Surgery for epilepsy

  • Protocol
  • Intervention



This is the protocol for a review and there is no abstract. The objectives are as follows:

The primary objective is to assess the overall outcome of epilepsy surgery.

The secondary objective is to identify the factors that correlate to seizure outcome.


Description of the condition

Epilepsy is defined by the International League Against Epilepsy as "a transient occurrence of signs and/or symptoms due to abnormal excessive or synchronous neuronal activity in the brain" (Fisher 2005). It is a common condition with a prevalence of around 1 in 200 people. Despite optimal pharmacotherapy, about 20% to 30% of individuals do not become seizure-free (Annegers 1979; Collaborative 1992; Cockerell 1995; Kwan 2000). For some of these people, surgery is a therapeutic option.

Description of the intervention

The intervention involves the localisation of the epileptogenic focus and then, if the potential benefit is assessed to outweigh the risk, its surgical resection. The first surgical intervention for focal epilepsy is attributed to Victor Horsley in 1886. Techniques for localising epileptogenic foci initially relied on detailed analysis of seizure semiology and comparison with animal data. Techniques for temporal lobe surgery were developed by Penfield and Jasper later refining these techniques to more limited resections of the antero-mesial temporal lobe. In the second half of the twentieth century, more refined techniques were developed using cortical mapping and imaging, initially with computerised tomography, latterly with magnetic resonance imaging (MRI) and most recently with single photon emission computed tomography (SPECT) and positron emission tomography (PET) (Feindel 2009).

Success of resective epilepsy surgery is estimated to have increased from 43% to 85% during the period 1986 to 1999 (National 1990a; Engel 1993; Engel Jr 2003). Data from multiple sources suggest that 55% to 70% of individuals undergoing temporal resection and 30% to 50% of individuals undergoing extratemporal resection become completely seizure-free. A prospective randomised controlled trial of surgery for temporal lobe epilepsy showed that 58% of individuals randomised to surgery were seizure-free compared to 8% of the medical group (Wiebe 2001).

Surgery is considered a valuable option for medically-intractable epilepsy, even in the absence of proven drug resistance (Engel Jr 1993).

How the intervention might work

The rationale of the intervention is the initial localisation of the epileptogenic focus and then its surgical resection.

Why it is important to do this review

Surgical outcomes may be greatly influenced by the presence of selected prognostic indicators (Tonini 1997; Berg 1998). However, there are still uncertainties about which patients are most likely to achieve good surgical outcomes. Good surgical outcomes appear to be associated with a number of factors (hippocampal sclerosis, anterior temporal localisation of interictal epileptiform activity, absence of preoperative generalised seizures, and absence of seizures in the first post-operative week) (McIntosh 2001). However, the published trial results are frequently confusing and contradictory, thus preventing inferences for clinical practice. This will be the first Cochrane review to look at these factors. It will complement the only systematic review to date which was performed by Tonini et al in 2004, and is therefore now out of date (Tonini 2004). It will inform the surgical selection process and allow the refinement of the risk/benefit analysis for surgical intervention.


The primary objective is to assess the overall outcome of epilepsy surgery.

The secondary objective is to identify the factors that correlate to seizure outcome.


Criteria for considering studies for this review

Types of studies

We will include studies if they satisfy the following criteria:

  1. Study design of randomised controlled trial (RCT), cohort, or case series, either prospective and/or retrospective;

  2. A sample size of at least 30 patients;

  3. A well-defined population (age, sex, seizure type and frequency, duration of epilepsy, etiology, magnetic resonance imaging (MRI) diagnosis, surgical findings);

  4. MRI performed in at least 90% of cases; and

  5. Expected duration of follow-up for at least one year.

We will exclude reports if they are in abstract form, in book chapters, or if they are not sufficiently clear about their methods, they are written in languages other than English, Italian, French, German, or Spanish, or they do not meet all the above inclusion criteria. We will also exclude repeated publications from the same institution (among which only the most recent was retained for review) unless they deal with different prognostic factors.

Types of participants

Children, adolescent and adults considered surgical candidates as having drug-resistant partial seizures and secondarily generalised seizures of temporal or extratemporal origin, i.e. seizures that continue despite treatment with anticonvulsant medication.

Types of interventions

Surgical treatment for drug-resistant partial seizures and secondarily generalised seizures of temporal or extratemporal origin.

Types of outcome measures

Primary outcomes
Seizure outcome

The outcome of seizures after epilepsy surgery is classified according to the Engel's four categories (Engel Jr 1987; see below) or it is reported as such when different definitions are used.

We will consider:

  • good outcome as seizure control or seizure-free status for at least one year or Engel class I;

  • improved outcome as near complete control or moderate improvement or Engel classes II and III;

  • worse outcome as slightly reduced or unchanged or worsened seizure frequency or Engel class IV.

We will also consider, where data for these time-points are available, results at 12 and 24 months. Reporting of this primary outcome is not an eligibility requirement for inclusion in this review. If the seizure outcome is not reported in a study which satisfies all inclusion criteria, where possible we will contact trial authors to enquire about seizure outcome data.

For the purposes of this review, improved and worse outcome will be combined into a single category (poor outcome).

Engel classDescription
Class 1 Free of disabling seizuresCompletely seizure-free since surgery, non-disabling simple partial seizures only since surgery, some disabling seizures after surgery but completely seizure-free for at least two years, and convulsions only when medications are withdrawn.
Class 2 Almost seizure-freeInitially seizure-free but has disabling seizures now, rare disabling seizures since surgery, more than rare seizures after surgery but now rare seizures for at least two years, nocturnal seizures only.
Class 3 Worthwhile improvementWorthwhile seizure reduction or prolonged seizure-free intervals amounting to half the follow-up period, but not less than two years
Class 4 No worthwhile improvementNo significant seizure reduction, no appreciable change, or seizures getting worse

Source: Engel Jr 1987

Search methods for identification of studies

Electronic searches

We intend to search:

  1. The Cochrane Epilepsy Group Specialized Register;

  2. The Cochrane Central Register of Controlled Trials (CENTRAL, The Cochrane Library) using the search strategy in outlined in Appendix 1;

  3. MEDLINE (Ovid) using the search strategy outlined in Appendix 2.


Language restrictions will be imposed (English, Italian, French, German, or Spanish).

Searching other resources

We will also examine the reference lists of included studies for further relevant studies for inclusion within this review.

Data collection and analysis

Selection of studies

Two review authors (SW and RN) will independently assess trials for inclusion. We will resolve any disagreements through mutual discussion and if agreement cannot be made, we will seek the opinion of a third author.

Data extraction and management

We will implement a database for the identification and inclusion of relevant articles, i.e. those which fulfil the inclusion criteria and have complete information about the outcome of epilepsy and prognostic factors.

SW, RN, JN, AS, SG will collect the data using a semi-structured form via a Microsoft Access database (created by SN) for each study.

All the following variables will be considered:

  1. methods of assessment of eligible studies;

  2. demographic and clinical characteristics (number of patients selected for surgery, age (with special attention to patients younger and older than 12 years), sex, disease duration, history of febrile seizures or relevant central nervous system (CNS) disorder);

  3. MRI pre-operative diagnosis (mesial temporal sclerosis, tumours, other CNS abnormalities, normal);

  4. surgical findings (age at surgery, side of resection, surgical procedure (temporal or extratemporal), extent of resection);

  5. histopathological diagnosis (same categories as MRI); duration of follow-up; post-surgery findings (drop-outs, adverse events);

  6. prognostic indicators: different indicators are described as factors affecting the outcome of epilepsy surgery in terms of seizure remission (neuromigration defects, febrile seizures, tumours, vascular disorders, CNS infections, mesial temporal sclerosis, abnormal MRI, electroencephalogram (EEG)/MRI concordance, interictal spikes, intracranial monitoring, extent of resection, post-operative discharges). Mesial temporal sclerosis and tumours require pathological confirmation.

The review is limited to prognostic factors which are clinically relevant and/or reported by at least two studies.

Consensus is also required on each variable reported in the data collection form; any disagreement will lead to a discussion of the issue by the two review authors and any persisting disagreement will be resolved by an independent third author; in selected cases, conflicting data will be resolved by an independent evaluator.

Assessment of risk of bias in included studies

For RCT evidence, we will assess all domains of the current Cochrane 'Risk of bias' tool (Higgins 2011)

For non-randomised evidence, we will employ the Effective Public Health Practice Project (EPHPP) tool, which is appropriate for case series study designs. See Appendix 3.

Risk of bias will be assessed by two review authors (SW and RN) and judgement independently checked by two review authors (SN and JP).

We will incorporate 'Risk of bias' assessments into the analysis if deemed appropriate, using sensitivity analysis, in that a secondary analysis of the data will include only studies rated as low in quality. Both results will be presented in the Results section of the review. Where applicable, we will create 'Summary of findings' tables for outcomes and grade each outcome using the GRADE approach.

Measures of treatment effect

We will measure the outcome of seizures after epilepsy surgery as good, improved and worse outcome (see Types of outcome measures) and then we will analyse these as a dichotomous outcome with improved and worse outcome combined into a single category (poor outcome) summarised with odds ratios (OR) with 95% confidence intervals.

Unit of analysis issues

We do not expect any unit of analysis issues, however in the event of inclusion of studies with unit of analysis issues (e.g. cluster randomised studies, cross over studies, repeated measures studies) we will investigate if the analysis in such studies was conducted appropriately and combine any extracted effect sizes from such studies using the generic inverse variance method of meta-analysis.

Dealing with missing data

We will seek missing statistics from studies through contact with the study authors. We will attempt to clarify the reasons for missing data to determine if missing at random or not and we will analyse the data according to intention-to-treat principles.

Assessment of heterogeneity

We will assess the existence of clinical heterogeneity by examining the differences in study characteristics and patient demographic factors in order to inform decisions regarding the combination of study data. Statistical heterogeneity will be assessed by visually inspecting forest plots and using a Chi2 test for heterogeneity (with a P value of 0.10 for significance) and the I2 statistic as a measure of inconsistency across studies (with an I2 value of 50% to 90% possibly representing substantial heterogeneity and an I2 value greater than 75% representing considerable heterogeneity). If significant statistical heterogeneity considerable heterogeneity according to I2 are found, we will perform meta-analysis with a random-effects model rather than a fixed-effect model, with sensitivity analyses investigating differences in study characteristics.

Assessment of reporting biases

We will investigate outcome reporting bias using the ORBIT classification system (Kirkham 2010). To enable comparison of outcomes of interest, we will request all protocols from study authors.

We will examine publication bias by identifying unpublished data by carrying out a comprehensive search of multiple sources and requesting any unpublished data from authors. We will look for small-study effects to establish the likelihood of publication bias. Funnel plots will be examined in the event an appropriate number of studies are able to be combined. The Cochrane Handbook (Higgins 2011) recommends a minimum of ten studies be combined when examining funnel plots.

Data synthesis

For each prognostic factor of interest individually, we will perform an aggregate data fixed-effect meta-analysis using an inverse variance method to assess the presence or absence of that factor on the outcome of surgery (good or poor outcome), analysed as a dichotomous outcome.

We will also investigate whether the effect of other prognostic factors on any individual prognostic factor has also been adjusted for (for example in multivariate regression models). In this case, we will perform separate meta-analyses of adjusted and unadjusted estimates and compare results.

In considerable statistical heterogeneity is found to be present (Chi2 test for heterogeneity, P value < 0.1 and/or I2 greater than 75%), we will use a random-effects meta-analysis using an inverse variance method and we will perform sensitivity analyses investigating differences in study characteristics.

Subgroup analysis and investigation of heterogeneity

In the case of a substantial amount of heterogeneity being found across prognostic factors, we will perform further analyses such as stratification, subgroup analyses and sensitivity analyses examining the differences in study characteristics (such as outcome definition, study design and quality) and patient demographic factors.

Sensitivity analysis

Good and poor outcomes are measured by varying definitions across studies (i.e. a good outcome can be defined as both seizure-free status for at least one year or Engel class I). The effect of the variation of outcome definition will be examined by sensitivity analysis or subgroup analysis depending on how commonly each definition of outcome is used.

We will also perform sensitivity (or subgroup) analyses to examine the effect of study quality from 'Risk of bias' and quality assessment tools.

Sensitivity analyses may be required if substantial missing data (greater than 10% of data missing) affect studies, for example using a best case/ worse case scenario assuming that all those with missing data (a) had a good outcome or (b) had a bad outcome.


Appendix 1. CENTRAL search strategy

#1           (epilep* NEAR/4 surg*):ti,ab

#2           (extratemporal NEAR/4 surg*):ti,ab

#3           (temporal NEAR/4 surg*):ti,ab

#4           (TLE NEAR/4 surg*):ti,ab

#5           (#1 OR #2 OR #3 OR #4)

#6           (epilep* or seizure* or convuls*):ti,ab

#7           MeSH descriptor Epilepsy explode all trees

#8           MeSH descriptor Seizures explode all trees

#9           (#6 OR #7 OR #8)

#10         MeSH descriptor Neurosurgical Procedures explode all trees

#11         (#9 AND #10)

#12         (#5 OR #11)

Appendix 2. MEDLINE search strategy

This strategy is based on the Cochrane Highly Sensitive Search Strategy for identifying randomized trials (Lefebvre 2011).

1. (epilep$ adj4 surg$).ti,ab.

2. ((extratemporal or temporal or TLE) adj4 surg$).ti,ab.

3. exp Epilepsy/

4. exp Seizures/

5. (epilep$ or seizure$ or convuls$).ti,ab.

6. 3 or 4 or 5

7. exp Neurosurgical Procedures/

8. 6 and 7

9. 1 or 2 or 8

10. (english or french or german or italian or spanish).lg.

11. 9 and 10

12. exp animals/ not

13. 11 not 12

14. randomized controlled

15. controlled clinical

16. randomized.ab.

17. clinical trials as

18. randomly.ab.

19. trial.ti.

20. exp Case-Control Studies/

21. Cohort Studies/

22. (case control*).tw.


24. (observational adj (study or studies)).tw.

25. 14 or 15 or 16 or 17 or 18 or 19 or 20 or 21 or 22 or 23 or 24

26. 13 and 25

Appendix 3. EPHPP Quality Assessment Tool



(Q1) Are the individuals selected to participate in the study likely to be representative of the target population?
1 Very likely 2 Somewhat likely 3 Not likely 4 Can’t tell

(Q2) What percentage of selected individuals agreed to participate?
1 80 - 100% agreement 2 60 – 79% agreement 3 less than 60% agreement 4 Not applicable 5 Can’t tell

See dictionary 1 2 3


Indicate the study design

1 Randomized controlled trial
2 Controlled clinical trial
3 Cohort analytic (two group pre + post)
4 Case-control
5 Cohort (one group pre + post (before and after))
6 Interrupted time series
7 Other specify ____________________________
8 Can’t tell

Was the study described as randomized? If NO, go to Component C.
No Yes
If Yes, was the method of randomization described? (See dictionary)
No Yes
If Yes, was the method appropriate? (See dictionary)
No Yes
See dictionary 1 2 3


(Q1) Were there important differences between groups prior to the intervention?
1 Yes
2 No
3 Can’t tell

The following are examples of confounders:
1 Race
2 Sex
3 Marital status/family
4 Age
5 SES (income or class)
6 Education
7 Health status
8 Pre-intervention score on outcome measure

(Q2) If yes, indicate the percentage of relevant confounders that were controlled (either in the design (e.g. stratification, matching) or analysis)?
1 80 – 100% (most)
2 60 – 79% (some)
3 Less than 60% (few or none)
4 Can’t Tell

See dictionary 1 2 3


(Q1) Was (were) the outcome assessor(s) aware of the intervention or exposure status of participants?
1 Yes
2 No
3 Can’t tell

(Q2) Were the study participants aware of the research question?
1 Yes
2 No
3 Can’t tell

See dictionary 1 2 3


(Q1) Were data collection tools shown to be valid?
1 Yes
2 No
3 Can’t tell

(Q2) Were data collection tools shown to be reliable?
1 Yes
2 No
3 Can’t tell

See dictionary 1 2 3


(Q1) Were withdrawals and drop-outs reported in terms of numbers and/or reasons per group?
1 Yes
2 No
3 Can’t tell
4 Not Applicable (i.e. one time surveys or interviews)

(Q2) Indicate the percentage of participants completing the study. (If the percentage differs by groups, record the lowest).
1 80 -100%
2 60 - 79%
3 less than 60%
4 Can’t tell
5 Not Applicable (i.e. Retrospective case-control) RATE THIS SECTION

See dictionary 1 2 3


(Q1) What percentage of participants received the allocated intervention or exposure of interest?
1 80 -100%
2 60 - 79%
3 less than 60%
4 Can’t tell

(Q2) Was the consistency of the intervention measured?
1 Yes
2 No
3 Can’t tell

(Q3) Is it likely that subjects received an unintended intervention (contamination or co-intervention) that may influence the results?
4 Yes
5 No
6 Can’t tell


(Q1) Indicate the unit of allocation (circle one)

community organization/institution practice/office individual

(Q2) Indicate the unit of analysis (circle one)

community organization/institution practice/office individual

(Q3) Are the statistical methods appropriate for the study design?
1 Yes
2 No
3 Can’t tell

(Q4) Is the analysis performed by intervention allocation status (i.e. intention to treat) rather than the actual intervention received?
1 Yes
2 No
3 Can’t tell

Please transcribe the information from the A-F boxes on pages 1-4 into this section.

1 2 3
1 2 3
1 2 3
1 2 3

1 2 3

1 2 3

1 STRONG (no WEAK ratings)
2 MODERATE (one WEAK rating)
3 WEAK (two or more WEAK ratings)

With both reviewers discussing the ratings:
Is there a discrepancy between the two reviewers with respect to the component (A-F) ratings?
No Yes

If yes, indicate the reason for the discrepancy
1 Oversight
2 Differences in interpretation of criteria
3 Differences in interpretation of study

Final decision of both reviewers (circle one): 1 STRONG

Contributions of authors

SW, RN, SN and JP contributed to the protocol.

SW, RN, JN, AS, SG will review the papers and abstract data.

SN created a Microsoft Database for data extraction and will analyse the data.

JP will perform independent quality and 'Risk of bias' assessment.

Declarations of interest

None known.