Article Text

Download PDFPDF

A systematic review of safety monitoring and drug toxicity in published randomised controlled trials of antiepileptic drugs in children over a 10-year period
  1. Mark Anderson,
  2. Imti Choonara
  1. University of Nottingham, Derbyshire Children's Hospital, Derby, UK
  1. Correspondence to Mark Anderson, Academic Division of Child Health, University of Nottingham, The Medical School, Derbyshire Children's Hospital, Uttoxeter Road, Derby DE22 3DT, UK; mark.anderson7{at}


Aim To determine the extent of drug toxicity experienced by children and methods to detect it in randomised controlled trials (RCTs) of antiepileptic drugs (AEDs) published over a 10-year period.

Methods EMBASE, MedLine and the Cochrane Library were searched for RCTs of AEDs which recruited children (age <16 years) and were published between 1998 and 2007. Methodological, efficacy and toxicity data were extracted.

Results 54 trials were identified (26 monotherapy and 28 adjunctive therapy designs). 29 trials recruited adults and children but only 10 of these reported the number of children recruited. Only three of the 29 trials analysed the paediatric patients separately. Adverse event rates were the primary toxicity measure, but the method of assessing incidence of side effects was frequently omitted and only six trials documented the use of a standardised method. Only three trials had an independent safety monitoring committee.

Conclusions RCTs of AEDs potentially provide the best source of information on relative efficacy, but rarely provide reliable information on side effects. Children are frequently exposed to the risks of RCTs but clinical information regarding efficacy and safety in paediatric patients is often not given. Trials involving paediatric patients alongside adult patients should always include information regarding the number of children recruited and details of efficacy and safety separately. All trials involving children should have an independent safety monitoring committee.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Epilepsy is the commonest chronic neurological disorder of childhood with a median reported incidence rate of about 80 cases per 100 000 children per year1 and a prevalence of approximately 1%.2 It is associated with significantly higher mortality in children compared with adults3 and therefore effective and safe therapy for this age group is vital.

The last 20 years have witnessed a dramatic expansion in the number of drugs available to treat epilepsy, with over seven newer antiepileptic drugs (AEDs) licensed for use in children.4 As the range of anticonvulsants widens, the broader choice makes drug selection for individual patients more complicated. Adequately constructed clinical trials to demonstrate and compare effectiveness are required to allow well-informed, evidence-based treatment decisions to be made. Unfortunately, as previous reviews have demonstrated,5 6 few trials of AEDs have been performed in children with epilepsy compared with their adult counterparts and, overall, the quality of the trials is poor. In addition, many of these trials were conducted for licensing purposes. Trials aimed at satisfying regulatory authorities with respect to efficacy are frequently inadequate for clinicians seeking to make pragmatic decisions regarding treatment.7

What is already known on this topic

  • ▶. The quality of randomised controlled trial (RCT) data relating to the efficacy of antiepileptic drugs (AEDs) in children is generally poor.

  • ▶. AED trials conducted for licensing purposes are often of limited use in clinical decision making.

What this study adds

  • ▶. Safety data reported in RCTs of AEDs are often of poor quality, and reporting is haphazard.

  • ▶. Children and adolescents are often included within adult trials of AEDs but paediatric efficacy and safety results are rarely given separately.

  • ▶. The majority of randomised clinical trials of AEDs lack an independent safety monitoring committee.

Treatment with AEDs is not without its risks. In a study of drug toxicity in children in the UK, they were the medication most likely to be associated with a fatal suspected adverse drug reaction (ADR),8 and in the USA, AEDs were associated with 23% of severe ADRs in a single children's hospital over a 10-year period.9 The adverse event profile of an AED is therefore important for clinical decision making and controlled clinical trials are the best initial source for this information. Frequently, these data for children are extrapolated from adult trials. As with efficacy data, however, this information is not readily generalisable – paediatric epilepsy can be distinct from adult epilepsy in aetiology, clinical features and natural history, and children may metabolise medications differently leading to altered efficacy and side effect profiles. It is therefore essential that AEDs are evaluated adequately in children in order to identify these important differences.

With these issues in mind, a systematic review of all published randomised controlled trials (RCTs) of AEDs involving children over a 10-year period was undertaken to assess the adequacy of safety assessment and reporting measures in the paediatric age group. The 10-year period 1998–2007 was chosen primarily as a sample of convenience. This period offers other advantages, however, in that it encompasses much of the period following the licensing of many of the new drugs and it follows the publication of guidance by the International League Against Epilepsy (ILAE) regarding the design of clinical trials in epilepsy.10


RCTs of AEDs in patients with epilepsy were identified via a literature search of the EMBASE and MedLine databases using the Ovid interface. The modified Cochrane search strategy11 was employed to identify trials combined with a paediatric filter using generic drug names combined with the keyword epilepsy as well as broader terms such as antiepileptic and anticonvulsant. The search strategy was based on that used by Glauser et al.5 A randomised, controlled clinical trial was defined as any parallel group study that recruited patients with epilepsy and randomly allocated them to groups comparing an AED with placebo, an active comparator or a lower dose of the AED under investigation.

Studies were included if they were published between 1998 and 2007, recruited children (aged under 16 years) prospectively, and considered seizure control as an outcome measure. The Cochrane Library of randomised controlled trials in epilepsy and the Cochrane Child Health Field Child Health Trials Register were also searched. Reference lists of studies identified were checked to ensure maximal yield. Trials were excluded if they involved treatment of status epilepticus or patients who had experienced only a single seizure or were not published in English.

The following data were extracted from the full text of the papers: age range and number of participants (including number of children in trials recruiting adults and children, if stated); type of epilepsy; duration of trial; trial design and comparator; efficacy outcome measures; and method of determination of side effects, if any, and details relating to reporting and interpretation of adverse events. In addition, relative frequencies of different classes of adverse effects were extracted and compared. Trials were initially grouped into monotherapy and adjunctive therapy designs, before being subdivided into target epilepsy types.


After removal of duplicates the initial search strategy produced 3085 results. Scanning of the abstracts of these results to exclude any publications which were obviously irrelevant or trials recruiting only adult patients aged over 16 years reduced these results to 281. Five trials which may have met the inclusion criteria were not published in English (one each in Spanish, Polish, Bulgarian, Russian and Japanese) which precluded their assessment and thus their incorporation into this review. The full text of the remaining studies was reviewed and 57 trials were identified which met the inclusion criteria. A further three trials were excluded as they were subgroup analyses of the children recruited to larger trials of adults and children. The data from these trials were incorporated into the analysis of their larger ‘parent’ trial. This resulted in a total of 54 trials included in the review. Twenty six of these were monotherapy designs and 28 were adjunctive therapy.

Monotherapy trials

There were 11 trials of AED monotherapy in children alone12,,22 (table 1) as first line treatment for newly or recently diagnosed epilepsy (focal epilepsy,12 13 absence epilepsy,16 17 benign epilepsy with centrotemporal spikes,18 19 infantile spasms20,,22 and mixed types14 15). All the trials involved a wide range of ages of children, except infantile spasms, for obvious reasons. In seven trials, seizure freedom was the primary outcome measure, with two trials using a composite measure of efficacy and tolerability, such as retention on treatment.

Table 1

Monotherapy trials involving children only

A further 15 trials of AED monotherapy recruited both adults and children. The commonest efficacy outcome measure was time to meet predefined exit criteria (nine trials). Nine trials did not even mention the number of children recruited.23,,31 Only six trials32,,37 (table 2) reported the number of children who were recruited. No trials analysed the children as a separate group in terms of efficacy, although two trials32 33 reported some adverse events in the paediatric age group separately. Two other trials35 36 published a subgroup analysis of the children as a separate publication in the review period.38 39 The children recruited to these trials were most frequently aged over 10 years, but six trials involved younger children.

Table 2

Monotherapy trials recruiting adults and children in which the number of children recruited could be determined

There were nine monotherapy trials involving lamotrigine, six with vigabatrin, six with topiramate and five with oxcarbazepine. Where a newer AED was being compared with an older drug, the comparator drug was usually carbamazepine (seven trials) or sodium valproate (four trials).

Adjunctive therapy trials

There were 14 trials of AEDs as adjunctive therapy for epilepsy recruiting children only40,,53 (table 3): seven of them were in focal epilepsy40,,46 and three in undifferentiated epilepsy.51,,53 A wide age range of children was recruited and 13 were placebo controlled. The most common outcome measure was change in seizure frequency or a related variable.

Table 3

Add-on trials involving children only

The remaining 14 trials in the adjunctive therapy group recruited both adults and children, 10 involving patients with focal epilepsy.54,,67 Only four of these trials reported the number of children who had been recruited64,,66 68 (table 4). Only one trial66 published a subgroup analysis of the children recruited in the review period.67 Nine trials were placebo controlled and three were dose controlled. As with the monotherapy trials recruiting adults and children, the age of the children was mostly over 10 years but four trials recruited younger children.

Table 4

Add-on trials recruiting adults and children in which the number of children recruited could be determined

Most of the drugs under investigation as adjunctive therapy were newer generation AEDs (most commonly, five trials involving lamotrigine, four with topiramate, three with gabapentin and three with tiagabine) and two trials studied more unusual compounds, namely antiepilepsirine (a piperine derivative, originally isolated from a traditional Chinese remedy) and black cumin seed aqueous extract (a herbal remedy, the active component of which is thymoquinone).


The most commonly used measure of tolerability and toxicity was the adverse event rate. Most studies reported that this was derived from physical examination, laboratory studies and simply ‘adverse events’. The exact means of assessing these adverse events was described in only 18 trials and most frequently this was only described as direct questioning or spontaneous report. Six trials were more systematic in their approach, three34 37 64 using the Liverpool Adverse Events Profile, an epilepsy-specific, 19-item questionnaire designed to monitor the frequency and severity of common adverse effects associated with AEDs, and one14 describing a checklist or a questionnaire, but details were sparse. The relative frequencies of the various reported adverse effects in children are classified by drug in table 5.

Table 5

Adverse event rates where reported classified by AED

Of the 25 trials recruiting children only, six did not report the number of children experiencing an adverse event, instead reporting the total number of adverse events in each group. Documented evaluation of adverse events as ADR was reported in only 11 trials and three trials failed to report the number of withdrawals related to suspected drug toxicity. Of the 29 trials recruiting both adults and children, in only one was it possible to extract information relating to adverse event rates in children. Five reported the number of children withdrawing due to suspected drug toxicity. Three trials documented the use of an independent safety monitoring committee or board.22 34 37


The inclusion of children in clinical trials of both new and old AEDs is welcomed. The exclusion, however, of a separate analysis of paediatric patients in the clinical trials that involved adults and children is ethically and scientifically unacceptable. It is of major concern that the majority of the trials that recruited both adults and children did not even mention the number of children recruited. One has to question what the rationale is for including children within a clinical trial if one is not going to report the data separately, particularly if the number of children involved is very small. It can only be assumed that it is to ensure licensing of the AED in older children.

RCTs of AEDs can demonstrate not only relative efficacy but also important information regarding safety. If two AEDs have similar efficacy, the frequency and types of expected drug toxicity are essential factors to consider when deciding which drug to use; some of this knowledge can be derived from parallel group trials. The derivation of information relating to drug adverse effects from RCTs is fraught. Firstly, these trials are not statistically powered to detect differences in adverse event rates69 and secondly, the closely-controlled dose increase regimens required by many trials in epilepsy may lead to an increase in adverse events that would not be seen in routine clinical practice. These concerns do not obviate the need for adverse event monitoring and analysis, however. It is vital that the well-being of all participants, and in particular of children, in clinical trials is safeguarded and this can only be achieved by continuous monitoring of adverse events and their subsequent evaluation as possible ADRs. The lack of independent safety monitoring committees in the vast majority of the trials, therefore, is of major concern particularly given that the drug toxicity of AEDs is well recognised.8 9 This lack of monitoring has been identified previously in other trials involving children.70 The EMEA and its Paediatric Committee now recommend an independent safety monitoring committee for any clinical trial involving children.71

In addition, adverse events in RCTs provide important initial signals for postmarketing drug safety studies, which cannot be provided by open-label studies, demonstrated particularly by the fact that a number of the double-blind studies identified in this review withdrew children from participation for suspected ADRs to placebo. Open-label studies therefore tend to overestimate ADRs. ADRs identified in RCTs, if detected in a standardised manner, are amenable to meta-analysis, partially overcoming concerns regarding statistical power.

We have attempted to analyse the relative frequency of different adverse effects experienced by the children involved in the trials (table 5). This information is limited, however, by the lack of standardised assessment of adverse events, with most trials not reporting the method of detection, and, in particular, in the trials recruiting adults and children, a complete failure to consider the fact that children may experience a different adverse event profile from adults. The VA Cooperative Studies72 73 in adult epilepsy, using a standardised questionnaire for systematic assessment of adverse events due to medication, found that drug toxicity contributed to trial exit in 40–60% of participants, much higher than most studies using only spontaneous reporting. Standardised assessment tools for adverse events in clinical trials of AEDs, either in the form of a questionnaire or a structured interview, for both adults and children, would allow further elucidation of these issues, and would allow incorporation into meta-analysis.

The reporting of data for AED trials involving children and adults is unsatisfactory from the viewpoint of children and health professionals wishing to ensure optimal management. This inadequate reporting is unacceptable, both scientifically and ethically. We urge journal editors to ensure that data for paediatric patients in publications are reported separately in order to allow clinicians to determine whether the treatment is safe and effective in children. We feel the Paediatric Committee of the EMEA should insist that the data for children recruited into clinical trials alongside adults should always be reported separately in relation to efficacy and safety. We feel that implementation of such a change will improve the evidence base for the management of children with a variety of medical conditions.



  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.