PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of bmjclinevidLink to Publisher's site
 
Clin Evid (Online). 2011; 2011: 1214.
Published online May 6, 2011.
PMCID: PMC3217777
Epilepsy (partial)
Melissa Maguire, Consultant Neurologist,# Professor Anthony G Marson, Professor of Neurology,# and Dr Sridharan Ramaratnam, Senior Consultant Neurologist#
Melissa Maguire, Leeds General Infirmary, Leeds, UK;
#Contributed equally.
AM has been paid for speaking at meetings by Johnson and Johnson, Janssen Cilag, Sanofi, and UCB SA, and by Glaxo Wellcome and UCB SA for attending conferences. SR has received hospitality from the following pharmaceutical companies: Sun Pharmaceuticals (India), Novartis (India), and Sanofi Aventis for attending conferences. SR is involved in ongoing clinical trials for UCB Pharma, Pfizer, and Johnson and Johnson. SR is also an author of some of the systematic reviews referenced in this review. MM declares that she has no competing interests.
Introduction
About 3% of people will be diagnosed with epilepsy during their lifetime, but about 70% of people with epilepsy eventually go into remission.
Methods and outcomes
We conducted a systematic review and aimed to answer the following clinical questions: What are the effects of starting antiepileptic drug treatment following a single seizure? What are the effects of drug monotherapy in people with partial epilepsy? What are the effects of additional drug treatments in people with drug-resistant partial epilepsy? What is the risk of relapse in people in remission when withdrawing antiepileptic drugs? What are the effects of behavioural and psychological treatments for people with epilepsy? What are the effects of surgery in people with drug-resistant temporal lobe epilepsy? We searched: Medline, Embase, The Cochrane Library, and other important databases up to July 2009 (Clinical Evidence reviews are updated periodically; please check our website for the most up-to-date version of this review). We included harms alerts from relevant organisations such as the US Food and Drug Administration (FDA) and the UK Medicines and Healthcare products Regulatory Agency (MHRA).
Results
We found 83 systematic reviews, RCTs, or observational studies that met our inclusion criteria. We performed a GRADE evaluation of the quality of evidence for interventions.
Conclusions
In this systematic review we present information relating to the effectiveness and safety of the following interventions: antiepileptic drugs after a single seizure; monotherapy for partial epilepsy using carbamazepine, gabapentin, lamotrigine, levetiracetam, phenobarbital, phenytoin, sodium valproate, or topiramate; addition of second-line drugs for drug-resistant partial epilepsy (allopurinol, eslicarbazepine, gabapentin, lacosamide, lamotrigine, levetiracetam, losigamone, oxcarbazepine, retigabine, tiagabine, topiramate, vigabatrin, or zonisamide); antiepileptic drug withdrawal for people with partial or generalised epilepsy who are in remission; behavioural and psychological treatments for partial or generalised epilepsy (biofeedback, cognitive behavioural therapy (CBT), educational programmes, family counselling, relaxation therapy (alone or plus behavioural modification therapy, yoga); and surgery for drug-resistant temporal lobe epilepsy ( lesionectomy, temporal lobectomy, vagus nerve stimulation as adjunctive therapy).
During their lifetime, about 3% of people will be diagnosed with epilepsy, but about 70% of people with epilepsy eventually go into remission.
After a first seizure, antiepileptic drugs may delay or prevent subsequent seizures, but they can cause adverse effects, and their long-term benefit is unknown. Antiepileptic drug treatment after a single seizure does not reduce the risk of drug refractory epilepsy in the long term.
Carbamazepine, gabapentin, lamotrigine, levetiracetam, phenobarbital, phenytoin, sodium valproate, and topiramate are widely considered effective in controlling seizures in newly diagnosed partial epilepsy, but we found no RCTs comparing them with placebo, and a placebo-controlled trial would now be considered unethical.
  • Systematic reviews found no reliable evidence on which to base a choice among antiepileptic drugs.
  • Adding second-line drugs to usual treatment reduces seizure frequency in people with drug-resistant partial epilepsy, but it increases adverse effects such as dizziness and somnolence. We don't know if any one antiepileptic drug is more likely to reduce seizures compared with the others.
CAUTION: Vigabatrin, which may be used as second-line treatment, causes concentric visual-field abnormalities in about 40% of people, which are probably irreversible.
In people who have been seizure free for at least 2 years on treatment, almost 60% of those with partial or generalised epilepsy who withdraw from antiepileptic treatment will remain seizure free, compared with almost 80% of people who continue treatment.
Educational programmes may reduce seizure frequency and improve psychosocial functioning in people with partial or generalised epilepsy, but we don't know whether relaxation, yoga, biofeedback, CBT, relaxation plus behavioural modification, or family counselling are beneficial.
There is consensus that temporal lobectomy or amygdalohippocampectomy can improve seizure control and quality of life in people with drug-resistant temporal lobe epilepsy, but they can cause neurological adverse effects.
High-level vagus nerve stimulation may reduce seizure frequency in people with drug-resistant partial seizures, but it may cause hoarseness and dyspnoea, and long-term effects are unknown. We don't know whether different stimulation cycles are more effective at reducing seizure frequency or at increasing the proportion of responders.
We don't know whether lesionectomy improves seizure control in people with drug-resistant temporal lobe epilepsy.
Definition
Epilepsy is a group of disorders rather than a single disease. Seizures can be classified by type as partial or focal (categorised as simple partial, complex partial, and secondary generalised tonic clonic seizure) or generalised (categorised as generalised tonic clonic, absence, myoclonic, tonic, and atonic seizures).[1] Temporal lobe epilepsy is a form of partial or focal epilepsy. A person is considered to have epilepsy if they have had two or more unprovoked seizures. This review deals with pharmacological treatment of single seizure that may progress to epilepsy, and with pharmacological and surgical treatments of partial epilepsy. This review also deals with behavioural and psychological treatments of any epilepsy (generalised or partial) and the risk of relapse in people with any epilepsy (generalised or partial) on withdrawing antiepileptic drugs. See also separate related review on Epilepsy (generalised) for information on pharmacological and surgical treatments of generalised epilepsy. Status epilepticus is not covered in this review.
Incidence/ Prevalence
Epilepsy (partial or generalised) is common, with an estimated average prevalence of 5.5/1000 people in Europe, [2] 6.8/1000 people in the United States, [3]and 7.5/1000 people in Australia. Prevalence rates in developing countries vary widely with studies carried in sub-Saharan Africa reporting rates of 5.2 to 74.4/1000 people, [4] studies in Asia reporting overall prevalence rates of 1.5 to 14.0/1000 people, [5]and Latin America reporting rates of 17 to 22/1000 people. [6]The annual incidence rates of epilepsy are 24 to 56/100,000 people in Europe,[2]44/100,000 in the United States, [7]63 to 158/100,000 people in sub-Saharan Africa, [4]113 to 190/100,000 people in Latin America, [6]and 28 to 60/100,000 people in Asia. [5] The worldwide incidence of single unprovoked seizures is 23 to 61/100,000 person-years. [8]About 3% of people will be diagnosed with epilepsy at some time in their lives.[9]
Aetiology/ Risk factors
Epilepsy is a symptom rather than a disease, and it may be caused by various disorders involving the brain. The causes/risk factors include birth/neonatal injuries, congenital or metabolic disorders, head injuries, tumours, infections of the brain or meninges, genetic defects, degenerative disease of the brain, cerebrovascular disease, or demyelinating disease. Epilepsy can be classified by cause.[1] Idiopathic generalised epilepsies (such as juvenile myoclonic epilepsy or childhood absence epilepsy) are largely genetic. Symptomatic epilepsies result from a known cerebral abnormality; for example, temporal lobe epilepsy may result from a congenital defect, mesial temporal sclerosis, or a tumour. Cryptogenic epilepsies are those that cannot be classified as idiopathic or symptomatic.
Prognosis
About 60% of untreated people have no further seizures during the 2 years after their first seizure.[10] Prognosis is good for most people with epilepsy. About 70% go into remission, defined as being seizure free for 5 years on or off treatment. This leaves 20% to 30% who develop chronic epilepsy, which is often treated with multiple antiepileptic drugs.[11]
Aims of intervention
To reduce the risk of subsequent seizures and to improve the prognosis of the seizure disorder; to improve quality of life; in people in remission, to withdraw antiepileptic drugs without causing seizure recurrence; to minimise adverse effects of treatment.
Outcomes
For treatment after a single seizure: Time to subsequent seizures, time to achieve remission, proportion of people achieving remission. For treatment of newly diagnosed epilepsy: Time to remission, time to first seizure after treatment, retention on allocated treatment or time to withdrawal of allocated treatment. For treatment of drug-resistant epilepsy: Percentage reduction in seizure frequency, proportion of responders (response defined as at least 50% reduction in seizure frequency). For drug withdrawal: Time to seizure recurrence. For behavioural treatments: Improvement in quality of life, reduction in anxiety, depression, and fear of seizures; coping or adjustment to epilepsy (assessed by validated measures). For all: quality of life, adverse effects.
Methods
Clinical Evidence search and appraisal July 2009. The following databases were used to identify studies for this systematic review: Medline 1966 to July 2009, Embase 1980 to July 2009, and The Cochrane Database of Systematic Reviews 2009, Issue 2 (1966 to date of issue). An additional search within The Cochrane Library was carried out for the Database of Abstracts of Reviews of Effects (DARE) and Health Technology Assessment (HTA). We also searched for retractions of studies included in the review. Abstracts of the studies retrieved from the initial search were assessed by an information specialist. Selected studies were then sent to the contributor for additional assessment, using pre-determined criteria to identify relevant studies. Study design criteria for inclusion in this review were: published systematic reviews of RCTs and RCTs in any language, at least double-blinded for drug trials, and containing more than 20 individuals of whom more than 80% were followed up. At least 3 months' follow-up was required to include studies. We excluded all studies described as "open", "open label", or not blinded unless blinding was impossible (e.g., behavioural and psychological treatments). For the questions on drug therapy and surgical interventions in people with partial epilepsy, we aimed to include studies in people with partial epilepsy only or subgroup analyses of people with partial epilepsy. However, where studies included a mixture of partial and generalised epilepsy, we included studies in which at least 60% of people had partial epilepsy. For the questions on starting antiepileptic drugs after a single seizure, stopping antiepileptic drugs in people in remission, and behavioural and psychological interventions, we included all studies on epilepsy, as the relative numbers of people with generalised or partial epilepsy were not described in these studies or were not relevant for this population. We included systematic reviews of RCTs and RCTs where harms of an included intervention were studied applying the same study design criteria for inclusion as we did for benefits. In addition we did an observational harms search for specific harms as highlighted by the contributor, peer reviewer, and editor. We searched for observational studies down to case series with a minimum of 20 cases. In addition, we use a regular surveillance protocol to capture harms alerts from organisations such as the US Food and Drug Administration (FDA) and the UK Medicines and Healthcare products Regulatory Agency (MHRA), which are added to the reviews as required. To aid readability of the numerical data in our reviews, we round many percentages to the nearest whole number. Readers should be aware of this when relating percentages to summary statistics such as relative risks (RRs) and odds ratios (ORs). We have performed a GRADE evaluation of the quality of evidence for interventions included in this review (see table ). The categorisation of the quality of the evidence (into high, moderate, low, or very low) reflects the quality of evidence available for our chosen outcomes in our defined populations of interest. These categorisations are not necessarily a reflection of the overall methodological quality of any individual study, because the Clinical Evidence population and outcome of choice may represent only a small subset of the total outcomes reported, and population included, in any individual trial. For further details of how we perform the GRADE evaluation and the scoring system we use, please see our website (www.clinicalevidence.com).
Table 1
Table 1
GRADE evaluation of interventions for Epilepsy (partial)
Glossary
Atonic seizure Momentary loss of limb muscle tone causing sudden falling to the ground or drooping of the head.
Beck Depression Inventory Standardised scale to assess depression. This instrument consists of 21 items to assess the intensity of depression. Each item is a list of 4 statements (rated 0, 1, 2, or 3), arranged in increasing severity, about a particular symptom of depression. The range of scores possible are 0 = least severe depression to 63 = most severe depression. It is recommended for people aged 13 to 80 years. Scores of more than 12 or 13 indicate the presence of depression.
Biofeedback Physiological responses such as heart rate, skin conductivity (galvanic skin response), or brain wave pattern (electroencephalogram) are monitored and visual or auditory online feedback is provided to the individual, to help him or her to actively control the physiological response.
Centers for Epidemiological Studies Depression (CES-D) Scale 20-item 4-point Likert scale, with scores that range from 0 to 60. Higher scores indicate more symptoms of depression.
Cognitive behavioural therapy A broad category of interventions designed to identify and control stress and minimise its effects, often by using intellectual experience to correct damaging thoughts and behaviour.
Crown Crisp Experiential Index Formerly known as Middlesex Hospital Questionnaire (MHQ), this is a self reported questionnaire providing information on psychoneurotic traits. It comprises 48 items with an overall score for neuroticism, with further subscores for free floating anxiety, phobic anxiety, obsessionality, somatic anxiety, depression, and hysterical anxiety. A higher score indicates more overall neurotic disorder.
Electroencephalographic (EEG) biofeedback A technique of making EEG activity apparent to a person, who is then taught to produce certain EEG waves that are believed to increase the threshold for seizures.
Hamilton Depression Rating Scale a measure of depressive symptoms using 17 items, with total scores from 0 to 54 (higher scores indicate increased severity of depression).
High-quality evidenceFurther research is very unlikely to change our confidence in the estimate of effect.
Lesionectomy Excision of a lesion consisting of a small area of abnormality (such as focal scar, vascular malformation, or tumour) in the brain where seizures originate.
Low-quality evidenceFurther research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Minnesota Multiphasic Personality Inventory (MMPI) A battery of standardised tests to assess personality (psychopathology).
Moderate-quality evidenceFurther research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Relaxation therapy Techniques to train people to control muscle tension.
SF-36 score A scale that assesses health related quality of life across eight domains: limitations in physical activities (physical component); limitations in social activities; limitations in usual role activities because of physical problems; pain; psychological distress and wellbeing (mental health component); limitations in usual role activities because of emotional problems; energy and fatigue; and general health perceptions.
Selective amygdalohippocampectomy Removal of the amygdala and hippocampus only.
Temporal lobectomy Removal of the lesion or epileptogenic area responsible for the development of complex partial seizures, which are the most common seizures associated with temporal lobe epilepsy. An en bloc anterior temporal lobectomy is a standardised operative procedure, in which 4.5–5.0 cm of the anterior lateral temporal lobe neocortex is removed, along with the amygdala, the anterior aspect of the parahippocampal gyrus, and the hippocampus in the medial portion of the temporal lobe.
Tonic clonic seizure Also known as a convulsion or "grand mal" attack. The person will become stiff (tonic) and collapse, and have generalised jerking (clonic) movements. Breathing might stop and the bladder might empty. Generalised jerking movements lasting typically for a few minutes are followed by relaxation and deep unconsciousness, before the person slowly comes round. People are often tired and confused, and may remember nothing. Tonic clonic seizures may follow simple partial or complex partial seizures (see above), where they are classified as secondary generalised tonic clonic seizures. Tonic clonic seizures occurring without warning and in the context of generalised epilepsy are classified as generalised tonic clonic seizures.
Vagus nerve stimulation A device similar to a pacemaker is implanted under the skin of the chest and connected to the left vagus nerve in the neck by a stimulator wire. By programming the device, the frequency, intensity, and duration of stimulation can be varied. Also, patients can activate the device by placing a magnet over it when a seizure occurs or is about to occur.
Very low-quality evidenceAny estimate of effect is very uncertain.
Washington Psychosocial Inventory (WPSI) A standardised battery of tests to assess adjustment in various spheres (measure of psychosocial difficulties) in people with epilepsy.

Notes
Disclaimer
The information contained in this publication is intended for medical professionals. Categories presented in Clinical Evidence indicate a judgement about the strength of the evidence available to our contributors prior to publication and the relevant importance of benefit and harms. We rely on our contributors to confirm the accuracy of the information presented and to adhere to describe accepted practices. Readers should be aware that professionals in the field may have different opinions. Because of this and regular advances in medical research we strongly recommend that readers' independently verify specified treatments and drugs including manufacturers' guidance. Also, the categories do not indicate whether a particular treatment is generally appropriate or whether it is suitable for a particular individual. Ultimately it is the readers' responsibility to make their own professional judgements, so to appropriately advise and treat their patients. To the fullest extent permitted by law, BMJ Publishing Group Limited and its editors are not responsible for any losses, injury or damage caused to any person or property (including under contract, by negligence, products liability or otherwise) whether they be direct or indirect, special, incidental or consequential, resulting from the application of the information in this publication.
Notes
Pharmacological and surgical treatments of generalised epilepsy, see review on Epilepsy (generalised). Treatment of typical absence seizures in children, see review on Absence seizures in children.
Contributor Information
Melissa Maguire, Leeds General Infirmary, Leeds, UK.
Professor Anthony G Marson, University of Liverpool, Liverpool, UK.
Dr Sridharan Ramaratnam, Apollo Hospitals, Chennai (Madras), India.
1. Commission on classification and terminology of the international league against epilepsy. Proposal for revised classification of epilepsies and epileptic syndromes. Epilepsia1989;30:389–399. [PubMed]
2. Forsgren L, Beghi E, Oun A, et al. The epidemiology of epilepsy in Europe – a systematic review. Eur J Neurol2005;12:245−253. [PubMed]
3. Hauser WA, Annegers JF, Kurland LT. Prevalence of epilepsy in Rochester, Minnesota: 1940−1980. Epilepsia1991;32:429−445. [PubMed]
4. Preux P, Druet-Cabanac M. Epidemiology and aetiology of epilepsy in sub-Saharan Africa. Lancet Neurol2005;4:21−31. [PubMed]
5. Mac TL, Tran DS, Quet F, et al. Epidemiology, aetiology, and clinical management of epilepsy in Asia: a systematic review. Lancet Neurol2007;6:533−543. [PubMed]
6. Burneo JG, Tellez-Zenteno J, Wiebe S, et al. Understanding the burden of epilepsy in Latin America: a systematic review of its prevalence and incidence. Epilepsy Res2005;66:63−74. [PubMed]
7. Hauser AW, Annegers JF, Kurland LT. Incidence of epilepsy and unprovoked seizures in Rochester, Minnesota 1935–84. Epilepsia1993;34:453–468. [PubMed]
8. Hauser WA, Beghi E, Hauser W, et al. First seizure definitions and worldwide incidence and mortality. Epilepsia2008;49(suppl 1):8−12. [PubMed]
9. Hauser WA, Kurland LT. The epidemiology of epilepsy in Rochester, Minnesota, 1935 through 1967. Epilepsia1975;16:1–66. [PubMed]
10. Berg AT, Shinnar S. The risk of seizure recurrence following a first unprovoked seizure: a quantitative review. Neurology1991;41:965–972. Search date not reported. [PubMed]
11. Cockerell OC, Johnson AL, Sander JW, et al. Remission of epilepsy: results from the national general practice study of epilepsy. Lancet1995;346:140–144. [PubMed]
12. Musicco M, Beghi E, Solari A, et al, for the FIRST Group. Treatment of first tonic clonic seizure does not improve the prognosis of epilepsy. Neurology1997;49:991–998. [PubMed]
13. Leone MA, Solari A, Beghi E, et al. Treatment of the first tonic-clonic seizure does not affect long-term remission of epilepsy. Neurology2006;67:2227–2229. [PubMed]
14. Gilad R, Lampl Y, Gabbay U, et al. Early treatment of a single generalized tonic-clonic seizure to prevent recurrence. Arch Neurol1996;53:1149–1152. [PubMed]
15. Camfield P, Camfield C, Dooley J, et al. A randomized study of carbamazepine versus no medication after a first unprovoked seizure in childhood. Neurology1989;39:851–852. [PubMed]
16. Chandra B. First seizure in adults: to treat or not to treat. Clin Neurol Neurosurg1992;94(Suppl):S61–S63. [PubMed]
17. Marson A, Jacoby A, Johnson A, et al., on behalf of the Medical Research Council MESS study Group. Immediate versus deferred antiepileptic drug treatment for early epilepsy and single seizures: a randomized controlled trial. Lancet2005;365:2007–2013. [PubMed]
18. U.S. Food and Drug Administration. Information for healthcare professionals: suicidal behavior and ideation and antiepileptic drugs. Available at: http://www.fda.gov/Drugs/DrugSafety/PostmarketDrugSafetyInformationforPatientsandProviders/ucm100192.htm (last accessed 5 May 2010).
19. First Seizure Trial Group (FIRST Group). Randomized clinical trial on the efficacy of antiepileptic drugs in reducing the risk of relapse after a first unprovoked tonic clonic seizure. Neurology1993;43:478–483. [PubMed]
20. Marson AG, Williamson PR, Hutton JL, et al. on behalf of the Epilepsy Monotherapy Trialists. Carbamazepine versus valproate monotherapy for epilepsy (Cochrane Review). In: The Cochrane Library, Issue 3, 2011. Chichester, UK: John Wiley & Sons, Ltd. Search date 2007.
21. Tudur Smith C, Marson AG, Williamson PR. Carbamazepine versus phenobarbitone monotherapy for epilepsy (Cochrane Review). In: The Cochrane Library, Issue 3, 2011. Chichester, UK: John Wiley & Sons, Ltd. Search date 2006.
22. Tudur Smith C, Marson AG, Clough HE, et al. Carbamazepine versus phenytoin monotherapy for epilepsy (Cochrane Review). In: The Cochrane Library, Issue 2, 2009. Chichester, UK: John Wiley & Sons, Ltd. Search date 2007.
23. Gamble CL, Williamson PR, Marson AG. Lamotrigine versus carbamazepine monotherapy for epilepsy. In: The Cochrane Library, Issue 2, 2009. Chichester, UK: John Wiley & Sons, Ltd. Search date 2007. [PubMed]
24. Rowan AJ, Ramsay RE, Collins JF, et al. New onset geriatric epilepsy: a randomized study of gabapentin, lamotrigine, and carbamazepine. Neurology2005;64:1868–1873. [PubMed]
25. Saetre E, Perucca E, Isojarvi J, et al. An international multicenter randomized double-blind controlled trial of lamotrigine and sustained-release carbamazepine in the treatment of newly diagnosed epilepsy in the elderly. Epilepsia2007;48:1292−1302. [PubMed]
26. Meador K, Reynolds MW, Crean S, et al. Pregnancy outcomes in women with epilepsy: a systematic review and meta-analysis of published pregnancy registries and cohorts. Epilepsy Res2008;81:1−13. [PMC free article] [PubMed]
27. Gissel T, Poulsen CS, Vestergaard P, et al. Adverse effects of antiepileptic drugs on bone mineral density in children. Expert Opin Drug Saf2007;6:267−278. [PubMed]
28. Kim SH, Lee JW, Choi KG, et al. A 6-month longitudinal study of bone mineral density with antiepileptic drug monotherapy. Epilepsy Behav2007;10:291−295. [PubMed]
29. Babayigit A, Dirik E, Bober E, et al. Adverse effects of antiepileptic drugs on bone mineral density. Pediatr Neurol2006;35:177−181. [PubMed]
30. El-Hajj FuleihanG, Dib L, Yamout B, et al. Predictors of bone density in ambulatory patients on antiepileptic drugs. Bone2008;43:149−155. [PubMed]
31. Ensrud KE, Walczak TS, Blackwell TL, et al. Antiepileptic drug use and rates of hip bone loss in older men: a prospective study. Neurology2008;71:723−730. [PMC free article] [PubMed]
32. Marson AG, Al-Kharusi AM, Alwaidh M, et al. The SANAD study of effectiveness of carbamazepine, gabapentin, lamotrigine, oxcarbazepine, or topiramate for treatment of partial epilepsy: an unblinded randomised controlled trial. Lancet2007;369:1000–1015. [PMC free article] [PubMed]
33. Brodie MJ, Perucca E, Ryvlin P, et al. Comparison of levetiracetam and controlled-release carbamazepine in newly diagnosed epilepsy. Neurology2007;68:402−408. [PubMed]
34. Taylor S, Tudur Smith C, Williamson PR, et al. Phenobarbitone versus phenytoin monotherapy for partial onset seizures and generalized onset tonic-clonic seizures. In: The Cochrane Library, Issue 2, 2009. Chichester, UK: John Wiley & Sons, Ltd. Search date 2006.
35. Tudur Smith C, Marson AG, Williamson PR. Phenytoin versus valproate monotherapy for partial onset seizures and generalized onset tonic-clonic seizures. In: The Cochrane Library, Issue 3, 2011. Chichester, UK: John Wiley & Sons, Ltd. Search date 2007.
36. Muller M, Marson AG, Williamson PR. Oxcarbazepine versus phenytoin monotherapy for epilepsy. In: The Cochrane Library, Issue 3, 2011. Chichester, UK: John Wiley & Sons, Ltd. Search date 2008. [PubMed]
37. Brodie MJ, Chadwick DW, Anhut H, et al. Gabapentin versus lamotrigine monotherapy: a double-blind comparison in newly diagnosed epilepsy. Epilepsia2002;43:993−1000. [PubMed]
38. Privitera MD, Brodie MJ, Mattson RH, et al. Topiramate, carbamazepine and valproate monotherapy: double-blind comparison in newly diagnosed epilepsy. Acta Neurol Scand2003;107:165–175. [PubMed]
39. Hunt S, Russell A, Smithson WH, et al. Topiramate in pregnancy: preliminary experience from the UK Epilepsy and Pregnancy Register. Neurology2008;71:272−276. [PubMed]
40. Hunt S, Craig J, Russell A, et al. Levetiracetam in pregnancy: preliminary experience from the UK Epilepsy and Pregnancy Register. Neurology2006;67:1876−1879. [PubMed]
41. Marson AG, Kadir ZA, Hutton JL, et al. Gabapentin add-on for drug-resistant partial epilepsy. In: The Cochrane Library, Issue 2, 2009. Chichester, UK: John Wiley & Sons, Ltd. Search date 2007.
42. Chaisewikul R, Privitera MD, Hutton JL, et al. Levetiracetam add-on for drug-resistant localization related (partial) epilepsy. In: The Cochrane Library, Issue 2, 2009. Chichester, UK: John Wiley & Sons, Ltd. Search date 2005.
43. Ramaratnam S, Marson AG, Baker GA. Lamotrigine add-on for drug-resistant partial epilepsy. In: The Cochrane Library, Issue 2, 2009. Chichester, UK: John Wiley & Sons, Ltd. Search date 2007.
44. Castillo S, Schmidt DB, White S. Oxcarbazepine add-on for drug-resistant partial epilepsy. In: The Cochrane Library, Issue 2, 2009. Chichester, UK: John Wiley & Sons, Ltd. Search date 2006.
45. Chadwick DW, Marson AG. Zonisamide add-on for drug-resistant partial epilepsy. In: The Cochrane Library, Issue 2, 2009. Chichester, UK: John Wiley & Sons, Ltd. Search date 2007.
46. Pereira J, Marson AG, Hutton JL. Tiagabine add-on for drug-resistant partial epilepsy. In: The Cochrane Library, Issue 2, 2009. Chichester, UK: John Wiley & Sons, Ltd. Search date 2008.
47. Lozsadi D, Hemming K, Marson AG, et al. Pregabalin add-on for drug-resistant partial epilepsy. In:The Cochrane Library, Issue 2, 2009. Chichester, UK: John Wiley & Sons, Ltd. Search date 2007.
48. Michael B, Marson AG, Michael B, et al. Clobazam as an add-on in the management of refractory epilepsy. In:The Cochrane Library, Issue 2, 2009. Chichester, UK: John Wiley & Sons, Ltd. Search date 2007.
49. Jette N, Hemming K, Hutton JL, et al. Topiramate add-on for drug-resistant partial epilepsy. In:The Cochrane Library, Issue 2, 2009. Chichester, UK: John Wiley & Sons, Ltd. Search date 2007.
50. Hemming K, Maguire MJ, Hutton JL, et al. Vigabatrin for refractory partial epilepsy. In:The Cochrane Library, Issue 2, 2009. Chichester, UK: John Wiley & Sons, Ltd. Search date 2008.
51. Glauser TA, Ayala R, Elterman RD, et al. Double-blind placebo-controlled trial of adjunctive levetiracetam in pediatric partial seizures. Neurology2006;66:1654–1660. [PubMed]
52. Porter RJ, Partiot A, Sachdeo R, et al. Randomized, multicenter, dose-ranging trial of retigabine for partial-onset seizures. Neurology2007;68:1197–1204. [PubMed]
53. Elger C, Bialer M, Cramer JA, et al. Eslicarbazepine acetate: a double-blind, add-on, placebo-controlled exploratory trial in adult patients with partial-onset seizures. Epilepsia2007;48:497–504. [PubMed]
54. Baulac M, Klement S, Losigamone Study Group, et al. Efficacy and safety of Losigamone in partial seizures: a randomized double-blind study. Epilepsy Res2003;55:177–189. [PubMed]
55. Togha M, Akhondzadeh S, Motamedi M, et al. Allopurinol as adjunctive therapy in intractable epilepsy: a double-blind and placebo-controlled trial. Arch Med Res2007;38:313–316. [PubMed]
56. Peltola J, Coetzee C, Jimenez F, et al. Once-daily extended-release levetiracetam as adjunctive treatment of partial-onset seizures in patients with epilepsy: a double-blind, randomized, placebo-controlled trial. Epilepsia2009;50:406−414. [PubMed]
57. Wu XY, Hong Z, Wu X, et al. Multicenter double-blind, randomized, placebo-controlled trial of levetiracetam as add-on therapy in Chinese patients with refractory partial-onset seizures. Epilepsia2009;50:398−405. [Erratum in: Epilepsia. 2009;50:971]. [PubMed]
58. Naritoku DK, Warnock CR, Messenheimer JA, et al. Lamotrigine extended-release as adjunctive therapy for partial seizures. Neurology2007;69:1610−1618. [PubMed]
59. Lee BI, Yi S, Hong SB, et al. Pregabalin add-on therapy using a flexible, optimized dose schedule in refractory partial epilepsies: a double-blind, randomized, placebo-controlled, multicenter trial. Epilepsia2009;50:464−474. [PubMed]
60. Elger C, Halasz P, Maia J, et al. Efficacy and safety of eslicarbazepine acetate as adjunctive treatment in adults with refractory partial-onset seizures: a randomized, double-blind, placebo-controlled, parallel-group phase III study. Epilepsia2009;50:454−463. [PubMed]
61. Ben-Menachem E, Biton V, Jatuzis D, et al. Efficacy and safety of oral lacosamide as adjunctive therapy in adults with partial-onset seizures. Epilepsia2007;48:1308−1317. [PubMed]
62. Halasz P, Kalviainen R, Mazurkiewicz-Beldzinska M, et al. Adjunctive lacosamide for partial-onset seizures: efficacy and safety results from a randomized controlled trial. Epilepsia2009;50:443−453. [PubMed]
63. Kalviainen R, Nousiainen I, Mantyjarvi M, et al. Vigabatrin, a gabaergic antiepileptic drug, causes concentric visual field defects. Neurology1999;53:922–926. [PubMed]
64. Otoul C, Arrigo C, van Rijckevorsel K, et al. Meta-analysis and indirect comparisons of levetiracetam with other second-generation antiepileptic drugs in partial epilepsy. Clin Neuropharmacol2005;28:72−78. [PubMed]
65. Medical Research Council Antiepileptic Drug Withdrawal Study Group. Prognostic index for recurrence of seizures after remission of epilepsy. BMJ1993;306:1374–1378. [PMC free article] [PubMed]
66. Lossius MI, Hessen E, Mowinckel P, et al. Consequences of antiepileptic drug withdrawal: a randomized, double-blind study (Akershus Study). Epilepsia2008;49:455−463. [PubMed]
67. Medical Research Council Antiepileptic Drug Withdrawal Study Group. Randomised study of antiepileptic drug withdrawal in patients in remission. Lancet1991;337:1175–1180. [PubMed]
68. Ranganathan LN, Ramaratnam S. Rapid versus slow withdrawal of antiepileptic drugs. In: The Cochrane Library, Issue 2, 2009. Chichester, UK: John Wiley & Sons, Ltd. Search date 2005. [PubMed]
69. Ramaratnam S, Baker GA, Goldstein L. Psychological treatments for epilepsy (Cochrane Review). In: The Cochrane Library, Issue 2, 2009. Chichester, UK: John Wiley & Sons, Ltd. Search date 2007.
70. Dahl J, Melin L, Lund L. Effects of a contingent relaxation treatment program on adults with refractory epileptic seizures. Epilepsia1987;28:125–132. [PubMed]
71. Puskarich CA, Whitman S, Dell J, et al. Controlled examination of effects of progressive relaxation training on seizure reduction. Epilepsia1992;33:675–680. [PubMed]
72. Rousseau A, Hermann B, Whitman S. Effects of progressive relaxation on epilepsy: analysis of a series of cases. Psychol Rep1985;57:1203–1212. [PubMed]
73. Ramaratnam S, Sridharan K. Yoga for epilepsy (Cochrane Review). In: The Cochrane Library, Issue 2, 2009. Chichester, UK: John Wiley & Sons, Ltd. Search date 2006. [PubMed]
74. Panjwani U, Selvamurthy W, Singh SH, et al. Effect of sahaja yoga practice on seizure control and EEG changes in patients of epilepsy. Indian J Med Res1996;103:165–172. [PubMed]
75. Lantz DL, Sterman MB. Neuropsychological assessment of subjects with uncontrolled epilepsy: effects of EEG feedback training. Epilepsia1988;29:163–171. [PubMed]
76. Tan SY, Bruni J. Cognitive behavior therapy with adult patients with epilepsy: a controlled outcome study. Epilepsia1986;27:225–233. [PubMed]
77. Lundgren T, Dahl J, Melin L, et al. Evaluation of acceptance and commitment therapy for drug refractory epilepsy: a randomized controlled trial in South Africa – a pilot study. Epilepsia2006;47:2173–2179. [PubMed]
78. Martinović Z, Simonović P, Djokić R, et al. Preventing depression in adolescents with epilepsy. Epilepsy Behav2006;9:619–624. [PubMed]
79. Davis GR, Armstrong HE Jr, Donovan DM, et al. Cognitive-behavioral treatment of depressed affect among epileptics: preliminary findings. J Clin Psychol1984;40:930–935. [PubMed]
80. Helgeson DC, Mittan R, Tan SY, et al. Sepulveda epilepsy education: the efficacy of a psychoeducational treatment program in treating medical and psychosocial aspects of epilepsy. Epilepsia1990;31:75–82. [PubMed]
81. Lewis MA, Salas I, De La Sota A, et al. Randomized trial of a program to enhance the competencies of children with epilepsy. Epilepsia1990;31:101–109. [PubMed]
82. May TW, Pifflin M. The efficacy of an educations treatment program for patients with epilepsy (MOSES): results of a controlled, randomized study. Epilepsia2002;43:539–549. [PubMed]
83. Olley BO, Osinowo HO, Brieger WR. Psycho-educational therapy among Nigerian adult patients with epilepsy: a controlled outcome study. Patient Educ Couns2001;42:25–33. [PubMed]
84. Dahl J, Melin L, Brorson LO, et al. Effects of a broad-spectrum behavior modification treatment program on children with refractory epileptic seizures. Epilepsia1985;26:303–309. [PubMed]
85. Sultana SM. A study on the psychological factors and the effect of psychological treatment in intractable epilepsy PhD Thesis, University of Madras, India, 1987.
86. Earl WL. Job stability and family counseling. Epilepsia1986;27:215–219. [PubMed]
87. Cucherat M. Meta-analysis of surgery trials in refractory epilepsy. Rev Neurol (Paris) 2004;160 Spec No 1:5S232–5S240. Search date 2003. [In French] [PubMed]
88. Wiebe S, Blume WT, Girvin JP, et al. A randomized, controlled trial of surgery for temporal lobe epilepsy. N Engl J Med2001;345:311–318. [PubMed]
89. Tellez-Zenteno JF, Dhar R, Wiebe S, et al. Long-term seizure outcomes following epilepsy surgery: a systematic review and meta-analysis. Brain2005;128:1188–1198. [PubMed]
90. Téllez-Zenteno JF, Dhar R, Hernandez-Ronquillo L, et al. Long-term outcomes in epilepsy surgery: antiepileptic drugs, mortality, cognitive and psychosocial aspects. Brain2007;130:334–345. [PubMed]
91. Engel J Jr, Wiebe S, French J, et al. Practice parameter: temporal lobe and localized neocortical resections for epilepsy: report of the Quality Standards Subcommittee of the American Academy of Neurology, in association with the American Epilepsy Society and the American Association of Neurological Surgeons. Neurology 2003;60:538–547. Search date 2001. [Erratum in: Neurology2003;60:1396]. [PubMed]
92. Privitera MD, Welty TE, Ficker DM, Welge J. Vagus nerve stimulation for partial seizures (Cochrane Review). In: The Cochrane Library, Issue 2, 2009. Chichester, UK: John Wiley & Sons, Ltd. Search date 2007. [PubMed]
93. DeGiorgio C, Heck C, Bunch S, et al. Vagus nerve stimulation for epilepsy: randomized comparison of three stimulation paradigms. Neurology2005;65:317–319. [PubMed]

Antiepileptic drugs after a single seizure
Summary
SEIZURE FREQUENCY Compared with no treatment/no immediate treatment: Treatment with antiepileptic drugs after a single seizure may be more effective at reducing the risk of relapse at 1 to 3 years, but not at increasing the proportion of people who achieve remission at 2 to 5 years ( very low-quality evidence ). NOTE We found no evidence that immediate treatment after a single seizure alters long-term prognosis. Long-term antiepileptic drug treatment may be potentially harmful. Antiepileptic drugs are associated with idiosyncratic reactions, teratogenesis, suicidality, and cognitive adverse effects.
Benefits
Immediate treatment versus no treatment/no immediate treatment:
We found no systematic review. We found four RCTs comparing immediate treatment versus no immediate treatment/no treatment after a first unprovoked seizure.[12] [13] [14] [15] [16] We found one RCT comparing immediate versus deferred antiepileptic drug treatment in people with one (56%) or more (44%) previous unprovoked seizures (see comment below).[17]
The first RCT compared immediate treatment after a first unprovoked seizure versus no immediate treatment.[12] People were randomised within 7 days of their first tonic clonic seizure. The RCT found half as many second seizures at 2 years with immediate treatment compared with no immediate treatment (419 people, 44% women, 27% aged under 16 years, 66% aged 16–60 years, 7% aged over 60 years; AR for relapse: 24% with immediate treatment v 42% with no immediate treatment; HR 0.5, 95% CI 0.3 to 0.6). However, the RCT found no significant difference in the proportion of people achieving a 2-year remission in seizures (146/215 [68%] with immediate treatment v 122/204 [60%] with no immediate treatment; RR 1.22, 95% CI 0.97 to 1.56; RR adjusted for time of starting treatment 1.04, 95% CI 0.82 to 1.30). People who had 2 seizure-free years after a relapse were included in the 2-year remission figures; this included people in the no-immediate-treatment group who started treatment after having a second seizure. A follow-up of this study compared treatment of a first unprovoked seizure (treated group) versus treatment only in the event of seizure recurrence (untreated group) over a 13-year period.[13] It found no significant difference in the proportion of people achieving 2-year or 5-year remission from seizures (2-year remission: 333 people; 174/215 [81%] with treatment v 159/204 [78%] with no treatment; RR 1.16, 95% CI 0.94 to 1.38; 5-year remission: 264 people; 82/128 [64%] with treatment v 86/136 [63%] with no treatment; RR 1.04, 95% CI 0.73 to 1.35). In the original study, 86 people were lost to follow-up, but analysis was by intention to treat. People who had 2 or 5 seizure-free years after a relapse were included in the remission figures; this included people in the no-immediate-treatment group who started treatment after having a second seizure. Follow-up after 3 years was <80%. The probability of achieving a 2-year remission and a 5-year remission was more favourable for the immediate-treatment group, but this difference was not sustained at 3 or at 10 years' follow-up.[13]
The second RCT compared immediate treatment (carbamazepine, or sodium valproate if carbamazepine not tolerated) versus no treatment.[14] The RCT found that immediate treatment significantly reduced the risk of relapse over 3 years of follow-up (91 people aged 18–50 years, presenting to hospital within 24 hours of a first unprovoked seizure; AR of recurrent epileptic attack after 36 months: 10/45 [22%] with immediate treatment v 29/42 [71%] with no treatment; P less than 0.05).
The third RCT compared immediate treatment with carbamazepine versus no treatment.[15] The RCT found that immediate treatment significantly reduced the risk of relapse over 12 months (31 children presenting within 1 month of a first unprovoked afebrile seizure; AR of recurrent unprovoked seizure: 2/14 [14%] with immediate treatment v 9/17 [53%] with no treatment; P = 0.03).
The fourth RCT compared immediate treatment with sodium valproate versus placebo.[16] It found that sodium valproate reduced the proportion of people having a recurrent seizure over 12 months, but the significance of this reduction was not assessed (228 adults aged 16–79 presenting within 2 weeks after their first seizure; seizure recurrence: 5/113 [4%] with sodium valproate v 63/115 [55%] with placebo; significance assessment not performed).[16] Although treatment allocation was not explicitly described in the fourth study, the contributors of this Clinical Evidence review have confirmed that allocation was random.
The fifth RCT found that immediate treatment increased time to first and second subsequent seizure (1443 people aged over 1 month old with a history of at least one unprovoked seizure; HR for first seizure: 1.4, 95% CI 1.2 to 1.7; HR for second seizure: 1.3, 95% CI 1.1 to 1.6).[17] Immediate treatment reduced the time to achieve 2-year remission of seizures (P = 0.023). However, at 5 years there was no difference in the proportion of people who had been seizure free for between 3 and 5 years (76% with immediate treatment v 77% with deferred treatment; ARR +0.2%, 95% CI –5.5% to +5.8%). It found that carbamazepine monotherapy significantly increased time to first seizure, but that the effects of sodium valproate were inconclusive.[17]
Harms
Antiepileptic drugs have been associated with increased risk of suicidal behaviour and ideation.[18] Some antiepileptic drugs have also been associated with adverse outcomes in pregnancy and adverse effects on bone mineral density (see harms described under each option in question on monotherapy in people with partial epilepsy).
Immediate treatment versus no treatment/no immediate treatment:
The adverse effects of antiepileptic drugs are well known, and include idiosyncratic reactions, teratogenesis, and cognitive effects. Interim analysis in the first RCT found that 14/204 (7%) of participants discontinued antiepileptic drug treatment owing to adverse events (not further specified).[19]No information on adverse events was reported in the final analyses.[12]
The second RCT did not report on adverse events.[14]
In the third RCT 4/16 (25%) children discontinued carbamazepine owing to adverse effects (somnolence or allergic rash).[15]
The fourth RCT reported the following adverse effects of sodium valproate: gastrointestinal effects, weight gain, and loss of hair (gastrointestinal effects: 3/113 [3%] with sodium valproate v 1/115 [1%] with placebo; weight gain: 5/113 [4%] with sodium valproate v 1/115 [1%] with placebo; loss of hair: 2/113 [2%] with sodium valproate v 0/115 [0%] with placebo; significance assessments not performed).[16]
In the fifth RCT, people in the immediate-treatment group were more likely to report at least one adverse event, including depression, dizziness, and gastrointestinal symptoms (AR for at least one adverse event: 270/685 [39%] with immediate treatment v 214/721 [31%] with deferred treatment; ARI 8.6%, 95% CI 3.6% to 13.6%).[17]
Comment
The RCT comparing immediate versus deferred antiepileptic drug treatment in people with one (56% people) or more (44% people) previous unprovoked seizures had a pragmatic design and included people who had had seizures where both the participant and the clinician were uncertain about the need for immediate antiepileptic drug treatment.[17] This RCT therefore addresses the same clinical question as RCTs purely in people with single seizures with respect to the balance of benefits and risks of initiating antiepileptic therapy.
Substantive changes
No new evidence

Carbamazepine for partial epilepsy
Summary
SEIZURE FREQUENCY Compared with sodium valproate: We don't know whether carbamazepine is more effective at achieving 12-month remission or at reducing the risk of first seizure in people with partial epilepsy ( low-quality evidence ). Compared with phenobarbital: Carbamazepine and phenobarbital seem equally effective at achieving 12-month remission, but phenobarbital seems more effective at increasing time to first seizure ( moderate-quality evidence ). Compared with phenytoin: Carbamazepine and phenytoin seem equally effective at achieving 12-month remission, and at increasing time to first seizure in people with partial epilepsy (moderate-quality evidence). Compared with lamotrigine: Carbamazepine standard release may be more effective at increasing the proportion of people with partial epilepsy who were seizure free at 6 months, but we don't know whether it is more effective at increasing the proportion of older people with new-onset epilepsy who were seizure free at 3 months, or at increasing time to first seizure in people with partial epilepsy. We don't know whether carbamazepine controlled release is more effective at increasing time to withdrawal (combined measure of efficacy and tolerability) in older people with epilepsy ( very low-quality evidence ). Compared with topiramate: We don't know whether carbamazepine is more effective at increasing the proportion of people seizure free, or whether it is associated with an increased time to first seizure in people with epilepsy (very low-quality evidence). Compared with gabapentin: We don't know whether carbamazepine is more effective at increasing the proportion of older people with new-onset epilepsy who are seizure free at 3 months (very low-quality evidence). Compared with levetiracetam: Carbamazepine controlled release and levetiracetam seem equally effective at achieving at least a 6-month and 12-month remission in people with epilepsy (moderate-quality evidence). NOTE We found no direct information from RCTs about whether carbamazepine used as monotherapy in people with partial epilepsy is better than no active treatment. However, there is consensus that carbamazepine reduces seizure rates.
Benefits
Carbamazepine versus placebo:
We found no systematic review or RCTs (see comment below).
Carbamazepine versus sodium valproate:
We found one systematic review (search date 2007, 5 RCTs, 1265 people; 830 [66%] with partial epilepsy, aged 3 to 83 years, at least 47% men, follow-up less than 5 years) comparing carbamazepine versus sodium valproate.[20] The systematic review included a meta-analysis of the subgroup of people with partial epilepsy. Carbamazepine significantly decreased time to 12-month remission compared with sodium valproate, and reduced the risk of first seizure (HR greater than 1 for an event more likely with sodium valproate; time to 12-month remission: HR 0.82, 95% CI 0.67 to 1.00; first seizure: HR 1.22, 95% CI 1.04 to 1.44). The review performed a test for statistical interaction between treatment and epilepsy type (partial v generalised), which was significant for time to first seizure but not for time to12-month remission. These subgroup analyses should therefore be treated with caution.
Carbamazepine versus phenobarbital:
We found one systematic review (search date 2006, 4 RCTs, 680 people aged 2 to 68 years, of whom 523 [77%] had partial epilepsy, at least 52% men) comparing carbamazepine versus phenobarbital.[21] For people with partial epilepsy, it found no significant difference in time to 12-month remission (HR greater than 1 for an event more likely with phenobarbital; HR 1.03, 95% CI 0.72 to 1.49). However, it found that phenobarbital significantly increased time to first seizure compared with carbamazepine (HR greater than 1 for an event more likely with phenobarbital; HR 0.71, 95% CI 0.55 to 0.91).
Carbamazepine versus phenytoin:
We found one systematic review (search date 2007, 3 RCTs, 552 adults and children, of whom 431 [78%] had partial epilepsy, at least 47% men) comparing carbamazepine versus phenytoin.[22] The review did not present results separately for people with partial epilepsy. Overall, it found no significant difference between carbamazepine and phenytoin for time to first seizure, or for 12-month remission (time to first seizure: HR 0.91, 95% CI 0.74 to 1.12; time to 12-month remission: HR 1.00, 95% CI 0.78 to 1.29).
Carbamazepine versus lamotrigine:
We found one systematic review (search date 2007) [23] and one additional RCT[24] comparing carbamazepine (standard release) versus lamotrigine. We found one subsequent RCT comparing carbamazepine controlled release (CR) with lamotrigine in people with newly diagnosed epilepsy.[25]
The systematic review (5 RCTs, 1384 people, of whom 1108 [80%] had partial epilepsy, 51% men) included a meta-analysis of the subgroup of people with partial epilepsy. The review found that carbamazepine significantly increased the proportion of people seizure free after 6 months (5 RCTs, 903 people; 145/302 [48%] with carbamazepine v 247/601 [41%] with lamotrigine; RR [lamotrigine v carbamazepine] 0.85, 95% CI 0.74 to 0.98). It found no significant difference in time to first seizure between lamotrigine and carbamazepine (HR less than 1 indicates a clinical advantage for lamotrigine; HR 1.28, 95% CI 0.98 to 1.66).[23]
The additional RCT (593 older people with new-onset epilepsy, 251/593 [42%] with complex partial seizures only, 76/593 [13%] with simple partial seizures only, 77/593 [13%] with generalised tonic clonic plus partial seizures, 30/593 [5%] with mixed partial seizures, 147/593 [25%] with generalised tonic clonic seizures only; average age 72 years) was a three-arm trial comparing carbamazepine (198 people) versus lamotrigine (200 people) versus gabapentin (195 people). [24] The RCT found no significant difference among the three treatment groups in the proportion of people who remained in the study at 3 months (402) and who were seizure free after start of treatment (seizure-free rates at 3 months: 64% with carbamazepine v 63% with lamotrigine v 62% with gabapentin; P = 0.09; absolute numbers not reported, results presented graphically). The RCT did not report on paired-way comparisons for these outcomes. About 57% of people were treatment naïve, and those who received treatment were treated for less than 4 weeks, or were treated with sub-therapeutic doses. Doses were titrated for the first 6 weeks of the trial to target doses (1500 mg/day for gabapentin, 150 mg/day for lamotrigine, and 600 mg/day for carbamazepine) although flexible dosing was allowed throughout the duration of the trial if seizure control was poor.[24]
The subsequent RCT (186 older people with newly diagnosed epilepsy, aged at least 65 years, 55% male, at least 2 partial onset or primary generalised tonic clonic seizure in the previous 6 months) comparing lamotrigine at a flexible dosing range 25 to 400 mg daily with carbamazepine CR at a flexible dosing range of 100 to 800 mg over 40 weeks. The RCT did not report subgroup analyses according to seizure type or epilepsy type and did not specify how many people had partial epilepsy. The RCT found no significant difference in time to withdrawal (combined measure of efficacy and tolerability) between lamotrigine and carbamazepine CR (184 people, absolute results shown graphically, HR [lamotrigine v carbamazepine] 0.77, 95% CI 0.45 to 1.31; P = 0.33, intention-to-treat analysis).[25]
Carbamazepine versus topiramate:
See benefits of topiramate.
Carbamazepine versus gabapentin:
We found one RCT (593 older people with new-onset epilepsy, 251/593 [42%] with complex partial seizures only, 76/593 [13%] with simple partial seizures only, 77/593 [13%] with generalised tonic clonic plus partial seizures, 30/593 [5%] with mixed partial seizures, 147/593 [25%] with generalised tonic clonic seizures only; average age 72 years) was a three-arm trial comparing carbamazepine (198 people) versus lamotrigine (200 people) versus gabapentin (195 people). [24] The RCT did not report paired-way comparisons for outcomes of interest (see carbamazepine v lamotrigine comparison, above, for detailed reporting of the RCT). The RCT found no significant difference among treatment groups in the proportion of people who remained in the study at 3 months (402) and were seizure free after start of treatment (seizure-free rates at 3 months: 64% with carbamazepine v 63% with lamotrigine v 62% with gabapentin; P = 0.09; absolute numbers not reported, results presented graphically).
Carbamazepine versus levetiracetam:
See benefits of levetiracetam.
Carbamazepine versus other antiepileptic drugs:
We found no systematic review or RCTs.
Harms
Antiepileptic drugs have been associated with increased risk of suicidal behaviour and ideation.[18]
Carbamazepine versus placebo:
We found no RCTs.
Carbamazepine versus sodium valproate:
The review found no significant difference in treatment withdrawal between carbamazepine and sodium valproate (HR greater than 1 for an event more likely with sodium valproate; HR 1.00, 95% CI 0.79 to 1.26).[20]
Carbamazepine versus phenobarbital:
The review found that carbamazepine was significantly less likely to be withdrawn than phenobarbital (HR greater than 1 for an event more likely on phenobarbital; HR 1.60, 95% CI 1.18 to 2.17).[21]
Carbamazepine versus phenytoin:
Overall, the review found no significant difference between carbamazepine and phenytoin in treatment withdrawal (HR 0.97, 95% CI 0.74 to 1.28). [22]
Carbamazepine versus lamotrigine:
The review found that lamotrigine significantly increased time to treatment withdrawal compared with carbamazepine (HR less than 1 indicates a clinical advantage for lamotrigine; HR 0.62, 95% CI 0.45 to 0.86).[23]
The additional RCT found that, at 12 months, carbamazepine was significantly more likely to lead to treatment withdrawal compared with lamotrigine (P less than 0.0001; absolute numbers not reported) and significantly increased the number of people who withdrew from treatment because of adverse effects (P less than 0.0001; absolute numbers not reported). The RCT found that carbamazepine significantly increased hypersensitivity reactions (rash of any degree) compared with lamotrigine (P = 0.007; absolute numbers not reported). [24]
The subsequent RCT found similar rates of treatment-emergent adverse effects in both groups (82/93 [88%] with lamotrigine v 79/92 [86%] with carbamazepine; statistical analysis not reported). It reported that central nervous system effects were the most common treatment-emergent adverse effect for both lamotrigine and carbamazepine (44/93 [47%] with lamotrigine v 45/92 [49%] with carbamazepine). The RCT found no significant difference in withdrawals due to adverse effects between groups (13/93 [14%] with lamotrigine v 23/92 [25%] with carbamazepine; P = 0.078).[25]
Carbamazepine versus topiramate:
See harms of topiramate.
Carbamazepine versus gabapentin:
The RCT found that, at 12 months, carbamazepine was significantly more likely to lead to treatment withdrawal compared with gabapentin (P = 0.008; absolute numbers not reported). The RCT found that, at 12 months, gabapentin significantly increased the proportion of people who gained weight compared with carbamazepine (P = 0.002; absolute numbers not reported). Hyponatraemia (sodium less than 130 mmol/L) occurred more frequently with carbamazepine compared with gabapentin.[24]
Carbamazepine versus levetiracetam:
See harms of levetiracetam.
Carbamazepine versus other antiepileptic drugs:
We found no RCTs.
Teratogenicity:
We found one systematic review (search date 2006, 59 observational studies, 65,533 pregnancies in women with epilepsy and 1,871,218 in healthy women, mean age of women with epilepsy 29 years, 57% monotherapy, 26% polytherapy) quantifying the incidence of congenital malformations as a function of in utero antiepileptic drug exposure.[26] See teratogenicity section in harms of valproate for full details.
Effect on bone mineral density (BMD):
We found one systematic review of observational data (search date 2006, 14 observational studies, children and adolescents aged up to 18 years old)[27] and one subsequent observational study, reporting separately on the effect of carbamazepine on BMD.[28] The review found six studies in children (in generally good physical condition with no other medical comorbidities potentially affecting vitamin D status) reporting on BMD evaluated by dual-energy-x-ray absorptiometry (DXA). The review was narrative and did not pool data. It concluded that carbamazepine was not associated with a reduction in BMD in the total body, lumbar spine, hip, or forearm.[27]The subsequent study (longitudinal design; 33 people with newly diagnosed epilepsy, aged 18 to 55 years, 60% men) assessed the effect of 6 months' treatment with carbamazepine, lamotrigine, or sodium valproate on BMD (measured by DXA at the right calcaneus) and biochemical markers of bone metabolism. It found significantly decreased BMD Z-score and vitamin D levels from baseline after 6 months' treatment with carbamazepine (change in BMD Z-score: +0.42 to −0.34; P = 0.043; change in vitamin D levels [nanograms/mL]: 33.9 to 23.0; P = 0.018).[28]
We found three subsequent observational studies reporting on the effect of a number of antiepileptic drugs (including carbamazepine) on BMD (data on carbamazepine not presented separately).[29] [30] [31]The first study (cross-sectional design; 68 children with primary idiopathic generalised epilepsy and no other medical comorbidities, 30 healthy controls) assessed the effect of treatment for 12 months or longer with carbamazepine, sodium valproate, or oxcarbazepine on lumbar BMD (measured by DXA) and biochemical markers of bone metabolism. It found significantly lower lumbar BMD and significantly higher alkaline phosphatase levels (P = 0.02) in people taking antiepileptic drugs than in healthy controls. It found no significant difference between groups in vitamin D levels. The duration or daily doses of antiepileptic drugs were not found to be predictors of bone loss.[29]
The second study (cross-sectional design; 137 adults [mean age 31 years, mean drug duration 11.4 years] and 88 children [mean age 13 years, mean drug duration 4.7 years], 323 controls) assessed the effect of enzyme-inducing (phenobarbital, carbamazepine, and primidone) and non-enzyme-inducing drugs (sodium valproate, lamotrigine, clonazepam, gabapentin, topiramate, ethosuximide, and vigabatrin) on BMD (lumbar spine, hip, and 1/3 radius for adults; lumbar spine and total body for children and adolescents, measured using DXA) and vitamin D levels. In adults, it found significantly lower bone density T-scores at all three sites in people taking any antiepileptic drug compared with controls. It found that adults taking enzyme-inducing drugs had significantly lower T-scores (at the lumbar spine and hip) than adults taking non-enzyme-inducing drugs. However these results should be interpreted with caution because adults taking enzyme-inducing drugs were also older and had been on treatment for a longer duration. It found that hypovitaminosis D (25[OH] D levels less than 20 nanograms/mL) was significantly more common in adults talking antiepileptic drugs compared with controls (74% with antiepileptic drugs v 53% with no antiepileptic drugs; P less than 0.01; absolute numbers not reported). In children and adolescents, it found no significant difference in spine BMD or total body BMD in people taking any antiepileptic drugs compared with controls. It also found that hypovitaminosis D was significantly less common in children talking antiepileptic drugs compared with controls (57% with antiepileptic drugs v 79% with no antiepileptic drugs; P less than 0.01; absolute numbers not reported). The study pooled results on adults and children/adolescents and found no significant difference in BMD between people on enzyme-inducing drugs or non-enzyme-inducing drugs. However, it found that people taking multiple drug treatments were more likely to have lower BMD compared with people taking only one drug.[30]The third study (longitudinal design; 4222 ambulant men aged over 65 years [100 using non-enzyme-inducing antiepileptic drugs, 62 using enzyme-inducing antiepileptic drugs, 4060 controls who did not use antiepileptic drugs]) assessed the effect of antiepileptic drug therapy on BMD (total hip and 2 sub-regions [femoral neck, trochanter] measured by DXA) at an average follow-up of 4.6 years. The study found that people taking non-enzyme-inducing drugs (most commonly gabapentin) had a significantly higher adjusted rate of bone loss at the total hip than controls (−0.53% a year with enzyme-inducing drugs v −0.35% a year with non-enzyme-inducing drugs; P = 0.04). The study found no significant difference in the adjusted rate of bone loss for people taking enzyme-inducing drugs (most commonly phenytoin) compared with controls (−0.46% a year with enzyme-inducing drugs v −0.35% a year with non-enzyme-inducing drugs; P = 0.31). The study found no significant difference in adjusted rates of bone loss for people taking enzyme-inducing drugs compared with non-enzyme-inducing drugs.[31]
Comment
Placebo-controlled trials of carbamazepine in partial epilepsy would now be considered unethical.
The meta-analysis comparing carbamazepine versus sodium valproate provides weak evidence in support of the consensus view to use carbamazepine as the drug of choice in people with partial epilepsy.[20] The systematic reviews did not present results separately in children and adults.[20] [21] [22] [23]
We found one large RCT (1721 people, 88% partial epilepsy, 10% unclassified) of a pragmatic design comparing the effectiveness of carbamazepine versus newer antiepileptic drugs (gabapentin, lamotrigine, oxcarbazepine, and topiramate) in the treatment of partial epilepsy.[32]The RCT was open label — to allow clinicians to determine what they considered the optimum rate of titration and dosing regime — and as such does not meet our inclusion criteria; however, because of a paucity of data comparing standard versus newer antiepileptic drugs, and the large size of the trial, we have reported the data here. The first date of randomisation in the RCT was 1 December 1999, and the last date was 31 August 2004: the number recruited each year was not reported. The last date of follow-up was reported as between 1 May 2005 and 31 August 2005. In addition, the oxcarbazepine arm had fewer people, as randomisation started after June 2001. For this reason, although the RCT reports data for 6-year follow-up, we have chosen to report data for only 12-month follow-up. The authors report that the study was underpowered to exclude clinically important differences in efficacy in terms of the number of people randomised to each arm, but study length was extended to compensate for this deficiency. The RCT found that carbamazepine was associated with a significantly higher risk of treatment failure (defined as unacceptable adverse effects after randomisation and inadequate seizure control) compared with lamotrigine (HR greater than 1 indicates that failure occurs more rapidly with lamotrigine; HR 0.78, 95% CI 0.63 to 0.97). However, there was no significant difference in time to treatment failure between carbamazepine and gabapentin (HR greater than 1 indicates that failure occurs more rapidly with gabapentin; HR 1.21, 95% CI 0.99 to 1.48), topiramate (HR greater than 1 indicates that failure occurs more rapidly with topiramate; HR 1.22, 95% CI 0.99 to 1.49), and oxcarbazepine (HR greater than 1 indicates that failure occurs more rapidly with oxcarbazepine; HR 1.04, 95% CI 0.78 to 1.39). The data reported constitute an intention-to-treat analysis, with follow-up data after a treatment failure included (people achieving a 1-year remission on drug regimens other than the one to which they were originally randomised). The RCT found that carbamazepine was significantly more effective at achieving 12-month remission compared with gabapentin (HR greater than 1 indicates that failure occurs more rapidly on gabapentin; HR 0.75, 95% CI 0.63 to 0.90). However, there was no significant difference in 12-month remission between carbamazepine and lamotrigine (HR greater than 1 indicates that failure occurs more rapidly with lamotrigine; HR 0.91, 95% CI 0.77 to 1.09), topiramate (HR greater than 1 indicates that failure occurs more rapidly with topiramate; HR 0.86, 95% CI 0.72 to 1.03), and oxcarbazepine (HR greater than 1 indicates that failure occurs more rapidly with oxcarbazepine; HR 0.92, 95% CI 0.73 to 1.18). The results for the oxcarbazepine comparisons include data only from people randomised after June 2001.
Substantive changes
Carbamazepine (partial epilepsy) Two RCT s added.[25] [33]One RCT found no significant difference in time to withdrawal (combined measure of efficacy and tolerability) between carbamazepine controlled release versus lamotrigine in older people with newly diagnosed epilepsy.[25] The other RCT found no significant difference between levetiracetam and carbamazepine controlled release in the proportion of people with newly diagnosed epilepsy who were seizure free for at least 6 months at 26 weeks, or for at least 12 months at 52 weeks.[33] Four systematic reviews, search date updated, no new evidence found.[20] [21] [22] [23] Observational harms data added on teratogenicity and bone mineral density changes. Categorisation unchanged (Likely to be beneficial).

Phenobarbital for partial epilepsy
Summary
SEIZURE FREQUENCY Compared with carbamazepine: Carbamazepine and phenobarbital seem equally effective at achieving 12-month remission, but phenobarbital seems associated with an increased time to first seizure ( moderate-quality evidence ). Compared with phenytoin: Phenobarbital and phenytoin seem equally effective at achieving 12-month remission and at increasing time to first seizure (moderate-quality evidence). NOTE We found no direct information from RCTs about whether phenobarbital used as monotherapy in people with partial epilepsy is better than no active treatment. However, there is consensus that phenobarbital reduces seizure rates.
Benefits
Phenobarbital versus placebo:
We found no systematic review or RCTs (see comment below).
Phenobarbital versus carbamazepine:
See benefits of carbamazepine.
Phenobarbital versus phenytoin:
We found one systematic review comparing phenobarbital versus phenytoin.[34] The review (search date 2006, 3 RCTs, 599 people with partial or generalised epilepsy, of whom about 78% had partial epilepsy, aged 3 to 77 years, at least 43% men) did not undertake subgroup analyses for people with partial or generalised epilepsy. Overall, it found no significant difference in 12-month remission or time to first seizure (HR greater than 1 for an event more likely with phenobarbital; 12-month remission: HR 0.93, 95% CI 0.70 to 1.23; first seizure: HR 0.84, 95% CI 0.68 to 1.05).
Phenobarbital versus other antiepileptic drugs:
We found no systematic review or RCTs.
Harms
Antiepileptic drugs have been associated with increased risk of suicidal behaviour and ideation.[18]
Phenobarbital versus placebo:
We found no RCTs.
Phenobarbital versus carbamazepine:
See harms of carbamazepine.
Phenobarbital versus phenytoin:
The review found more treatment withdrawal with phenobarbital than with phenytoin (HR greater than 1 for an event more likely on phenobarbital; HR 1.62, 95% CI 1.22 to 2.14). This could be because it was less well tolerated.[34]
Phenobarbital versus other antiepileptic drugs:
We found no RCTs.
Teratogenicity:
We found one systematic review (search date 2006, 59 observational studies, 65,533 pregnancies in women with epilepsy and 1,871,218 in healthy women, mean age of women with epilepsy 29 years, 57% monotherapy, 26% polytherapy) quantifying the incidence of congenital malformations as a function of in utero antiepileptic drug exposure.[26] See teratogenicity section in harms of valproate for full details.
Effect on bone mineral density (BMD):
We found one systematic review of observational data (search date 2006, 14 observational studies, children and adolescents aged up to 18 years old),[27] which reported separately on the effect of phenobarbital on BMD. The review was narrative. It found one study in children reporting separately the effects of phenobarbital on BMD. The study (78 generally healthy children taking either phenobarbital or phenytoin, 78 controls) found a significant decrease in spine and total-body BMD with phenobarbital, but only among those who had used phenobarbital for more than 2 years (total body BMD [g/cm2]: 0.85 with phenobarbital v 0.96 with no phenobarbital; P less than 0.01; spine BMD: 0.75 with phenobarbital v 0.82 with no phenobarbital; P = 0.03; number of children in each group not reported).[27]
We found three subsequent observational studies reporting on the effect of a number of antiepileptic drugs (2 including phenobarbital) on BMD (data on phenobarbital not presented separately — see harms of carbamazepine for more details).[29] [30] [31]
Comment
Placebo-controlled trials of phenobarbital would now be considered unethical. The review did not present results separately for adults and children.[34]
Substantive changes
Phenobarbital (partial epilepsy) Two systematic reviews, search date updated, no new evidence found.[21] [34]Observational harms data added on teratogenicity and bone mineral density changes. Categorisation unchanged (Likely to be beneficial).

Phenytoin for partial epilepsy
Summary
SEIZURE FREQUENCY Compared with sodium valproate: Phenytoin and sodium valproate seem equally effective at achieving 12-month remission and at increasing time to first seizure in people with partial epilepsy ( moderate-quality evidence ). Compared with phenobarbital: Phenytoin and phenobarbital seem equally effective at achieving 12-month remission and at increasing time to first seizure (moderate-quality evidence). Compared with carbamazepine: Phenytoin and carbamazepine seem equally effective at achieving 12-month remission, and at increasing time to first seizure in people with partial epilepsy (moderate-quality evidence). Compared with oxcarbazepine: Phenytoin and oxcarbazepine seem equally effective at achieving 6-month or 12-month remission in people with partial epilepsy (moderate-quality evidence). NOTE We found no direct information from RCTs about whether phenytoin used as monotherapy in people with partial epilepsy is better than no active treatment. However, there is consensus that phenytoin reduces seizure rates.
Benefits
Phenytoin versus placebo:
We found no systematic review or RCTs (see comment below).
Phenytoin versus sodium valproate:
We found one systematic review (search date 2007, 5 RCTs, 250 people with partial epilepsy and 395 with generalised epilepsy, aged 3 to 95 years, at least 36% men, follow-up less than 5 years) comparing sodium valproate versus phenytoin.[35] It included a meta-analysis in people with partial epilepsy. It found no significant difference in 12-month remission or time to first seizure (HR greater than 1 for an event more likely with phenytoin; 12-month remission: HR 1.02, 95% CI 0.68 to 1.54; first seizure: HR 0.81, 95% CI 0.59 to 1.10).
Phenytoin versus phenobarbital:
See benefits of phenobarbital.
Phenytoin versus carbamazepine:
See benefits of carbamazepine.
Phenytoin versus oxcarbazepine:
We found one systematic review (search date 2008, 2 RCTs, 480 people, of whom 333 [69%] had partial epilepsy) comparing oxcarbazepine versus phenytoin. The review carried out a subgroup analysis in people with partial epilepsy. It found no significant difference between oxcarbazepine and phenytoin in time to first seizure (HR greater than 1 indicates a clinical advantage for oxcarbazepine; HR 1.08, 95% CI 0.80 to 1.47), or in achieving 6-month or 12-month remission (HR greater than 1 indicates a clinical advantage for phenytoin; time to achieve 6-month remission: HR 0.84, 95% CI 0.58 to 1.21; time to achieve 12-month remission: HR 0.93, 95% CI 0.58 to 1.50). [36]
Phenytoin versus other antiepileptic drugs:
We found no systematic review or RCTs.
Harms
Antiepileptic drugs have been associated with increased risk of suicidal behaviour and ideation.[18]
Phenytoin versus placebo:
We found no RCTs.
Phenytoin versus sodium valproate:
The review found no significant difference in treatment withdrawal between phenytoin and sodium valproate (HR greater than 1 for an event more likely with phenytoin; 1.23, 95% CI 0.77 to 1.98).[35]
Phenytoin versus phenobarbital:
See harms of phenobarbital.
Phenytoin versus carbamazepine:
See harms of carbamazepine.
Phenytoin versus oxcarbazepine:
The review performed a meta-analysis for people with partial epilepsy and found a significantly lower risk of treatment withdrawal with oxcarbazepine compared with phenytoin (HR greater than 1 indicates a clinical advantage for oxcarbazepine; HR 1.92, 95% CI 1.17 to 3.16). The review pooled results for time to treatment withdrawal for people with partial and generalised epilepsy, and found a significant difference between oxcarbazepine and phenytoin (HR greater than 1 indicates a clinical advantage for oxcarbazepine; HR 1.64, 95% CI 1.09 to 2.47). However, a test for statistical interaction between partial and generalised epilepsy was not significant. These subgroup analyses must therefore be treated with caution. [36]
Phenytoin versus other antiepileptic drugs:
We found no RCTs.
Teratogenicity:
We found one systematic review (search date 2006, 59 observational studies, 65,533 pregnancies in women with epilepsy and 1,871,218 in healthy women, mean age of women with epilepsy 29 years, 57% monotherapy, 26% polytherapy) quantifying the incidence of congenital malformations as a function of in utero antiepileptic drug exposure.[26] See teratogenicity section in harms of valproate for full details.
Effect on bone mineral density (BMD):
We found one systematic review of observational data (search date 2006, 14 observational studies, children and adolescents aged up to 18 years old),[27] which reported separately on the effect of phenytoin on BMD. The review was narrative. One identified study failed to show a decrease in femur BMD, whereas another study reported a decrease in total body and spine BMD, but only with the use of phenytoin for more than 2 years. The medical and vitamin D status of the children taking phenytoin varied between studies.[27]
We found three subsequent observational studies reporting on the effect of a number of antiepileptic drugs (2 including phenytoin) on BMD (data on phenytoin not presented separately — see harms of carbamazepine for more details).[29] [30] [31]
Comment
Placebo-controlled trials of phenytoin would now be considered unethical. The reviews did not present results separately for adults and children.[35] [36]
Substantive changes
Phenytoin (partial epilepsy) Four systematic reviews, search date updated, no new evidence found.[35] [34] [22] [36]Observational harms data added on teratogenicity and bone mineral density changes. Categorisation unchanged (Likely to be beneficial).

Sodium valproate for partial epilepsy
Summary
SEIZURE FREQUENCY Compared with carbamazepine: We don't know whether sodium valproate is more effective at increasing 12-month remission or at reducing the risk of first seizure in people with partial epilepsy ( low-quality evidence ). Compared with phenytoin: Sodium valproate and phenytoin are equally effective at achieving 12-month remission and at reducing time to first seizure in people with partial epilepsy ( moderate-quality evidence ). Compared with topiramate: We don't know whether sodium valproate is more effective at increasing the proportion of people seizure free, or whether it is associated with an increased time to first seizure in people with epilepsy ( very low-quality evidence ). NOTE We found no direct information from RCTs about whether sodium valproate used as monotherapy in people with partial epilepsy is better than no active treatment. However, there is consensus that sodium valproate reduces seizure rates.
Benefits
Sodium valproate versus placebo:
We found no systematic review or RCTs (see comment below).
Sodium valproate versus carbamazepine:
See benefits of carbamazepine.
Sodium valproate versus phenytoin:
See benefits of phenytoin.
Sodium valproate versus topiramate:
See benefits of topiramate.
Sodium valproate versus other antiepileptic drugs:
We found no systematic review or RCTs.
Harms
Antiepileptic drugs have been associated with increased risk of suicidal behaviour and ideation.[18]
Sodium valproate versus placebo:
We found no RCTs.
Sodium valproate versus carbamazepine:
See harms of carbamazepine.
Sodium valproate versus phenytoin:
See harms of phenytoin.
Sodium valproate versus topiramate:
See harms of topiramate.
Sodium valproate versus other antiepileptic drugs:
We found no RCTs.
Teratogenicity:
We found one systematic review (search date 2006, 59 studies, 65,533 pregnancies in women with epilepsy and 1,871,218 in healthy women, mean age of women with epilepsy 29.1 years, 57% monotherapy, 26% polytherapy) quantifying the incidence of congenital malformations as a function of in utero antiepileptic drug exposure. Incidence rates were stratified for carbamazepine, lamotrigine, phenobarbital, phenytoin, and valproate. The review found that the incidence of births with congenital malformations was higher in women with epilepsy compared with women without epilepsy (7%, 95% CI 5.62 to 8.54 in women with epilepsy v 2%, 95% CI 1.46 to 3.10 in women without epilepsy). The review also found that the incidence of congenital malformations was higher for women with epilepsy taking polytherapy compared with monotherapy (17%, 95% CI 0.51 to 33.05 with polytherapy v 10%, 95% CI 1.96 to 18.28 with monotherapy). The incidence of congenital malformations was higher with valproate monotherapy compared with other antiepileptic drug monotherapy (11%, 95% CI 8.16 to 13.29 with valproate v 7%, 95% CI 3.60 to 11.00 with phenytoin v 4.6%, 95% CI 3.48 to 5.76 with carbamazepine v 4.9%, 95% CI 3.22 to 6.59 with phenobarbital v 2.9%, 95% CI 2.00 to 3.82 with lamotrigine). Polytherapy regimes including valproate plus two or more other drugs had the highest rate of congenital malformations (25%, 95% CI 5.97 to 44.0). The most common malformations were cardiovascular defects followed by musculoskeletal defects.[26]
Effect on bone mineral density (BMD):
We found one systematic review[27] and one observational study, which presented data separately on sodium valproate.[28]
The review (search date 2006, 14 observational studies, children and adolescents aged up to 18 years old) assessed the effect of antiepileptic drugs on BMD. The review was narrative. It presented descriptive data for individual drugs and did not carry out a comparative meta-analysis. For valproate, two controlled studies reported a reduction in spine BMD, whereas two other studies did not. Two studies reported a decrease in hip BMD with valproate, whereas one did not. Three studies reporting on forearm BMD in users of valproate identified a decrease. The children in these studies were reported not to have any other concomitant diseases or malnutrition that may have affected vitamin D status. In summary the review reported no conclusive evidence that valproate had a clinically significant detrimental effect on bone health, particularly in children with normal vitamin D status.[27]
The observational study (longitudinal design; 33 people with newly diagnosed epilepsy, aged 18 to 55 years, 60% men) assessed the effect of 6 months' treatment with carbamazepine, lamotrigine, or sodium valproate on BMD (measured by dual-energy x-ray absorptiometry [DXA] at the right calcaneus) and biochemical markers of bone metabolism. It found no significant decrease in BMD Z-score or vitamin D levels with sodium valproate from baseline, after 6 months' treatment (change in BMD Z-score: 0.61 to 0.06; P = 0.068; change in vitamin D levels [nanograms/mL]: 34.6 to 38.0; P = 0.593).[28]
We found three subsequent observational studies reporting on the effect of a number of antiepileptic drugs (including valproate) on BMD (data on valproate not presented separately — see harms of carbamazepine for more details).[29] [30] [31]
Comment
Placebo-controlled trials of sodium valproate would now be considered unethical. Valproate is associated with a greater incidence of congenital malformations when compared with phenytoin, carbamazepine, phenobarbital, and lamotrigine.
Substantive changes
Sodium valproate (partial epilepsy) Two systematic reviews, search date updated, no new evidence found.[20] [35]Observational harms data added on teratogenicity and bone mineral density changes. Categorisation unchanged (Likely to be beneficial).

Lamotrigine for partial epilepsy
Summary
SEIZURE FREQUENCY Compared with carbamazepine: Lamotrigine may be more effective than carbamazepine (standard release) at increasing the proportion of people seizure free at 6 months, but we don't know whether it is associated with an increased time to a first seizure. We don't know whether lamotrigine is more effective than carbamazepine controlled release at increasing time to withdrawal (combined measure of efficacy and tolerability) in older people with epilepsy ( very low-quality evidence ). Compared with gabapentin: We don't know whether lamotrigine is more effective at increasing the proportion of older people with new-onset epilepsy who are seizure free, or whether it is associated with an increased time to first seizure in people with partial epilepsy (very low-quality evidence). NOTE We found no direct information from RCTs about whether lamotrigine used as monotherapy in people with partial epilepsy is better than no active treatment. However there is widespread consensus that lamotrigine reduces seizure rates.
Benefits
Lamotrigine versus placebo:
We found no systematic review or RCTs.
Lamotrigine versus carbamazepine (standard or controlled release):
See benefits of carbamazepine.
Lamotrigine versus gabapentin:
See benefits of gabapentin.
Lamotrigine versus other antiepileptic drugs:
We found no systematic review or RCTs.
Harms
Antiepileptic drugs have been associated with increased risk of suicidal behaviour and ideation.[18]
Lamotrigine versus placebo:
We found no RCTs.
Lamotrigine versus carbamazepine:
See harms of carbamazepine.
Lamotrigine versus gabapentin:
See harms of gabapentin.
Lamotrigine versus other antiepileptic drugs:
We found no RCTs.
Teratogenicity:
We found one systematic review (search date 2006, 59 observational studies, 65,533 pregnancies in women with epilepsy and 1,871,218 in healthy women, mean age of women with epilepsy 29 years, 57% monotherapy, 26% polytherapy) quantifying the incidence of congenital malformations as a function of in utero antiepileptic drug exposure.[26] See teratogenicity section in harms of valproate for full details.
Effect on bone mineral density (BMD):
We found one observational study (longitudinal design; 33 people with newly diagnosed epilepsy, aged 18 to 55 years, 60% men), which presented data separately on sodium valproate.[28]The study assessed the effect of 6 months' treatment with carbamazepine, lamotrigine, or sodium valproate on BMD (measured by dual-energy x-ray absorptiometry [DXA] at the right calcaneus) and biochemical markers of bone metabolism. It found no significant decrease in BMD Z-score or vitamin D levels with lamotrigine from baseline (change in BMD Z-score: 0.60 to 0.48; P = 0.100; change in vitamin D levels [nanograms/mL]: 36.9 to 33.8; P = 0.668). [28]
We found three subsequent observational studies reporting on the effect of a number of antiepileptic drugs (including lamotrigine) on BMD (data on lamotrigine not presented separately — see harms of carbamazepine for more details).[29] [30] [31]
Drug safety alert
FDA issues drug safety alert on the risk of aseptic meningitis associated with lamotrigine (12 August 2010).
A drug safety alert has been issued on the risk of aseptic meningitis associated with lamotrigine. (www.fda.gov)
Comment
Placebo-controlled trials of lamotrigine would now be considered unethical. We found one large RCT of pragmatic design comparing carbamazepine versus newer antiepileptic drugs (gabapentin, lamotrigine, oxcarbazepine, and topiramate) in the treatment of partial epilepsy. [32]The RCT was open label, and as such does not meet our inclusion criteria; however, because of a paucity of data comparing standard versus newer antiepileptic drugs, and the large size of the trial, we have reported the data here. For more details on study design, and comparisons versus carbamazepine, see comments on carbamazepine. The RCT found that lamotrigine was associated with a significantly lower risk of treatment failure compared with topiramate (HR greater than 1 indicates that failure occurs more rapidly with topiramate; HR 1.56, 95% CI 1.26 to 1.93) and gabapentin (HR greater than 1 indicates that failure occurs more rapidly with lamotrigine; HR 0.65, 95% CI 0.52 to 0.80). However, there was no significant difference in time to treatment failure compared with oxcarbazepine (HR greater than 1 indicates that failure occurs more rapidly with oxcarbazepine; HR 1.15, 95% CI 0.86 to 1.54). The RCT found that lamotrigine was significantly less effective at achieving 12-month remission compared with gabapentin (HR greater than 1 indicates that failure occurs more rapidly with lamotrigine; HR 1.21, 95% CI 1.01 to 1.46). However, there were no significant differences in time to achieve 12-month remission between lamotrigine versus topiramate (HR greater than 1 indicates that failure occurs more rapidly with topiramate; HR 0.94, 95% CI 0.78 to 1.113) versus oxcarbazepine (HR greater than 1 indicates that failure occurs more rapidly with oxcarbazepine; HR 1.15, 95% CI 0.89 to 1.47).
Substantive changes
Lamotrigine (partial epilepsy) Two RCTs added.[25] [37] One RCT found no significant differences in time to withdrawal (combined measure of efficacy and tolerability) between carbamazepine CR and lamotrigine in older people with newly diagnosed epilepsy.[25] One RCT found no significant differences in time to exit between gabapentin and lamotrigine.[37] One systematic review, search date updated, no new evidence found.[23] Observational harms data added on teratogenicity and bone mineral density changes. Categorisation unchanged (Likely to be beneficial).

Topiramate for partial epilepsy
Summary
SEIZURE FREQUENCY Compared with carbamazepine: We don't know whether topiramate is more effective at increasing the proportion of people seizure free, or whether it is associated with an increased time to first seizure in people with epilepsy ( very low-quality evidence ). Compared with sodium valproate: We don't know whether topiramate is more effective at increasing the proportion of people seizure free, or whether it is associated with an increased time to first seizure in people with epilepsy (very low-quality evidence). NOTE We found no direct information from RCTs about whether topiramate used as monotherapy in people with partial epilepsy is better than no active treatment. However, there is consensus that topiramate reduces seizure rates.
Benefits
Topiramate versus placebo:
We found no systematic review or RCTs.
Topiramate versus carbamazepine:
We found one RCT (621 people, 19% children [6–16 years], 53% male, 39% with generalised seizures, 63% partial epilepsy) comparing topiramate (210 people taking 100 mg and 199 people taking 200 mg) versus carbamazepine (126 people) versus sodium valproate (78 people).[38]The RCT did not undertake subgroup analyses for people with partial or generalised epilepsy. The initial efficacy analysis of the RCT compared topiramate 100 mg versus topiramate 200 mg. The RCT found no difference in efficacy between the two doses of topiramate, and so combined data for these groups for analysis of effectiveness compared with carbamazepine and sodium valproate. The RCT found no significant difference in time to exit for any reason (ineffective treatment, adverse effect, loss to follow-up or patient choice) between topiramate and carbamazepine (P = 0.53; absolute numbers not reported) . It found no significant difference in time to first seizure between topiramate and carbamazepine (P = 0.35; absolute numbers not reported) . The follow-up period of the study is unclear. People continued treatment either until they exited the RCT or until completion 6 months after the last person was randomised. The RCT reported that some people were treated for up to 806 days, and that the median treatment duration was 244 days. The RCT found no significant difference in the proportion of people without seizures during the last 6 months of the study between topiramate and carbamazepine (49% with topiramate 100 mg v 44% with topiramate 200 mg v 44% with carbamazepine; absolute numbers and P value not reported; reported as not significant). The RCT carried out an intention-to-treat analysis (population was not defined). However, only 45% of people randomised completed the study (105 people in topiramate 100 mg group, 87 people in topiramate 200 mg group, 63 people in carbamazepine group, and 30 people in sodium valproate group), and it was not clear how many people were followed up.[38]
Topiramate versus sodium valproate:
We found one RCT (621 people, 19% children [6–16 years], 53% male, 39% with generalised seizures, 63% partial epilepsy) comparing topiramate (210 people taking 100 mg and 199 people taking 200 mg) versus carbamazepine (126 people) versus sodium valproate (78 people).[38]The RCT did not undertake subgroup analyses for people with partial or generalised epilepsy. See topiramate versus carbamazepine comparison, above, for detailed reporting of this RCT. The RCT found no significant difference in time to exit for any reason (ineffective treatment, adverse effect, loss to follow-up or patient choice) between topiramate and sodium valproate (P = 0.53; absolute numbers not reported). It found no significant difference in time to first seizure between topiramate and sodium valproate (P = 0.35; absolute numbers not reported). The RCT found no significant difference in the proportion of people without seizures during the last 6 months of the study between topiramate and sodium valproate (49% with topiramate 100 mg v 44% with topiramate 200 mg v 44% with sodium valproate; absolute numbers and P value not reported; reported as not significant).[38]
Topiramate versus other antiepileptic drugs:
We found no systematic review or RCTs.
Harms
Antiepileptic drugs have been associated with increased risk of suicidal behaviour and ideation.[18]
Topiramate versus placebo:
We found no RCTs.
Topiramate versus carbamazepine or sodium valproate:
The RCT reported that the most common adverse effects were: paraesthesia, fatigue, headache, dizziness, and upper respiratory tract infections (paraesthesia: 25% with topiramate 100 mg v 33% with topiramate 200 mg v 4% with carbamazepine v 3% with sodium valproate; fatigue: 20% with topiramate 100 mg v 23% with topiramate 200 mg v 29% with carbamazepine v 18% with sodium valproate; headache: 25% with topiramate 100 mg v 18% with topiramate 200 mg v 29% with carbamazepine v 18% with sodium valproate; dizziness: 13% with topiramate 100 mg v 12% with topiramate 200 mg v 16% with carbamazepine v 10% with sodium valproate; upper respiratory tract infections: 18% with topiramate 100 mg v 17% with topiramate 200 mg v 15% with carbamazepine v 12% with sodium valproate).[38]
Topiramate versus other antiepileptic drugs:
We found no RCTs.
Teratogenicity:
We found one UK prospective observational study reporting major congenital malformation rates in pregnancies exposed to topiramate (203 pregnancies; 70 monotherapy [50–800 mg/day], 133 polytherapy [25–1000 mg/day]). The study found that 16 live births had major congenital malformations (9%, 95% CI 5.6% to 14.1%). Of these, three occurred following exposure to topiramate monotherapy (5%, 95% CI 1.7% to 13.3%) and 13 in cases exposed to topiramate as part of a polytherapy regimen (11%, 95% CI 6.7% to 18.2%].[39]
Effect on bone mineral density (BMD):
We found no observational studies that reported separately on the effects of topiramate on BMD. We found three observational studies reporting on the effect of a antiepileptic drugs in general on BMD (see harms of carbamazepine for more details).[29] [30] [31]
Comment
Placebo-controlled trials would now be considered unethical. We found one large RCT of pragmatic design comparing carbamazepine versus newer antiepileptic drugs (gabapentin, lamotrigine, oxcarbazepine, and topiramate) in the treatment of partial epilepsy.[32] The RCT was open label and as such does not meet our inclusion criteria; however, because of a paucity of data comparing standard epileptics versus newer epileptics, and the large size of the trial, we have reported the data here. For more details on study design, and comparisons versus carbamazepine, see comments on carbamazepine. For more details about the comparison versus lamotrigine, see comments on lamotrigine. The RCT found no significant differences in time to treatment failure between topiramate and gabapentin (HR greater than 1 indicates that failure occurs more rapidly with topiramate; HR 1.01, 95% CI 0.83 to 1.23) and oxcarbazepine (HR greater than 1 indicates that failure occurs more rapidly with oxcarbazepine; HR 0.90, 95% CI 0.68 to 1.19).There were no significant differences in achieving12-month remission between topiramate and gabapentin (HR greater than 1 indicates that failure occurs more rapidly with topiramate; HR 1.14, 95% CI 0.95 to 1.37) and oxcarbazepine (HR greater than 1 indicates that failure occurs more rapidly with oxcarbazepine; HR 1.10, 95% CI 0.86 to 1.42).
Substantive changes
Topiramate (partial epilepsy) Observational harms data added on teratogenicity.

Levetiracetam for partial epilepsy
Summary
SEIZURE FREQUENCY Compared with carbamazepine controlled release: Levetiracetam and carbamazepine controlled-release seem equally effective at achieving at least a 6-month or 12-month remission in people with epilepsy ( moderate-quality evidence ). NOTE We found no direct information from RCTs about whether levetiracetam used as monotherapy in people with partial epilepsy is better than no active treatment. However, there is consensus that levetiracetam reduces seizure rates.
Benefits
Levetiracetam versus placebo:
We found no systematic review or RCTs.
Levetiracetam versus carbamazepine:
We found one RCT (579 people, mean age 39 years, 54.7% male, at least 2 partial or generalised tonic clonic seizure in previous year, 80% with partial epilepsy, 20% with primary generalised seizures) comparing levetiracetam (1000−3000 mg/day) with carbamazepine (controlled release, 400−1200 mg/day) in people with newly diagnosed epilepsy. The RCT did not report subgroup analyses for seizure types or epilepsy type. The RCT found no significant difference in the proportion of people seizure free for at least 6 months at 26 weeks (intention-to-treat [ITT] analysis: 190/285 [67%] with levetiracetam v 194/291 [67%] with carbamazepine; difference +0.1%, 95% CI −7.4% to +7.5%; per protocol analysis: 173/237 [72.9%] with levetiracetam v 171/235 [72.8%] with carbamazepine; difference +0.2%, 95% CI −7.8% to +8.2%). The RCT also found no significant difference in the proportion seizure free for at least 1 year at 52 weeks (ITT analysis: 142/285 [50%] with levetiracetam v 155/291 [53%] with carbamazepine; difference −3.2%, 95% CI −11.3% to +4.8%; per protocol analysis: 129/228 [57%] with levetiracetam v 131/224 [59%] with carbamazepine; difference −1.8%, 95% CI −10.8% to +7.2%). The RCT found that the majority of people achieving at least 6 months' seizure freedom were taking either levetiracetam at a dose of 1000 mg daily or carbamazepine controlled release at a dose of 400 mg daily.[33]
Levetiracetam versus other antiepileptic drugs:
We found no systematic review or RCTs.
Harms
Antiepileptic drugs have been associated with increased risk of suicidal behaviour and ideation.[18]
Levetiracetam versus placebo:
We found no RCTs.
Levetiracetam versus carbamazepine:
The RCT found that similar rates of any adverse effect were reported in both groups (80% with levetiracetam v 81% with carbamazepine; absolute numbers and significance assessment not reported). It found that depression and insomnia were reported significantly more often with levetiracetam compared with carbamazepine controlled release (depression: 6% with levetiracetam v 2% with carbamazepine; RR 3.06, 95% CI 1.23 to 7.61; insomnia: 6% with levetiracetam v 2% with carbamazepine; RR 2.48, 95% CI 1.04 to 5.89). It found that back pain was reported significantly more often with carbamazepine controlled release compared with levetiracetam (3% with levetiracetam v 7% with carbamazepine; RR 0.41, 95% CI 0.18 to 0.91).[33]
Levetiracetam versus other antiepileptic drugs:
We found no RCTs.
Teratogenicity:
We found one prospective observational case series study assessing major congenital malformation rates in pregnancies exposed to levetiracetam (117 pregnancies in the UK; 39 women receiving levetiracetam monotherapy [500–4000 mg/day], 78 women receiving levetiracetam in polytherapy with at least one other antiepileptic drug [250–4000 mg/day levetiracetam]). The study found major congenital malformations in three pregnancies (2.7%, 95% CI 0.9% to 7.7%) all of which occurred with levetiracetam in polytherapy (one case of pyloric stenosis with lamotrigine, levetiracetam plus topiramate, one case of spina bifida with carbamazepine plus levetiracetam, and one case of spina bifida with clobazam, lamotrigine, levetiracetam plus valproate).[40]
Effect on bone mineral density (BMD):
We found no observational studies which reported separately on the effects of levetiracetam on BMD. We found three observational studies reporting on the effect of antiepileptic drugs in general on BMD (see harms of carbamazepine for more details).[29] [30] [31]
Comment
Placebo-controlled trials would now be considered unethical. The RCT did not present data separately for seizure type.
Substantive changes
Levetiracetam (partial epilepsy) New option added, for which we found one RCT.[33]The RCT found no significant difference between levetiracetam and carbamazepine controlled release in the proportion of people with newly diagnosed epilepsy who were seizure free for at least 6 months at 26 weeks, or for at least 12 months at 52 weeks. Observational harms data added on teratogenicity. Categorised as Likely to be beneficial, based on consensus.

Gabapentin for partial epilepsy
Summary
SEIZURE FREQUENCY Compared with lamotrigine: We don't know whether gabapentin is more effective at increasing the proportion of people who are seizure free, or whether it is associated with an increased time to first seizure in people with partial epilepsy ( very low-quality evidence ). Compared with carbamazepine: We don't know whether gabapentin is more effective at increasing the proportion of older people with new-onset epilepsy who are seizure free at 3 months (very low-quality evidence). NOTE We found no direct information from RCTs about whether gabapentin used as monotherapy in people with generalised epilepsy is better than no active treatment. However, there is consensus that gabapentin reduces seizure rates.
Benefits
Gabapentin versus placebo:
We found no systematic review or RCTs.
Gabapentin versus lamotrigine:
We found no systematic review but we found two RCTs.[24] [37]
The first RCT (593 older people with new-onset epilepsy, 251/593 [42%] with complex partial seizures only, 76/593 [13%] with simple partial seizures only, 77/593 [13%] with generalised tonic clonic plus partial seizures, 30/593 [5%] with mixed partial seizures, 147/593 [25%] with generalised tonic clonic seizures only; average age 72 years) was a three-arm trial comparing carbamazepine (198 people) versus lamotrigine (200 people) versus gabapentin (195 people). [24] The RCT found no significant difference among the three treatment groups in the proportion of people who remained in the study at 3 months (402) and who were seizure free after start of treatment (seizure-free rates at 3 months: 64% with carbamazepine v 63% with lamotrigine v 62% with gabapentin; P = 0.09; absolute numbers not reported, results presented graphically). The RCT did not report on paired way comparisons for these outcomes. About 57% of people were treatment naïve, and those who received treatment were treated for less than 4 weeks, or were treated with sub-therapeutic doses. Doses were titrated for the first 6 weeks of the trial to target doses (1500 mg/day for gabapentin, 150 mg/day for lamotrigine, and 600 mg/day for carbamazepine) although flexible dosing was allowed throughout the duration of the trial if seizure control was poor.[24]
The second RCT (291 people [233 with partial seizures], 52% male, 13–78 years, more than 2 seizures/12 months) comparing gabapentin (1200–3600 mg/day) and lamotrigine (100–300 mg/day) in people with newly diagnosed epilepsy. The RCT found no significant difference in time to exit, a composite outcome of efficacy and safety between gabapentin and lamotrigine at 24 weeks (Kaplan–Meier analysis, 291 people: HR 1.043, 95% CI 0.602 to 1.809). The RCT carried out a subgroup analysis in people with partial seizures, and reported no significant difference between groups in time to exit (exit events: 14/117 [12%] with gabapentin v 19/116 [16%] with lamotrigine; reported as no significant difference in time to exit, details of Kaplan–Meier analysis not reported).[37]
Gabapentin versus carbamazepine:
See benefits of carbamazepine.
Gabapentin versus other antiepileptic drugs:
We found no systematic review or RCTs.
Harms
Antiepileptic drugs have been associated with increased risk of suicidal behaviour and ideation.[18]
Gabapentin versus placebo:
We found no RCTs.
Gabapentin versus lamotrigine:
The first RCT found that, at 12 months, lamotrigine was significantly less likely to lead to adverse effect-related treatment withdrawals compared with gabapentin (P = 0.015; absolute numbers not reported). The RCT found that gabapentin significantly increased the proportion of people who gained weight at 12 months compared with lamotrigine (P = 0.001; absolute numbers not reported). [24]
The second RCT found no significant difference in the proportion of people withdrawing due to adverse events between gabapentin and lamotrigine (17/158 [11%] with gabapentin v 23/151 [15%] with lamotrigine; difference –4.4%, 95% CI –12% to +3%). The most frequent treatment-related adverse effect in both treatment groups were dizziness, asthenia, and headache.[37]
Gabapentin versus carbamazepine:
See harms of carbamazepine.
Gabapentin versus other antiepileptic drugs:
We found no RCTs.
Teratogenicity:
We found no observational studies which reported separately on teratogenicity of gabapentin.
Effect on bone mineral density (BMD):
We found no observational studies which reported separately on the effects of gabapentin on BMD. We found three observational studies reporting on the effect of a antiepileptic drugs in general (including gabapentin) on BMD (see harms of carbamazepine for more details).[29] [30] [31]
Comment
Placebo-controlled trials would now be considered unethical. The RCT did not present data separately for adults and children.
Substantive changes
Gabapentin (partial epilepsy) New option for which we found two RCTs.[24] [37] One RCT was previously reported in this Clinical Evidence review.[24] The second RCT found no significant difference in time to exit between gabapentin and lamotrigine.[37]Categorised as Likely to be beneficial, based on consensus.

Addition of second-line antiepileptic drugs (partial epilepsy)
Summary
SEIZURE FREQUENCY Allopurinol compared with placebo: Adding allopurinol to usual care may be more effective at reducing the total number of seizures over 6 months in people with epilepsy ( low-quality evidence ). Eslicarbazepine compared with placebo: Adding eslicarbazepine to usual care seems more effective at increasing the proportion of 50% responders in people with partial epilepsy ( moderate-quality evidence ). Gabapentin compared with placebo: Adding gabapentin to usual treatment seems more effective at reducing seizure frequency in people with partial epilepsy (moderate-quality evidence). Lacosamide compared with placebo: Adding lacosamide to usual treatment seems more effective at reducing seizure frequency and at increasing the proportion of 50% responders in people with partial epilepsy (moderate-quality evidence). Lamotrigine compared with placebo: Adding lamotrigine to usual treatment seems more effective at reducing seizure frequency in people with partial epilepsy (moderate-quality evidence). Levetiracetam compared with placebo: Adjunctive levetiracetam seems more effective at reducing partial-onset seizure frequency/week and at increasing the proportion of 50% responders in people with partial epilepsy (moderate-quality evidence). Losigamone compared with placebo: Adding losigamone to usual care seems more effective at reducing partial seizure frequency and at increasing the proportion of 50% responders in people with partial epilepsy (moderate-quality evidence). Oxcarbazepine compared with placebo: Adding oxcarbazepine to usual treatment seems more effective at reducing seizure frequency in people with partial epilepsy ( high-quality evidence ). Pregabalin compared with placebo: Adding pregabalin is more effective at reducing seizure frequency in people with partial epilepsy (high-quality evidence). Retigabine compared with placebo: Adjunctive retigabine seems more effective at reducing the frequency of monthly seizures and at increasing the proportion of 50% responders in people with partial epilepsy (moderate-quality evidence). Tiagabine compared with placebo: Adding tiagabine to usual treatment is more effective at reducing seizure frequency in people with partial epilepsy (high-quality evidence). Topiramate compared with placebo: Adding topiramate to usual treatment seems more effective at reducing seizure frequency in people with partial epilepsy (moderate-quality evidence). Vigabatrin compared with placebo: Adding vigabatrin to usual treatment seems more effective at reducing seizure frequency in people with partial epilepsy (moderate-quality evidence). Zonisamide compared with placebo: Adding zonisamide to usual treatment is more effective at reducing seizure frequency in people with partial epilepsy (high-quality evidence). NOTE We found no clinically important results from RCTs about second-line treatment antiepileptic drugs compared with each other in people with drug-resistant partial epilepsy.
Benefits
Adding second-line antiepileptic drugs versus adding placebo:
We found 10 systematic reviews [41] [42] [43] [44] [45] [46] [47] [48] [49] [50] and 12 subsequent RCTs[51] [52] [53] [54] [55] [56] [57] [58] [59] [60] [61] [62] comparing the addition of active drugs versus placebo in people who had not responded to usual drug treatment.
Allopurinol versus placebo:
One RCT (38 people, of whom 26 [68%] had complex partial seizures only, 10/38 [26%] had generalised seizures (tonic, tonic clonic, or absence only), mean age 25 years, 3 seizures/month) compared adjunctive allopurinol 300 mg daily as a three times-daily regimen versus placebo. The RCT found that allopurinol 300 mg daily significantly decreased the total number of seizures over 6 months compared with placebo (32 people, mean decrease in seizure frequency per month; 5.1 with allopurinol v 0.1 with placebo; P = 0.001).[55]
Clobazam versus placebo:
One systematic review (search date 2007, 4 crossover RCTs, 196 people) compared adding clobazam versus placebo in people with refractory partial-onset or generalised-onset seizures. The review did not pool data because of significant methodological differences between studies and differences in outcome measures, and reported the studies narratively. The review commented that all of the studies had methodological weaknesses, including not stating the method of randomisation and incomplete reporting of results. One of the RCTs satisfied Clinical Evidence inclusion criteria (see harms below). The authors of the review concluded that clobazam may be an effective add-on therapy in partial epilepsy.[48]
Eslicarbazepine acetate versus placebo:
One RCT (143 people, 18–65 years, at least 4 partial-onset seizures/month) compared adding once-daily and twice-daily regimens of eslicarbazepine (400–1200 mg/day) versus adding placebo to usual treatment. The RCT found that adding eslicarbazepine once daily significantly increased the proportion of 50% responders compared with placebo at 12 weeks (143 people; 54% with eslicarbazepine once daily v 28% with placebo; absolute numbers not reported; P = 0.008), but there was no significant difference between adding the twice-daily regimen and placebo (41% with eslicarbazepine twice daily v 28% with placebo; absolute numbers not reported; P = 0.12).[53]
A second RCT (402 people, 18−76 years, at least 4 partial-onset seizures/month) compared adding eslicarbazepine (400, 800, or 1200 mg once daily) versus adding placebo to usual treatment. It found that eslicarbazepine (800 and 1200 mg once daily) significantly reduced mean seizure frequency compared with placebo during the 12-week maintenance phase, but found no significant difference between eslicarbazepine (400 mg once daily) and placebo (mean seizure frequency adjusted/4 weeks: 5.35 with eslicarbazepine 1200 mg v 5.66 with eslicarbazepine 800 mg v 6.73 with eslicarbazepine 400 mg v 7.64 with placebo; P = 0.0003 [eslicarbazepine 1200 mg v placebo]; P = 0.0028 [eslicarbazepine 800 mg v placebo]; P value reported as not significant, intention-to-treat [ITT] analysis [eslicarbazepine 400 mg v placebo]). The RCT also found that adding eslicarbazepine (800 and 1200 mg once daily) significantly increased the responder rate (proportion of people with at least 50% decrease in seizure frequency) compared with adding placebo during the 12-week maintenance period, but found no significant difference between adding eslicarbazepine (400 mg once daily) and placebo (43% with eslicarbazepine 1200 mg v 34% with eslicarbazepine 800 mg v 23% with eslicarbazepine 400 mg v 20% with placebo; absolute results not reported; P less than 0.001 [eslicarbazepine 1200 mg v placebo]; P less than 0.05 [eslicarbazepine 800 mg v placebo]; P value reported as not significant; ITT analysis [eslicarbazepine 400 mg v placebo]). The RCT also found that eslicarbazepine (1200 mg once daily) significantly increased the proportion of people seizure free compared with placebo, but found no significant difference between eslicarbazepine (800 or 400 mg once daily) and placebo (8/102 [8%] with eslicarbazepine 1200 mg v 4/98 [4%] eslicarbazepine 800 mg v 2/100 [2%] with eslicarbazepine 400 mg v 2/102 [2%] with placebo; P less than 0.05 [eslicarbazepine 1200 mg v placebo]; P value reported as not significant [eslicarbazepine 800 mg or 400 mg v placebo]).[60]
Gabapentin versus placebo:
One systematic review (search date 2007, 5 RCTs, 997 adolescents and adults all with drug-resistant partial epilepsy, 39–66% men) found that adding gabapentin to usual treatment significantly reduced seizure frequency compared with adding placebo, and that efficacy increased with increasing dose (see table 1 ).[41]
Table 1
Table 1
Effects of additional drug treatment in people not responding to usual treatment: results of systematic reviews only (see text).
Lacosamide versus placebo:
One RCT (421 people, mean age 39.9 years, 46% male, at least 4 partial seizures/month) compared adjunctive lacosamide (200, 400, or 600 mg/day) with placebo. The RCT found that lacosamide (400 and 600 mg/day) significantly reduced partial seizure frequency (relative median % reduction in partial seizure frequency: 39% with lacosamide 400 mg/day v 10% with placebo; P less than 0.01; 40% with lacosamide 600 mg/day v 10% with placebo; P less than 0.01). The RCT also found that lacosamide (400 and 600 mg/day) significantly increased 50% responders compared with placebo (proportion of people with reduction in seizure frequency by at least 50%: 41% with lacosamide 400 mg/day v 22% with placebo; P = 0.0038; 38% with lacosamide 600 mg/day v 22% with placebo; absolute results not reported; P = 0.014). [61]
A second RCT (485 people, mean age 37.8 years, 50% male, at least 4 partial seizures/month) compared adjunctive lacosamide 200 mg daily and 400 mg daily with placebo. The RCT found that both doses of lacosamide significantly reduced partial seizure frequency (relative median % reduction in partial seizure frequency: 35% with lacosamide 200 mg/day v 20% with placebo; P = 0.02; 36% with lacosamide 400 mg/day v 20% with placebo; absolute results not reported; P = 0.03). The RCT also found that lacosamide at 400 mg daily significantly increased the proportion of 50% responders compared with placebo (reduction in seizure frequency by at least 50%; 40% with lacosamide 400 mg/day v 26% with placebo; absolute results not reported; P = 0.01).[62]
Lamotrigine versus placebo:
One systematic review (search date 2007, 11 RCTs, 1243 adults and children, about 50% men) found that adding lamotrigine to usual treatment significantly reduced seizure frequency compared with adding placebo (see table 1 ).[43]One subsequent RCT (243 people, aged at least 12 years [mean age 36 years], 50% male, at least 8 partial seizures/8-week baseline) compared adding lamotrigine extended release (XR) (200–500 mg/day, single daily dose) with adding placebo to usual care. Lamotrigine was titrated to a target dose (depending on other concomitant treatment, mean dose 231 mg/day) over a 7-week period. The RCT found that adding lamotrigine XR significantly reduced partial seizure frequency compared with adding placebo to usual care (median % reduction in weekly seizure frequency from baseline to 19 weeks: 46% with lamotrigine XR v 24% with placebo; P = 0.0004). The RCT also found that lamotrigine XR significantly increased the proportion of 50% responders compared with placebo over 19 weeks' treatment (42% with lamotrigine XR v 24% with placebo; P = 0.0007; absolute numbers not reported).[58]
Levetiracetam versus placebo:
One systematic review (search date 2005, 4 RCTs, 1023 people aged 16 years or over, 49–61% men) found that adding levetiracetam to usual treatment significantly reduced seizure frequency compared with adding placebo (see table 1 ).[42] One subsequent double-blind RCT (198 children, aged 4–16 years, 4 partial-onset seizures/month, 54% male) compared adding levetiracetam at a target dose of 60 mg/kg/day versus adding placebo over 14 weeks. The RCT found that levetiracetam as adjunctive therapy significantly reduced partial-onset seizure frequency/week compared with placebo (AR: 26.8%, 95% CI 14.0% to 37.6%; P = 0.0002) and significantly increased the proportion of 50% responders (proportion of people with at least 50% reduction in partial seizure frequency per week: 45/101 [45%] with levetiracetam v 19/97 [20%] with placebo; P = 0.0002).[51]
A second subsequent RCT (158 people, aged 12–67 years, more than 2 partial-onset seizures/month, 63% male) compared adding levetiracetam (once-daily XR, 1000 mg/day) versus adding placebo to usual care, over 12 weeks. The RCT found that levetiracetam significantly reduced partial seizure frequency (relative median % reduction per week: 46% with levetiracetam v 33% with placebo; P = 0.038). The RCT also found that levetiracetam significantly increased the proportion of 50% responders compared with placebo at 12 weeks (34/79 [43%] with levetiracetam v 23/79 [29%]) with placebo; P = 0.07).[56]
A third subsequent RCT (202 Chinese people, aged 16–70 years, 52% male, at least 8 partial-onset seizures/8-week baseline) compared adding levetiracetam (3000 mg/day) with adding placebo to usual care over 16 weeks. The RCT found that levetiracetam significantly reduced partial seizure frequency (% reduction in median weekly seizure frequency compared with placebo: 27%, 95% CI 14% to 38%; P less than 0.001). The RCT also found that levetiracetam significantly increased the proportion of 50% responders compared with placebo at 16 weeks (56% with levetiracetam v 26% placebo; P less than 0.001). Levetiracetam also increased the proportion of people seizure free compared with placebo at 16 weeks (11% with levetiracetam v 2% placebo; P = 0.012).[57]
Losigamone versus placebo:
One RCT (264 people, mean age 35.7 years, 4 partial-onset seizures/month) compared losigamone 1200 and 1500 mg three times daily versus placebo. The RCT found that both doses of losigamone significantly reduced partial seizure frequency (relative median % reduction in partial seizure frequency: 20% [95% CI 15.0% to 25.5%] with losigamone 1200 mg/day v +3% [95% CI –5.7% to +14.1%] with placebo; P less than 0.01; 25% [95% CI 15.5% to 35.1%] with 1500 mg/day losigamone v +3% [95% CI –5.7% to +14.1%] with placebo; P less than 0.01). The RCT also found that losigamone 1500 mg daily significantly increased the proportion of 50% responders compared with placebo (reduction in seizure frequency by at least 50%: 27/92 [29%] with losigamone 1500 mg/day v 10/85 [12%] with placebo; P = 0.004; 15/87 [17%] with losigamone 1200 mg/day v 10/85 [12%] with placebo; P value not reported). [54]
Oxcarbazepine versus placebo:
One systematic review (search date 2006, 2 RCTs, 961 adults and children, sex not specified) found that adding oxcarbazepine to usual treatment significantly reduced seizure frequency compared with adding placebo (see table 1 ).[44]
Pregabalin versus placebo:
One systematic review (search date 2007, 4 RCTs, 1397 people, all with treatment-resistant partial epilepsy, aged 12–82 years, about 50% male) found that adding pregabalin to usual treatment (1–4 antiepileptic drugs) significantly reduced seizures frequency compared with adding placebo (see table 1 ).[47]
One subsequent RCT (178 Korean people with partial seizures, mean age 33 years, 48% male, at least 4 seizures/6-week baseline) compared adding pregabalin (150–600 mg/day, median dose 367 mg/day) with adding placebo to usual treatment (1–3 antiepileptic drugs). The RCT found that adding pregabalin significantly reduced partial seizure frequency compared with adding placebo during 12 weeks' treatment (mean % change in partial seizure frequency/month from baseline [a positive value indicates worsening of seizures]: –35.8 with pregabalin v –23.2 with placebo; P = 0.015). The RCT found no significant difference between groups in the proportion of 50% responders; however, this was higher with pregabalin (46% with pregabalin v 32% with placebo; P = 0.068). [59]
Retigabine versus placebo:
One RCT (396 people with partial seizures, aged 16–66 years, 52% male) compared adjunctive retigabine (600, 900, and 1200 mg/day) versus placebo over 16 weeks. It found that retigabine (900 and 1200 mg/day) significantly reduced partial seizure frequency compared with placebo (mean % change in monthly total partial seizure frequency from baseline [a positive value indicates worsening of seizures] –14% with retigabine 900 mg v –3% with placebo; P = 0.0387; –24% with retigabine 1200 mg v –3% with placebo; P = 0.0024; 9% with retigabine 600 mg v –3% with placebo; P value not reported). It also found that retigabine 900 and 1200 mg daily significantly increased the proportion of 50% responders compared with placebo, but found no significant difference between retigabine 600 mg daily and placebo (30/95 [32%] with retigabine 900 mg/day v 15/96 [16%] with placebo; P = 0.021; 35/106 [33%] with retigabine 1200 mg/day v 15/96 [16%] with placebo; P = 0.016; 23/99 [23%] with retigabine 600 mg/day v 15/96 [16%] with placebo; reported as not significant; P value not reported). [52]
Tiagabine versus placebo:
One systematic review (search date 2008, 3 RCTs, 769 people aged 12–77 years, at least 56% men) found that adding tiagabine to usual treatment significantly reduced seizure frequency compared with adding placebo (see table 1 ).[46]
Topiramate versus placebo:
One systematic review (search date 2007, 10 RCTs, 1312 adults and children, 49–85% men) found that adding topiramate to usual treatment significantly reduced seizure frequency compared with adding placebo (see table 1 ).[49]
Vigabatrin versus placebo:
One systematic review (search date 2008, 11 RCTs, 745 people aged 6–64 years, 47% men) found that adding vigabatrin to usual treatment significantly reduced seizure frequency compared with adding placebo (see table 1 ).[50]
Zonisamide versus placebo:
One systematic review (search date 2007, 4 RCTs, 850 people aged 12–77 years, 51–66% men) found that adding zonisamide to usual treatment significantly reduced seizure frequency compared with adding placebo (see table 1 ).[45]
Other epileptic drugs versus placebo:
We found no RCTs.
Antiepileptic drugs versus other antiepileptic drugs:
We found no RCTs directly comparing adjuvant antiepileptic drug therapy. We found one systematic review (search date 2002, 37 RCTs, 6541 people) indirectly comparing treatment outcomes in adjuvant antiepileptic drug trials (see comment).
Harms
Adding second-line antiepileptic drugs versus adding placebo:
Antiepileptic drugs have been associated with increased risk of suicidal behaviour and ideation.[18] Some antiepileptic drugs have also been associated with adverse outcomes in pregnancy and adverse effects on bone mineral density (see alson harms described under each option in question on monotherapy in people with partial epilepsy).
Allopurinol versus placebo:
The RCT found no significant difference in withdrawal due to adverse effects between adding allopurinol and adding placebo (1 reported in each group; reported as no significant difference; absolute numbers not reported). It reported the following adverse effects in people taking allopurinol: transient rash (2 people), nausea (2 people), and dizziness (2 people); and in people taking placebo: rash (1 person), nausea (1 person).[55]
Clobazam versus placebo:
The review reported that one RCT found higher rates of adverse effects with adding clobazam than with adding placebo (38/106 [36%] with clobazam v 13/106 [12%] with placebo; significance assessment not reported). It found that drowsiness and dizziness were the adverse effects most commonly reported with clobazam (drowsiness: 71% with clobazam v 16% with placebo; dizziness: 26% with clobazam v 6% with placebo; significance assessments not reported).[48]
Eslicarbazepine acetate versus placebo:
The first RCT reported similar rates of adverse effects in the three groups (proportion of people reporting adverse effects: 19/50 [38%] with eslicarbazepine once daily v 19/46 [41%] with eslicarbazepine twice daily v 21/47 [45%] with placebo; significance assessment not reported). Central nervous system effects (headache, dizziness) and nausea were the most commonly reported adverse effects.[53]
The second RCT reported that treatment-emergent adverse effects (most commonly dizziness, headache, and somnolence) were more common with higher doses of eslicarbazepine compared with lower doses or placebo (proportion of people with any treatment-related adverse effect: 62/102 [61%] with eslicarbazepine 1200 mg/day v 49/98 [50%] with eslicarbazepine 800 mg/day v 44/100 [44%] with eslicarbazepine 400 mg/day v 32/102 [31%] with placebo; significance assessment not reported). It found similar rates of serious treatment-emergent adverse effects (not further defined) between groups (proportion of people with serious treatment-related adverse effect: 6/102 [6%] with eslicarbazepine 1200 mg/day v 4/98 [4%] with eslicarbazepine 800 mg/day v 5/100 [5%] with eslicarbazepine 400 mg/day v 4/102 [4%] with placebo; significance assessment not reported).[60]
Gabapentin versus placebo:
The review found that adverse effects (dizziness, fatigue, somnolence) were reported significantly more frequently with adding gabapentin than with placebo (see table 1 ).[41]
Lacosamide versus placebo:
One RCT found higher rates of any treatment-emergent adverse effect with higher doses of lacosamide (proportion of people with any treatment-related adverse effect: 98/106 [92%] with lacosamide 600 mg/day v 87/108 [81%] with lacosamide 400 mg/day v 85/107 [79%] with lacosamide 200 mg/day v 68/97 [70%] with placebo; significance assessment not reported). The most common treatment related adverse effects in all groups were dizziness, headache, nausea, fatigue, and ataxia.[61]
The second RCT reported that the most common treatment-emergent adverse effects in all groups were dizziness, headache, diplopia, nausea, vertigo, fatigue, nasopharyngitis, abnormal coordination, and vomiting. It found that more people withdrew owing to adverse effects with the higher dose of lacosamide compared with the lower dose or with placebo (24/159 [15%] with lacosamide 400 mg/day v 10/163 [6%] with lacosamide 200 mg v 8/163 [5%] with placebo; significance assessment not reported).[62]
Lamotrigine versus placebo:
The review found that adverse effects (ataxia, dizziness, diplopia, nausea) were reported significantly more frequently with adding lamotrigine than with adding placebo to usual care (see table 1 ).[43]
The subsequent RCT reported similar rates of any adverse effects reported in both groups (proportion of people with at least one adverse effect: 69% with lamotrigine v 62% with placebo; absolute numbers and significance assessment not reported). It found that central nervous system effects (headache, dizziness) were the most common treatment-emergent adverse effects with lamotrigine (headache: 20/118 [17%] with lamotrigine v 18/121 [15%] with placebo; dizziness: 21/118 [18%] with lamotrigine v 6/121 [5%] with placebo; significance assessment not reported).[58]
Levetiracetam versus placebo:
The review found that adverse effects (dizziness, infection) were reported significantly more frequently with adding levetiracetam than with placebo (see table 1 ).[42] The first subsequent RCT reported similar rates of people reporting treatment-emergent adverse effects (89/101 [88%] with levetiracetam v 89/97 [92%] with placebo; significance assessment not reported). It found that somnolence, accidental injury, and vomiting were the most common treatment-emergent adverse effects reported with levetiracetam.[51] The second subsequent RCT reported similar rates of people reporting treatment-emergent adverse effects (43/79 [54%] with levetiracetam v 41/77 [53%] with placebo; significance assessment not reported). It found that central nervous system effects (somnolence, irritability, headache, dizziness) were the most common treatment-emergent adverse effects reported with levetiracetam.[56]The third subsequent RCT reported similar rates of treatment-emergent adverse effects (65/103 [63%] with levetiracetam v 62/103 [60%] with placebo; significance assessment not reported). [57]
Losigamone versus placebo:
The RCT reported that central nervous system effects were the most common treatment-emergent adverse effects for losigamone.[54]
Oxcarbazepine versus placebo:
The review found that adverse effects (ataxia, dizziness, fatigue, nausea, somnolence, diplopia) were reported significantly more frequently with adding oxcarbazepine than with placebo (see table 1 ).[44]
Pregabalin versus placebo:
The review found that adverse effects (ataxia, dizziness, somnolence, and weight gain) were reported significantly more frequently with adding pregabalin than with placebo (see table 1 ).[47] It found no significant difference between groups in the proportion of people reporting fatigue, headache, or nausea.The subsequent RCT found higher rates of any treatment-related adverse effect with adding pregabalin versus adding placebo to usual care, but it did not assess the significance of the difference (77/119 [65%] with pregabalin v 18/59 [31%] with placebo; significance not assessed). It reported that central nervous system effects (dizziness, somnolence) and weight gain were the most common treatment-emergent adverse effects with pregabalin.[59]
Retigabine versus placebo:
The RCT reported that central nervous system effects were the most common treatment-emergent adverse effects for retigabine.[52]
Tiagabine versus placebo:
The review found that dizziness was reported significantly more frequently with adding tiagabine than with placebo (see table 1 ).[46]
Topiramate versus placebo:
The review found that adverse effects (ataxia, dizziness, fatigue, nausea, somnolence, "thinking abnormally") were reported significantly more frequently with adding tiagabine than with placebo (see table 1 ).[49]
Vigabatrin versus placebo:
The review found that adverse effects (fatigue/drowsiness, depression) were reported significantly more frequently with adding vigabatrin than with placebo (see table 1 ).[50] Vigabatrin causes concentric visual-field abnormalities in about 40% of people, which are probably irreversible.[63]
Zonisamide versus placebo:
The review found that adverse effects (ataxia, dizziness, somnolence, agitation/irritability, anorexia) were reported significantly more frequently with adding zonisamide than with placebo (see table 1 ).[45] A drug safety alert has been issued on the risk of metabolic acidosis associated with zonisamide (http://www.fda.gov).
Other epileptic drugs versus placebo:
We found no RCTs.
Antiepileptic drugs versus other antiepileptic drugs:
We found no RCTs.
Comment
Few RCTs have compared second-line drugs directly versus each other. One systematic review carried out a meta-analysis of placebo-controlled trials of similar design, and indirectly compared levetiracetam with gabapentin, lamotrigine, oxcarbazepine, tiagabine, topiramate, and zonisamide. The review found that the response rate was significantly higher with levetiracetam compared with gabapentin or lamotrigine (Mantel–Haenszel OR greater than 1 depicting the outcome was more likely with levetiracetam than another antiepileptic drug; levetiracetam v gabapentin: OR 2.64, 95% CI 1.51 to 4.63; P = 0.001; levetiracetam v lamotrigine: OR 1.86, 95% CI 1.04 to 3.34; P = 0.0036). Levetiracetam had a significantly lower withdrawal rate compared with oxcarbazepine or topiramate (levetiracetam v oxcarbazepine: OR 0.55, 95% CI 0.33 to 0.92; P = 0.022; levetiracetam v topiramate: OR 0.52, 95% CI 0.29 to 0.93; P = 0.027).[64]Caution must be exercised with this meta-analysis as the results are only valid for the drug doses used within the trials. Because of the irreversible visual-field abnormalities associated with vigabatrin, consensus among neurologists is not to recommend this drug. The reviews did not present results separately for adults and children.[41] [42] [43] [44] [49] [45] [46] [53]
Substantive changes
Addition of second-line antiepileptic drugs (in drug-resistant partial epilepsy) Four systematic reviews added.[47] [48] [50] [49]Two of these systematic reviews supersede two previously reported reviews. Seven RCTs added.[56] [57] [58] [59] [60] [61] [62] The newly added reviews found that adding pregabalin, vigabatrin, and topiramate reduced seizure frequency compared with adding placebo in people with drug-resistant partial epilepsy. The newly added RCTs found that adding levetiracetam, lamotrigine (extended release), pregabalin, eslicarbazepine, and lacosamide reduced seizure frequency compared with adding placebo in people with drug-resistant partial epilepsy. Four other systematic reviews updated, search dates updated, no new evidence included.[41] [43] [45] [46] Categorisation unchanged (Beneficial).

Antiepileptic drug withdrawal for people in remission from epilepsy
Summary
SEIZURE FREQUENCY Continued antiepileptic treatment compared with slow antiepileptic drug withdrawal: We don't know whether continuing treatment with antiepileptic drugs is more effective at increasing the proportion of people who remain seizure free at 2 years ( moderate-quality evidence ). NOTE Clinical predictors of relapse after drug withdrawal include age, seizure type, number of antiepileptic drugs being taken, whether seizures have occurred since antiepileptic drugs have started, and the period of remission before drug withdrawal. Antiepileptic drugs are associated with idiosyncratic reactions, teratogenesis, and cognitive effects.
Benefits
We found one systematic review, which identified one open-label RCT (149 children, mean age 11 years, 53% male, seizure free for over 18 months) comparing rapid withdrawal over 6 weeks versus slow withdrawal over 9 months of antiepileptic drugs in children who had been seizure free for 2 to 4 years. The RCT did not meet Clinical Evidence inclusion criteria and is not discussed further. We found two RCTs.[65] [66]
The first large RCT (1013 people aged at least 26 years who had been seizure free for more than 2 years, 49% men) was reported in two publications and compared continued antiepileptic treatment versus slow antiepileptic drug withdrawal.[67] [65] At 2 years, 78% of people who continued treatment remained seizure free compared with 59% in the withdrawal group. There were no significant differences in psychosocial outcomes between groups. Risk reductions with 95% confidence intervals for the main factors predicting recurrence of seizures are tabulated (see table 2 ).[65]
Table 1
Table 1
Relative risks of seizure recurrence within 2 years of treatment withdrawal, according to prognostic variable.[67] [65]
The second RCT (160 people, aged 18–66 years, who had been seizure free for at least 2 years on monotherapy, 47% men, 76% partial epilepsy, see comment below) compared continued antiepileptic treatment (carbamazepine, valproate, phenytoin, phenobarbital, lamotrigine) versus drug withdrawal at a rate of 20% dose reduction every 6 weeks. It found no significant difference between groups in the proportion of people who had seizure relapse after 1 year (5/77 [7%] with continued treatment v 11/72 [15%] with withdrawal; RR 2.46, 95% CI 0.85 to 7.08; P = 0.095). It found no significant difference between groups in health-related quality of life (assessed by QOLIE-89, EQ-5D, or 15D questionnaire) after 1 year (QOLIE-89 overall score, change from baseline: from 58.1 to 60.0 with continued treatment v from 56.9 to 59.8 with withdrawal; mean difference in changes between group 0.3, 95% CI –1.55 to +2.07; P = 0.78; 15D utility score, change from baseline: from 0.92 to 0.94 with continued treatment v from 0.91 to 0.93 with withdrawal; mean difference in changes between group –0.011, 95% CI –0.027 to +0.005; P = 0.22; EQ visual analogue scale [VAS] mean, change from baseline: from 77.5 to 78.5 with continued treatment v from 78.9 to 83.0 with withdrawal; mean difference in changes between group +3.03, 95% CI –0.99 to +7.06; P = 0.14). Predictors for remaining seizure free after antiepileptic drug withdrawal over 1 year were normal neurological examination (OR 2.77, 95% CI 1.18 to 142.86; P = 0.036) and use of carbamazepine prior to withdrawal (OR 2.86, 95% CI 1.31 to 6.26; P = 0.01).[66]
Harms
Antiepileptic drugs have been associated with increased risk of suicidal behaviour and ideation.[18] Some antiepileptic drugs have also been associated with adverse outcomes in pregnancy and adverse effects on bone mineral density (see harms described under each option in question on monotherapy in people with partial epilepsy).
In the first RCT, 16 people died during the trial, 10 in the continued-treatment group and six in the withdrawal group.[67] Only two deaths were attributed to epilepsy, and both of these occurred in people randomised to continued treatment.
The second RCT did not report on adverse effects during the study period (1 year). It reported in further follow-up (median 47 months, and 41 months for people off medication) that four people died, two from apparently sudden unexpected death in epilepsy (one a few weeks after withdrawal, and one 4 years after withdrawal).[66]
Comment
The second RCT also examined the effects of drug withdrawal on neuropsychological performance 4 months after withdrawal from treatment was completed. It found a significant improvement in the following tests with withdrawal compared with continuing treatment: verbal fluency under time pressure, complex motor coordination with non-dominant hand, response inhibition under time pressure, choice reaction time with lexical stimuli, and form discrimination after antiepileptic drug withdrawal.[66]People with a seizure recurrence were less likely to be in paid employment at 2 years.[67] [65]There is currently no reliable evidence to guide the rate at which antiepileptic drugs should be withdrawn. [68]
Substantive changes
Antiepileptic drug withdrawal for people in remission from partial or generalised epilepsy One RCT added comparing continued antiepileptic treatment versus drug withdrawal at a rate of 20% dose reduction every 6 weeks. It found no significant difference between groups in the proportion of people who had seizure relapse or in health-related quality of life, after 1 year.[66]Categorisation unchanged (Trade-off between benefits and harms).

Relaxation therapy for people with epilepsy
Summary
SEIZURE FREQUENCY Compared with no relaxation therapy: We don't know whether relaxation therapy is more effective at reducing seizure frequency in people with epilepsy ( very low-quality evidence ).
Benefits
We found one systematic review (search date 2007,[69] 3 small open-label controlled trials,[70] [71] [72] 50 adults including 32 women). The trials used weak methods (see comment below). Two of the studies found a non-significant reduction in seizure frequency with relaxation therapy compared with no relaxation therapy, and one study found a significantly reduced seizure frequency. The weak methods preclude reliable conclusions.
Harms
The RCTs gave no information on adverse effects.[70] [71] [72]
Comment
All three trials used weak methods.[70] [71] [72] The treatment allocation methods were strict alternation,[72] alternation in blocks of five,[71] or were not reported.[70] The baseline seizure frequency varied considerably among the allocated groups in all of the trials. In one trial, two people in the treatment group had new antiepileptic medication added during the study period and one of these had a greater than 50% reduction in seizure frequency; another person discontinued antiepileptic medication.[71] Antiepileptic drug treatment was also adjusted during the trial, making it difficult to conclude whether the observed results were due to changes in drug treatment or due to the intervention. The trial duration and follow-up were short. The possibility of publication bias cannot be excluded. The effects of relaxation therapy remain unclear.
Substantive changes
Relaxation therapy for people with partial or generalised epilepsy One systematic review,[69] search date updated, no new evidence found. Categorisation unchanged (Unknown effectiveness).

Yoga for people with epilepsy
Summary
SEIZURE FREQUENCY Compared with control: We don't know whether yoga is more effective at reducing seizure frequency in people with epilepsy ( very low-quality evidence ).
Benefits
We found one systematic review (search date 2006,[73]1 quasi-randomised trial,[74] 32 adults, including 30 women). The trial compared sahaja yoga (10 people) versus control (sham yoga [10 people], no intervention [12 people]). It found that yoga reduced seizure frequency compared with control, but it used weak methods, which precludes reliable conclusions.
Harms
The trial gave no information on adverse effects.[74]
Comment
The baseline seizure frequency and duration varied among the groups, making results difficult to interpret.[74]
Substantive changes
Yoga for people with partial or generalised epilepsy One systematic review,[73] search date updated, no new evidence found. Categorisation unchanged (Unknown effectiveness).

Biofeedback for people with epilepsy
Summary
SEIZURE FREQUENCY Electroencephalographic biofeedback compared with control/no feedback: We don't know whether electroencephalographic feedback is more effective at reducing seizure frequency in people with uncontrolled epilepsy ( very low-quality evidence ).
Benefits
Electroencephalographic biofeedback:
We found one systematic review (search date 2007),[69]which identified one controlled trial[75] (24 adults, including 15 men, with uncontrolled epilepsy) of electroencephalographic biofeedback compared with control treatment. The trial compared three treatments: electroencephalographic biofeedback, sham (non-contingent) feedback, and no intervention (8 people in each group). It found a significant reduction in seizure frequency compared with the baseline frequency in people given biofeedback (median seizure reduction with biofeedback 61% v baseline; P less than 0.005; see comment below).
Galvanic skin response biofeedback:
We found no systematic review or RCTs that met our inclusion criteria.
Harms
Electroencephalographic biofeedback:
The trial gave no information on harms.[75]
Galvanic skin response biofeedback:
We found no RCTs.
Comment
Electroencephalographic biofeedback:
The RCT did not provide data about seizure frequency in the control group.[75] We were therefore unable to compare the electroencephalographic-biofeedback and control groups. The RCT did not report on the proportion of people who had a greater than 50% reduction in seizure frequency. The study was not blinded and the randomisation method is not clear. The duration of follow-up was only 6 weeks. The evidence is insufficient to draw reliable conclusions about the effects of electroencephalographic biofeedback.
Substantive changes
Biofeedback for people with partial or generalised epilepsy One systematic review, [69] search date updated, no new evidence found. Categorisation unchanged (Unknown effectiveness).

CBT for people with epilepsy
Summary
SEIZURE FREQUENCY Compared with control: We don't know whether CBT is more effective at reducing seizure frequency ( very low-quality evidence ). QUALITY OF LIFE Compared with control: We don't know whether CBT is more effective at improving pyschosocial functioning or quality of life measures (very low-quality evidence).
Benefits
Cognitive behavioural therapy (CBT) versus control:
We found one systematic review (search date 2007).[69]
Seizure frequency:
The systematic review identified two RCTs comparing CBT versus control treatments, which examined seizure frequency.[76] [77]
The first RCT (30 adults with significant psychological problems and inadequate seizure control, as judged by the treating neurologist) included in the review, was a three-arm unblinded trial comparing CBT versus supportive counselling control versus no intervention. It found no significant difference between CBT and either supportive counselling control or no treatment in reduction in seizure frequency (proportion of people with greater than 67% reduction in seizure frequency: 1/10 [10%] with CBT v 2/10 [20%] with supportive counselling v 1/10 [10%] with control; OR [CBT v supportive counselling] 0.47, 95% CI 0.04 to 5.19; OR [CBT v no treatment] 1.00, 95% CI 0.06 to 17.25).[76] However, the RCT was too small to detect a clinically important difference. The method of randomisation concealment was also not known.
The second RCT included in the review (27 adults, aged 21–55 years, refractory epilepsy, 4 seizures/3 months) compared acceptance commitment therapy (ACT, which is developed from the same theory of learning as CBT) versus supportive therapy (people instructed to reflect on their lives and problems in a non-judgemental empathic and accepting environment, without any active advice).[77] The RCT found that ACT significantly reduced frequency of seizures at 6 and at 12 months' follow-up compared with supportive therapy (change in mean seizure frequency [from pre-treatment] at 6 months: 3.79 to 0.70 with ACT v 5.84 to 5.86 with supportive therapy; WMD –5.16, 95% CI –7.18 to –3.14; at 12 months: 3.79 to 0.62 with ACT v 5.84 to 5.80 with supportive therapy; WMD –5.18, 95% CI –7.14 to –3.22; P less than 0.05 for both comparisons).
Psychosocial functioning:
The review[69] identified four RCTs assessing the effects of CBT on psychosocial functioning.[76] [77] [78] [79] One of these RCTs (15 people with epilepsy and depression) did not meet our inclusion criteria and is not discussed further.[79]
The first RCT (30 adults with significant psychological problems and inadequate seizure control, as judged by the treating neurologist) included in the review found no significant difference between CBT and control treatments in various psychological scales, such as the Washington Psychosocial Inventory, the Minnesota Multiphasic Personality Inventory, and the Beck Depression Inventory.[76]
The second RCT (27 adults, 21–55 years, refractory epilepsy, 4 seizures/3 months) found that CBT significantly improved quality-of-life scores (measured on the WHO Quality of Life [WHOQOL-BREF] scale and the satisfaction with life scale [SWLS]) at 6 months and at 12 months compared with supportive therapy (mean WHOQOL-BREF scores at 6 months: 61.21 with CBT v 56.08 with supportive therapy; P less than 0.001; mean SWLS scores: 27.07 with CBT v 14.46 with supportive therapy; P less than 0.001; mean WHOQOL-BREF scores at 12 months: 66.07 with CBT v 51.85 with supportive therapy; P less than 0.001; mean SWLS scores: 27.07 with ACT v 15.77 with supportive therapy; P less than 0.001).[77]
The third RCT (30 people, 13–19 years, newly diagnosed epilepsy and with sub-threshold depression) compared CBT versus treatment with counselling as usual (TAU).[78] Assessments were made at baseline, and at 6 and at 9 months, using the Beck Depression inventory (BDI), the Center for Epidemiological Study on depression (CES-D), the Hamilton Depression Scale (HAMD), and the Quality-of-Life Epilepsy Inventory Total Scores (QOLIE-31 Total). The RCT found that CBT significantly improved subthreshold depression compared with TAU during follow-up ([higher QOLIE-31 Total scores, and lower BDI, and CES-D scores indicate an improvement]; mean BDI scores at 9 months: 5.60 with CBT v 7.70 with TAU; CES-D scores at 9 months: 10.5 with CBT v 13.8 with TAU; QOLIE-31 Total scores at 9 months: 56.40 with CBT v 42.23 with TAU; P less than 0.05 between the 2 groups for all comparisons). The RCT found that depression was diagnosed in three people in the TAU group and none in the CBT group after 7 to 9 months (reported as not significant; P value not reported).[78]
Harms
The review gave no information on harms.[69]
Comment
Publication bias cannot be excluded. The evidence is insufficient to define the effects of CBT in people with epilepsy.
Substantive changes
CBT for people with partial or generalised epilepsy One systematic review updated.[69] It now includes two RCTs previously reported separately in this Clinical Evidence review. Conclusion unchanged. Categorisation unchanged (Unknown effectiveness).

Educational programmes for people with epilepsy
Summary
SEIZURE FREQUENCY Compared with control: We don't know whether a 2-day educational programme is more effective at reducing seizure frequency at 6 months ( low-quality evidence ). QUALITY OF LIFE Compared with control: Educational programmes may be more effective at improving knowledge and understanding of epilepsy, adjustment to epilepsy, and psychosocial functioning, but not at improving health-related quality-of-life scores ( very low-quality evidence ).
Benefits
We found one systematic review (search date 2007,[69] 3 RCTs,[80] [81] [82] and one quasi-randomised trial).[83]
Seizure frequency:
One RCT included in the review (242 people aged 16–80 years, 57% women) reported on seizure frequency.[82] It found that a 2-day educational programme significantly reduced seizure frequency at 6 months compared with waiting list control (proportion of people with at least 2-point reduction in seizure frequency on a 6-point scale [0 = no seizures in last 6 months, 5 = at least 1 seizure daily]: 19% with education v 7% with control; P value not reported).[82] However, the clinical importance of this effect is unclear.
Psychosocial functioning:
The review[69] identified four RCTs assessing psychological functioning.[80] [81] [82] [83]
The first RCT included in the review (43 adults, including 28 women) found that a specific 2-day educational programme significantly improved responses to a 50-item true/false questionnaire compared with control intervention (overall understanding of epilepsy, significant decrease in fear of seizures, significant decrease in hazardous medical self-management) and significantly improved compliance with current medication (shown by serum antiepileptic drug levels).[80]
The second RCT included in the review (252 children aged 7–14 years) found that a child-centred, family-focused educational programme significantly improved questionnaire responses compared with control intervention (knowledge about what to do during a seizure, purpose of the electroencephalographic examination, and minimal restriction in activities), increased the proportion of children likely to participate in normal activities, improved perceived academic and social competencies of the children, and reduced the anxiety of parents (see comment below).[81]
The third RCT (242 people aged 16–80 years, 57% women) in the review that also reported seizure frequency found that a 2-day educational programme had no significant effect on SF-36 questionnaire scores 6 months after the programme compared with waiting list control (SF-36 mental health component score: 43.7 with educational package v 42.5 with control; P value not reported; SF-36 physical component score: 50.4 with educational package v 52.0 with control; P value not reported).[82] Scales validated using the study population revealed significant improvement in epilepsy knowledge and coping with epilepsy.
The quasi-randomised trial in the review (30 adults, number of women not stated) compared a 2-day modular didactic psychoeducational programme on adjustment to epilepsy, stigma, psychoneurotic traits, depression, and knowledge about epilepsy versus waiting list control.[83] It found that the educational programme significantly improved depression and neurotic disorders at 2 months compared with control (change in depression measured using Beck Depression Inventory scores: from 15.0 at baseline to 1.5 with psychoeducational programme v from 15.1 at baseline to 10.0 with control; P less than 0.0001; neurotic disorders assessed using change in Crown Crisp Experiential Index scores: from 36.4 at baseline to 7.3 with psychoeducational programme v from 35.6 at baseline to 34.1 with control; P less than 0.0001).[83]
Harms
The review gave no information on adverse effects.[69]
Comment
All of the RCTs had weak methods.[80] [81] [82] [83] In one RCT, randomisation was by random number assignment, but only a proportion of medical records were available to the authors (65% in the psychoeducational programme group v 47% in the control group).[80] In the second RCT, the method of randomisation was not reported.[81] A minority of the people in the first RCT actively participated in the interventions (23/50 [46%] in the psychoeducational programme group v 20/50 [40%] in the control group) or completed the study (20/50 [40%] in the psychoeducational programme group v 18/50 [36%] in the control group).[80] In the third RCT, the method of randomisation was not reported, and, among 383 people randomised, 242 (113 in the psychoeducational programme group and 119 in the control group) completed the study.[82] In the fourth RCT, randomisation was by alternate allocation.[83] The interventions used in these studies were similar, but there was some variation, which is outlined below. The first RCT used the Sepulveda Epilepsy Education program — a 2-day psychoeducational treatment programme designed to provide medical education and psychosocial therapy.[80] In the second RCT, children with epilepsy and their parents attended four 1.5-hour sessions at weekly intervals.[81] The children's sessions included: understanding body messages and how seizures occur; controlling seizures with medication; straightforward methods of talking about seizures; and coping and adapting to epilepsy. The third RCT used MOSES (Modular service package Epilepsy), a programme developed for use in German-speaking countries.[82] It included nine units: living with epilepsy, epidemiology, basic knowledge, diagnostics, therapy, self-control, prognosis, psychosocial aspects, and network. The programme was delivered in 14 hours over 2 days. The fourth RCT used a 2-day modular, didactic, psychoeducational programme, which comprised preliminary assessment and establishment of rapport on day 1 and a didactic modular educational session (covering background information, diagnosis and management of epilepsy, day-to-day adjustments, seizure control, and stigma), with time for group discussions and clarifications, on day 2.[83]
Substantive changes
Educational programmes for people with partial or generalised epilepsy One systematic review,[69] search date updated, no new evidence found. Categorisation unchanged (Likely to be beneficial).

Relaxation plus behavioural modification therapy for people with epilepsy
Summary
SEIZURE FREQUENCY Compared with control: We don't know whether combined relaxation and behavioural modification treatment is more effective at reducing seizure frequencies ( very low-quality evidence ).
Benefits
We found one systematic review (search date 2007),[69] which identified two RCTs comparing relaxation plus behavioural modification therapy versus control.[84] [85]The first small RCT (18 children with uncontrolled epilepsy) compared three interventions for 6 weeks. This RCT is not discussed further as it did not meet our inclusion criteria. [84]The second RCT (150 adults with uncontrolled epilepsy, including 68 women and 67 men who completed the study) compared Jacobson's muscle relaxation plus behavioural therapy versus control treatment.[85]
Seizure frequency:
The second RCT reported separately the mean seizure frequencies for each seizure type but did not specify the number in each category. It reported separately the mean seizure frequencies for those people with fewer than 20 seizures/month and those with more than 20 seizures/month at baseline. We were unable to analyse these results in a meaningful way. [85]
Psychological outcomes:
The second RCT found that relaxation plus behavioural modification therapy (conditioning, desensitisation, aversion, assertive training, and supportive psychotherapy) significantly improved anxiety (Spielberger's self-assessment questionnaire for trait and state anxiety; P less than 0.01); and home, health, social, and emotional adjustment compared with control (assessed by adjustment inventory; P values not reported).[85]
Harms
The RCT gave no information on adverse effects.[85]
Comment
It is possible that the results of the psychological interventions on psychosocial functioning may depend on the baseline personalities of the people included in the study, and on their education and intelligence.
Substantive changes
Relaxation plus behavioural modification therapy for people with partial or generalised epilepsy One systematic review,[69] search date updated, no new evidence found. Categorisation unchanged (Unknown effectiveness).

Family counselling for people with epilepsy
Summary
QUALITY OF LIFE Compared with control: We don't know whether family counselling is more effective at improving psychosocial inventory scores ( very low-quality evidence ).
Benefits
We found no systematic review but found one small RCT (36 adults, including 26 men, with epilepsy and job loss), which compared three interventions: family therapy (no detailed description, but it seems that the family was present for discussion of problems for a mean of 7.8 sessions), one family session (in which information about the seizure profile was given), and usual care (vocational assistance in obtaining a job, with no follow-up other than site visit).[86] It did not report on seizure frequencies, but found a significantly improved psychosocial inventory score with family therapy (Washington Psychosocial Inventory, 27 completers: improved perceived acceptance by family, emotional adjustment, interpersonal adjustment, adjustment to seizures, and overall psychosocial function). It found a trend towards improvement in job stability.
Harms
The RCT gave no information on adverse effects.[86]
Comment
The method of concealment of randomisation was not described in the RCT.[86] Nine of the 36 people did not complete the study, and withdrawal was uneven across the groups (2 with family therapy, 6 with 1 family session, 1 with no intervention). The available evidence is insufficient to define the effects of family counselling.
Substantive changes
No new evidence

Temporal lobectomy for drug-resistant temporal lobe epilepsy
Summary
SEIZURE FREQUENCY Compared with medical treatment: Temporal lobectomy seems more effective at 1 year at increasing the proportion of people completely free of seizures, and at improving the proportion of seizure-free people with or without auras ( moderate-quality evidence ). QUALITY OF LIFE Compared with medical treatment: Temporal lobectomy may be more effective at improving quality of life at 1 year in people with poorly controlled temporal lobe epilepsy (moderate-quality evidence). NOTE There is consensus that temporal lobectomy is beneficial for people with drug-resistant temporal lobe epilepsy.
Benefits
We found one systematic review (search date 2003, 1 RCT, 80 adults, at least 52% women, with poorly controlled temporal lobe epilepsy).[87] The RCT compared temporal lobectomy versus medical treatment for 1 year.[88]
The RCT found that temporal lobectomy significantly increased the proportion of people completely free of seizures, and the proportion free of seizures with or without auras, compared with medical treatment after 1 year (seizure free: 15/40 [38%] with surgery v 1/40 [2%] with control; NNT 3, 95% CI 2 to 5; seizure free with or without auras: 23/40 [58%] with surgery v 3/40 [7%] with control; NNT 2, 95% CI 2 to 3).[88]It found that surgery improved quality of life compared with medical treatment at 1 year (Quality of Life in Epilepsy Inventory-89 [range 0 to maximum quality of 100]: 73.8 with surgery v 64.3 with medical treatment; P less than 0.001 after adjusting for baseline differences).[88]However, it found no significant difference between surgery and medical treatment in the proportion of people employed or attending school at 1 year (56% with surgery v 38% with medical treatment; P = 0.11).[88]
Harms
The RCT found no deaths at 1 year after surgery and one death, of unknown cause, with medical treatment.[88] The RCT found that neurological adverse effects were more common with surgery than with medical treatment at 1 year (4/40 [10%] with surgery [1 person with small thalamic infarct causing thigh dysaesthesia, 1 person with an infected wound, 2 people with decline in verbal memory affecting occupation for 1 year] v 0/40 [0%] with medical treatment; P value not reported).[88] It found that 22/40 (55%) people had asymptomatic superior subquadrantic visual-field defects after surgery. The RCT found similar rates of depression with surgical and medical treatment (18% with surgery v 20% with medical treatment; P value not reported).[88]It found transient psychosis in one person (1/40 [3%]) in each treatment group.[88]
Comment
There is consensus that temporal lobectomy is beneficial for people with drug-resistant temporal lobe epilepsy. We found two systematic reviews of non-randomised studies reporting long-term outcomes after resective epilepsy surgery. [89] [90]The first review found that, in the long term (over 5 years), seizure freedom after temporal lobectomy was 66% (19 studies, 1803 people, 95% CI 64% to 69%) using Engel’s classification of seizure freedom. [89]On average, 14% (95% CI 11% to 17%) of people were able to discontinue their antiepileptic drug treatment, 50% (95% CI 45% to 55%) achieved antiepileptic monotherapy, and 33% (95% CI 29% to 38%) remained on polytherapy.[90] Although non-controlled studies consistently reported an improvement in the long-term psychosocial outcomes (such as driving status, quality of life, educational and employment status, interpersonal relationships, and social behaviour), the effect was less clear in studies that included a medically treated control group. Mortality was lower in people who become seizure free after epilepsy surgery, but was still higher compared with the general population. Five studies found no worsening in intelligence scores. Three studies found conflicting information on long-term memory outcomes, depending on factors such as side of surgery and seizure freedom.[90]
Substantive changes
No new evidence

Amygdalohippocampectomy for drug-resistant temporal lobe epilepsy
Summary
We found no clinically important results from RCTs about amygdalohippocampectomy in people with drug-resistant temporal lobe epilepsy. However, there is consensus that amygdalohippocampectomy is likely to be beneficial for people with drug-resistant temporal lobe epilepsy.
Benefits
We found no systematic review or RCTs that examined the effect of amygdalohippocampectomy for people with drug-resistant temporal lobe epilepsy.
Harms
We found no RCTs.
Comment
There is consensus that amygdalohippocampectomy is likely to be beneficial for people with drug-resistant temporal lobe epilepsy.
Substantive changes
No new evidence

Lesionectomy for drug-resistant temporal lobe epilepsy
Summary
We found no clinically important results from RCTs about lesionectomy in people with drug-resistant temporal lobe epilepsy thought to be caused by a known cerebral lesion.
Benefits
We found no systematic review or RCTs on the effects of lesionectomy in people with drug-resistant temporal lobe epilepsy thought to be caused by a known cerebral lesion.
Harms
We found no RCTs.
Comment
We found one systematic review of observational studies (search date 2001, 8 studies, 131 adults with lesions).[91] It found that, at between 1 and 4 years, 63% of the 131 people who had lesionectomy were free of disabling seizures. Surgical removal of tumours and vascular lesions may be indicated to prevent bleeding, herniation, or paralysis.
Substantive changes
No new evidence

Vagus nerve stimulation for drug-resistant temporal lobe epilepsy
Summary
SEIZURE FREQUENCY High-level compared with low-level vagus nerve stimulation: High-level vagus nerve stimulation may be more effective at 12 to 16 weeks at reducing the frequency of seizures in people with medication-resistant partial seizures ( low-quality evidence ). Different stimulation cycles compared with each other: We don't know which stimulation cycle (7 seconds on/18 seconds off [rapid cycle]; 30 seconds on/30 seconds off; 30 seconds on/3 minutes off) is more effective at reducing seizure frequency or at increasing the proportion of 50% responders (low-quality evidence).
Benefits
We found one systematic review (search date 2007, 2 RCTs, 312 adults)[92]and one subsequent RCT.[93]
The two included studies in the systematic review were double-blind, parallel RCTs comparing high-level vagus nerve stimulation (VNS) versus low-level VNS (control) over 12 to 16 weeks in people with medication-resistant partial seizures. All participants were implanted with a stimulator, but the control group received less-frequent and lower-intensity stimulation. In addition, the control group did not receive any electrical current when the device was manually activated by the participant. The review found that high-level VNS significantly reduced seizure frequency compared with low-level VNS (AR for at least 50% reduction in seizure frequency: 39/151 [26%] with high-level VNS v 24/159 [15%] with low-level VNS; RR 1.71, 95% CI 1.08 to 2.70).[92]
The subsequent RCT (61 people) compared three distinctive stimulation cycles; 7 seconds on and 18 seconds off (rapid cycle) versus 30 seconds on and 30 seconds off versus 30 seconds on and 3 minutes off.[93] The RCT found no differences in seizure frequency between different stimulation cycles during the 3-month treatment period (median % reduction in seizures: –25.5 with 7 seconds on/18 seconds off cycle v –27.3 with 30 seconds on/30 seconds off cycle v –29 with 30 seconds on/3 minutes off cycle; significance assessment between groups not reported, reported as not significant). The proportion of 50% responders was similar among the groups (31.6% for 7 seconds on/18 seconds off cycle v 31.7% for 30 seconds on/30 seconds off cycle v 26.1% for 30 seconds on/3 minutes off cycle; significance assessment between groups not reported).
Harms
The review found no significant difference in treatment withdrawal between high-level VNS and low-level VNS (RR 1.08, 95% CI 0.07 to 17.09).[92] One participant in the second RCT included in the review had an infection at the site of device implantation and was not randomised. Hoarseness and dyspnoea were significantly more frequent in the high-level than in the low-level stimulation group, and were therefore thought to be attributable to stimulation (high-level VNS v low-level VNS; hoarseness: RR 2.36, 99% CI 1.62 to 3.45; dyspnoea: RR 2.25, 99% CI 1.09 to 4.67). Hoarseness, cough, and paraesthesia were significantly more frequent in the low-level stimulation group compared with baseline, and were therefore thought to be attributable to device implantation (low-level VNS v baseline; hoarseness: RR 4.37, 99% CI 1.83 to 10.44; cough: RR 2.15, 99% CI 1.29 to 3.61; paraesthesia: RR 7.00, 99% CI 2.27 to 21.63). Hoarseness, cough, dyspnoea, pain, and paraesthesia were more frequent in the high-level stimulation group compared with baseline, and are probably attributable to implantation plus stimulation (high-level VNS v baseline; hoarseness: RR 14.08, 99% CI 5.20 to 38.11; cough: RR 1.81, 99% CI 1.12 to 2.92; dyspnoea: RR 4.39, 99% CI 1.48 to 13.03; pain: RR 2.87, 99% CI 1.46 to 5.64; paraesthesia: RR 12.02, 99% CI 1.85 to 77.94). No haemorrhages were reported in any of the participants.
The subsequent RCT found that cough and voice alteration was more common with rapid-cycle stimulation compared with less-rapid cycles of stimulation (26% with rapid cycles v 5% with 30 seconds on/30 seconds off cycle v 9% for 30 seconds on/3 minutes off cycle; absolute numbers and P value not reported). Three participants withdrew from the study: one could not tolerate rapid-cycle stimulation, one developed infection, and the third was lost to follow-up. [93]
Comment
The two RCTs identified by the review were active control trials, in which the control group had a stimulator implanted, but the stimulation was at a much lower amplitude and much less frequent than in the treatment group.[92] It is possible that investigators and participants were not adequately blinded to the treatment groups, because individuals in the control group may have been able to detect that the stimulations were very infrequent. These VNS studies were also of short duration.
Substantive changes
No new evidence
Articles from Clinical Evidence are provided here courtesy of
BMJ Publishing Group