Treatment of infantile spasms

  • Review
  • Intervention

Authors


Abstract

Background

Infantile spasms (West's Syndrome) is a syndrome that includes a peculiar type of epileptic seizure—the spasms—and an electroencephalographic (EEG) abnormality often called hypsarrhythmia. Psychomotor retardation is frequently found at follow-up. Approximately two-thirds of affected infants will have a detectable underlying neurological abnormality, but still little is known about the pathophysiological basis for infantile spasms, and treatment remains problematic.

Objectives

To compare the effects of single pharmaceutical therapies used to treat infantile spasms in terms of control of the spasms, resolution of the EEG, relapse rates, psychomotor development, subsequent epilepsy, side effects, and mortality.

Search methods

To identify published data, we searched the Cochrane Epilepsy Group Specialised Register (October 2012), CENTRAL (The Cochrane Library 2012, Issue 9), MEDLINE (1946 to September Week 4, 2012), EMBASE (1980 to March 2003), and the reference lists of all retrieved articles.
To identify unpublished data, we searched the ISRCTN Register (www.controlled-trials.com), corresponded with colleagues and drug companies, and made requests at international conferences.

Selection criteria

All randomised controlled trials (RCTs) of the administration of drug therapy to patients with infantile spasms.

Data collection and analysis

Data collection from all relevant publications was independently undertaken by three review authors (before 2010) or by two review authors using a standard proforma. Analysis included assessment of study quality and a search for sources of heterogeneity.

Main results

We found 16 small RCTs (fewer than 100 patients enrolled) and 2 larger RCTs (more than 100 patients enrolled). These 18 studies looked at a total of 916 patients treated with a total of 12 different pharmaceutical agents. Overall methodology of the studies was poor, in part because of ethical dilemmas such as giving placebo injections to children. Two studies showed that placebo was not as good as active treatment in resolving the spasms. The strongest evidence suggested that hormonal treatment (prednisolone or tetracosactide depot) leads to resolution of spasms faster and in more infants than does vigabatrin. Responses without subsequent relapse may be no different. The same study suggests that hormonal treatments might improve the long-term developmental outcome compared with vigabatrin in infants not found to have an underlying cause for their infantile spasms.

Authors' conclusions

To date, few well-designed RCTs have considered the treatment of infantile spasms, and the numbers of patients enrolled have been small. In the majority, methodology has been poor, hence it is not clear which treatment is optimal in the treatment of this epilepsy syndrome. Hormonal treatment resolves spasms in more infants than vigabatrin, but this may or may not translate into better long-term outcomes. If prednisolone or vigabatrin is used, high dosage is recommended. Vigabatrin may be the treatment of choice in tuberous sclerosis. Resolution of the EEG features may be important, but this has not been proven. Further research using large studies with robust methodology is required.

Résumé scientifique

Traitement des spasmes infantiles

Contexte

Les spasmes infantiles (syndrome de West) sont un syndrome qui inclut un type spécifique de crise épileptique -les spasmes- et une anomalie de l'électroencéphalogramme (EEG) souvent appelée hypsarythmie. Le retard psychomoteur est fréquemment rencontré lors du suivi. Environ deux tiers des nourrissons affectés auront une anomalie neurologique sous-jacente détectable, mais on n'en sait encore que très peu sur la base patho-psychologique des spasmes infantiles, et le traitement reste problématique.

Objectifs

Comparer les effets des thérapies pharmaceutiques uniques utilisées pour traiter les spasmes infantiles en termes de contrôle des spasmes, résolution de l'EEG, taux de rechute, développement psychomoteur, épilepsie ultérieure, effets secondaires et mortalité.

Stratégie de recherche documentaire

Pour identifier les données publiées, nous avons effectué des recherches dans le registre spécialisé du groupe Cochrane sur l'épilepsie (octobre 2012), CENTRAL (The Cochrane Library 2012, Numéro 9), MEDLINE (de 1946 à la semaine 4 de septembre 2012), EMBASE (de 1980 à mars 2003), et dans les listes bibliographique de tous les articles analysés.
Pour identifier les données non publiées, nous avons effectué des recherches dans le registre ISRCTN (www.controlled-trials.com), correspondu avec nos collègues et les fabricants de médicaments et fait des demandes lors des conférences internationales.

Critères de sélection

Tous les essais contrôlés randomisés (ECR) portant sur l'administration d'un traitement médicamenteux chez des patients souffrant de spasmes infantiles.

Recueil et analyse des données

Toutes les données issues des publications pertinentes ont été recueillies de manière indépendante par trois auteurs (avant 2010) ou deux auteurs utilisant un schéma standard. L'analyse incluait une évaluation de la qualité de l'étude et une recherche sur les sources d'hétérogénéité.

Résultats principaux

Nous avons trouvé 16 petits ECR (moins de 100 patients inscrits) et 2 ECR plus étendus (plus de 100 patients inscrits). Ces 18 études totalisaient 916 patients, traités avec un total de 12 agents pharmaceutiques différents. La méthodologie globale des études était mauvaise, en partie en raison des dilemmes éthiques comme par exemple administrer des injections placebo à des enfants. Deux études ont montré que le placebo ne donnait pas d'aussi bons résultats que le traitement actif pour résoudre les spasmes. Les données les plus solides suggéraient qu'un traitement hormonal (prednisolone ou tétracosactide) permettait de résoudre les spasmes plus rapidement et chez plus de nourrissons que le vigabatrine. Les réponses sans rechute ultérieure ne sont pas forcément différentes. La même étude suggère que les traitements hormonaux peuvent améliorer le résultat concernant le développement à long terme par rapport au vigabatrine chez les enfants chez qui aucune cause sous-jacente expliquant leurs spasmes infantiles n'a été trouvée.

Conclusions des auteurs

A ce jour, peu d'ECR bien conçus ont pris en compte le traitement des spasmes infantiles, et le nombre de patients inscrits était restreint. En majorité, la méthodologie était mauvaise, c'est pour cela qu'il subsiste des incertitudes quant au meilleur traitement de ce syndrome épileptique. Le traitement hormonal résout les spasmes chez plus d'enfants que le vigabatrine, mais cela peut ou pas se traduire par de meilleurs résultats à long terme. Si le prednisolone ou le vigabatrine est utilisé, un dosage élevé est recommandé. Le vigabatrine peut être le traitement de choix pour la sclérose tubéreuse. La résolution des fonctionnalités de l'EEG peut être importante, mais cela n'a pas été démontré. Des recherches complémentaires utilisant des études à grand échelle avec une méthodologie robuste sont nécessaires.

Plain language summary

Treatment of infantile spasms

The optimum treatment for infantile spasms has yet to be proven with confidence, in part because of the different aims of existing studies. However, some useful conclusions can be drawn from current evidence.

Infantile spasms is a rare seizure disorder commonly associated with severe learning difficulties. Many different treatments are currently used worldwide in the treatment of this disorder, and many more have been tried in the past, often with little success. Not all treatments are licensed for use in all countries. Most treatments are associated with significant adverse effects. Additional research is needed to explore the long-term benefits of different therapies for seizure control and for neurodevelopment. Two studies have shown that a placebo is not as good as an active treatment in resolving the spasms. The strongest evidence suggests that hormonal treatment (prednisolone or tetracosactide depot) leads to resolution of spasms faster and in more infants than does vigabatrin. Responses without subsequent relapse may be no different, but one study suggested that hormonal treatment (prednisolone or tetracosactide) might improve long-term neurodevelopmental outcomes in infants and young children for whom no underlying cause for their infantile spasms has been identified. This makes hormonal treatment more attractive, at least for this group of infants. More information and further research are needed to compare currently available therapies.

Résumé simplifié

Traitement des spasmes infantiles

Le meilleur traitement pour les spasmes infantiles doit encore être démontré avec certitude, en partie à cause des différents objectifs des études existantes. Pour autant, certaines conclusions utiles peuvent être tirées des données actuelles.

Les spasmes infantiles sont un trouble de convulsions rare généralement associé à de graves difficultés d'apprentissage. Des traitements différentes sont actuellement utilisés à travers le monde pour traiter ce trouble, et bien d'autres l'ont été dans le passé, souvent sans grand succès. Les traitements ne sont pas tous autorisés dans tous les pays. Les plupart des traitements sont associés à des effets indésirables importants. Des recherches supplémentaires sont nécessaires pour explorer les bénéfices à long terme des différentes thérapies de contrôle des convulsions et de neurodéveloppement. Deux études ont montré qu'un placebo n'est pas aussi efficace qu'un traitement actif pour résoudre les spasmes. Les données les plus solides suggèrent qu'un traitement hormonal (prednisolone ou tétracosactide) permet de résoudre les spasmes plus rapidement et chez plus de nourrissons que le vigabatrine. Les réponses sans rechute ultérieure peuvent ne pas être différentes, mais une étude suggérait que le traitement hormonal (prednisolone ou tétracosactide) était susceptible d'améliorer les résultats de neurodéveloppement à long terme chez les nourrissons et les jeunes enfants pour qui aucune cause sous-jacente pour leurs spasmes infantiles n'a été identifiée. Cela rend le traitement hormonal plus attrayant, au moins pour ce groupe de nourrissons. Des informations complémentaires et des recherches ultérieures sont nécessaires pour comparer les thérapies actuellement disponibles.

Notes de traduction

Traduit par: French Cochrane Centre 13th February, 2014
Traduction financée par: Pour la France : Minist�re de la Sant�. Pour le Canada : Instituts de recherche en sant� du Canada, minist�re de la Sant� du Qu�bec, Fonds de recherche de Qu�bec-Sant� et Institut national d'excellence en sant� et en services sociaux.

Background

Infantile spasms, also known as West's syndrome or salaam spasms, were first described by Dr West in a letter he wrote to The Lancet in 1841. It is a syndrome that includes a peculiar type of seizure, high risk of psychomotor retardation, and usually a characteristic electroencephalographic (EEG) pattern known as hypsarrhythmia (ILAE 1989; Roger 1985). It is relatively uncommon, with an estimated incidence of 0.16 to 0.42 per 1000 live births (Cowan 1991). An association with several disorders has been noted, some of which are known at disease onset (e.g. cerebral palsy, Down syndrome), although others (e.g. tuberous sclerosis, neuronal migration disorders) are discovered on investigation after the onset of spasms. However, in a significant minority of cases, the aetiology remains unknown (Aicardi 1994). Despite awareness of the condition for over 150 years, little progress has been made in our understanding of the pathophysiology of the condition, and treatment of the disorder remained largely empirical for a long time.

Onset of seizures usually occurs within the first year of life, with a peak age of onset of three to five months. The seizures usually consist of sudden, generally bilateral, and symmetrical contractions of the neck, trunk, and extremities that are associated with a brief loss of consciousness. Less commonly, they may consist of an extensor spasm of the legs and spine, or simple head nodding. Rarely they are asymmetrical. Seizures often occur in clusters or runs; commonly 20 or so but as many as 100 spasms can occur in a single cluster, with each individual spasm lasting 1 to 2 seconds only. These clusters frequently occur as the infant is waking from sleep and are commonly associated with a cry. In most cases, they resolve by the age of three, although rarely they can persist up to 10 to 15 years of age (Aicardi 1994).

Hypsarrhythmia is used to describe an EEG pattern that is characterized by random, high-voltage spikes and slow waves. The most striking features of hypsarrhythmia are high-voltage slow waves with variable amplitude; spikes and waves from many foci, varying with time; and lack of synchrony, with a generally 'chaotic' appearance. The typical appearance is more likely to be noted in earlier stages of infantile spasms and when onset occurs at a younger age. The hypsarrhythmic pattern may disappear during rapid-eye-movement (REM) sleep, but it may be observed with greater sensitivity in some other stages of sleep (Lux 2004b).

The seizures are refractory to treatment with most conventional antiepileptic drugs. Although the spasms resolve with time, the long-term prognosis is poor. Many children develop other forms of severe epilepsy, and most (80% to 90%) have psychomotor retardation (Riikonen 1996). Some children have delayed development before the onset of their seizures as part of a predisposing condition, for example, Down syndrome. Nevertheless, even in these patients, further regression of development is often seen after the onset of spasms. The degree of psychomotor delay is severe in approximately 70% of children, placing a great burden on both caregivers and the health system. Although few studies have been undertaken to look at long-term outcomes for these patients, it is a widely accepted view that earlier diagnosis, along with quicker control of the spasms, might improve the prognosis. Evidence now supports this, in that a longer lead time to treatment (time from onset to start of treatment) leads to a worse developmental outcome (O'Callaghan 2011).

In the past, numerous clinical trials have investigated different treatment regimens. The main problem is that few of these trials were randomised controlled trials, and the drug under investigation has frequently been used long after the onset of seizures and in addition to, or after, other anticonvulsants. Wide variation in the drug dosages used and in the duration of their use has been reported. As a result, it is confusing when looking at the literature to know which is the best treatment for this disorder, with the consequence that many different treatment modalities have been used. Most trials have considered infantile spasms as a single entity rather than considering the underlying aetiology, but it is possible that infantile spasms resulting from one particular disorder (e.g. Down syndrome, tuberous sclerosis) might respond better to one form of treatment than another. Other factors than treatment that might affect the outcome also need to be considered, for example, the belief that the sooner the spasms are brought under control (i.e. the shorter the lead time), the better the outcome. Likewise, if infants entered into a trial have been previously treated for infantile spasms, this might also affect outcome. Recently, it has been suggested that age at onset affects developmental outcome, with improvement noted for each increase in age of one month (O'Callaghan 2011).

Infantile spasms are an unusual epilepsy syndrome. The interictal EEG is very abnormal in these infants, but occasionally it is abnormal only during sleep. The ictal EEG is variable, often with suppression or fast activity, but the ictal event is frequently (but not always) typical. Agreement on the exact definition of the EEG abnormality required to make the diagnosis has not been achieved, but fortunately, the combination of the unusual ictal event, the clinical semiology, and the very abnormal EEG makes any other diagnosis unlikely, even when either the EEG or the ictal event is not typical.

The fact that ictal episodes are individually very brief (lasting between one and two seconds), their occurrence in clusters or batches (many of which can occur in a day), and the frequency of the ictal episodes during a cluster (often many per cluster) are features also contributing to the fact that this is an unusual epilepsy. Absence of ictal activity can, therefore, be detected with some accuracy (depending on the accuracy of the witness) on a daily basis by direct observation. Cessation of all ictal activity is generally accompanied by resolution of the features usually seen on the EEG, but many infants will not then have a normal EEG because of other seizure types or underlying structural brain abnormalities. It is not practical to record the EEG on a daily basis; therefore, the relationship between resolution of ictal activity and improvement of the EEG is not clearly established. Therefore, disagreement continues over the importance of continuing treatment if the EEG remains severely abnormal in the absence of clinical evidence of continuing infantile spasms. Some clinicians believe that persistence of the EEG abnormality must indicate continuation of the epileptic encephalopathy, and that the EEG findings should, therefore, be treated even in the absence of observed ictal episodes.

Another unusual feature of this condition is the developmental regression that so often accompanies, or quickly follows, the onset of seizures. This, along with the severity of the EEG abnormality, has led many to the conclusion that infantile spasms are an epileptic encephalopathy. This theory lends support to the hypothesis that it may be important to continue treatment until the EEG abnormalities have improved. Most treatments are believed to work if a response occurs within a short time of the commencement of treatment, usually within 14 days. Such claims need to be backed up with evidence, especially since because the tendency in the literature is for claims of rapid response not to be supported by the evidence. However, the practical outcome of the developmental regression and the usual rapid response (viz other epilepsy syndromes) mean that continuation of spasms beyond 14 days of treatment is usually an indication that one should try an alternative treatment.

These issues, definitions, and outcomes have been subject to review by an independent group using Delphi methodology. This review concluded that clinical evidence of cessation of spasms should be reported independently of cessation of spasms combined with resolution of the EEG features of the condition. It was agreed that subsequent developmental progress was an important outcome, and that this might be a more important outcome than cessation of spasms. It is proposed that the protocol for the present review will need to be reviewed before the next review, if one is to take into account this internationally agreed consensus.

Additional difficulties have arisen from the use of terms such as 'high dose' and 'low dose' in publications. In some studies, the 'high dose' has been lower than the dose used in other studies as the 'low dose.' In addition, lack of study by pharmaceutical companies, a common problem in paediatric practice, means that the most effective dose often is not known. Some studies have corrected for body size through an adjustment for body weight; others have adjusted for surface area. However, such adjustment is not usually justified, and the scientific basis for or against adjustment is not as robust as it could be.

Therefore, it is clear that resolution of the clinical evidence of spasms is one outcome, and that this may occur quickly, in contrast to most epilepsy syndromes. Resolution of the EEG features also needs to be reported, and subsequent developmental progress is an important later outcome. Other seizure types occur before the spasms in many infants and may follow spasms, even when these resolve and other seizure types did not occur before the onset of spasms; therefore, epilepsy outcomes for all seizure types need to be reported.

Significant conventional reviews of the literature on the treatment of infantile spasms include Haines 1994, Riikonen 1996, Mackay 2004, and Go 2012.

Objectives

To compare the effects of single pharmaceutical therapies used to treat infantile spasms in terms of control of spasms, resolution of hypsarrhythmia, long-term psychomotor development, and subsequent epilepsy rates.

The following hypotheses were tested.

Therapy versus placebo

(1) Therapy A* is more effective than placebo treatment in controlling infantile spasms (in terms of spasm cessation, reduction in total number of spasms, and relapse). .
(2) Therapy A* is more effective than placebo treatment in resolving the EEG appearance.
(3) Therapy A* leads to improved long-term psychomotor development compared with placebo treatment.
(4) Treatment of infantile spasms with therapy A* reduces subsequent epilepsy rates compared with placebo treatment.

Therapy versus no treatment

(1) Therapy A* is more effective than no treatment in controlling infantile spasms (in terms of spasm cessation, reduction in total number of spasms, and relapse).
(2) Therapy A* is more effective than no treatment in resolving the EEG appearance.
(3) Therapy A* leads to improved long-term psychomotor development compared with no treatment.
(4) Treatment of infantile spasms with therapy A* reduces subsequent epilepsy rates compared with no treatment.

Comparisons between therapies

(1) Therapy A* is more effective in controlling infantile spasms (in terms of spasm cessation, reduction in total number of spasms, and relapse) compared with any other single pharmaceutical therapy.
(2) Therapy A* is more effective in resolving the EEG appearance compared with any other single pharmaceutical therapy.
(3) Therapy A* leads to improved long-term psychomotor development compared with any other single pharmaceutical therapy.
(4) Treatment of infantile spasms with therapy A* reduces subsequent epilepsy rates compared with any other single pharmaceutical therapy.

*Therapy A = ACTH** or hydrocortisone or prednisone or prednisolone or carbamazepine or ethosuximide or gabapentin or lamotrigine or phenobarbitone or phenytoin or topiramate or vigabatrin or valproate or clonazepam or diazepam or nitrazepam or pyridoxine (vitamin B6), or methysergide, or methylparatyrosine, or sulthiame, or any other single therapeutic agent studied in the literature.

**ACTH – at the time this review was undertaken, two ACTH preparations were in widespread use: ACTH (adrenocorticotrophin hormone) and tetracosactide. ACTH is naturally occurring, and the therapeutic product is derived from a bovine or porcine source and is administered as an intramuscular injection. However in the UK, with existing concerns surrounding bovine spongiform encephalopathy (BSE), ACTH has been withdrawn from the market. Tetracosactide is a synthetic alternative to ACTH and consists of the first 24 amino acids occurring in ACTH. It displays the same physiological properties as ACTH. It can be provided in depot and non-depot preparations' the Depot preparation is usually given on alternate days.

Methods

Criteria for considering studies for this review

Types of studies

All randomised controlled trials (RCTs) of the administration of drug therapy to infants and children with infantile spasms were included in this review, including trials that compared a therapy with no treatment or placebo, and trials that compared one drug with another or with different doses of the same drug.

Definition of an RCT: trials in which participants are prospectively allocated to treatment groups by a random (e.g. random number generation, coin flips) or a quasi random (e.g. by date of birth) process.

If the study was not an RCT, it was not included in the review. The existence of such studies was documented.

We considered studies looking at drug therapy as second-line therapy, as well as studies looking at drug therapy as first-line therapy.

Types of participants

Any infant or child treated for infantile spasms regardless of whether or not EEGs were performed, or whether therapy had been given before trial entry.

Most published studies on infantile spasms did not give the definition of infantile spasms used for their participants; therefore, for the purpose of this review, we assumed that a clinical diagnosis had been made for any participant that was entered into a trial. We documented any definitions that were given and considered whether differences in definition might account for variation between studies.

Types of interventions

(1) Any trial that compared at least one therapy against placebo treatment.
(2) Any trial that compared at least one therapy against no therapy.
(3) Any trial that compared at least one therapy against another therapy or a different dose of the same therapy.

Therapies included:
(1) 'Steroids': ACTH, tetracosactide, hydrocortisone, prednisone or prednisolone.
(2) Antiepileptic drugs: carbamazepine, clonazepam, diazepam, ethosuximide, flunarizine, gabapentin, ganaxolone lamotrigine, methysergide, methylparatyrosine, nitrazepam, phenobarbitone, phenytoin, pyridoxine, sodium valproate, sulthiame, topiramate, vigabatrin.
(3) Any other single therapeutic agent studied in the literature.

Any dose regimen of the above therapies was included.

Types of outcome measures

(1) Cessation of spasms

This was defined as total cessation of spasms for at least 48 hours after the start of therapy but occurring within a month of commencement of therapy. It was measured as a dichotomous variable (i.e. ceased or continuing). Time taken from commencement of therapy to cessation of spasms was also to be measured as a continuous variable (measured in days).

(2) Quantitative reduction of spasms

This was measured as the number of spasms occurring before treatment was commenced compared with the number occurring after treatment and was to be measured as a dichotomous variable (i.e. greater than 50% reduction or less than 50% reduction in the number of spasms occurring per day over the seven-day period before trial entry and over a seven-day period at one month after the start of therapy).

(3) Resolution of EEG abnormality

Participants were divided into those in whom the EEG abnormality remained and those in whom there was resolution. This was measured as a dichotomous variable (i.e. resolved or unresolved).

(4) Relapse rates of spasms

A single spasm occurring by one year of age or within the study period but after cessation of spasms constituted a relapse. It was measured as a dichotomous variable (i.e. relapse occurred or no relapse occurred). Time taken from cessation of spasms to relapse was also to be measured as a continuous variable (measured in days) in those patients in whom spasms had ceased.

(5) Long-term psychomotor development

This was to be analysed in months as a measure of deviation from the chronological age. It was to be measured at three months post entry into the trial and, where possible, at five years of age. It was measured as a continuous variable.

(6) Subsequent epilepsy rates

The diagnosis of any epileptic seizure type other than infantile spasms after commencement of therapy constituted subsequent epilepsy. This was to be measured at three months post entry into the trial and, where possible, at five years of age. Any seizure type plus each seizure type (e.g. primary generalised, partial) was to be measured as a dichotomous variable (i.e. present or absent) over the previous one month at three months post entry, and over the previous six months at five years of age.
We used the original authors' diagnosis because of the difficulty involved in defining epileptic seizures in children.

(7) Adverse effects

Only side effects that were considered severe enough to warrant discontinuation of the test treatment were measured. They were measured as a dichotomous variable (i.e. therapy stopped vs therapy not stopped). They were also qualitatively summarised.

(8) Deaths

All deaths were measured as a dichotomous variable (i.e. alive or deceased).

Search methods for identification of studies

We searched the Epilepsy Group Specialised Register (5 October 2012). This register contains reports of trials identified from regular searches of the Cochrane Central Register of Controlled Trials (CENTRAL) and of MEDLINE. Relevant reports were also identified by handsearching selected journals and conference proceedings.

In addition, we carried out searching as follows.

Electronic searches

We searched the following databases. No language restrictions were applied.

(a) CENTRAL (The Cochrane Library 2012, Issue 9) was searched using the strategy outlined in Appendix 1.
(b) MEDLINE (Ovid) (1946 to September Week 4, 2012) was searched using the strategy outlined in Appendix 2.
(c) EMBASE (1980 to March 2003) was searched in a similar manner to MEDLINE. We no longer have direct access to that database. However, a project to identify reports of trials in EMBASE is being carried out by the UK Cochrane Centre. This search is updated annually and these records are published in CENTRAL. These records are therefore available to us via our searches of CENTRAL.
(d) ISRCTN Register (www.controlled-trials.com) was searched using the term "infantile spasms" (4 October 2012).

Searching other resources

The reference lists of the RCTs and of rejected articles were scanned (by ECH) to identify possible articles missed by the computerised search. We contacted pharmaceutical companies and colleagues, and made appeals at international conferences to try to identify unpublished data and studies not published in English.

Data collection and analysis

Relevant publications were reviewed independently by at least two review authors (ECH or SE and JO) and before 2010 by three review authors (ECH, JO, and Professor Milner or SE). Discrepancies were resolved by discussion. There was no blinding of authorship or results. All RCTs were considered, as were all non-English studies.

Assessment of methodological quality

  • Selection bias: the studies were assessed as to whether allocation concealment was adequate, unclear, or inadequate.

  • Performance bias: the studies were assessed as to whether recipients and those measuring outcome were unaware of the assigned therapy (however, we accepted that most of the studies in this review would not be double-blinded trials, except where blind assessment of outcomes was possible).

  • Attrition bias: assessed as to whether patients were lost to follow-up in the studies.

Using these criteria, studies were then divided into: (1) those with a low risk of bias; (2) those with a moderate risk of bias; and (3) those with a high risk of bias (see the Cochrane Handbook for Systematic Reviews of Interventions, Section 8.5) (Higgins 2005).

Exclusion criteria

  • Any trial that was not a RCT was excluded from analysis but was documented.

  • If a clinical definition for infantile spasms was given, any trial in which there was doubt about the clinical diagnosis was excluded from analysis but was documented.

In studies in which it is unclear if the above criteria have been met, the review authors are endeavouring to obtain additional information by contacting the first author on up to three occasions. This is documented.

The following data were originally extracted independently by three review authors (ECH, JO, and Professor Milner or SE), and discrepancies were resolved by discussion. For this update, only two authors were used (ECH, JO).

Participants (i.e. those characteristics of the population that may affect outcome regardless of treatment):

  • Age at spasm onset, at diagnosis, and at start of treatment (mean or median and range measured to the nearest completed week of age).

  • Time from onset of spasms to initiation of treatment (mean or median and range measured to the nearest completed week).

  • Sex (male, female).

  • Previous treatment for infantile spasms (e.g. prednisolone, ACTH, vigabatrin, valproate) – but not previous treatment for other seizure types.

Interventions:

  • Type of pharmaceutical agent used (e.g. prednisolone, ACTH, vigabatrin, valproate).

  • Dose (measured in internationally accepted units (e.g. mg for prednisolone, IU for ACTH).

  • Frequency (measured as the number of times the pharmaceutical agent was given in a 24 hour period).

  • Route of administration (i.e. oral, intramuscular, intravenous).

  • Treatment length (measured to the nearest completed day).

Outcome measures:

  • Cessation of spasms and the time taken to cessation of spasms.

  • Resolution of EEG and time taken for resolution.

  • Reduction in spasms.

  • Relapse rates and time taken to relapse.

  • Psychomotor development.

  • Subsequent epilepsy rates.

  • Side effects.

  • Deaths.

Analysis plan

  • Study quality. This was done with a table of met or unmet criteria for selection, performance, and attrition bias.

  • Dichotomous data. For each item of data requiring dichotomous analysis, the following were recorded: number of patients who experienced the event (or outcome) in each group for each comparison and the total number in each group. These data were analysed using Peto odds ratios (ORs) in the Cochrane RevMan software.

  • Continuous data. For each item of data requiring continuous data analysis (except psychomotor development), the following were recorded: number of patients in each group, mean or median value for the outcome in each group, and standard deviation. These values were analysed using weighted means in the Cochrane RevMan software.

  • When data for the same outcome were presented in some studies as dichotomous and in others as continuous data, we endeavoured to obtain continuous data from the investigators. If it was not possible to obtain continuous data, for example, because they had not been recorded, data were analysed either as dichotomous data with a 'cut-off' point agreed by the three review authors, or as a mixture of dichotomous and continuous data, using two separate tables.

  • We looked for sources of heterogeneity between trials: methodological and clinical differences, including previous treatment for spasms, age at trial entry, and single underlying cause. We also looked for subgroup analysis of clinical features, for example, developmental delay before onset of spasms, differences in drug dosages, timing and length of treatment, and specific underlying diagnosis, such as tuberous sclerosis.

Results

Description of studies

The literature search identified 96 studies to be evaluated for inclusion in this review. Some studies were identified from more than one source. Review of the studies eliminated 64 studies because they were not RCTs and a further six because the participants did not suffer from infantile spasms; one paper was excluded because the data referred to the effects of hormonal treatment on the brain when imaged using computerised tomography and did not report any other outcome measure. This left a total of 27 papers reporting 18 different RCTs conducted to look at the treatment of infantile spasms, yielding a total of 858 patients. These 18 trials looked at 12 different pharmacological agents: vigabatrin; ACTH (nine different treatment regimens and different preparations); prednisolone; prednisone; hydrocortisone; magnesium sulphate; nitrazepam; sodium valproate; sulthiame; flunarizine; ganaxolone methysergide; and alpha-methylparatyrosine (see table Characteristics of included studies and Table 1, Table 2, Table 3, Table 4). We have been careful to amalgamate for meta-analysis only those treatments for which a rationale for combining treatments was provided.

Table 1. Methodological quality of included studies
Study IDNo. patientsPlacebo RCTAllocation blindedRecipients blindedOutcome blindedLoss of follow-upMulti-centreRisk of bias
Appleton 199940yesyesyesyesnoyeslow
Askalan 20039nonot clearnosomenonohigh
Baram 199629noyesnoyesnonenohigh
Chiron 199722nonot clearnonononeyeshigh
Debus 200451yesyesyesnonoyeslow
Dreifuss 198652nonot clearnoyesyesyeshigh
Dyken 198517yesnot clearyesyesyesnohigh
Elterman 2010221nonot clearnonoyesyeshigh
Hrachovy 198324nonot clearyesnot clearnoneunclearhigh
Hrachovy 198924nonot clearnot clearnot clearnoneunclearhigh
Hrachovy 199459nonot clearnoyesyesunclearhigh
Lux 2004107noyesnono except developmental outcomeyesyeshigh
Bitton 201268yesnot clearyesyesyesyeslow
Shu 200930nonot clearnot clearnot clearnot clearnot clearhigh
Tsai 200956yesnot clearyesyesnot clearnot clearhigh
Vigevano 199742nonot clearnonot clearnonenohigh
Yanagaki 199926nonot clearnot clearnot clearyesnohigh
Zou 201038nonot clearnonononounclear
Table 2. Participants
Study IDInterventionMale:FemaleAge at onsetAge at diagnosisAge at trial entryDelay to treatmentPrevious treatment
Appleton 1999Vigabatrin11:930 weeksnot available35 weeks6 weeksno
 Placebo8:1226 weeksnot available35 weeks7 weeksno
Askalan 2003ACTH1:2not availablenot availablenot availablenot availableno
 Vigabatrin3:3not availablenot availablenot availablenot availableno
Baram 1996ACTH4:11not availablenot available22 weeksnot availablenot available
 Prednisone8:6not availablenot available32 weeksnot availablenot available
Chiron 1997Vigabatrin5:625 weeksnot available29 weeks3 weeksnot available
 Hydrocortisone5:626 weeksnot available34 weeks8 weeksnot available
Debus 2004Sulthiame14:11 (3 unclear)31 weeks31 weeks35 weeks4 weeksyes
 Placebo7:12 (4 unclear)29 weeks29 weeks33 weeks4 weeksyes
Dreifuss 1986Nitrazepam14:13not availablenot available37 weeksnot availablenot available
 ACTH15:10not availablenot available35 weeksnot availablenot available
Dyken 1985Valproatenot availablenot availablenot availablenot availablenot availableyes
 Placebonot availablenot availablenot availablenot availablenot availableyes
Elterman 2010Vigabatrin (low dose)42:33not available29.5 weeks32.9 weeksnot availableyes
 Vigabatrin (high dose)29:38not available28.1 weeks31.6 weeksnot availableyes
Hrachovy 1983ACTHnot availablenot availablenot availablenot availablenot availablenot available
 Prednisonenot availablenot availablenot availablenot availablenot availablenot available
Hrachovy 1989Methysergidenot availablenot availablenot availablenot availablenot availablenot available
 Methylparatyrosinenot availablenot availablenot availablenot availablenot availablenot available
Hrachovy 1994ACTH (high-dose)not availablenot availablenot availablenot availablenot availablenot available
 ACTH (low-dose)not availablenot availablenot availablenot availablenot availablenot available
Lux 2004Vigabatrin32:2022 weeks26 weeks26 weeks4 weeksno
 Hormonal32:2322 weeks26 weeks26 weeks4 weeksno
Bitton 2012Flunarizine25:9Not available28 weeks28weeksNot availableno
 Placebo22:12Not available31 weeks31 weeksNot availableno
Shu 2009ACTH (high - dose)Not availableNot availableNot availableNot availableNot availableNot available
 ACTH (low dose)Not availableNot availableNot availableNot availableNot availableNot available
Tsai 2009GanaxoloneNot availableNot availableNot available17.3 to 104 weeksNot availableNot available
 PlaceboNot availableNot availableNot available17.3 to 104 weeksNot availableNot available
Vigevano 1997Vigabatrin14:925 weeksnot available27 weeksnot availableno
 ACTH7:1226 weeksnot availablenot availablenot availableno
Yanagaki 1999ACTH (high-dose)8:520 weeksnot available28 weeks7 weeksyes
 ACTH (low-dose)7:523 weeksnot available41 weeks16 weeksyes
Zou 2010ACTH14:56.6 monthsnot availableunclearnot availableyes, some
 ACTH + magnesium sulphate9:105.5 monthsnot availableunclearnot availableyes, some
Table 3. Outcomes
Study IDNo. participantsInterventionDevelopmentSeizuresSpasms stoppedSpasms reducedRelapse ratesResolution of EEG
Appleton 199920Vigabatrinnot reportednot reported35%40% reduction57%71%
 20Placebonot reportednot reported10%15% reduction100%50%
Askalan 20033ACTHnot reportednot reported100%not reportednot reported100%
 6Vigabatrinnot reportednot reported100%not reportednot reported100%
Baram 199615ACTHnot reported50% (2-48 month F.U.)87%not reported14%87%
 14Prednisonenot reported50% (2-46 month F.U.)29%not reported29%29%
Chiron 199711Vigabatrinnot reportednot reported100%not reported9%not reported
 11Hydrocortisonenot reportednot reported45%not reportednot reportednot reported
Debus 200420Sulthiamenot reportednot reported29%not reportednone29%
 17Placebonot reportednot reported<1%not reportednone<1%
Dreifuss 198627Nitrazepamnot reportednot reportednot reported66%not reportednot reported
 25ACTHnot reportednot reportednot reported50%not reportednot reported
Dyken 198513 (/17)Valproatenot reportednot reportednot reportedp<0.04not reportednot reported
 17 (/17)Placebonot reportednot reportednot reportedp<0.04not reportednot reported
Elterman 2010114Vigabatrin (low-dose)not reportednot reported11%not reportednot reported11%
 107Vigabatrin (high-dose)not reportednot reported36%not reportednot reported36%
Hrachovy 198312ACTHnot reportednot reported42%not reported60%not reported
 12Prednisonenot reportednot reported33%not reported25%not reported
Hrachovy 198912Methysergideno significant differencenot reported8%25%100%not reported
 12Methylparatyrosineno significant differencenot reported16%17%50%not reported
Hrachovy 199430ACTH (high-dose)not reportednot reported43%not reported15%23%
 29ACTH (low-dose)not reportednot reported48%not reported21%21%
Lux 200452Vigabatrinsee textsee text54%not reported 56%
 55Hormonalsee textsee text72%not reported 88%
Bitton 201234Flunarazineno significant improvementnot reported50%not reportednot reportednot reported
 34Placebono significant improvementnot reported62%not reportednot reportednot reported
Shu 200930ACTH (high-dose)not reportednot reportednot reportednot reportednot reportednot reported
 ACTH (low-dose)not reportednot reportednot reportednot reportednot reportednot reported
Tsai 200956Ganaxolonenot reportednot reportednot reportednot reportednot reportednot reported
 Placebonot reportednot reportednot reportednot reportednot reportednot reported
Vigevano 199723Vigabatrinno significance difference25% (9-44 months F.U.)48%not reportednot reported36%
 19ACTHno significance difference25% (9-44 months F.U.)74%not reportednot reported79%
Yanagaki 199913ACTH (high-dose)no significance differencenot reported84%not reported27%61%
 13ACTH (low-dose)no significance differencenot reported69%not reported33%33%
Zou 201019ACTH alonenot clearnot reported42%7 of 11 remainingnone26%
 19ACTH + magnesium sulphatenot clearnot reported63%3 of 7 remaining17%47%
Table 4. Adverse effects and deaths
Study IDInterventionNo. adverse effectsTypes of side effectsNo. deathsCause of death
Appleton 1999Vigabatrinnone none 
 Placebonone none 
Askalan 2003ACTHnot reported not reported 
 Vigabatrinnot reported not reported 
Baram 1996ACTH  none 
 Prednisonenot reported none 
Chiron 1997Vigabatrinnone none 
 Hydrocortisone1not givennone 
Debus 2004Sulthiame1somnolencenone 
 Placebo0 none 
Dreifuss 1986Nitrazepamnone none 
 ACTH61 = meleana, 1 = hypertension, 4 = not given1unknown
Dyken 1985Valproatenot reported not reported 
 Placebonot reported not reported 
Elterman 2010Low-dose vigabatrin9, unclear which groupnot given2 deaths occurred, unclear which groupone unknown, one due to pulmonary haemorrhage
 High-dose vigabatrin9, unclear which groupnot given2 deaths occurred, unclear which groupone unknown, one due to pulmonary haemorrhage
Hrachovy 1983ACTHnot reported none 
 Prednisonenot reported none 
Hrachovy 1989Methysergidenone none 
 Methylparatyrosinenone none 
Hrachovy 1994High-dose ACTHnot reported not available 
 Low-dose ACTHnot reported not available 
Lux 2004Vigabatrin2drowsiness and vomiting3 
 Hormonal2rash and irritability, vomiting, and high BP2 
Bitton 2012Flunarizinenot reportednot reported3not reported
 Placebonot reportednot reported3not reported
Shu 2009High-dose ACTH93%not reportednot reportednot reported
 High-dose ACTH20%not reportednot reportednot reported
Tsai 2009Ganaxolonemore frequentsomnolence, lethargy, and irritabilitynot reportednot reported
 Placeboless frequentnot reportednot reported
Vigevano 1997Vigabatrin1excessive irritability and agitation1 death occurred, unclear which groupunknown
 ACTH1not given1 death occurred, unclear which groupunknown
Yanagaki 1999High-dose ACTHnone none 
 Low-dose ACTHnone none 
Zou 2010ACTH16pyrexia, URTI, diarrhoea, anorexia, hypertension, insomnia, irritability, decreased heart rate with prolonged PR interval.none 
 ACTH + magnesium sulphate10pyrexia, URTI, diarrhoea, anorexia, vomitingnone 

Appleton 1999
This was a short-term, randomised, double-blind, placebo-controlled multicentre trial of 40 participants. Inclusion criteria included participants between 1 and 20 months of age with newly diagnosed and previously untreated infantile spasms in whom the EEG demonstrated classical or modified hypsarrhythmia. A spasm was defined as a sudden, generally bilateral and symmetrical contraction of the muscles of the neck, trunk, and extremities (flexor, extensor, or mixed). The principal exclusion criterion was the use of any medication, including steroids, that could be considered to be an anti-epileptic drug within a two-month period before entry into the trial. The trial consisted of a five-day double-blind phase, during which patients received vigabatrin or placebo. Twenty participants received vigabatrin and 20 received placebo. The initial starting dose was 50 mg/kg/day for 24 hours; if spasms did not cease completely, the dose was increased to 100 mg/kg/day, and this dose was maintained for a further 48 hours. The dose could then be increased further to 150 mg/kg/day. Randomisation was pre-determined by a code held by the pharmacy department for each participating hospital. Both recipients and assessors were blinded. Outcomes reported were cessation of spasms, reduction in spasms, resolution of hypsarrhythmia, and relapse rates. Some patients were lost to follow-up, but not during the double-blind phase.

Askalan 2003
This was an open-label, randomised, single-centred trial of nine participants. Inclusion criteria included participants between 3 and 16 months of age who presented with infantile spasms. The diagnosis of infantile spasms was confirmed by the presence of myoclonic flexor spasms, extensor spasms, or both, with video-EEG showing hypsarrhythmia or modified hypsarrhythmia. Individuals were excluded if they had previously received vigabatrin or corticosteroids; had a known visual disturbance; or had a known medical condition for which corticosteroids would be contraindicated. The trial consisted of three participants receiving 150 IU/m2/day ACTH intramuscularly in two divided doses for one week, followed by 75 IU/m2/day ACTH for one week; and six participants receiving 100 mg/kg/day vigabatrin orally in two divided doses for two days, then increasing to 150 mg/kg/day for the remainder of the two-week period. After the initial two-week period, responders remained on their initial trial treatment and completed either a 12-week tapering protocol for ACTH or 18 months of vigabatrin. Non-responders were crossed over to the alternate trial drug while being tapered off the original study drug over one week. The method of randomisation was not stated. Assessors interpreting the EEGs and psychologists were blinded. Outcomes reported were cessation of spasms, resolution of hypsarrhythmia, short-term cognitive outcomes, language scores, and the evolution of epilepsy and autism.

Baram 1996
This was a randomised, single-blinded, single-centred trial of 29 participants. Infants with clinical infantile spasms were considered for the study and underwent 24-hour video-EEG to ascertain hypsarrhythmia and clinical spasms. The trial consisted of a two-week period during which 15 participants were treated with ACTH 150 units/m2/day intramuscularly in two divided doses; and 14 participants were treated with prednisone 2 mg/kg/day given orally in two divided doses. A computer-generated random number list determined treatment, and allocation was concealed (as confirmed by Dr Baram, direct correspondence). Recipients were not blinded, but blinded assessment of outcomes was performed. Outcomes reported were cessation of spasms, resolution of hypsarrhythmia, relapse rates, and development of other seizure types within the follow-up period.

Chiron 1997
This was a randomised, multicentre trial of 22 participants with infantile spasms, all of whom had tuberous sclerosis. Inclusion criteria included epileptic spasms recorded on EEG or seen by an experienced clinician, diffuse interictal paroxysmal activity, and age ranging from one month to two years. Participants were excluded if they had been previously treated with steroids or vigabatrin but not with other anticonvulsant medications. The trial spanned a one-month period, during which 11 infants received vigabatrin at a dose of 150 mg/kg/day, and 11 infants received hydrocortisone at a dose of 15 mg/kg/day. The method of randomisation was not stated. Neither recipients nor assessors were blinded. Outcomes reported were cessation of spasms, resolution of hypsarrhythmia, and relapse rates.

Debus 2004
This was a randomised, double-blind, multicentre trial of 51 participants. Inclusion criteria included children between 3 and 18 months of age with newly diagnosed infantile spasms. The diagnosis was confirmed by the presence of hypsarrhythmia or hemi-hypsarrhythmia on an EEG that had to contain periods of sleep and wakefulness. Participants were excluded if they had a history of epilepsy before the development of infantile spasms, unless they were not treated or had been treated with phenytoin or phenobarbitone; or if they had other ongoing or acute diseases. The trial consisted of a nine-day (days one to nine) treatment period with pyridoxine for all participants. On days one to three, all participants received just pyridoxine. From day four of the trial, 28 participants received sulthiame at 5 mg/kg/day in two divided doses, and 23 participants received placebo. If no response was seen after three days, the dose of the trial drug was doubled for three days. Randomisation was concealed by sealed envelopes (personal communication). The recipients were blinded, but the assessors were not blinded to treatment allocation. Outcomes reported were cessation of spasms, resolution of hypsarrhythmia, and relapse rates.

Dreifuss 1986
This was a randomised multicentre trial of 52 participants. Inclusion criteria included participants between 1 and 24 months of age with infantile spasms documented by a hypsarrhythmic or a modified hypsarrhythmic pattern on EEG. None of the participants had received steroids or nitrazepam before entry into the trial. The trial consisted of a four-week period during which 27 participants received nitrazepam, and 25 participants ACTH. Nitrazepam was started at 0.2 mg/kg/day in two divided doses or at 1 mg twice daily, whichever was greater, and was adjusted twice weekly by increments of 0.3 to 0.4 mg/kg/day. Maintenance dosage ranged from 4.8 to 9 mg/day. ACTH was given as a once-daily intramuscular injection of 40 units. A computer-generated randomisation code determined treatment, but concealment of allocation was not clearly stated. Recipients were not blinded, but blinded assessment of outcome was performed. The only outcome reported was reduction in spasm frequency. A discrepancy between the review authors was noted over the assessment of how many participants were lost to follow-up in this trial; this discrepancy arose because the authors were unclear in their paper as to how many participants were lost to follow-up. We have been unable to contact the author to clarify this point, so the number of participants lost to follow-up remains unclear.

Dyken 1985
This was a randomised, placebo-controlled, cross-over, single-centre trial of 17 participants. Inclusion criteria included participants between 8 and 83 months of age with a clinical diagnosis of infantile spasms and hypsarrhythmia on EEG, who had previously failed to respond to steroids. Many participants had received other anticonvulsant medications. The trial compared sodium valproate against placebo, but the length of the trial and the dosages used were not given. A computer-generated random numbers table was used to allocate participants, but concealment of allocation was not clearly stated. Recipients were blinded, and blinded assessment of outcome was performed. Only reduction in spasm frequency was reported as an outcome. Four participants were lost to follow-up.

Elterman 2010
This was a randomised, multicentre trial of 221 participants. Inclusion criteria included participants younger than two years of age but weighing at least 3.5 kg. Diagnostic criteria included infantile spasms and an EEG pattern typical of hypsarrhythmia or modified hypsarrhythmia, or multifocal spike wave discharges, or a video-EEG recording capturing an event confirming the diagnosis of infantile spasms electroencephalographically. Participants were receiving stable doses of antiepilepsy medication, if applicable, and were unresponsive to a 50-mg bolus of pyridoxine, when appropriate. Participants were excluded if they had received an investigational drug within 30 days of enrolment (one-week washout for ganaxolone): and had been or were currently taking corticosteroids (brief exposure to low-dose corticosteroids for inflammatory conditions or replacement therapy was acceptable but required a one-week washout), valproic acid; or felbamate within 60 days of enrolment; or were receiving more than two standard antiepilepsy drugs to treat non–infantile spasm seizures. Participants were also excluded if they had a treatable or progressive cause of seizures; had current active medical disorders; had a caregiver who was unable to reliably record seizures or recall adverse events; or had showed poor compliance with medication. The trial consisted of a two-week period during which participants received low-dose vigabatrin 18 to 36 mg/kg/day or high-dose vigabatrin 100 to 148 mg/kg/day. The dose was based on the participant's weight and, in the high-dose group, was titrated over the first seven days of the study. Medication was given twice daily unless the total dose was less than 250 mg per day, in which case daily dosing was permitted. Participants in either group who continued to have infantile spasms after seven days of the full dose of vigabatrin specified for their weight could have the dose increased weekly by 25 to 50 mg/kg/day to a maximum dose of 200 mg/kg/day. The method of randomisation was not stated. The authors stated that this was a single-blinded study, and that the recipients were blinded but the assessors were not. However, the recipients were not blinded to the dose of vigabatrin given (it was supplied as 500-mg tablets in open-labelled bottles), and caregivers were simply 'masked' as to whether their children were assigned to the high-dose or the low-dose group. Outcomes reported were cessation of spasms;resolution of hypsarrhythmia;time to cessation of spasms; and relapse rates. Safety data were also published. A total of 221 participants were randomly assigned: all were followed up as intention to treat, none were lost to follow-up.

Hrachovy 1983
This was a randomised single-centre trial (using a double-dummy technique) of 24 participants. Inclusion criteria included participants with infantile spasms and hypsarrhythmic EEG patterns on serial 24-hour video and polygraphic monitoring. No participant had been previously treated with steroids, but some had received other anticonvulsant medications. Twelve participants received ACTH 20 to 30 units/day for two weeks, which was then tapered over one week if a response was seen. If a response was not seen, participants continued a further four weeks before the dose was tapered. These participants also received a prednisone placebo. Twelve participants received prednisone, 2 mg/kg/day for two weeks, which was then tapered over one week if a response was seen. If a response was not seen, they continued a further four weeks before the dose was tapered. These participants also received an ACTH gel placebo. The method of randomisation was not given. Recipients were not blinded, but it was unclear if blinded assessment of outcome was performed. Outcomes reported were complete cessation of spasms and relapse rates.

Hrachovy 1989
This was a randomised trial of 24 participants. Inclusion criteria included newly diagnosed participants with infantile spasms and hypsarrhythmic EEGs. The trial consisted of a three-week period during which 12 participants were treated with methysergide 2 mg/m2/day for the first seven days, then 5 mg/m2/day for the next 14 days, and 12 participants were treated with alpha-methylparatyrosine 500 mg/m2/day, increasing by 150 mg/m2/day on day two and thereafter until a maximum dosage of 1250 mg/m2/day was reached. The method of randomisation was not given. It was not clear whether recipients and assessors were blinded. Authors reported cessation of spasms, a reduction in the number of spasms, and developmental status at three weeks as outcome measures.

Hrachovy 1994
This was a randomised controlled trial of 59 participants. Inclusion criteria included infants in whom a diagnosis of infantile spasms had recently been made, who had hypsarrhythmic EEG findings. They were excluded if they had previously received steroids. Thirty participants received 150 units/m2 of ACTH per day for three weeks, with a tapering dose over a further nine weeks, and 29 participants received 20 to 30 units/day of ACTH for two to six weeks, tapering over one week. The method of randomisation was not stated. Recipients were not blinded, but blinded assessment of outcomes was performed. Outcomes reported were complete cessation of spasms, resolution of hypsarrhythmia, and relapse rates. Nine participants were lost to follow-up.

Lux 2004
This was an open, randomised, controlled, multicentre trial of 107 participants. Inclusion criteria were participants between 2 and 12 months of age with a clinical diagnosis of infantile spasms and a hypsarrhythmic or similar EEG. Exclusion criteria included a diagnosis or high risk of tuberous sclerosis; previous treatment with or a contraindication to vigabatrin or hormonal treatments; a lethal or potentially lethal disorder other than infantile spasms; inability of parents or guardians to give informed, signed consent or to know when spasms stop; leaving the UK unexpectedly within one month of randomisation; enrolment in a concurrent trial that either used treatment that might affect the outcome measures or was labour-intensive for participants, guardians, or medical practitioners. Fifty-two participants received vigabatrin given orally in two divided doses: 50 mg/kg/day for the first two doses, increasing to 100 mg/kg/day and then to 150 mg/kg/day at 96 hours if no response was seen. Twenty-five participants received tetracosactide depot given intramuscularly, 0.5 mg (40 IU) on alternate days for two weeks, and increased to 0.75 mg (60 IU) on alternate days if no response was seen after one week. Thirty participants received prednisolone given orally, 10 mg four times daily for two weeks, increasing to 20 mg three times a day after one week if no response was seen. Randomisation was concealed by sealed envelopes. Neither recipients nor assessors were blinded. Outcomes reported were cessation of spasms and resolution of hypsarrhythmia, relapse rates, development at 14 months of age, seizure rates at 14 months of age, development at 4 years of age, and seizure rates at 4 years of age. One participant was lost to developmental follow-up at 14 months of age, and 30 participants were lost to follow-up at 4 years of age.

Bitton 2012

This was a study in which flunarizine or placebo was administered as an add-on drug to vigabatrin to 68 infants aged 3 to 18 months, with new onset of infantile spasms in a double-blind, randomised, controlled trial. Infants were included if they had 'de novo' onset of infantile spasms and an EEG with hypsarrhythmia or modified hypsarrhythmia Children were excluded if they had a known neurometabolic or neurodegenerative disease at study entry or a pre-existing medical condition contraindicating steroid treatment, if they could not obtain the neuropsychological evaluation before enrolment, or if families were unable to comply with follow-up visits.All children received vigabatrin 100 mg/kg/day for 3 days followed by 150 mg/kg/day when spasms persisted, as first-line treatment. Responders continued to take vigabatrin for 6 months. If after 2 weeks, vigabatrin failed to arrest clinical spasms and abolish hypsarrhythmia, children were switched to synthetic adrenocorticotropic hormone (sACTH) given intramuscularly, and vigabatrin was tapered over a week. The dose of sACTH given every 2 days was calculated to be equivalent to 150 IU of natural ACTH/m2/day for 2 weeks, followed by a 10-week tapering schedule. If clinical spasms and/or hypsarrhythmia persisted after 2 weeks of treatment with sACTH,the children were switched to topiramate (10 mg/kg/day).Outcome was reported as development at 24 months of age using the Vineland Adaptive Behaviour Scale and the Bayley Scales of Infant Development. Some participants were lost to follow-up.

Shu 2009

This was a prospective, randomised, controlled trial of 30 children with West's syndrome. Participants received either ACTH 50 IU/day for two weeks, then tapered to zero over two weeks, or 0.4 IU/kg for two weeks, increased to 1 IU/ kg for a further two weeks if no reponse was seen, then tapered to zero over two weeks. The method of randomisation is not clear. It is not clear whether both participants and assessors were blinded. Outcomes reported were response rates, relapse rates, time to relapse, long-term outcome at 12 months, and adverse effects.

Tsai 2009

Thisdouble-blind, placebo-controlled randomised trial of 56 participants with infantile spasms compared ganaxolone with placebo. Infants were randomly assigned to receive either ganaxolone or placebo for one week, then all infants received ganaxolone while still blinded to the treatment of the previous week. Subjects were allowed two concomitant antiepileptic drugs (AEDs). The method of randomisation is not clear. Outcomes reported were the reduction of spasm cluster frequency in the first week, spasm frequency after two weeks of treatment, investigator/caregiver global assessment, and responder rates (defined as greater than 50% reduction in spasm frequency). Side effects were reported. It is not clear whether any subjects were lost to follow-up.

Vigevano 1997
This was a randomised (response-mediated, cross-over) single-centre trial of 42 participants. Inclusion criteria included participants with newly diagnosed and previously untreated infantile spasms diagnosed according to the International League Against Epilepsy (ILAE) classification. Twenty-three participants received vigabatrin at 100 to 150 mg/kg/day, and 19 received ACTH at 10 units/day. The method of randomisation was not stated, but direct correspondence with the authors revealed a quasi-random method (i.e. alternate allocation of therapy). Recipients were not blinded, and it was not clear whether assessors of outcomes were blinded. Outcomes reported were cessation of spasms, resolution of hypsarrhythmia, and short-term effects on subsequent epilepsy rates and cognitive function.

Yanagaki 1999
This was a randomised single-centre trial of 26 participants. Inclusion criteria included participants with infantile spasms and hypsarrhythmia on video-EEG. Individuals were excluded if they had received previous steroid or intravenous gamma globulin treatment. Thirteen participants received 1 unit/kg/day of synthetic ACTH for two weeks, tapering over a further two weeks; and 13 participants received 0.2 units/kg/day of synthetic ACTH for two weeks, tapering over a further two weeks. The method of randomisation was not stated. It was not clear whether recipients and assessors were blinded. Outcomes reported were cessation of spasms, resolution of hypsarrhythmia, relapse rates, and the short-term effect on development. One participant was lost to follow-up.

Zou 2010

This was a single-centre trial of 38 participants. Inclusion criteria included participants with infantile spasms and hypsarrhythmia. Exclusions included acute associated illness and previous treatment with ACTH, prednisolone, hydrocortisone, or magnesium sulphate. All participants received ACTH 25 units/day, and half received magnesium sulphate 0.25 g/kg/day for three weeks. Allocation was not concealed, but randomisation was applied. Recipients and outcomes were not blinded. Outcomes reported were cessation of spasms and resolution of hypsarrhythmia: some developmental outcome data were given, but no between group comparisons were done. No deaths occurred.

Risk of bias in included studies

Overall, methodological quality of the included studies was poor. All but two of the studies had small numbers of participants—the largest included 221 and the second largest 107 participants. All the other studies enrolled fewer than 70 participants, and subsequently the power of most of the studies was low. There were five placebo-controlled trials (Appleton 1999; Debus 2004; Dyken 1985; Bitton 2012; Tsai 2009 ). One trial used a double-dummy technique (Hrachovy 1983). Although all studies stated that they were RCTs, only six stated the method of randomisation, and only four indicated that concealment of allocation had been performed. Recipients were blinded in five studies; blinding did not take place in eight studies, and it was unclear whether blinding was performed in the remaining studies. Six studies clearly stated that those measuring outcomes were unaware of the assigned therapy. Six studies had losses to follow-up. Six studies were multicentre trials, and six recruited participants from a single centre; it was unclear how many centres were involved in recruiting participants in the remaining studies. Of the 18 RCTs considered in this review, all but two had a high risk of bias (see Table 1: Methodological quality of included studies).

Also of importance for interpretation of the meta-analyses below, the results of longer-term outcomes were complicated by the fact that infants initially randomly assigned to receive one therapy were likely to have gone on to receive the comparative or other therapies, or both, within the follow-up period if they failed to respond to the first therapy.

Effects of interventions

There were characteristics of the population that may affect outcome regardless of the treatment (see Table 2: Participants).
The male to female ratio was given for 12 of the studies and was 1.1:1.
The mean age at spasm onset was given in seven studies and ranged from 20 to 31 weeks.
The mean age at which diagnosis of spasms was made was given for four studies and ranged from 26 to 31 weeks.
The mean age at trial entry (i.e. the age at which the trial treatment was started) was given in 11 studies and ranged from 15.1 to 104 weeks.
The time delay between onset of spasms and start of treatment was given in three studies and ranged from 3 to 16 weeks.
In five studies, other treatments had been previously tried; in five studies, other treatments had not been used; and in eight studies, it was unclear whether previous treatment had been received.

We had also planned to look for heterogeneity according to different diagnostic groupings, for example, tuberous sclerosis and Down syndrome; and we planned to look at whether participants had demonstrated developmental delay before the onset of spasms. Unfortunately, for most of the studies selected, this information was not available and therefore could not be analysed.

Vigabatrin versus placebo

One study was included (Appleton 1999).

Effects on cessation of spasms: The Appleton 1999 study enrolled 40 participants and showed complete cessation of spasms in 7 of 20 (35%) participants treated with vigabatrin compared with 2 of 20 (10%) treated with placebo (Peto OR 4.1, 95% confidence interval (CI) 0.9 to 17.5).
Effects on time taken to achieve cessation of spasms were not reported as an outcome in this study.
Effects on reduction in the number of spasms: The Appleton 1999 study showed > 70% reduction in spasms in 40% of the group treated with vigabatrin compared with 15% in the group treated with placebo. However, it was not clear from the article what proportion of the two groups of participants was reflected in these figures, and whether the figures applied to the whole group, or just to those participants in whom complete cessation of spasms was not achieved.
Effects on relapse rates: Four of seven participants who responded to vigabatrin relapsed, and all participants successfully treated with placebo relapsed.
Overall, only three participants treated with vigabatrin and no participant treated with placebo remained spasm free within the four-week study period (Peto OR 8.2, 95% CI 0.8 to 84).
Effects on time taken to relapse were not reported as an outcome in this study.
Effects on resolution of EEG: Five of the seven participants who were spasm free with vigabatrin showed resolution of EEG, compared with one of the two participants who had become spasm free while taking placebo (Peto OR 2.4, 95% CI 0.1 to 54.6).
Effects on psychomotor development were not reported as an outcome in this study.
Effects on subsequent seizure rates were not reported as an outcome in this study.
No side effects severe enough to warrant stopping treatment were reported in this study.
No deaths were reported.

Sodium valproate versus placebo

One study was included (Dyken 1985).

Effects on cessation of spasms were not reported as an outcome in this study.
Effects on time taken to achieve cessation of spasms were not reported as an outcome in this study.
Effects on reduction in the number of spasms: The Dyken 1985 study used a spasm index (SI) = (spasm frequency x spasm duration/total observation time) x 1000 to calculate the mean reduction in spasm frequency from baseline. At the end of the four-week treatment period, valproate treatment was found to have a lower mean spasm index than placebo when valproate was administered first (P < 0.04). No significant difference between placebo and valproate was noted during the second level of treatment. It was not possible to calculate a Peto OR.
Effects on relapse rates were not reported as an outcome in this study.
Effects on time taken to relapse were not reported as an outcome in this study.
Effects on resolution of EEG were not reported as an outcome in this study.
Effects on psychomotor development were not reported as an outcome in this study.
Effects on subsequent seizure rates were not reported as an outcome in this study.
Side effects were not reported as an outcome in this study.
Deaths were not reported as an outcome in this study.

Sulthiame versus placebo

One study was included (Debus 2004).

Effects on cessation of spasms:The Debus 2004 study found 8 of 28 participants treated with sulthiame to be spasm free as compared with 1 of the 23 participants given placebo (Peto OR 5.13, 95% CI 1.22 to 21.47).
Effects on time taken to cessation of spasms were not reported as an outcome in this study.
Effects on reduction in the number of spasms were not reported as an outcome in this study.
Effects on relapse rates were not reported as an outcome in this study.
Effects on time taken to relapse were not reported as an outcome in this study.
Effects on resolution of EEG: The Debus 2004 study stated that 8 of 28 participants treated with sulthiame had resolution of their EEG as compared with 1 of 23 participants given placebo (Peto OR 5.13, 95% CI 1.22 to 21.47).
Effects on psychomotor development were not reported as an outcome in this study.
Effects on subsequent seizure rates were not reported as an outcome in this study.
Side effects: Sulthaime had to be withdrawn in one patient because of somnolence.
No deaths occurred in either group.

Low-dose vigabatrin versus high-dose vigabatrin

One study was included (Elterman 2010).

Effects on cessation of spasms: On the basis of primary criteria and an intention-to-treat approach (please see below), 17 of 107 participants treated with high-dose vigabatrin became spasm free compared with 8 of 114 participants treated with low-dose vigabatrin (Peto OR 0.41, 95% CI 0.18 to 0.95).
Effects on time taken to achieve cessation of spasms within the initial two-week study period were not reported as an outcome in this study.
Effects on reduction in the number of spasms were not reported as an outcome in this study.
Effects on relapse rates: Again on the basis of primary criteria, 2 of 17 responders in the high-dose group subsequently relapsed compared with 2 of 8 participants in the low-dose group.
Effects on time taken to relapse: In the high-dose group, the time taken to relapse was 162 days (range 53 to 270 days) compared with 45 days (range 31 to 58 days) in the low-dose group (Peto OR 0.25, 95% CI 0.25 to 24.32).
Effects on resolution of EEG: 8 of 75 participants treated with low-dose vigabatrin had resolution of their EEG compared with 24 of 67 participants treated with high-dose vigabatrin (Peto OR 0.24, 95% CI 0.11 to 0.52).
Effects on psychomotor development were not reported as an outcome in this study.
Effects on subsequent seizure rates were not reported as an outcome in this study.
Side effects: In total, vigabatrin was stopped in five participants as a result of an adverse event directly attributed to the vigabatrin therapy; the reasons were not given, and the dose that had been given was not stated.
Three deaths were reported in this study; all three were judged to be unrelated to vigabatrin.

However, it is important to note the following when assessing the results of this study.

  • This study was initially published in 2001. In the initial paper, many participants had been 'lost to follow-up' because they were excluded from the analysis for violating the protocol. However, in the final article, published in 2010, the authors report the findings on an intention-to-treat basis (now more widely accepted as best practise). This explains why the numbers presented in this updated review are different from those presented previously, with a lower rate of participants lost to follow-up.

  • In their original study, primary responders were defined as subjects who were spasm free and obtained a video-EEG within 3 days of the seventh day of spasm freedom, which showed no indication of spasms or hypsarrhythmia. During the course of the study, it became apparent that obtaining the video-EEG within 3 days of spasm cessation was not feasible for most subjects. A post hoc analysis was therefore performed, relaxing the timing of vidoe-EEG to any subsequent visit. This analysis yielded response rates of 33/107 in the high-dose group and 15/114 in the low-dose group (P = 0.0014).

Vigabatrin versus hydrocortisone

One study (Chiron 1997) considered only participants with tuberous sclerosis.

Effects on cessation of spasms: The Chiron 1997 study compared vigabatrin (150 mg/kg/day) and hydrocortisone (15 mg/kg/day) in 22 infants with infantile spasms due to tuberous sclerosis and found in the initial phase that 11 of the 11 participants (100%) treated with vigabatrin were spasm free as compared with 5 of 11 participants (45%) treated with hydrocortisone (Peto OR 13.8, 95% CI 2.21 to 86.35).
Effects on time taken to achieve cessation of spasms: On average, the 11 responders to vigabatrin took 4 days (range 0.5 to 14 days, median 2 days) to achieve complete cessation of spasms, and the 5 responders to hydrocortisone took an average of 13 days (range 3 to 30 days, median 23.5 days) (weighted mean difference (WMD) -8.8, 95% CI -19.2 to 1.6).
Effects on reduction in the number of spasms were not reported as an outcome in this study.
Effects on relapse rates: 10 of the 11 participants who responded to vigabatrin remained spasm free; this information was not given for the five responders to hydrocortisone.
Effects on time taken to relapse were not reported as an outcome in this study.
Effects on resolution of EEG were not reported as an outcome in this study.
Effects on psychomotor development were not reported as an outcome in this study.
Effects on subsequent seizure rates were not reported as an outcome in this study.
Side effects: Hydrocortisone had to be stopped in one patient; no reason was given.
No deaths were reported in this study.

Vigabatrin versus hormonal treatment (ACTH, tetracosactide, or high-dose prednisolone), excluding hydrocortisone

Three studies were included (Askalan 2003, Lux 2004, and Vigevano 1997).

Effects on cessation of spasms: The Askalan 2003 trial of nine participants showed that all six of the participants randomly assigned to vigabatrin had cessation of spasms, as did the three participants randomly assigned to ACTH. The Lux 2004 trial showed that 28 of the 52 participants randomly assigned to receive vigabatrin had cessation of their spasms compared with 40 of the 55 participants randomly assigned to receive either tetracosactide or high-dose prednisolone. The Vigevano 1997 trial of 42 participants showed cessation of spasms in 11 of the 23 participants randomly assigned to vigabatrin compared with 14 of the 19 participants randomly assigned to ACTH. Combining these three studies revealed that 45 of the 81 participants randomly assigned to vigabatrin had cessation of their spasms compared with 57 of the 77 randomly assigned to ACTH, tetracosactide, or high-dose prednisolone (Peto OR 0.42, 95% CI 0.21 to 0.80).

Effects on time taken to achieve cessation of spasms: The Askalan 2003 study did not report on this outcome. The Lux 2004 study reported that the 28 responders to vigabatrin took a median of 11 and a half days to respond, but the 40 responders to tetracosactide or high-dose prednisolone had a median response time of three days. In the Vigevano 1997 trial, the 11 responders to vigabatrin took between 1 and 14 days to achieve complete cessation of spasms, and the 14 responders to ACTH took between 2 and 12 days.

Effects on reduction in the number of spasms were not reported as an outcome in these three studies.

Effects on relapse rates were not reported as an outcome in the Askalan 2003 and Vigevano 1997 studies. The Lux 2004 study reported relapse in 9 of 28 responders on vigabatrin compared with 18 of 40 responders on tetracosactide or high-dose prednisolone (Peto OR 0.59, 95% CI 0.22 to 1.57). Overall, 19 of the 52 participants randomly assigned to receive vigabatrin remained spasm free as compared with 22 of the 55 participants randomly assigned to receive tetracosactide or prednisolone.

Effects on time taken to first relapse were reported by Lux 2004 using actuarial curves. This outcome was not reported in two of the studies.

Effects on resolution of EEG: In the Askalan 2003 study, resolution occurred in all six responders to vigabatrin and in all three responders to ACTH. In the Lux 2004 investigation, resolution occurred in 20 of 28 responders to vigabatrin for whom it was measured, compared with 26 of 32 responders to tetracosactide or high-dose prednisolone for whom it was measured. In the Vigevano 1997 study, resolution occurred in 4 of 11 participants responding to vigabatrin and in 11 of 14 participants responding to ACTH. Combining these three studies revealed that resolution occurred in 30 of 45 participants responding to vigabatrin and in 40 of 49 participants responding to ACTH (Peto OR 0.38, 95% CI 0.15 to 0.99).

Effects on psychomotor development:

Investigators in the Askalan 2003 study administered the Bayley Scales of Infant Development at trial entry and subsequently at 3, 12, and 24 months' follow-up.

The Lux 2004 study reported that Vineland Adaptive Behaviour Scales (VABS) were obtained in 101 of 102 survivors (Figure 1) at 14 months of age. This study reported that no significant differences were noted in mean composite scores between tetracosactide and high-dose prednisolone and vigabatrin (78.6 (SD 16.8) vs 77.5 (SD 12.7); difference 1.0, 95% CI -4.9 to 7.0; t99 = 0.35, P = 0.73). Mean composite scores were significantly higher in infants with no identified aetiology than in infants with a known aetiology (83.8 (SD 16.5) vs 73.3 (SD 11.4); difference 10.5, 95% CI 4.9 to 16.1; t97 = 3.74, P = 0.0003). Composite scores were also significantly higher in infants who had cessation of spasms than in those who did not (80.9 (SD 15.0) vs 72.7 (SD 13.4); difference 8.2, 95% CI 2.2 to 14.2; t99 = 2.70, P = 0.008). No significant differences in composite scores were noted between those with an interval between onset of spasms and treatment (lead time to treatment) of less than one month and those with an interval greater than one month (where this interval was known) for all infants (80.3 (SD 14.9) vs 76.1 (SD 15.1); difference 4.3, 95% CI -2.2 to 10.7; t89 = 1.32, P = 0.19) and for the subgroup with no identified underlying aetiology (87.4 (SD 16.2) vs 81.9 (SD 16.1); difference 5.5, 95% CI -5.1 to 16.1; t38 = 1.05, P = 0.30). Analysis of composite scores using ANOVA showed a significant interaction between treatment and underlying aetiology (F1,95 = 6.78, P = 0.011). For the subgroup of infants with no identified underlying aetiology, mean composite scores were higher in infants allocated hormone treatment than in those allocated vigabatrin (88.2 (SD 17.3) vs 78.9 (SD 14.3); difference 9.3, 95% CI 1.2 to 17.3; least significant difference test t95 = 2.28, P = 0.025). If the two infants whose aetiology was not fully investigated were added to the no identified aetiology group (the sensitivity analysis with the most extreme data against the detected effect), the interaction remained significant (F1,97 = 5.02, P = 0.027), but the difference between means was not significant (86.9 (SD 18.1) vs 79.5 (SD 14.3); difference 7.4, 95% CI -0.7 to 15.4; t97 = 1.82, P = 0.071). The mean composite scores in the group with an identified aetiology did not differ significantly between treatment groups (hormone 70.8 (SD 11.1) vs vigabatrin 75.9 (SD 11.3); difference 5.0, 95% CI -2.3 to 12.4; least significance difference test t95 = 1.36, P = 0.18).

The Lux study (Lux 2004) reported that VABS were obtained at four years of age in 77 infants of the original cohort. VABS scores for the 39 infants allocated hormonal treatment were 60 compared with 50 for the 38 infants allocated vigabatrin (Mann-Whitney U = 575; P = .091; median difference 8 (95%CI -1 to 19). For those with no identified aetiology, VABS scores were 96 for the 21 infants allocated hormonal treatment compared with 63 for the 16 allocated vigabatrin (Mann Whitney U = 98.5; P = 0.033; median difference 14, 95% CI 1 to 42. VABS scores for those with proven aetiology were 45 for the 18 infants allocated hormonal treatment compared with 50 for the 21 allocated vigabatrin (Mann-Whitney U = 179.5; P = 0.79; median difference -1, 95% CI -11 to 7).

The Vigevano report (Vigevano 1997) stated that there was no apparent difference in cognitive development between the two groups during follow-up, which ranged from 9 to 44 months. They did not state how they assessed development and did not quantify it any further.

Effects on subsequent seizure rates:

The Askalan study (Askalan 2003) stated that the ACTH responder and two of the vigabatrin responders subsequently developed epilepsy during the two-year follow-up period; no further information was available. The Lux 2004 study (reported 2005) reported seizure rates at 14 months of age. They indicated that in the vigabatrin group, seven participants continued to have infantile spasms (no other seizures present) compared with six in the hormonal group. In the vigabatrin group, seven participants were free of spasms but had other seizures present, compared with 13 in the hormonal group; and five participants in the vigabatrin group had both infantile spasms and other seizures present compared with eight in the hormonal group. None of these comparisons were statistically significant. The Lux study (Lux 2004) also reported seizure rates at 4 years of age. They indicated that in the vigabatrin group, five participants continued to have infantile spasms, as did five participants in the hormonal group. All ten participants had seizures of other types. In addition, in the vigabatrin group, 16 participants were free of spasms but had other seizures present, compared with 12 in the hormonal group. None of these comparisons were statistically significant.

The Vigevano study (Vigevano 1997) found the incidence of other types of epileptic seizures to be 25% in both groups during follow-up at 9 to 44 months. No study considered subsequent epilepsy rates at five years of age. Combining these studies revealed that 27 of 81 participants receiving vigabatrin had seizures at follow-up compared with t33 of 77 participants receiving hormonal treatment, Peto OR ratio 0.69 (95% CI 0.36 to 1.32).

Side effects:

The Lux trial (Lux 2004) stopped vigabatrin in two participants because of drowsiness and vomiting; tetracosactide in one participant because of a rash; and high-dose prednisolone in one participant because of irritability, vomiting, and high blood pressure. In the Vigevano trial (Vigevano 1997), investigators stopped vigabatrin in one participant because of excessive irritability, and ACTH in one participant with no reason given.

Deaths:

The Lux study (Lux 2004) reported nine deaths during follow-up to four years of age; five occurred in the vigabatrin group and four in the hormonal group. One participant in the Vigevano 1997 study died, but the cause was not given, and it was unclear to which treatment group they had been randomly assigned.

ACTH versus high-dose prednisolone

One study (Lux 2004) was included (nested within the comparison of vigabatrin with hormonal treatment).

Effects on cessation of spasms: 19 of 25 participants treated with ACTH (40 to 60 units/alternate days) had cessation of their spasms compared with 21 of 30 participants treated with prednisolone (40 to 60 mg/day) (Peto OR 1.36, 95% CI 0.41 to 4.53).
Effects on time taken to achieve cessation were not reported as an outcome in this study.
Effects on reduction in the number of spasms were not reported as an outcome in this study.
Effects on relapse rates were not reported individually for these two treatments in this study.
Effects on time taken to relapse were not reported as an outcome in this study.
Effects on resolution of EEG: 16 of 18 ACTH responders, in whom it was measured, showed resolution of their EEG compared with 10 of 14 of the prednisolone responders (Peto OR 3.20, 95% CI 0.49 to 20.81).
Effects on psychomotor development were not reported individually for these two treatments in this study.
Effects on subsequent seizure rates were not reported individually for these two treatments in this study.

Side effects:

ACTH was withdrawn in one participant because of a rash, and prednisolone in one participant because of vomiting.

Deaths:

Two deaths were reported during follow-up to 14 months.

ACTH versus low-dose prednisone

Two studies (Baram 1996; Hrachovy 1983) were included. However, it is known that prednisone requires metabolism to convert it to the active form, prednisolone. It is now known that at birth, the newborn infant has poor capacity of HSD11B1 to reduce prednisone to prednisolone, and that this ability only slowly increases over the first six months of life. Hence, prednisone and prednisolone should not be considered equivalent treatments for infants, and it is likely that prednisone will not be as effective as prednisolone in the treatment of infantile spasms.

Effects on cessation of spasms:

Baram (Baram 1996) compared 15 participants treated with natural ACTH (150 units/m2/day) with 14 treated with prednisone (2 mg/kg/day), and showed ACTH to be superior to prednisone, with cessation of spasms in 13 of 15 participants and 4 of 14 participants, respectively. Hrachovy (Hrachovy 1983) compared 12 participants treated with ACTH (20 to 30 units/day) with 12 participants treated with prednisone (2 mg/kg/day). In the initial phase of the trial, 5 of 12 participants treated with ACTH had complete cessation of spasms compared with 4 of 12 treated with prednisone. Combining these two studies revealed that ACTH stopped the spasms in 18 of 27 participants compared with prednisone in 8 of 26 participants (Peto OR 4.19, 95% CI 1.42 to 12.35).
Effects on time taken to achieve cessation:

In the Barram study (Baram 1996), on average, the 13 responders to ACTH took 3.2 days (range 1 to 7 days, median 2 days) to achieve complete cessation of spasms, and the 4 responders to prednisone took an average of 4 days (range 2 to 7 days, median 3.5 days; WMD -0.8, 95% CI -3.3 to 1.7).
Effects on reduction in the number of spasms were not reported as an outcome in these studies.
Effects on relapse rates:

In the Barram study (Baram 1996), 2 of the 13 participants who responded to ACTH relapsed, and none of the four responders to prednisone relapsed. In the Hrachovy trial (Hrachovy 1983), three of the five participants who responded to ACTH relapsed, and one of the four responders to prednisone also relapsed.

Overall, for Barram (Baram 1996), 11 participants who responded to ACTH remained spasm free and the four responders to prednisone also remained spasm free. For Hrachovy (Hrachovy 1983), two participants successfully treated with ACTH remained spasm free, and three successfully treated with prednisone also remained spasm free within the study period. The combined Peto OR for these two studies was 2.6 (95% CI 0.8 to 8.1).
Effects on time taken to relapse were not reported as an outcome in these studies.
Effects on resolution of EEG:

For Baram (Baram 1996), this study showed resolution of their EEG in 13 of 15 participants treated with ACTH compared with 4 of 14 participants treated with prednisone (Peto OR 10.10, 95% CI 2.36 to 43.19). For Hrachovy (Hrachovy 1983), 5 of 12 participants treated with ACTH had resolution of their EEG, but this was not reported for the group treated with prednisone, and their results were not therefore included in the OR calculation.

Effects on psychomotor development were not reported as an outcome in these studies.
Effects on subsequent seizure rates:

In the Baram study (Baram 1996), investigators compared ACTH with prednisone and found that seven participants in both groups developed other seizure types over the period of follow-up, 2 to 48 months. They did not report subsequent epilepsy rates at five years of age.
Side effects were not reported in these studies.
Deaths were not reported in these studies.

High-dose ACTH versus low-dose ACTH

Three studies were included (Hrachovy 1994; Yanagaki 1999; Shu 2009 ). Hrachovy and Shu used natural ACTH, and Yanagaki used synthetic ACTH. This made direct comparison difficult. Yanagaki stated that the dose equivalent of 1 IU of natural ACTH is 0.025 IU of synthetic ACTH. The calculation of the dose to be given also varied: some doses were calculated on body weight, some on surface area, and others were given a standard dose regardless of weight, height, or age. The results must be interpreted with caution.

Effects on cessation of spasms: The Hrachovy study (Hrachovy 1994) randomly assigned 30 participants to high-dose ACTH (150 units/m2/day for three weeks with a tapering dose over a further nine weeks), and 29 participants to low-dose ACTH (20 to 30 units/day for two to six weeks, tapering over one week). Of 30 participants treated with high-dose therapy, 13 had complete cessation of spasms, and of 29 participants treated with low-dose therapy, 14 had complete cessation of spasms. The Yanagaki trial (Yanagaki 1999) randomly assigned 13 participants to receive 1 unit /kg/day (equivalent to 40 units/kg/day of natural ACTH) for two weeks, tapering over a further two weeks, and 13 participants to receive 0.2 units/kg/day (equivalent to 8 units/kg/day of natural ACTH) for two weeks, tapering over a further two weeks. Among those treated with high-dose ACTH, spasm cessation occurred in 11 of 13 participants compared with 9 of 13 participants treated with low-dose ACTH. Combining these two studies revealed that high-dose ACTH stopped the spasms in 79.5% participants compared with 76.5% of participants with low-dose ACTH (Peto OR 1.1, 95% CI 0.4 to 2.6). The Shu study (Shu 2009) randomly assigned 30 children to receive either conventional-dose (high-dose) ACTH of 50 IU/day for 2 weeks, then tapered off to zero over 2 weeks, or low-dose ACTH of 0.4 IU/kg/day for 2 weeks, increased to 1 IU/kg /day if no response was seen, and then tapered to zero over 2 weeks. Investigators report that 53% of the high-dose group had a "good initial response" compared with 60% of participants in the low-dose group.
Effects on time taken to achieve cessation of spasms were not reported as an outcome in these studies.
Effects on reduction in the number of spasms were not considered as an outcome in these studies.
Effects on relapse rates: In the Hrachovy study (Hrachovy 1994), 2 of the 13 participants who responded to high-dose ACTH relapsed, and 2 of the 14 participants who responded to low-dose ACTH also relapsed. In the Yanagaki study (Yanagaki 1999), 3 of the 11 participants who responded to high-dose ACTH relapsed, and three of the nine responders to low-dose ACTH relapsed. The Shu study (Shu 2009) reported no difference in relapse rates between the two groups.
The Hrachovy study (Hrachovy 1994) reported that 11 participants who responded to high-dose ACTH remained spasm free and 12 participants who responded to low-dose ACTH also remained spasm free. In the Yanagaki study (Yanagaki 1999), eight participants who responded to high-dose ACTH remained spasm free and six participants who responded to low-dose ACTH remained spasm free. Combining these two studies yields a Peto OR of 0.8 (95% CI 0.3 to 1.9).
Effects on time taken to relapse were not reported in these studies.
Effects on resolution of the EEG: In the Hrachovy study (Hrachovy 1994), 3 of 13 responders to high-dose ACTH showed resolution of their EEG, as did 3 of 14 responders to low-dose ACTH. Yanagaki (Yanagaki 1999) reported that 8 of 13 treated with high-dose ACTH showed resolution of their EEG, as did 7 of 12 treated with low-dose ACTH. Combining these two studies yields a Peto OR for remaining spasm free participants of 1.9 (95% CI 0.6 to 6.2). The Shu study (Shu 2009) reported no significant differences between the two groups.
Effects on psychomotor development: The Yanagaki study (Yanagaki 1999) compared high-dose ACTH with low-dose ACTH and used the Japanese Tumor Scale to evaluate developmental status at entry into the trial as well as in 17 responders who were followed up for longer than one year. Investigators found no significant differences in developmental quotients between the two groups at the end of the follow-up period. However, they considered only psychomotor development in the responders, which gives rise to a biased analysis; this should be interpreted with caution. Investigators did not consider long-term developmental status at five years of age.
Effects on subsequent seizure rates were not reported as an outcome in these studies.
Side effects were not reported as an outcome in the Hrachovy and Yanagaki studies (Hrachovy 1994, Yanagaki 1999). The Shu study (Shu 2009) reported that ACTH therapy–related adverse effects were seen in 20% of the conventional (high-dose) group as compared with 93% of the low-dose group.
No deaths were reported in these studies.

Nitrazepam versus ACTH

One study was included (Dreifuss 1986).

Effects on cessation of spasms were not reported as an outcome in this study.
Effects on time taken to achieve cessation of spasms were not reported as an outcome in this study.
Effects on reduction in the number of spasms: The Dreifuss study (Dreifuss 1986) showed greater than 50% reduction in spasms in 66% of the group treated with nitrazepam compared with 50% in the group treated with ACTH (Peto OR 2.0, 95% CI 0.7 to 5.9).
Effects on relapse rates were not reported as an outcome in this study.
Effects on time taken to relapse were not reported as an outcome in this study.
Effects on resolution of EEG were not reported as an outcome in this study.
Effects on psychomotor development were not reported as an outcome in this study.
Effects on subsequent seizure rates were not reported as an outcome in this study.
Side effects: Investigators in the Dreifuss study (Dreifuss 1986) had to withdraw six patients who were receiving ACTH; hypertension and meleana were given as two causes for withdrawal.
Deaths: One patient in the ACTH group died, but no cause was found at autopsy.

Methysergide versus alpha-methylparatyrosine

One study was included (Hrachovy 1989).

Effects on cessation of spasms: The Hrachovy study (Hrachovy 1989) found that only one patient (8%) treated with methysergide and two (16%) treated with alpha-methylparatyrosine responded to therapy (Peto OR 0.5, 95% CI 0.1 to 5.2).
Effects on time taken to achieve cessation of spasms were not reported as an outcome in this study.
Effects on reduction in the number of spasms: The Hrachovy study (Hrachovy 1989) showed a greater than 50% reduction in spasms in 25% of the group treated with methysergide compared with 17% in the group treated with alpha-methylparatyrosine (Peto OR 1.6, 95% CI 0.2 to 11.2).
Effects on relapse rates: No patient treated with methysergide remained spasm free, and only one of the two patients successfully treated with alpha-methylparatyrosine remained spasm free (Peto OR 0.14, 95% CI 0.0 to 6.8).
Effects on time taken to relapse were not reported as an outcome in this study.
Effects on resolution of EEG were not reported as an outcome in this study.
Effects on psychomotor development: Investigators in the Hrachovy study (Hrachovy 1989) used the Denver Developmental Screening Test to evaluate developmental status at entry into the trial and at three weeks post entry. They stated, in their results section, that two patients in each group showed developmental improvement, but they did not quantify this further. They did not report developmental status at three months post entry or at five years of age.
Effects on subsequent seizure rates were not reported as an outcome in this study.
No side effects severe enough to warrant stopping treatment were reported in this study.
No deaths were reported in this study.

Ganaxolone versus placebo

One study was included (Tsai 2009).

The results of this study have been presented only as an abstract.

The authors state that ganaxolone did not significantly decrease spasm cluster frequency, nor did it increase responder rate, compared with placebo. Somnolence, lethargy, and irritability were reported slightly more frequently in ganaxolone-treated participants.

Vigabatrin with Flunarizine versus vigabatrin alone

One study was included (Bitton 2012).

Effects on cessation of spasms: The Bitton study (Bitton 2012) found that in 17 (50%) of 34 patients treated with vigabatrin and flunarizine, spasms stopped within a two-week period compared with 21 (62%) of 34 patients treated with vigabatrin alone.

Effects on time taken to achieve cessation of spasms was not reported as an outcome in this study.

Effects on reduction in the number of spasms were not reported as an outcome in this study.

Effects on relapse rates were not reported as an outcome in this study.

Effects on resolution of EEG were not reported as an outcome in this study.

Effects on psychomotor development: Bayley and Vineland results were available at baseline and at 24 months in 45 children. No significant difference in the BSID developmental quotient were noted between flunarizine and placebo-treated children at baseline (44.3 ± 35.5 vs 30.9 ± 29.8; P = 0.18) or 24 months later (56.9 ± 33.3 vs 46 ± 34.2; P = 0.29). However, the 10 flunarizine-treated children with no identified etiology had better outcomes than eight controls at 24 months on both the Vineland Scale (84.1 ± 11.3 vs 72.3 ± 9.8; P = 0.03) and the Bayley Scale (87.6 ± 14.7 vs 69.9 ± 25.3; P = 0.07).

Effects on subsequent seizure rates were not reported as an outcome in this study.

Side effects were not reported in this study.

Six deaths were reported in this study—three in each group—but the causes of death are not given.

ACTH with magnesium sulphate versus ACTH alone

One study was included (Zou 2010).

Effects on cessation of spasms: a total of 38 participants had complete cessation of spasms at 4 weeks among 8 of 19 (42%) treated with ACTH alone compared with 12 of 19 (63%) treated with ACTH and magnesium sulphate (Peto OR, 95% CI).
Effects on time taken to achieve cessation of spasms were not reported as an outcome in this study.
Effects on reduction in the number of spasms: At 4 weeks, seven participants treated with ACTH alone had greater than 50% reduction in spasms among ten still with spasms, and three treated with ACTH and magnesium sulphate had greater than 50% reduction in spasms among the six still with spasms.

Effects on relapse rates: None of eight responders receiving ACTH alone but two of twelve responders given ACTH and magnesium sulphate relapsed.

Effects on time taken to relapse were not reported as an outcome in this study.
Effects on resolution of EEG: Five of the 19 patients treated with ACTH alone had "normalisation" of the EEG, compared with 9 of 19 receiving ACTH and magnesium sulphate.

Effects on psychomotor development were reported as an outcome in this study, but were given as within-group comparisons before and after treatment.
Effects on subsequent seizure rates were not clearly reported as an outcome in this study.
No side effects severe enough to warrant stopping treatment were reported in this study.
No deaths were reported.

Discussion

Infantile spasms were first described in 1841 in a letter that Dr West wrote to The Lancet. Since that time, at least 30 drugs have been tried and reported in the literature, but treatment remains problematic because of the variable availability of treatments, adverse reactions, incomplete response rates, and the long-term prognosis, which remains poor for both psychomotor development and subsequent development of other seizure types.

Despite the wealth of literature on this subject, most studies have been open prospective or retrospective trials. Most of the RCTs that have been undertaken include only small numbers of participants, with the result that individually, they have little power. In addition, studies have used widely varying treatments, dosage regimens, and outcome measures, making meta-analysis difficult to perform. We found no RCTs comparing therapies such as pyridoxine that are first-line therapies in some countries, including Japan. Some therapies (e.g. gamma-globulins, pyridoxine, lamotrigine, topiramate) have been described in case reports or in uncontrolled trials. In addition, many different preparations have been used. This is due in part to the availability of different compounds in different countries. In Europe and Japan, synthetic ACTH may be available, whilst in the United States, natural ACTH derivatives have been used until recently, when costs made them less attractive. Vigabatrin was not licensed for use in the United States or in Japan until recently. ACTH can be given only as an injection (intramuscularly, intravenously, or into the peritoneal cavity). Infants find injections distressing, and many centres use oral preparations (usually prednisolone) in preference: The use of prednisone probably should not be recommended. This makes analysis of the efficacy of individual 'steroid' preparations difficult.

Despite an extensive literature search, we found only 18 RCTs, and the overall reported methodology was poor. Only two studies had considered the number of participants required to determine whether one treatment was more efficacious than another. One study based its calculations on the assumption that the trial drug would be 50% more efficacious than the comparator drug in stopping the spasms (this trial did not find such a difference). The other study based its calculations on the assumption that 250 infants would be required to give 90% power for finding a 20% difference between the two treatment groups. Investigators were able to recruit 107 infants, giving nearly 80% power to see the reported difference of 23%.

Although all studies claimed to be randomised, less than half stated how randomisation had been performed, and only four studies reported adequate concealment of randomisation. Ideally, to minimise performance bias, trials should be double-blinded, although perhaps it is not surprising that only four such trials have been performed. Blinding recipients (or parents) and staff administering the therapies remains problematic. ACTH can be given only by intramuscular injection, but the other drugs used in the included RCTs are administered orally; nowadays, it is generally considered to be unethical to administer placebo intramuscular injections to young infants. In addition, many of the side effects of steroids, for example, weight gain and hypertension, are common and are easily recognised, making blinding difficult.

Some would argue that it is now proved that spasms are causally linked to the development of severe learning difficulties. Thus delaying treatment for long or administering a placebo would now be considered unethical. Finally, six studies stated that they had lost participants to follow-up. Not all remaining studies clearly stated that all participants entered into the trials had been successfully followed up for the duration of the study. We had planned to look for possible sources of clinical heterogeneity between studies, for example, sex, age at spasm onset, age at trial entry, and delay to treatment, and whether or not participants had received previous treatment. Unfortunately, this information often was not available, and complex analysis would have been required for any conclusions to be drawn. However, it is possible that some of these sources of heterogeneity are important in determining the outcome. For example, researchers in the Baram study (Baram 1996) found ACTH to be more efficacious than prednisone but had three times as many girls as boys in the ACTH group and roughly equal numbers in the prednisone group. It is possible, therefore, that if girls have a better outcome than boys, this may have contributed to the differences in response noted between the two groups.

When developing our protocol, we tried to provide definitions for terms such as infantile spasms, cessation of spasms, relapse, and so forth. We had planned to establish whether different definitions used by authors for infantile spasms might affect which participants would be included in a trial, thereby affecting outcomes. It became clear from our review of the literature that few authors had clearly defined the term infantile spasms. It was clear that some trials using entry criteria that defined the need for hypsarrhythmia in fact entered infants who did not meet this criterion. Some of these issues have now been considered by an expert group using Delphi methodology to avoid bias (Lux 2004b). No consensus was obtained on a definition of hypsarrhythmia, but other recommendations will be incorporated in our next update. This will require a revised protocol.

Inclusion and exclusion criteria varied, and some studies excluded infants with other seizure types. This is likely to have an impact on the outcome. For the purpose of this review, we accepted that any RCT stating that its participants had infantile spasms would be included, whether or not other seizure types were included.

The exact relationship between clinical spasms and typical or even modified (or atypical) hypsarrhythmia on the EEG is not clear. Not all infants with clinical evidence of infantile spasms are reported to have hypsarrhythmia. For those in whom it is present, it is not always present all of the time. It is not known whether the presence or absence of hypsarrhythmia affects long-term outcomes. Again, we had hoped that collecting and combining data from studies might help clarify these areas of uncertainty. Unfortunately, few studies provided sufficient data to allow any conclusions to be drawn. Neither will meta-analysis be reliable until hypsarrhythmia is reliably defined.

Infantile spasms are associated with a poor long-term prognosis. Most affected infants will also have severe learning difficulties, with high cost implications not only for patients and their families and their quality of life, but also for health services. Therefore, if any single treatment was shown to improve long-term psychomotor development and to reduce the risk of development of further seizure types, it would probably be considered to be more efficacious than other treatments, regardless of its effect on the spasms themselves or on EEG features. Although with time the spasms will resolve, many patients continue to have other forms of epilepsy into adult life, and about one-fifth will progress to Lennox Gastaut syndrome (Beaumanoir 1992). Therefore, these patients will require anticonvulsants and will continue to suffer the effects of epilepsy, often for life. Very little information on developmental outcome is available; the most important information has come from the Lux studies (Lux 2004, 2005). This suggests that early control of spasms might improve development in those with no proven underlying aetiology, but the finding is not robust and longer-term follow-up is awaited. The effect of time from onset of spasms to the start of treatment has not been reported, except in Lux 2005, in which no difference between less than one month and longer than one month was reported. Little information is available on subsequent seizure rates, but the literature suggests that this may not be influenced by treatment.

The spasms themselves are certainly distressing for parents and for caregivers. It is possible that they are also distressing for the infant, as many cry immediately following a spasm or appear to be in pain and distressed. The long-term benefits of different therapies remain uncertain, and more research on this topic is needed. Two studies have shown that placebo is not as good as active treatment in resolving the spasms. The strongest evidence we found suggests that hormonal treatment leads to resolution of spasms faster and in more infants than treatment with vigabatrin. Response without subsequent relapse may be no different, but if child development might be better after hormonal treatment, this might make hormonal treatments more attractive. More information and further research are required. We found no other RCTs comparing other forms of therapies, such as sodium valproate, benzodiazepines, pyridoxine, or phenobarbitone, in terms of cessation of spasms. Although four studies did consider reduction in the number of spasms, it is no longer clear to us that a reduction in spasms is of any real benefit to a patient. Complete control of spasms has to be the main objective.

One small study that looked at participants with infantile spasms and an underlying diagnosis of tuberous sclerosis found vigabatrin to be more efficacious than hydrocortisone in stopping the spasms in this group of participants—a finding also reported in some open studies (e.g. Hancock 1999). One study (Elterman 2010) suggested that vigabatrin was more effective in stopping the spasms in participants with tuberous sclerosis as compared with participants without a diagnosis of tuberous sclerosis. However, in participants without a diagnosis of tuberous sclerosis, the best available therapy remains unclear. Another study excluded participants with a known diagnosis of tuberous sclerosis or at high risk for tuberous sclerosis, and researchers suggested that hormonal treatment (tetracosactide or prednisolone) might be more efficacious in people who do not have an underlying diagnosis of tuberous sclerosis. No study has looked specifically at any other subgroup(s) of participants.

Some open, prospective studies have suggested that vigabatrin might control spasms more quickly than steroids. Five RCTs considered the time taken to cessation of spasms. Two small studies showed no difference between vigabatrin and a hormonal treatment, but one study found hormonal treatments to be quicker at resolving spasms than vigabatrin. This finding would be supported by the speed of response to hormonal treatment described by Baram 1996.

When we developed our protocol for this review, we decided to look at relapse rates as it was felt that this would be a useful outcome for clinicians. One treatment may be more efficacious in stopping the spasms in the short term, but if most of those patients subsequently relapse, another treatment with a lower but more sustained success rate might be considered more beneficial. When developing our analysis plan, we decided that we would analyse this information as dichotomous data. However, when we came to extract and analyse these data from the trials, it became apparent that they are censored data, that is, not all participants entered into a trial would be eligible for this outcome; only those in whom the spasms stopped could potentially relapse. We have, therefore, attempted to clarify this by also reporting how many participants remained spasm free, that is, never relapsed (an outcome for which all participants entered into any one trial were eligible). However, we believe that a more appropriate method of analysis would be to look at survival curves, and we will consider this as a more appropriate method of analysis in future updates of this review if individual patient data become available.

We had also planned to look at subgroup analysis by aetiological subgroup and by clinical features, but unfortunately, because of the small numbers of trials and patients and the lack of available detail, this was not possible. It was possible to consider participants with an underlying diagnosis of tuberous sclerosis as a subgroup in only two studies. One underpowered study (Chiron 1997) showed vigabatrin to be more efficacious than hydrocortisone in stopping the spasms in this group of participants. All participants in this study who were treated with vigabatrin had complete cessation of spasms, and only one participant relapsed within the study period. However, the numbers were small, with only eleven participants in each group, and these results should be treated with caution, even though this result mimics the findings of open studies. Further trials with larger numbers of patients are required to confirm (or refute) the findings of this study. The second study (Elterman 2010) found that 52% of 25 participants with tuberous sclerosis responded to vigabatrin within two weeks of treatment compared with 16% of 117 participants without tuberous sclerosis. Researchers stated that within three months, all participants with tuberous sclerosis who continued with vigabatrin were free of infantile spasms; however, two participants had been withdrawn from vigabatrin at this point, at least one because of lack of efficacy. These participants might also have received other treatments by three months. It is not clear, therefore, whether vigabatrin was responsible for the cessation of their spasms.

Side effects (adverse reactions) must always be taken into account. Any drug with unacceptable side effects will not be used in clinical practice. Likewise, if two drugs show similar beneficial effects, the one with the safer drug profile should be used in preference. Within these studies, few side effects or deaths were reported. However, both steroids and vigabatrin have potentially serious side effects. Steroids are associated with potentially life-threatening problems, including depression of the immune system and modified response to infection leading to overwhelming sepsis. Less serious side effects, for example, hypertension, are often transient but nevertheless have the potential to cause morbidity. Minor side effects, estimated to occur in two-thirds of patients, include behavioural changes, especially irritability, changes in appetite, weight gain, and alteration in sleep patterns. In addition, some forms of steroids (e.g. ACTH) involve daily or alternate-day intramuscular injections. However, reports have described both asymptomatic and symptomatic visual field defects with loss of peripheral vision to varying degrees in adults and children treated with vigabatrin. This appears to occur most commonly in patients who have been treated with vigabatrin for longer than six months; it does not appear to be reversible on withdrawal of the vigabatrin. The main difficulty in infants is that they cannot be tested or monitored for visual field defects (a child must be at least 11 years old to be able to co-operate and complete a reliable test). However, on balance, the devastating effects of infantile spasms and the poor response rate on placebo suggest that the risks of developing the severe side effects of these therapies are less important than the risk of not using the therapies.

The overwhelming conclusion is that more studies are required on this disorder, in particular studies that report long-term developmental and epilepsy outcomes.

Authors' conclusions

Implications for practice

Hormonal treatment (prednisolone (not prednisone), tetracosactide depot, and ACTH) resolves spasms faster than vigabatrin and in more infants, but it is not yet clear whether this will result in better long-term outcomes. Hormonal treatment might improve development in those with no proven underlying neurological disease, but the evidence is not robust. It is important to note that if prednisolone or vigabatrin is used, high dosage is recommended. The optimum dose of ACTH (or tetracosactide) is not yet known. Vigabatrin may be the treatment of choice in tuberous sclerosis. Other treatments cannot be recommended on the basis of the evidence reviewed. No evidence from the RCTs indicates that it is necessary to treat EEG features rather than spasms. A clear statement on the optimum treatment for infantile spasms requires further research. Until additional study results become available, other factors (in particular, underlying aetiology and adverse reactions) will influence parents and clinicians about individual treatment choices.

Implications for research

Further trials are needed with good methodology, standardised and valid outcome measures, larger numbers of participants, and detailed reporting, including other underlying neurological diseases, gender, time to treatment, time to response, EEG outcome, deaths, epilepsy outcome, and development. Adverse reactions need to be fully reported. Long-term follow-up of development is essential. As has been stated, infantile spasms are a relatively rare disorder, so if investigators are to recruit large numbers of patients into randomised controlled trials, continuing multi-centre collaboration is required.

Acknowledgements

We would like to acknowledge Philip Milner who was an author on a previous version of this review (see Contributions of authors).

Bath Unit for Research in Paediatrics.

Cow and Gate.

The Tuberous Sclerosis Association.

The Castang Foundation.

The NIHR through local CLRN funding.

Data and analyses

Download statistical data

Comparison 1. Vigabatrin versus placebo
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Cessation of spasms140Peto Odds Ratio (Peto, Fixed, 95% CI)4.05 [0.93, 17.52]
2 Resolution of hypsarrhythmia19Peto Odds Ratio (Peto, Fixed, 95% CI)2.36 [0.10, 54.60]
3 Relapse rates - number of patients who remain spasm free140Peto Odds Ratio (Peto, Fixed, 95% CI)8.23 [0.81, 84.07]
Analysis 1.1.

Comparison 1 Vigabatrin versus placebo, Outcome 1 Cessation of spasms.

Analysis 1.2.

Comparison 1 Vigabatrin versus placebo, Outcome 2 Resolution of hypsarrhythmia.

Analysis 1.3.

Comparison 1 Vigabatrin versus placebo, Outcome 3 Relapse rates - number of patients who remain spasm free.

Comparison 2. Vigabatrin versus hydrocortisone
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Cessation of spasms122Peto Odds Ratio (Peto, Fixed, 95% CI)13.80 [2.21, 86.35]
2 Time taken to cessation of spasms116Mean Difference (IV, Fixed, 95% CI)-8.8 [-19.22, 1.62]
Analysis 2.1.

Comparison 2 Vigabatrin versus hydrocortisone, Outcome 1 Cessation of spasms.

Analysis 2.2.

Comparison 2 Vigabatrin versus hydrocortisone, Outcome 2 Time taken to cessation of spasms.

Comparison 3. ACTH versus low-dose prednisone
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Cessation of spasms253Peto Odds Ratio (Peto, Fixed, 95% CI)4.19 [1.42, 12.34]
2 Time taken to cessation of spasms117Mean Difference (IV, Fixed, 95% CI)-0.80 [-3.29, 1.69]
3 Relapse rates - number of patients who remain spasm free253Peto Odds Ratio (Peto, Fixed, 95% CI)2.57 [0.81, 8.11]
4 Resolution of hypsarrhythmia129Peto Odds Ratio (Peto, Fixed, 95% CI)10.10 [2.36, 43.19]
Analysis 3.1.

Comparison 3 ACTH versus low-dose prednisone, Outcome 1 Cessation of spasms.

Analysis 3.2.

Comparison 3 ACTH versus low-dose prednisone, Outcome 2 Time taken to cessation of spasms.

Analysis 3.3.

Comparison 3 ACTH versus low-dose prednisone, Outcome 3 Relapse rates - number of patients who remain spasm free.

Analysis 3.4.

Comparison 3 ACTH versus low-dose prednisone, Outcome 4 Resolution of hypsarrhythmia.

Comparison 4. High-dose ACTH versus low-dose ACTH
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Cessation of spasms285Peto Odds Ratio (Peto, Fixed, 95% CI)1.06 [0.44, 2.56]
2 Relapse rates - number of patients who remain spasm free285Peto Odds Ratio (Peto, Fixed, 95% CI)0.79 [0.33, 1.88]
3 Resolution of hypsarrhythmia252Peto Odds Ratio (Peto, Fixed, 95% CI)1.94 [0.60, 6.20]
Analysis 4.1.

Comparison 4 High-dose ACTH versus low-dose ACTH, Outcome 1 Cessation of spasms.

Analysis 4.2.

Comparison 4 High-dose ACTH versus low-dose ACTH, Outcome 2 Relapse rates - number of patients who remain spasm free.

Analysis 4.3.

Comparison 4 High-dose ACTH versus low-dose ACTH, Outcome 3 Resolution of hypsarrhythmia.

Comparison 5. Nitrazepam versus ACTH
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Reduction in spasms151Peto Odds Ratio (Peto, Fixed, 95% CI)1.96 [0.65, 5.93]
Analysis 5.1.

Comparison 5 Nitrazepam versus ACTH, Outcome 1 Reduction in spasms.

Comparison 6. Methysergide versus methylparatyrosine
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Cessation of spasms124Peto Odds Ratio (Peto, Fixed, 95% CI)0.48 [0.05, 5.15]
2 Reduction in spasms124Peto Odds Ratio (Peto, Fixed, 95% CI)1.62 [0.24, 11.17]
3 Relapse rates - number of patients who remain spasm free124Peto Odds Ratio (Peto, Fixed, 95% CI)0.14 [0.00, 6.82]
Analysis 6.1.

Comparison 6 Methysergide versus methylparatyrosine, Outcome 1 Cessation of spasms.

Analysis 6.2.

Comparison 6 Methysergide versus methylparatyrosine, Outcome 2 Reduction in spasms.

Analysis 6.3.

Comparison 6 Methysergide versus methylparatyrosine, Outcome 3 Relapse rates - number of patients who remain spasm free.

Comparison 7. Low-dose vigabatrin versus high-dose vigabatrin
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Cessation of spasms1221Peto Odds Ratio (Peto, Fixed, 95% CI)0.41 [0.18, 0.95]
2 Resolution of EEG1142Peto Odds Ratio (Peto, Fixed, 95% CI)0.24 [0.11, 0.52]
3 Relapse rates125Peto Odds Ratio (Peto, Fixed, 95% CI)2.57 [0.27, 24.32]
Analysis 7.1.

Comparison 7 Low-dose vigabatrin versus high-dose vigabatrin, Outcome 1 Cessation of spasms.

Analysis 7.2.

Comparison 7 Low-dose vigabatrin versus high-dose vigabatrin, Outcome 2 Resolution of EEG.

Analysis 7.3.

Comparison 7 Low-dose vigabatrin versus high-dose vigabatrin, Outcome 3 Relapse rates.

Comparison 8. Sulthiame versus placebo
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Cessation of spasms151Peto Odds Ratio (Peto, Fixed, 95% CI)5.13 [1.22, 21.47]
2 Resolution of EEG151Peto Odds Ratio (Peto, Fixed, 95% CI)5.13 [1.22, 21.47]
Analysis 8.1.

Comparison 8 Sulthiame versus placebo, Outcome 1 Cessation of spasms.

Analysis 8.2.

Comparison 8 Sulthiame versus placebo, Outcome 2 Resolution of EEG.

Comparison 9. Vigabatrin versus hormonal treatment
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Cessation of spasms3158Peto Odds Ratio (Peto, Fixed, 95% CI)0.42 [0.21, 0.80]
2 Resolution of EEG394Peto Odds Ratio (Peto, Fixed, 95% CI)0.38 [0.15, 0.99]
3 Seizures at follow-up3158Peto Odds Ratio (Peto, Fixed, 95% CI)0.69 [0.36, 1.32]
Analysis 9.1.

Comparison 9 Vigabatrin versus hormonal treatment, Outcome 1 Cessation of spasms.

Analysis 9.2.

Comparison 9 Vigabatrin versus hormonal treatment, Outcome 2 Resolution of EEG.

Analysis 9.3.

Comparison 9 Vigabatrin versus hormonal treatment, Outcome 3 Seizures at follow-up.

Comparison 10. ACTH versus high-dose prednisolone
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Cessation of infantile spasms155Peto Odds Ratio (Peto, Fixed, 95% CI)1.35 [0.41, 4.38]
2 Resolution of EEG132Peto Odds Ratio (Peto, Fixed, 95% CI)3.04 [0.52, 17.66]
Analysis 10.1.

Comparison 10 ACTH versus high-dose prednisolone, Outcome 1 Cessation of infantile spasms.

Analysis 10.2.

Comparison 10 ACTH versus high-dose prednisolone, Outcome 2 Resolution of EEG.

Comparison 11. ACTH with magnesium sulphate versus ACTH alone
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Cessation of spasms138Peto Odds Ratio (Peto, Fixed, 95% CI)0.34 [0.09, 1.24]
2 Resolution of EEG138Peto Odds Ratio (Peto, Fixed, 95% CI)0.41 [0.11, 1.52]
Analysis 11.1.

Comparison 11 ACTH with magnesium sulphate versus ACTH alone, Outcome 1 Cessation of spasms.

Analysis 11.2.

Comparison 11 ACTH with magnesium sulphate versus ACTH alone, Outcome 2 Resolution of EEG.

Comparison 12. Vigabatrin +Flunarizine vs Vigabatrin + Placebo
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Cessation of spasms168Peto Odds Ratio (Peto, Fixed, 95% CI)0.62 [0.24, 1.62]
Analysis 12.1.

Comparison 12 Vigabatrin +Flunarizine vs Vigabatrin + Placebo, Outcome 1 Cessation of spasms.

Appendices

Appendix 1. CENTRAL search strategy

#1 MeSH descriptor Spasms, Infantile explode all trees
#2 (infantile spasm*)
#3 "West syndrome" or "West's syndrome"
#4 (salaam spasm*)
#5 (hypsarrhythmia)
#6 (#1 OR #2 OR #3 OR #4 OR #5)

Appendix 2. MEDLINE search strategy

The following search is based on the Cochrane highly sensitive search strategy for identifying randomised trials in MEDLINE as set out in Appendix 5b of the Cochrane Handbook for Systematic Reviews of Interventions (version 4.2.4, updated March 2005) (Higgins 2005).

1. Spasms, Infantile/

2. West$ syndrome.tw.

3. salaam spasm$.tw.

4. hypsarrhythmia.tw.

5. infantile spasm$.tw.

6. 1 or 2 or 3 or 4 or 5

7. randomized controlled trial.pt.

8. controlled clinical trial.pt.

9. exp Randomized Controlled Trials/

10. exp Random Allocation/

11. exp Double-Blind Method/

12. exp Single-Blind Method/

13. 7 or 8 or 9 or 10 or 11 or 12

14. (animals not human).sh.

15. 13 not 14

16. clinical trial.pt.

17. Clinical Trial/

18. (clin$ adj trial$).ab,ti.

19. ((singl$ or doubl$ or trebl$ or tripl$) adj (blind$ or mask$)).ab,ti.

20. exp PLACEBOS/

21. placebo$.ab,ti.

22. random$.ab,ti.

23. exp Research Design/

24. 16 or 17 or 18 or 19 or 20 or 21 or 22 or 23

25. (animals not human).sh.

26. 24 not 25

27. 15 or 26

28. 27 and 6

What's new

DateEventDescription
15 January 2014AmendedErrors in 'Data and Analyses' section checked and corrected. Conclusions remain the same.

Contributions of authors

Eleanor Hancock was primarily responsible for all aspects of this review including protocol design, undertaking the searches, the database of studies, data collection and extraction, data analysis, and except for this review, presentation of the results.

John Osborne agreed and approved the review at all stages, independently evaluated which studies should be included and excluded from the review, and also independently extracted the data from the included studies. He edited the presentation of the results for this review.

Philip Milner agreed on and approved the review, independently evaluated which studies should be included in and excluded from the review, and also independently extracted the data from the included studies up until February 2002 but including some in this review.

Stuart Edwards replaced Philip Milner and has agreed and approved the review, independently evaluated which studies should be included and excluded from the review, and also independently extracted data from the included studies from February 2002 until February 2010.

Declarations of interest

Dr Hancock, Professor Osborne, and Dr Edwards are members of the Trial Steering Committee for two multi-centre, randomised, controlled trials in infantile spasms (United Kingdom Infantile Spasms Study (UKISS) and the International Collaborative Infantile Spasms Study).

Professor Milner was the chairman of the Data Monitoring and Ethics Committee for UKISS.

Professor Osborne is also a co-author for the Consensus Statement of the West Delphi Group.

Sources of support

Internal sources

  • Royal United Hospital, Combe Park, Bath, UK.

  • Chelsea and Westminster Hospital, Fulham Road, London, UK.

  • University of Bath, School of Health, UK.

  • Great Ormond Street Hospital, Great Ormond Street, London, UK.

  • Surrey Heath and Woking Primary Care Trust, UK.

External sources

  • Bath Unit for Research into Paediatrics, UK.

    Salary funding for all reviews

  • The Tuberous Sclerosis Association, UK.

    Salary funding for first review

  • Cow and Gate, UK.

    Salary funding for first review

  • NHS Research and Development Funding, UK.

    Salary funding for latest review

  • Castang Foundation, UK.

    Funding for salary

Differences between protocol and review

We have included information on studies comparing combination treatment with a single treatment as this is now an obvious way forward for improving the treatment of infantile spasms.

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Appleton 1999

MethodsRCT
Participants40 participants with a clinical definition of infantile spasms and classical or modified hypsarrhythmia on EEG
InterventionsVigabatrin
Placebo
Outcomes(1) Cessation of spasms.
(2) Reduction in spasms.
(3) Relapse rate.
(4) Resolution of EEG.
Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Low riskA - Adequate

Askalan 2003

MethodsRCT
Participants9 participants with a clinical definition of infantile spasms and video-EEG with hypsarrhythmia
InterventionsACTH
Vigabatrin
Outcomes(1) Cessation of spasms.
(2) Resolution of EEG.
(3) Cognitive outcomes - short term.
(4) Language scores.
(5) Evolution of epilepsy and autism.
Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Baram 1996

MethodsRCT
Participants29 participants with a diagnosis of infantile spasms confirmed by hypsarrhythmia on video-EEG
InterventionsACTH
Prednisone
Outcomes(1) Cessation of spasms.
(2) Relapse rates.
(3) Resolution of EEG.
Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Low riskA - Adequate

Bitton 2012

MethodsRCT
Participants68 patients with infantile spasms
Interventions

Flunarizine

Placebo

OutcomesDevelopmental outcome at 24 months after initiation of treatment
Notes 

Chiron 1997

MethodsRCT
Participants22 participants with a clinical diagnosis of infantile spasms or epileptic spasms or diffuse interictal activity on EEG
InterventionsVigabatrin
Hydrocortisone
Outcomes(1) Cessation of spasms.
(2) Relapse rates.
(3) Resolution of EEG.
Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Debus 2004

MethodsRCT
Participants51 participants with a clinical diagnosis of infantile spasms and hypsarrhythmia on EEG
InterventionsSulthiame
Placebo
Outcomes(1) Cessation of spasms.
(2) Resolution of EEG.
(3) Relapse rates.
Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Low riskA - Adequate

Dreifuss 1986

MethodsRCT
Participants52 participants with a diagnosis of hypsarrhythmia or modified hypsarrhythmia on EEG
InterventionsACTH
Nitrazepam
OutcomesReduction in seizures
Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Dyken 1985

MethodsRCT
Participants17 participants with a clinical definition of infantile spasms and hypsarrhythmia on EEG
InterventionsValproate
Placebo
OutcomesReduction in spasms
Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Elterman 2010

MethodsRCT
Participants221 participants with a clinical diagnoses of infantile spasms and hypsarrhythmia, modified hypsarrhythmia, or multifocal spike wave discharges on EEG
InterventionsLow-dose vigabatrin
High-dose Vigabatrin
Outcomes(1) Cessation of spasms.
(2) Time to cessation of spasms.
Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Hrachovy 1983

MethodsRCT
Participants24 participants with a clinical definition of infantile spasms and hypsarrhythmia on EEG
InterventionsACTH
Prednisone
OutcomesCessation of spasms
Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Hrachovy 1989

MethodsRCT
Participants24 participants with a clinical definition of infantile spasms and hypsarrhythmia on EEG
InterventionsMethysergide
Alpha-methylparatyrosine
Outcomes(1) Cessation of spasms.
(2) Reduction in spasms.
Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Hrachovy 1994

MethodsRCT
Participants59 participants with a clinical definition of infantile spasms and hypsarrhythmia on EEG
InterventionsACTH high dose
ACTH low dose
Outcomes(1) Cessation of spasms.
(2) Relapse rates.
(3) Resolution of EEG.
Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Lux 2004

MethodsRCT
Participants107 participants with a clinical definition of infantile spasms and hypsarrhythmia on EEG
InterventionsVigabatrin
Hormonal treatment
Outcomes(1) Cessation of spasms.
(2) Resolution of EEG.
(3) Relapse rates.
(4) Development 14 months.
(5) Seizure rates.
Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Low riskA - Adequate

Shu 2009

MethodsRCT
Participants30 patients with infantile spasms
Interventions

Conventional (high)-dose ACTH

Low-dose ACTH

Outcomes

(1) Initial response.

(2) EEG findings.

(3) Relapse rates.

(4) Time to relapse.

(5) Long-term outcomes.

(6) Adverse effects.

Notes 

Tsai 2009

Methods 
Participants56 infants with infantile spasms
Interventions

Ganaxolone

Placebo

Outcomes

(1) Spasm cluster frequency.

(2) Adverse effects.

Notes 

Vigevano 1997

MethodsRCT
Participants42 participants with a diagnosis according to the ILAE classification
InterventionsVigabatrin
ACTH
Outcomes(1) Cessation of spasms.
(2) Relapse rates.
(3) Subsequent epilepsy rates.
(4) Resolution of hypsarrhythmia.
Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Yanagaki 1999

MethodsRCT
Participants26 participants with a diagnosis confirmed by hypsarrhythmia on video-EEG
InterventionsACTH high dose
ACTH low dose
Outcomes(1) Cessation of spasms.
(2) Relapse rates.
(3) Resolution of hypsarrhythmia.
(4) Developmental outcome.
Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Zou 2010

  1. a

    EEG: electroencephalogram
    RCT: randomised controlled trial

MethodsRCT
Participants38 infants with infantile spasms and hypsarrhythmia
InterventionsACTH alone compared with ACTH with magnesium sulphate
OutcomesCessation of spasms, normalisation of EEG and development
Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear risk"Treatment was assigned using a table of numbers following a central 1:1 randomisation."

Characteristics of excluded studies [ordered by study ID]

StudyReason for exclusion
  1. a

    RCT: randomised controlled trial

Al Ajlouni 2005Not an RCT
Albsoul-Younes 2004Not an RCT
Appleton 1993Not an RCT
Arvio 2005Not an RCT
Azam 2005Not an RCT
Bachman 1982Not an RCT
Bensch 1977Not infantile spasms
Besag 2004Not an RCT
Bingel 2003Not an RCT
Blennow 1986Not an RCT
Bobele 1994Not an RCT
Boel 2004Not an RCT
Brodie 1996Not an RCT
Browne 1978Not an RCT
Buoni 2004Not an RCT
Chamberlain 1996Not an RCT
Chevrie 1984Not an RCT
Connelly 1993Not an RCT
Debus 2002Not an RCT
Duncan 2002Not an RCT
Duse 1996Not an RCT
Francois 2003Not an RCT
Gibbs 1965Not an RCT
Glauser 2002Not an RCT
Glaze 1985Not an RCT
Glaze 1986Considered only changes on CT as an outcome measure
Grosso 2005Not an RCT
Haines 1994Not an RCT
Hosain 2006Not an RCT
Hrachovy 1979Not an RCT
Inanaga 1989Not infantile spasms
Ishii 2002Not an RCT
Klein 1970Not an RCT
Li 2004Not an RCT
Lombroso 1983Not an RCT
Lortie 1997Not an RCT
Lotze 2004Not an RCT
Luna 1989Not infantile spasms
Mackay 2004Not an RCT
Matsumoto 1987Not an RCT
Mikaeloff 2003Not an RCT
Mikati 2002Not an RCT
Mimaki 1998Not an RCT
Mumford 1989Not an RCT
Mumford 1994Not an RCT
Nabbout 2003Not an RCT
Nolte 1988Not an RCT
O'Donohoe 1977Not an RCT
Oguni 2006Not an RCT
Ohtahara 2006Not an RCT
Oka 2004Not an RCT
Pietz 1993Not an RCT
Ramsay 1984Not an RCT
Scher 2003Not an RCT
Schlanger 2002Not an RCT
Schlumberger 1994Not an RCT
Suzuki 1996Not an RCT
Suzuki 2002Not an RCT
Tada 1991Not an RCT
Trevanthan 2003Not an RCT
Trompetter 1977Not infantile spasms
Vassella 1973Not an RCT
Veggiotti 1994Not an RCT
Wallace 1995Not an RCT
Weinmann 1967Not an RCT
Willig 1980Not an RCT
Wohlrab 1998Not an RCT
Yagi 2004Not an RCT
Yamamoto 1998Not an RCT
Yamauchi 2004Not an RCT
Zafeiriou 1996Not an RCT

Characteristics of ongoing studies [ordered by study ID]

ICISS

Trial name or titleInternational Collaboration Infantile Spasms Study
Methods 
ParticipantsInfants with a diagnosis of infantile spasms
InterventionsVigabatrin
High-dose prednisolone
Tetracosactide
OutcomesCessation of spasms
Long-term development
Long-term seizures
Starting dateMay 2007
Contact information

Dr Finbar O'Callaghan,

Chief Investigator, ICISS

Children's Centre,
Royal United Hospital
Bath BA1 3NG
UK

Notes 

ISRCTN 36757519

Trial name or title 
Methods 
Participants 
Interventions 
Outcomes 
Starting date 
Contact information 
Notes 

ISRCTN Surgery/AEDs

  1. a

    EEG: electroencephalogram

Trial name or title 
Methods 
Participants 
Interventions 
Outcomes 
Starting date 
Contact information 
Notes 

Ancillary