis evaluated by the mRS, the cerebral oedema and mass effect evaluated by comparing the admission brain CT scan with the control CT scan performed at 72 hours, the neurological status as assessed by the NIHSS and the quality of life measured by the Stroke Impact Scale. The frequency of side effects related to treatment with levetiracetam (anxiety and depression assessed by the Hospital Anxiety and Depression Scale) are also to be included among reported outcomes.
Risk of bias in included studies
See Figure 2 and Figure 3 for summaries of the risk of bias in each included study. We allocated each study an overall rating for risk of bias. All studies included in the review were individually rated as low risk of bias. Please see below for specific domain ratings.
Risk of bias summary: review authors' judgements about each risk of bias domain for each included study
Risk of bias graph: review authors' judgements about each risk of bias domain presented as percentages across all included studies
Allocation
Gilad 2011 used a computer‐generated randomisation scheme to randomise 72 consecutively admitted participants into the valproic acid treatment group and the placebo control group. van Tuijl 2021 included 784 out of 879 subjects from the EGASIS study (Lodder 2006). Participants were randomised by a computer‐generated randomisation scheme to receive a three‐day course of diazepam or placebo. Overall, the risk of selection biases in random sequence generation and allocation concealment were low in both studies.
Blinding
Gilad 2011 was a double‐blinded study. Participants and outcome assessors were blinded to the group allocation. The neurologist monitoring the valproic acid drug level was also blinded to the group allocation and medication adjustment. van Tuijl 2021, which was a subgroup analysis of EGASIS (Lodder 2006), was also a double‐blinded study. Participants, investigators, treating physicians, and nurses were blinded to trial medication. The risk of performance bias and detection bias were considered to be low in the two studies.
Incomplete outcome data
In Gilad 2011, the reasons for subject exclusion were reported in detail. The study had low attrition due to loss to follow‐up. In van Tuijl 2021, the authors had stated that the patients lost to follow‐up had been excluded. We considered the risk of attrition bias to be low in both studies.
Selective reporting
The outcomes stated in the method sections of the publications of both studies were reported. We considered the risk of reporting bias to be low.
Other potential sources of bias
We did not detect any other sources of bias.
Effects of interventions
See: Summary of findings 1 Summary of findings
Primary outcomes
Primary prevention of post‐stroke seizure by AEDs
In Gilad 2011, the one‐month treatment with valproic acid with initial dose of 800 mg/d (400 mg twice daily, entirely oral) and then with maintenance dose to aim at mid therapeutic serum levels (50 to 100 g/dl), showed no difference in both early and late seizure prevention after one year, when compared with placebo. The authors defined any seizure within two weeks after stroke onset as early seizure, and after two weeks as late seizure. This study randomised 36 patients into the valproic acid group and 36 patients into the placebo group. The total number of patients with seizures in the valproic acid group was seven (19.4%) while that in the placebo group was eight (22.2%), (RR 0.88, 95% CI 0.35 to 2.16). The number of patients with early seizures in the valproic acid group was one (2.8%) while that in the placebo group was 4 (11.1%), (RR 0.25, 95% CI 0.03 to 2.13). The number of patients with late seizures in the valproic acid group was six (16.6%) while that in the placebo group was four (11.1%) (RR 1.5, 95% CI 0.46 to 4.87).
|
| Early and late seizure | No seizure | Total |
| Valproic acid group | 7 | 29 | 36 |
| Placebo group | 8 | 28 | 36 |
| Risk of early and late seizure in VPA group/risk of early and late seizure in placebo group= (7/29)/(8/28): RR 0.88, 95% CI 0.35 to 2.16 | |||
|
| Early seizure (seizure within 2 weeks) | No early seizure | Total |
| Valproic acid group | 1 | 35 | 36 |
| Placebo group | 4 | 32 | 36 |
| Risk of early seizure in VPA group/risk of early seizure in placebo group= (1/35)/(4/32): RR 0.25, 95% CI 0.03 to 2.13 | |||
|
| Late seizure (seizure beyond 2 weeks) | No late seizure | Total |
| Valproic acid group | 6 | 30 | 36 |
| Placebo group | 4 | 32 | 36 |
| Risk of late seizure in VPA group/risk of late seizure in placebo group= (6/30)/(4/32): RR 1.5, 95% CI 0.46 to 4.87 | |||
In van Tuijl 2021, 784 patients were randomised into diazepam or placebo groups. The diazepam group received 10 mg diazepam administered rectally within 12 hours after stroke onset, followed by 10 mg diazepam administered orally twice daily at 12‐hour intervals for the following three days (in total, six 10 mg doses) or until earlier discharge. There was no significant difference in primary post‐stroke seizure prevention in total stroke group nor in a subgroup of patients with an intracerebral haemorrhage or ischaemic stroke after three months follow‐up. Overall, the post‐stroke seizure rate was 1.5% (6 out of 389) in the diazepam group compared with 3.3% (13 out of 395) in the placebo group (RR 0.47, 95% CI 0.18 to 1.22). In the haemorrhagic stroke subgroup, post‐stroke seizure rate was 4.7% (2 out of 43) in the diazepam group compared with 6.3% (3 out of 48) in the placebo group (RR 0.74, 95% CI 0.13 to 4.25). In the ischaemic stroke subgroup, the post‐stroke seizure rate was 1.2% (4 out of 346) in the diazepam group compared with 2.9% (10 out of 347) in the placebo group (RR 0.40, 95% CI 0.13 to 1.27). There was no evidence of difference between these two groups.
However, in a subgroup analysis of 408 ischaemic stroke patients with anterior circulation infarction (including total and partial anterior circulation infarction), the risk of post‐stroke seizure was significantly lower in the diazepam group than in the placebo group. In the diazepam group, 2 patients out of 211 (0.9%) had at least one seizure while in the placebo group, 9 patients out of 197 (4.6%) had at least one seizure (RR 0.21, 95% CI 0.05 to 0.95).
| Overall | Seizure | No seizure | Total |
| Diazepam group | 6 | 383 | 389 |
| Placebo group | 13 | 382 | 395 |
| RR = risk of seizure in diazepam group/risk of seizure in placebo group = (6/389)/(13/395) = RR 0.47, 95% CI 0.18 to 1.22 | |||
| Intracerebral haemorrhage subgroup | Seizure | No seizure | Total |
| Diazepam group | 2 | 41 | 43 |
| Placebo group | 3 | 45 | 48 |
| Risk of seizure in diazepam group/risk of seizure in placebo group = (2/43)/(3/48): RR 0.74, 95% CI 0.13 to 4.25 | |||
| Ischaemic stroke subgroup | Seizure | No seizure | Total |
| Diazepam group | 4 | 342 | 346 |
| Placebo group | 10 | 337 | 347 |
| Risk of seizure in diazepam group/risk of seizure in placebo group = (4/346)/(10/347): RR 0.40, 95% CI 0.13 to 1.27 | |||
| Anterior circulation infarct subgroup | Seizure | No seizure | Total |
| Diazepam group | 2 | 209 | 211 |
| Placebo group | 9 | 188 | 197 |
| Risk of seizure in diazepam group/risk of seizure in placebo group = (2/211)/(9/197): RR 0.21, 95% CI 0.05 to 0.95 | |||
Pooled analysis of the two studies suggests a non‐significant trend towards a lower risk of post‐stroke seizure when AEDs were used for primary prevention of post‐stroke seizure compared to placebo (RR 0.65, 95% CI 0.34 to 1.26; 2 RCTs, 856 participants, moderate‐certainty evidence; Tau2 = 0.00, I2 = 0%, low heterogeneity). We downgraded the certainty of evidence from high to moderate because of imprecision of effect estimates in view of the wide confidence interval.
Secondary prevention of post‐stroke seizure by AEDs
Neither study reported data on this outcome.
Secondary outcomes
Proportion of participants who achieved remission
Neither study reported data on this outcome.
Proportion of participants who withdrew from the allocated treatment within the scheduled follow‐up period
Neither study reported data on this outcome.
Proportion of participants who had died or become dependent at the end of the scheduled follow‐up period
In Gilad 2011, 6 participants out of 36 in the valproic acid group died before the 12 months follow‐up while 5 out of 36 died in the placebo group (RR 1.2, 95% CI 0.40 to 3.58).
| Intervention | Death | Survival | Total |
| Valproic acid group | 6 | 30 | 36 |
| Placebo group | 5 | 31 | 36 |
| Risk of death in diazepam group/risk of death in placebo group: RR 1.2, 95% CI 0.40 to 3.58 | |||
In van Tuijl 2021, mortality rates did not differ between the diazepam group and the placebo group at 2 weeks (RR 0.84, 95% CI 0.56 to 1.26) and 3 months follow‐up (RR 0.95, 95% CI 0.72 to 1.26).
| Intervention | Death within 2 weeks | Survival | Total |
| Diazepam group | 39 | 350 | 389 |
| Placebo group | 47 | 348 | 395 |
| Risk of death within 3 months in diazepam group/risk of death in placebo group: RR 0.84, 95% CI 0.56 to 1.26 | |||
| Intervention | Death within 3 months | Survival | Total |
| Diazepam group | 74 | 315 | 389 |
| Placebo group | 79 | 316 | 395 |
| Risk of death within 2 weeks in diazepam group/risk of death in placebo group: RR 0.95, 95% CI 0.72 to 1.26 | |||
Pooled analysis of the two studies suggests AEDs did not alter the mortality rate after stroke ( RR 0.97, 95% CI 0.73 to 1.27; 2 RCTs, 856 participants, moderate‐certainty evidence; Tau2 = 0.00, I2 = 0%, low heterogeneity). We downgraded the certainty of evidence from high to moderate because of imprecision of effect estimates in view of the wide confidence interval.
NIHSS upon follow‐up
In Gilad 2011, the mean NIHSS of the valproic acid group was lower at 12 months after spontaneous intracerebral haemorrhage than in the placebo group. The mean NIHSS was 4.4 (+/‐ 4.1) in the treatment group, compared with 8.6 (+/‐ 6.1) In the placebo group (P = 0.002). In van Tuijl 2021, this outcome was not included.
Quality of life
Neither study reported data on this outcome.
Duration of stay for the acute phase of stroke recovery
Neither study reported data on this outcome.
Optimal duration of treatment
Neither study reported data on this outcome.
Discussion
Summary of main results
This review aimed to assess the effects of AEDs for the primary and secondary prevention of seizures after stroke. Using our review criteria, only two studies, with a total of 856 subjects, could be included for further analysis.
The first study (72 participants) did not produce a statistically significant result when comparing valproic acid with placebo for the primary prevention of seizures after spontaneous non‐aneurysmal, non‐traumatic intracerebral haemorrhage. However, the treatment group had a lower, non‐statistically significant incidence of early seizures (less than 14 days after onset of haemorrhage) compared to the placebo group. Notably, the time definition of early post‐stroke seizure was different from that of ILAE. Mortality rates were similar in both groups. The valproic acid treatment group also demonstrated a probable benefit in the secondary outcome of lower NIHSS score at one year compared to the placebo group. This supports the hypothesis that valproic acid might have a neuroprotective or neuro‐remodelling effect, but this requires further confirmation. Whether these results can be applied to other forms of stroke (e.g. ischaemic, subarachnoid haemorrhage) is not certain.
The second study (784 participants) demonstrated that primary prophylaxis with three‐day diazepam after acute stroke prevents seizures in the first three months but did not show evidence in prophylactic effect in post‐stroke seizure, in overall strokes and in the subgroup of haemorrhagic and ischaemic strokes. However, it was shown that diazepam had prophylactic effect for post‐stroke seizure in the subgroup of total or partial anterior circulation infarction. There was no difference in mortality between diazepam and placebo at two weeks and three months follow‐up.
In the meta‐analysis, AEDs were not shown to be effective in the primary prophylaxis of post‐stroke seizure (RR 0.65, 95% CI 0.34 to 1.26; 2 studies, 856 participants, moderate‐certainty evidence). AEDs did not affect the mortality rate of patients after stroke (RR 1.03, 95% CI 0.78 to 1.36; 2 RCTs, 856 participants, moderate‐certainty evidence).
Overall completeness and applicability of evidence
This is the second update of the original 2010 Cochrane Review on this topic (Kwan 2010; Sykes 2014). We have thoroughly searched for and critically analysed all available evidence. We identified only two eligible trials, one of which focused on intracerebral haemorrhagic stroke and the other on both haemorrhagic and ischaemic strokes. In view of the selective focus in these studies, the results may not be applicable to other settings or stroke subtypes. The included studies only reported on primary prevention of seizures after stroke. We found no evidence on secondary prevention.
The European Stroke Organisation does not recommend primary prophylactic therapy for haemorrhagic stroke (ESO 2014). The latest American Stroke Association guideline does not suggest any routine primary prophylaxis for post‐stroke seizures (ASA 2016). It recommends standard management approaches to any patient who develops a seizure after stroke. Some other experts go further and recommend that early and late post‐stroke seizures should receive long‐term prophylactic treatment with AEDs (Asconape 1991). In some countries, such as the United Kingdom, sodium valproate remains a very popular AED for the treatment of post‐stroke seizures (Stephen 2003), although according to previous trials, there is no conclusive evidence to support this practice (SANAD 2007).
Many clinicians would agree that repeated unprovoked post‐stroke seizures would require treatment with AEDs, but again there is no good evidence to inform which drug(s) should be initiated, at what dosage, and for how long. Furthermore, there is some evidence that there may be potential risks with using AEDs in the post‐stroke recovery period. For example, there are concerns that the use of phenytoin, phenobarbital, and benzodiazepines in the post‐stroke period may adversely affect motor recovery (Goldstein 1990). Therefore, the management of post‐stroke seizures remains controversial.
Quality of the evidence
We used the GRADE approach to assess the certainty of the evidence in the included studies, using the criteria outlined in the Cochrane Handbook (Higgins 2011). The certainty of evidence for the reported outcomes was low to moderate. The number of recruited participants was relatively large in the study of van Tuijl 2021 (n = 784), though the number of Gilad 2011 was relatively small (n = 72). We judged both studies to be at an overall low risk of bias. Both included studies had detailed descriptions on randomisation, sequence generation and blinding. However, the two studies only investigated the effectiveness of primary seizure prevention after stroke. No data were provided on secondary post‐stroke seizure prevention. We downgraded the evidence from high to moderate due to imprecision as the confidence interval of the relative effect of primary post‐stroke seizure prophylaxis by AEDs was wide. Overall, there was a paucity of data on the secondary outcomes. Both studies reported the mortality rates. Only Gilad 2011 included the outcome of change in NIHSS score.
Potential biases in the review process
We followed the guidance provided in the Cochrane Handbook to minimise potential biases in the review process (Higgins 2011). Multiple databases were covered in the literature searches. The search outputs were independently screened by two review authors to assess their eligibility. As only two trials were included in this review, we were unable to use funnel plots to assess the risk of publication bias. Three ongoing studies were identified that may be reported in further updates of this review.
Agreements and disagreements with other studies or reviews
One meta‐analytic narrative review evaluated the evidence on primary post‐stroke seizure prevention with AEDs (Angriman 2019). The authors focused on haemorrhagic stroke. They included seven studies in qualitative synthesis. Their selection criteria were different from those of this review, as they included observational studies and randomised controlled trials (including Gilad 2011), while we included only studies of the latter design. Angriman 2019 concluded that the use of AEDs as primary prophylaxis for post‐stroke seizure was not associated with improved neurological function or decreased seizure incidence. Although the scope of their review and study inclusion criteria were different from the current review, the conclusions of Angriman 2019 did not contradict with ours.
Another systematic review with a network meta‐analysis evaluated the evidence on secondary post‐stroke seizure prevention with AEDs (Brigo 2018). Adjusted indirect comparisons were made between each AED using controlled‐release carbamazepine (CR‐CBZ) as a common comparator. Two randomised controlled trials were included, one comparing levetiracetam with CR‐CBZ and the other comparing lamotrigine with CR‐CBZ. No significant difference was found in seizure freedom with either agent against CR‐CBZ. Our review did not include these two studies as they did not have a placebo comparison group, and thus failed to fulfil our inclusion criteria.
Study flow diagram
Risk of bias summary: review authors' judgements about each risk of bias domain for each included study
Risk of bias graph: review authors' judgements about each risk of bias domain presented as percentages across all included studies
Comparison 1: Effect of AEDs on primary prevention of post‐stroke seizure (for all types of stroke), Outcome 1: Post‐stroke seizure
Comparison 2: Effect of AEDs on mortality, Outcome 1: Mortality
| AEDs compared with placebo for post‐stroke seizure prophylaxis | ||||||
| Patient or population: patients with stroke Settings: clinical Intervention: AEDs Comparison: placebo | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect (95% CI) | No. of participants | Certainty of the evidence | Comments | |
| Assumed risk | Corresponding risk | |||||
| Primary outcomes | ||||||
| Primary seizure prophylaxis after stroke | 49 per 1000 | 31 per 1000 | RR 0.65 (0.34 to 1.26) | 856 (2 studies) | ⊕⊕⊕⊝a | In subgroup analysis of van Tuijl 2021, 408 ischaemic stroke patients with a total or partial anterior circulation infarction, the risk of post‐stroke was lower in the diazepam group (2 out of 211) than the placebo group (9 out of 197) (RR 0.21, 95% CI 0.05 to 0.95). |
| Secondary seizure prophylaxis after stroke | No evidence identified | |||||
| Secondary outcomes | ||||||
| Proportion of participants who had died or become dependent at the end of the scheduled follow‐up period | 195 per 1000 (mortality) | 188 per 1000 (mortality) | RR 0.97 (0.73 to 1.27) | 856 (2 studies) | ⊕⊕⊕⊝a | |
| NIHSS from admission for SICH to 12 months follow‐up | mean 4.4 (+/‐ 4.1) out of 42 (30 participants) | mean 8.6 (+/‐ 6.1) out of 42 (31 participants) | 72 (1 study) | ⊕⊕⊝⊝b, c | 6 patients in the VPA group and 5 patients in the placebo group died before 12 months follow‐up | |
| *The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect Moderate certainty: we are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different Low certainty: our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect Very low certainty: we have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect a Downgraded one level: impression for wide confidence interval bDowngraded one level: information from one study. c Downgraded one level: small number of participants included in this study. | ||||||
| Study | Participants randomised/ analysed | Experimental treatment | Control treatment | Primary outcomes | Secondary outcomes |
| 71 adults with late post‐stroke seizures | Gabapentin | Nil | Efficacy as measured by seizure recurrence average and the recurrence rate (estimated recurrence per year) of seizures after the start of medication and after completing the dose titration period | Tolerability as measured by incidence of side effects and withdrawal | |
| Meta‐analytic narrative review | |||||
| 35 patients with post‐stroke seizures | Levetiracetam | Sustained‐release carbamazepine | Compare time to second seizure | Cognitive function and quality of life, EEG changes, seizure frequency and safety of the drugs | |
| 128 adults with post‐stroke seizures | Levetiracetam | Sustained‐release carbamazepine | Efficacy as measured by seizure freedom | Time recurrence to the first seizure, EEG tracings, cognitive functions and side effects | |
| 54 patients with early post‐stroke seizures | Levetiracetam | Control (details not mentioned) | Efficacy as measured by seizure recurrence | Nil | |
| Nil (ongoing trial) | Levetiracetam | Valproic acid | Efficacy as measured by seizure freedom during the 12 months |
| |
| 64 patients with post‐stroke seizures | Lamotrigine | Sustained‐release carbamazepine | Efficacy as measured by secure freedom | Tolerability as measured by withdrawal due to side effects | |
| 97 with TBI or stroke, 32 patients in the phenytoin arm and 65 patients in the levetiracetam arm | Levetiracetam | Phenytoin | Seizure frequency | Cost, adverse drug reactions | |
| Nil (Review, not a study) | |||||
| 522 patients with TBI were included, 272 patients in study arm and 250 patients in control arm | Levetiracetam | No treatment | Incidence of early seizures | Nil | |
| 60 adults with late post‐stroke seizures after cerebral haemorrhage | Traditional Chinese medicine combined with valproic acid | Valproic acid | TCM syndrome efficacy as defined by Guiding Principles for Clinical Research of Treatment of Epilepsy Using New TCM; seizure frequency | EEG performance; Quality of life measured by Minnesota life quality scale; cognitive function as measured by Wechsler Adult Intelligence Scale revised; safety indicators as confirmed by routine laboratory blood, urine and stool tests, etc. | |
| 295 patients with intracerebral haemorrhage | AEDs | No treatment | mRS on day 90 | Nil | |
| 84 SAH patients | Extended levetiracetam | Brief levetiracetam | In‐hospital seizures | Nil | |
| Nil (ongoing trial) | Levetiracetam | No treatment | Incidence of seizure | Nil | |
| Nil (terminated trial), targeted patients with aneurysmal subarachnoid haemorrhage | Levetiracetam | No treatment | Randomisation yield | Protocol adherence yield | |
| Nil (terminated trial), targeted patients undergoing surgical repair of unruptured intracranial aneurysms | Levetiracetam | No treatment | Incidence of seizure | Return to daily activities and return to work | |
| 178 patients with traumatic or spontaneous subarachnoid haemorrhage | Levetiracetam Phenytoin | No treatment | Seizure preceding and during hospitalisation | Hospital length of stay Mortality | |
| 462 patients with acute ischaemic stroke | Fosphenytoin | Placebo | Stroke outcome assessed at months 1 and 3 by the mRS | Stroke outcome assessments at months 1 and 3 including the BI, Glasgow Outcome Scale, and NIHSS | |
| 1. Initial monotherapy trial: 61 patients with post‐stroke epilepsy | Lacosamide | Carbamazepine | Treatment‐emergent adverse events (TEAEs) during treatment | Seizure free for 6 and 12 consecutive months | |
| 2. Conversion to lacosamide monotherapy trial: 30 patients with post‐stroke epilepsy | Lacosamide | Nil | Treatment‐emergent adverse events (TEAEs) during treatment | 50% and 75% seizure reduction and seizure freedom | |
| 3. Lacosamide adjective to one baseline AED: 83 patients with post‐stroke epilepsy | Lacosamide in addition to one AED | Nil | Treatment‐emergent adverse events (TEAEs) during treatment | 50% and 75% seizure reduction and seizure freedom | |
| 593 elderly patients of age 60 or above | Lamotrigine and gabapentin | Carbamazepine | Retention in the trial for 12 months | Seizure freedom at 12 months, time to first seizure, and drug toxicity | |
| 716 patients with epilepsy | Valproate, topiramate and lamotrigine | Nil | Time to treatment failure and time to 1‐year remission | Time from randomisation to a first seizure; time to achieve a 2‐year remission; and the frequency of clinically important adverse events and side effects emerging after randomisation; quality of life outcomes as measured by Newly Diagnosed Epilepsy Quality of Life battery; and cost‐effectiveness | |
| 1721 patients with epilepsy | Carbamazepine, gabapentin, lamotrigine, oxcarbazepine, or topiramate | Nil | Time to treatment failure and time to 1‐year remission | Time from randomisation to a first seizure; time to achieve a 2‐year remission; and the frequency of clinically important adverse events and side effects emerging after randomisation; quality of life outcomes as measured by Newly Diagnosed Epilepsy Quality of Life battery; and cost‐effectiveness | |
| 89 adult patients with late onset post‐stroke epilepsy | Phenobarbital or levetiracetam | No treatment | PR and QTc intervals in ECG | Nil | |
| Nil (ongoing trial), targeted cerebrovascular disease patients | Antiepileptic drug | No treatment | Seizure | Nil | |
| Abstract or full text could not be retrieved | Unknown | Unknown | Unknown | Unknown | |
| Nil (ongoing trial), targeted 80 patients with poor‐grade subarachnoid haemorrhage | Levetiracetam | Placebo | Functional outcome using the mRS | 1. Frequency of convulsion 2. Frequency of abnormal findings on electroencephalograph 3. Frequency of delayed cerebral ischaemia and occurrence of performing endovascular therapy against cerebral vasospasm 4. Rate of brain atrophy at 6 months after the onset 5. Functional improvement using Mini‐Mental State Examination (MMSE), Frontal Assessment Battery (FAB), and Barthel Index (BI) at 1 and 6 months after onset 6. Safety | |
| 25 patients with non‐convulsive status epilepticus | Lacosamide | Nil | Time for Termination of non‐convulsive status epilepticus | Time to improvement in the EEG and in clinical observation Side effects of treatment | |
| 92 patients with epilepsy after stroke | Gastrodin in combination with folate and vitamin‐B12 in additional to regular antiepileptics | Regular antiepileptics | Frequency of epileptic seizure and Montreal cognitive assessment (MoCA) scores |
Changes in serum high‐mobility group protein B1, interleukin‐2 and interleukin‐6 | |
| BI: Barthel scale; LDH: lactate dehydrogenase; NIHSS: National Institutes of Health Stroke Scale; mRS: modified Rankin Scale; PDH: Pirovate dehydrogenase; SAH: subarachnoid haemorrhage; TBI: traumatic brain injury | |||||
| Setting | Hospital, Israel | Multi‐centre, International |
| Design | Prospective randomised placebo controlled trial | Prospective randomised placebo controlled trial |
| Total number of participants | 72 | 784 |
| Treatment group: placebo group participants | 36:36 | 389:395 |
| Mean age in treatment group | 68.5 | 70.53 |
| Mean age in placebo group | 71 | 70.73 |
| Gender in treatment group (male number (%)) | 24 (66.7) | 205 (52.7) |
| Gender in placebo group (male number (%)) | 23 (63.9) | 223(56.5) |
| Primary Outcome | ||
| Total number of seizures | ||
| Treatment group Placebo group | 7 8 | 6a 13b |
| Number of early seizures | ||
| Treatment group Placebo group | 1 4 | NA NA |
| Number of late seizures | ||
| Treatment group Placebo group | 6 4 | NA NA |
| Secondary Outcome | ||
| NIHSS score | ||
| Treatment group Placebo group | 4.4 8.6 | NA NA |
| Mortality | ||
| Treatment group Placebo group | 6 5 | 74c 79c |
| a 2 out of 43 ICH, 4 out of 346 IS, 2 out of 211 PACI/TACI b 3 out of 48 ICH, 10 out of 347 IS, 9 out of 197 PACI/TACI c upon 3 months follow‐up NA: Not available ICH: intracerebral haemorrhage IS: ischaemic stroke PACI: partial anterior circulation infarction TACI: total anterior circulation infarction | ||
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1.1 Post‐stroke seizure Show forest plot | 2 | 856 | Risk Ratio (M‐H, Random, 95% CI) | 0.65 [0.34, 1.26] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 2.1 Mortality Show forest plot | 2 | 856 | Risk Ratio (M‐H, Random, 95% CI) | 1.03 [0.78, 1.36] |
