Impact of epilepsy surgery on seizure control and quality of life: A 26-year follow-up study


Address correspondence to Hussan S. Mohammed, University of Maryland School of Medicine, 655 W. Baltimore St., Baltimore, MD 21201, U.S.A. E-mail:


Purpose:  The short-term efficacy and safety of epilepsy surgery relative to medical therapy has been established, but it remains underutilized. There is a lack of data regarding the long-term seizure-control rates and quality of life outcomes after epilepsy surgery. This study represents the longest follow-up study to date, with a mean follow-up duration of 26 years.

Methods:  We studied the seizure and health-related quality of life outcomes of patients who underwent epilepsy surgery by Dr. Sidney Goldring from 1967 to 1990. Retrospective clinical chart reviews gathered perioperative data and surveys obtained follow-up data. Seizure outcome was evaluated using the Engel classification system.

Key Findings:  Of 361 patients, 117 (32.4%) completed follow-up interviews. Fifty-six patients (48%) were Engel class I. Mean overall Quality of Life in Epilepsy (QOLIE-31) questionnaire score for the cohort was 68.2 ± 16. Eighty percent of patients reported their overall quality of life now as being better than before surgery. Seizure freedom was associated with better quality of life. We did not observe a statistically significant association between postoperative complications and long-term outcome. Patients who underwent temporal lobe resection achieved better seizure outcomes than those who underwent other types of procedures. Astatic seizures and bilateral surgery were associated with a worse Engel class outcome.

Significance:  Our study demonstrates that the beneficial effects of epilepsy surgery are sustained over decades, and that these beneficial effects are correlated with an improved quality of life. The confirmation of its durability makes us optimistic that the outcomes from modern epilepsy surgery will be even better and that our present enthusiasm for this treatment modality is not misplaced.

More than 50 million people worldwide have epilepsy (World Health Organization, 2001). For patients, the burden is extensive and multifaceted, comprising obvious physiologic dysfunction as well as psychosocial impairments. Patients with epilepsy have significantly poorer health-related quality of life and higher rates of comorbidities compared to the general population. The physical hazards of unpredictable seizures and social stigma associated with seizures contributes to the lower rates of employment and marriage, and lower education levels that exist in this population (Strine et al., 2005; Elliott et al., 2009).

Although the majority of patients with epilepsy respond to medical therapy, greater than 30% are refractory (Sanders, 1993; Brodie & Dichter, 1996). A study by Kwan et al. showed that only 3% of patients became seizure-free when taking more than one drug, confirming the assertion that for patients with correctable structural abnormalities, surgery should be considered as soon as treatment with two first-line medications fails (Engel, 1996; Wieser, 1998; Kwan & Brodie, 2000).

Despite the increase in epilepsy surgeries performed in recent decades, many argue that surgery remains underutilized (Engel, 2001; Wiebe et al., 2001; Choi et al., 2008). The short-term efficacy and safety of epilepsy surgery relative to medication has been established through one randomized, controlled trial (Wiebe et al., 2001) as well as hundreds of published reports. Yet in 2001, it was estimated that of the more than 4 million people worldwide who could benefit from epilepsy surgery, <0.1% of the potential candidates actually receive surgery annually (Sakamoto et al., 2001).

There are scant data available on long-term seizure control and quality of life outcomes after surgery. In addition, several limitations of the existing outcome studies have been described in the literature (Spencer, 1996). These include small patient populations, neglecting to measure nonseizure outcomes, a lack of standardization in measuring outcomes, and short follow-up durations (McIntosh et al., 2001; Wieser et al., 2001; Tellez-Zenteno et al., 2005). In this report, we present a long-term follow-up study of a cohort of patients of Sidney Goldring, a neurosurgeon who helped develop the technologies used in modern epilepsy surgery. We hypothesized that the beneficial effects of epilepsy surgery on seizure burden and quality of life are durable over a patient’s lifetime. To this end, we performed follow-up interviews on a large cohort of patients who underwent epilepsy surgery up to 40 years ago. With a mean follow-up duration of 26 years, this study provides truly unique clinically relevant outcomes data with major implications for the long-term efficacy of epilepsy surgery, even from procedures performed during the nascent stages of the field in the 1960s, 1970s, and 1980s.


General description

We studied the seizure and health-related quality of life outcomes of patients who underwent epilepsy surgery by Dr. Sidney Goldring at Barnes/St. Louis Children’s Hospital from 1967 to 1990. The methodology of this study involved retrospective clinical chart reviews and prospective follow-up interviews.

Standard protocol approvals, registrations, and consents

Approval by the Washington University Human Research Protection Office (HRPO # 09-0393) was obtained prior to the start of this study. Written informed patient consent was not required in this study given its low-risk and retrospective nature. However, verbal informed patient consent was obtained prior to the start of every patient interview.

Study population

The patient files of Dr. Sidney Goldring were reviewed to identify study candidates. We initially included all 361 patients who were operated on by Dr. Goldring from 1967 to 1990 for seizures poorly controlled by medication.

Patients in this study received surgery during a period in which new imaging technologies were emerging, most notably the clinical introduction of the magnetic resonance imaging (MRI) in the 1980s. Therefore, there was significant variability in the types of presurgical evaluation across the patient population.

Data collection

A telephone survey obtained current information regarding each patient’s seizure and health-related quality of life outcomes. Contact information for as many patients as possible was found using patient charts, our electronic records system, and public search websites (i.e.,, If the patient could not be reached via telephone but an address was found, a questionnaire was mailed. Any patient who had died was considered as lost to follow-up because they could not contribute data for the primary outcomes.

Primary outcomes assessment

Seizure outcome was evaluated using the Engel classification system. Engel class I indicates freedom from disabling seizures, Engel class II indicates rare disabling seizures, and Engel classes III and IV indicate worthwhile improvement and no worthwhile improvement, respectively (Engel et al.,1993).

In order to measure health-related quality of life outcomes, we incorporated the Quality of Life in Epilepsy 31 (QOLIE-31) questionnaire in our survey (Cramer et al., 1998). Scores range from 0 to 100, with higher scores reflecting better quality of life (Cramer et al., 1998). No scale score was calculated if the respondent answered <70% of the items on a scale; otherwise the mean of the answered items was used to calculate the scale score. In addition to the QOLIE-31 questions, we asked each patient how they would rate their overall quality of life now compared to before surgery.

Index procedure was defined as the first resective/disconnective surgery done by Dr. Goldring. If no resection/disconnection was performed, then the index procedure indicates surgery done for electrode placement only.

Statistical analysis

Characteristics of patients with and without follow-up data were compared using Wilcoxon’s test (for continuous variables) and chi-square test (for categorical variables). However, when cell sample sizes in the contingency table were small, Fisher’s exact test was used instead of the chi-square test.

To determine if patient and surgical characteristics changed between 1967 and 1988, a tertiary split was performed on the distribution of index surgical procedure dates. This created three categories of surgical eras for comparison, with an equivalent number of patients in each. Characteristics were compared across the three surgical eras with the Kruskal-Wallis test (for age) or chi-square test. When there was a suspected linear pattern in proportions across the surgical eras, the Cochran-Armitage test for trend was also calculated; however, these results are not reported because conclusions were similar to those achieved by the chi-square test.

Mean QOLIE-31 scores were compared across Engel classes by analysis of variance (ANOVA). When the overall ANOVA was significant, Tukey’s test was used to determine which Engel class groups were significantly different. The linear association between QOLIE-31 scores and Engel class was assessed with Spearman correlations; however, these results are not reported as they do not provide information in addition to that provided by the ANOVA. ANOVA (for variables with a normal distribution) or Wilcoxon’s test (for variables with a nonnormal distribution) was used to compare select QOLIE-31 scales (i.e., emotional well-being, cognitive, and overall scale scores) for patients with and without complications, by employment, education level, and marital history. Engel class was compared for these patient subgroups by Wilcoxon’s test. Patient characteristics were compared across Engel class by ANOVA (for continuous variables) or chi-square test (for categorical variables).

Not all patients contributed data for every variable; therefore, sample sizes may vary across analyses. Unless otherwise indicated, data are reported as mean ± standard deviation (SD). The data analysis was generated using SAS software (SAS System for Linux, v.9.1.3; SAS Institute Inc., Cary, NC, U.S.A.).


Study patients

Forty-nine (14%) of the 361 patients who were initially eligible had died. Mean age at death was 45.0 ± 19, with a range between 2.9 and 85.2 years. We were unable to locate 186 patients, and 9 patients declined to participate. One hundred seventeen patients (32%) successfully completed the long-term follow-up questionnaire. Mean duration of follow-up from index procedure was 26.5 ± 5 years (range 21.3–41.9 years).

For the entire cohort (N = 361) at the time of surgery, 158 patients (44%) had daily seizures, 64 (18%) had weekly seizures, 59 (16%) had seizures less than once per week, and 80 (22%) had an unknown seizure frequency. Seventy-five percent (269 of 361) of patients had undergone computed tomography (CT) as part of their index procedure’s preoperative workup. For the 361 patients, the prevalence of other preoperative assessments include: electroencephalography (EEG) (100%), angiography (74%), WADA test (50%), MRI (30%), pneumoencephalography (28%), visual field examination (12%), positron emission tomography (PET) (9%), and neuropsychological testing (2%). The characteristics of patients with and without long-term follow-up were similar in almost all variables of comparison (Table 1).

Table 1.   Descriptive data for the whole cohort and by follow-up status (N = 361)
VariableTotal sample
(N = 361)
By follow-up statusp-Value
No follow-up
(n = 244)
Has follow-up
(n = 117)
  1. Data are mean ± SD (range) or the no. of patients (% of group).

  2. p-Values compare patients without and with follow-up data by Wilcoxon’s test (for age and number of seizure types) or chi-square test (for categorical variables), unless otherwise indicated.

  3. Index procedure was defined as the first resective/disconnective surgery done by Dr. Goldring. If no resection/disconnection was performed, then the index procedure indicates surgery done for electrode placement only.

  4. *p-value by Fisher’s exact test.

  5. a Created from a tertiary split of dates from the entire cohort of 361 patients.

Age of onset (year)9.1 ± 10 (0–60)9.6 ± 11 (0–60)8.2 ± 8 (0–44)0.84
Preadolescent (≤12 years) age of onset (%)271 (76)180 (74)91 (78)0.40
Age at first surgery (year)21.0 ± 12 (0.1–64)21.6 ± 12 (0.1–64)19.9 ± 11 (0.2–51)0.25
Female gender (%)138 (38)93 (38)45 (38)0.95
History (%)    
 Structural33 (9)21 (9)12 (10)0.91
 Febrile seizure37 (10)24 (10)13 (11)
 Unknown cause218 (60)149 (61)69 (59)
 Infection25 (7)16 (7)9 (8)
 Trauma35 (10)26 (11)9 (8)
 Other13 (4)8 (3)5 (4)
Seizure type (%)    
 Complex partial261 (72)168 (69)93 (79)0.03
 Generalized178 (49)123 (50)55 (47)0.55
 Focal motor99 (27)71 (29)28 (24)0.30
 Absence48 (13)32 (13)16 (14)0.88
 Astatic26 (7)20 (8)6 (5)0.29
Total no. of seizure types (%)    
 1148 (41)97 (40)51 (44)0.70
 2177 (49)124 (51)53 (45)
 334 (9)23 (9)11 (9)
 42 (1)02 (2)
Surgery types (includes index and nonindex procedures) (%)    
 Electrode placement284 (79)182 (75)102 (87)0.006
 Temporal resections177 (49)116 (48)61 (52)0.41
 Extratemporal cortical resection101 (28)77 (32)24 (21)0.03
 Hemispherectomy15 (4)7 (3)8 (7)0.09*
 Callosotomy15 (4)9 (4)6 (5)0.58*
 Other10 (3)7 (3)3 (3)1.0*
 Electrode placement only78 (22)55 (23)23 (20)0.53
Surgical data (%)    
 Side of major surgery    
  Bilateral74 (21)47 (19)27 (23)0.57
  Left126 (35)89 (37)37 (32)
  Right160 (44)107 (44)53 (45)
  Malformation of cortical development (MCD)26 (7)20 (8)6 (5)0.02
  Mesial temporal sclerosis (MTS)44 (12)21 (9)23 (20)
  Tumor65 (18)48 (20)17 (15)
  Other226 (63)155 (64)71 (61)
Complications (includes index and nonindex procedures) (%)    
 Surgical39 (11)30 (12)9 (8)0.18
 Expected neurologic deficits85 (24)59 (25)26 (22)0.62
 Unexpected neurologic deficits53 (15)32 (13)21 (18)0.25
Surgical era (%)    
 Era of index surgerya    
  1/30/1967–2/18/1980120 (33)93 (38)27 (23)<0.001
  2/19/1980–2/5/1985121 (33)85 (35)36 (31)
  2/6/1985–8/9/1988120 (33)66 (27)54 (46)

Not surprisingly, more patients whose index procedure occurred in the early surgical eras were lost to follow-up. There was a marked decrease in the number of patients with daily seizures preoperatively over time (p = 0.002). Table 2 demonstrates that surgical procedures became easier to categorize during the later surgical eras, as fewer mixed/combination types of surgeries were performed. As expected, there was a significant decline over time in the occurrence of surgical complications (p < 0.001) and mortality during long-term follow-up (p < 0.001) after epilepsy surgery.

Table 2.   Descriptive data for the whole cohort by surgical era (N = 361)
VariableEra of index surgeryp-Value
(n = 120)
(n = 121)
(n = 120)
  1. Data are mean ± SD or the no. of patients (% of group).

  2. p-values compare patients across surgical era by Kruskal–Wallis test (for age) or chi-square test (for categorical variables), unless otherwise indicated.

  3. *p-value by Fisher’s exact test.

Age of onset (year)9.7 ± 119.5 ± 108.1 ± 90.36
Age at index surgery (year)21.6 ± 1221.8 ± 1019.8 ± 120.14
History (%)    
 Structural11 (9)13 (11)9 (8)0.03
 Febrile seizure6 (5)11 (9)20 (17)
 Unknown cause72 (60)72 (60)74 (62)
 Infection9 (8)6 (5)10 (8)
 Trauma17 (14)12 (10)6 (5)
 Other5 (4)7 (6)1 (1)
Seizure frequency (%)    
 Daily64 (53)59 (49)35 (29)0.002
 Weekly14 (12)17 (14)33 (28)
 Less than once per week16 (13)19 (16)24 (20)
 Unknown26 (22)26 (21)28 (23)
Surgery types (includes index and nonindex procedures) (%)    
 Electrode placement103 (86)85 (70)96 (80)0.01
 Temporal resections57 (48)58 (48)62 (52)0.78
 Extratemporal cortical resection35 (29)38 (31)28 (23)0.35
 Hemispherectomy9 (8)2 (2)4 (3)0.07*
 Callosotomy7 (6)6 (5)2 (2)0.22*
 Other8 (7)2 (2)00.004*
 Electrode placement only27 (22)21 (17)30 (25)0.34
Complications (includes index and nonindex procedures) (%)    
 Surgical25 (21)5 (4)9 (8)<0.001
 Expected neurological deficits32 (27)32 (27)21 (18)0.15
 Unexpected neurological deficits22 (19)18 (15)13 (11)0.25
Death (excludes unknowns) (%)25/55 (45)16/41 (29)8/65 (12)<0.001

Long-term follow-up and quality of life

Of the 117 patients who completed follow-up interviews, 56 (48%) were Engel Class I (Table 3). One patient did not provide enough information to allow Engel classification.

Table 3.   Seizure-related outcome data for the patients completing a long-term follow-up questionnaire (n = 117)
VariableMean ± SD or no. of patients (%)
  1. Data are no. of patients (% of group), unless otherwise indicated.

Hospitalizations for seizures after surgery 
 None55 (49)
 One21 (19)
 Multiple36 (32)
Currently taking AEDs87 (76)
Auras40 (40)
Aura frequency 
 Every seizure (E)13 (35)
 Less than every seizure (L)18 (49)
 More than every seizure (M)6 (16)
Total no. of epilepsy surgeriesMedian = 1
Range =  1–13
>1 total no. of epilepsy surgeries43 (38)
Patients receiving VNS treatment at some point after surgery21 (19)
Engel class 
 I56 (48)
  IA23 (20)
  IB6 (5)
  IC23 (20)
  ID4 (3)
 II11 (9)
 III26 (22)
 IV23 (20)

Fifty-two (45%) of 116 patients were employed at the time of follow-up, and 100 (86%) reported previous employment, which included sheltered workshops. Fourteen (12%) of 114 patients never graduated high school, whereas 57 patients (50%) went on to reach a postsecondary level. Fifty-two (45%) of 115 patients never married. Fifty-one (44%) of 115 patients had children.

The mean overall QOLIE-31 score for the entire group was 68.2 ± 16. When patients were asked to compare postoperative to preoperative quality of life, 72% (79 of 110) reported it as much better, 8% (9 of 110) as a little better, 9% (10 of 110) as about the same, and 11% (12 of 110) as worse. Better Engel class was associated with better QOL scores (Fig. 1).

Figure 1.

Association between seizure and quality of life outcomes. (A) Bar graph comparing the percentages of patients in each Engel class when answering the question, “Compared to before your epilepsy surgery, how would you rate your overall quality of life now?” Answers were scored on a 0–3 scale where 0 = worse, 1 = about the same, 2 = a little better, and 3 = much better. (B) Bar graph showing the mean QOLIE-31 overall scores of each Engel class. Scores are out of a 0–100 scale, with higher scores reflecting better quality of life. Engel subclass IA patients are shown separately from the other Engel class I patients. Engel class IA patients had a mean score of 78.0 ± 14; Engel class IB, IC, and ID scored 73.7 ± 14; Engel class II scored 71.8 ± 15; Engel class III scored 61.3 ± 14; and Engel class IV patients scored 54.2 ± 16. (C) Bar graph that depicts the association of demographic factors with seizure outcome. Patients who reported an employment history, a postsecondary education level, marital history, or children were more likely to be in a better Engel class. Of the 100 patients who reported previous or current employment, 62 (62%) were in Engel class I or II, whereas 5 (31%) of the 16 patients who have never been employed were in Engel class I or II (p = 0.02). Thirty-eight (67%) of the 57 patients who completed some postsecondary education were in Engel class I or II, 23 (53%) of the 43 patients who graduated high school were in Engel class I or II, and only 4 (29%) of the 14 patients who did not graduate high school were in Engel class I or II (p = 0.03). Forty-one (65%) of the 63 patients who were married, divorced, or widowed were in Engel class I or II, whereas 25 (48%) of the 52 patients who never married were in Engel class I or II (p = 0.07). Thirty-five (70%) of the 50 patients with children were in Engel class I or II, whereas 31 (48%) of the 65 patients without children were in Engel class I or II (p = 0.02).


Postoperative complications were assigned to three different categories: surgical complications (i.e., infection, hematoma, etc.), expected neurologic deficits (i.e., contralateral superior quadrant hemianopsia after temporal lobectomy), and unexpected neurologic deficits (i.e., cranial nerve paresis after temporal lobectomy). The distinction between expected and unexpected neurologic deficit was made after careful review of each patient’s operative report and discharge summary. Of the 117 patients, 9 (8%) suffered surgical complications, 26 (22%) experienced expected neurologic deficits, and 21 (18%) acquired unexpected neurologic deficits. We did not observe a statistically significant association between postoperative complications and long-term seizure or quality of life outcomes.

Prognostic factors of seizure outcome

Table 4 compares clinical characteristics of patients for each Engel class. Astatic seizures or “drop attacks” were associated with a worse Engel class (p = 0.05). Of interest, undergoing preoperative MRI was not shown to be a statistically significant predictor of Engel class. We found several correlations between surgical procedure and long-term seizure outcome. Patients who had temporal lobe resections tended to do considerably better (p = 0.03), whereas patients who had electrode placement without resection or disconnection had worse outcomes (p = 0.02). Those who had bilateral resections were more likely to have poor seizure outcome (p = 0.02) and patients with malformation of cortical development (MCDs), mesial temporal sclerosis (MTS), or tumors, tended to have a better long-term seizure outcome. Forty-eight percent (32 of 67) of patients with Engel class I or II had MCDs, MTS or a tumor compared to 27% (13 of 49) of patients with Engel class III or IV (p = 0.02 by chi-square test).

Table 4.   Association of patient characteristics with Engel class
CharacteristicEngel classp-value*
I (n = 56)II (n = 11)III (n = 26)IV (n = 23)
  1. Unless otherwise noted, data are mean ± SD or the no. of patients (% of Engel class).

  2. *p-value compared the characteristic across Engel class by ANOVA (for continuous variables) or chi-square test (for categorical variables).

  3. ‡p-value by Fisher’s exact test.

  4. a Data rank-transformed prior to analysis.

  5. b Categories formed using tertiary spilt of the dates from the entire cohort of 361 patients.

Age at first surgery (year)20.8 ± 1221.8 ± 917.8 ± 1219.8 ± 90.66
Time from index procedure to follow-up (year)26.9 ± 526.7 ± 626.0 ± 425.9 ± 40.91a
Female gender (%)20 (36)5 (45)10 (38)10 (43)0.89
History, unknown cause (%)29 (52)8 (73)16 (62)15 (65)0.48
Seizure types (%)     
 Complex partial43 (77)8 (73)19 (73)22 (96)0.13‡
 Generalized25 (45)5 (45)13 (50)11 (48)0.97
 Focal motor15 (27)3 (27)6 (23)4 (17)0.83
 Absence6 (11)1 (9)3 (12)6 (26)0.34‡
 Astatic1 (2)04 (15)1 (4)0.05‡
 Total no. of seizure typesMedian = 2Median = 2Median = 2Median = 20.51a
Pre-op workup prior to index procedure (%)     
 MRI21 (38)6 (55)10 (38)10 (43)0.74
Surgery types (%)     
 Electrode placement47 (84)10 (91)24 (92)21 (91)0.78‡
 Temporal resections34 (61)8 (73)11 (42)7 (30)0.03
 Extratemporal cortical resection11 (20)2 (18)8 (31)3 (13)0.50‡
 Hemispherectomy5 (9)02 (8)1 (4)0.90‡
 Callosotomy1 (2)03 (12)2 (9)0.18‡
 Electrode placement ONLY8 (14)2 (18)3 (12)10 (43)0.02‡
 Resection + electrode34 (61)8 (73)16 (62)8 (35)0.10
 More than one surgery type39 (70)8 (73)21 (81)11 (48)0.09
Surgical data (%)     
 Bilateral major surgery9 (16)2 (18)5 (19)11 (48)0.02
  MCD, MTS, or tumor26 (46)6 (55)7 (27)6 (26)0.13
  Other30 (54)5 (45)19 (73)17 (74)
Era of index surgeryb (%)     
 1/30/1967–2/18/198014 (25)4 (36)5 (19)4 (17)0.65
 2/19/1980–2/5/198516 (29)2 (18)11 (42)6 (26)
 2/6/1985–8/9/198826 (46)5 (45)10 (38)13 (57)


Historical context of epilepsy surgery at Washington University

Dr. Sidney Goldring (1923–2004), former chairman of neurosurgery at Washington University in 1974–1989, was at the forefront of epilepsy surgery advancement. His most significant contribution was the development of techniques that allowed all surgical manipulations to be performed under general anesthesia, making it possible to operate on adults and children who would not have tolerated a craniotomy under local anesthesia. Prior to that time, epilepsy surgery for intractable seizures had been performed with local anesthesia, permitting electrical stimulation to map the neocortex controlling movement, sensation, and language. Goldring’s pioneering use of indwelling epidural electrode arrays for extraoperative electrocorticography to localize spontaneously occurring seizures and regions of eloquent cortex made this transition from local to general anesthesia possible (Weller, 1976; Coxe, 1990). Between 1967 and 1990, a novel split-screen video-monitoring system developed in Goldring’s lab was used in every patient undergoing preoperative evaluation and during the surgical procedures. Goldring’s surgical method became widely used in epilepsy surgery and in research in neuroprosthetics (Grubb, 2005).

Strengths and limitations of the study

Limitations of this study include its retrospective design, incomplete follow-up, follow-up data acquired via surveys without clinical reevaluation, and the large number of statistical tests performed. Multiple statistical analyses were performed, which increases the likelihood that any one of these tests is significant by chance alone. Caution must be used in evaluating the significance of any individual test. More confidence can be had in patterns of similar results. Nevertheless, this study represents the most sensible way to ascertain the very long-term follow-up of a large series of epilepsy surgery patients. Although our follow-up rate of 32% appears low, it still provides meaningful insights into the effects of epilepsy surgery on seizure control and quality of life.

As an epilepsy surgery outcome study, this study is unprecedented in terms of its duration and length of clinical follow-up (average of 26 years). Most published epilepsy surgery outcomes studies have reported prospectively and/or retrospectively, on the order of 5 years. Schmidt & Stavern (2009) performed a meta-analysis of epilepsy surgery outcome studies and found 20, of which 13 had some patients with follow-up of over 4 years. A meta-analysis by Tellez-Zenteno et al. (2005) found four studies with follow-up of over 10 years (Guldvog et al., 1991a,b; Cohen-Gadol et al., 2004; Elsharkawy et al., 2008; Tanriverdi et al., 2008; Elsharkawy et al., 2009). A well-designed, prospective, multicenter study by Spencer et al. (2005) defined long-term follow-up as >2 years. Recently, Dunlea et al. (2010) published the outcomes of epilepsy surgery performed in Ireland between 1975 and 2005. Although this study did have 199 patients, the mean duration of follow-up was 7 years with a range of 1–24 years. A recent study of pediatric epilepsy surgery outcomes at UCLA covered a 22-year time period (1986–2008). Although there were 571 patients in the study, the longest time point to follow-up was 5 years (Hemb et al., 2010).

One of the values of a longer term follow-up study is its ability to assess enduring impact on quality of life. Epilepsy is a common disease with an estimated prevalence of 40–200 per 100,000 and 0.5–1% incidence (Sanders & Shorvon, 1996). The burden of disease is also quite high. Individuals with epilepsy in developed countries have an up to threefold increase in mortality compared to the general population (Forsgren et al., 2005). A study by Sillanpää & Shinnar (2010) showed that a cohort of patients with childhood-onset epilepsy had a mortality rate three times as high as the expected age- and sex-adjusted mortality in the general population. In our study population, the mean age of death was 45. Furthermore, our study again reinforces that the quality of life with epilepsy is low. Of importance, we demonstrate that a lower Engel score correlates with a higher QOLIE-31 scale, extending results of shorter duration studies that improvement in seizure control is associated with a better quality of life.

Reflections on changes in medical technologies and state of epilepsy management over time

In this study, the underlying etiology of the epilepsy predicted surgical outcome. Tumor, mesial temporal sclerosis (MTS), and malformations of cortical development were more likely to have a lower Engel class score. A recent meta-analysis by Tonini et al. (2004) similarly found an association between low Engel class and MTS, tumor, and abnormal MRI independently. Modern high-resolution MRI can now detect MTS and subtle cortical dysplasias; these findings may have been undetectable by MRI during Dr. Goldring’s time. Thus our study did not demonstrate a correlation between receiving a MRI and seizure outcome.

Dr. Goldring’s records show a more uniform surgical approach over time, likely reflecting the maturation and standardization of epilepsy surgery during his career. In addition, there were increased applications of imaging techniques and improvements in surgical complication and mortality rates over time, likely reflecting advances in technology and surgical technique.

Patients presenting for epilepsy surgery between 1967 and 1980 had a much higher frequency of daily seizures than those between 1985 and 1988. This raises the question of whether the use of newer antiepileptic medications (largely carbamazepine and valproate) had improved the quality of medical management. An alternative possibility is that epilepsy surgery was progressively considered for less severely affected patients, especially as better imaging became available. Other factors could have also contributed to this trend.

We also found that fewer hemispherectomies were performed over time, likely reflecting that better imaging made more targeted resections feasible. Our suspicion is that this is a trend that is reversing. Our center’s experience is that functional hemispherotomy can be a highly effective surgery, with 78% achieving seizure freedom (Limbrick et al., 2009). Moreover, recent papers have demonstrated that if there is a congenital or early acquired lesion, hemispherotomy can be effective even in the setting of generalized EEG findings or mild contralateral MRI abnormalities (Wyllie et al., 2007; Loddenkemper et al., 2009; Hallbook et al., 2010).


Although limited by study design, this study demonstrates that the beneficial effects of epilepsy surgery are sustained over long periods of time. This is the longest follow-up study of epilepsy surgery to date. We found an enduring improvement in quality of life with improvement in seizure control. Moreover, the excellent short-term outcomes that have been reported in more modern series are likely to be lifelong.


This project was funded by the Epilepsy Foundation Health Sciences Student Fellowship (HM).


Hussan Mohammed received the Health Sciences Student Fellowship grant from the Epilepsy Foundation for this study. He reports no other disclosures. Dr. Kaufman and Rebecca Munro report no disclosures. Dr. Limbrick received research support from EKR Therapeutics, the NIH/NINDS, the National Institute of Child Health & Human Development/NICHD, the Children’s Surgical Sciences Institute at St. Louis Children’s Hospital, and the Park/Reeves Center for Syringomyelia Research. Karen Steger-May is a statistical reviewer for the Journal of Orthopedic Trauma. She is funded by NIH grants # R01 AR051026, UL1 RR024992, R34 10517082, NIH/NHLBI grant # P50 HL084922, and NIH/NIDDK grant # P30 DK56341. She also receives institutional support for her role as a consultant. Dr. Grubb Jr. serves on the advisory board for the Journal of Neurosurgery. He is funded by NIH/NINDS grants # 5 UOI NS42167-05 and 5 P50NS05597703. Dr. Rothman received a travel stipend from UCB Pharma to present research at an AES meeting in 2009, received patent #6,978,183 in 2005 for “system and method for cooling the cortex to treat neocortical seizures,” and shares a pending patent to Dr. Smyth, “depth cooling implant system” (#US12/164857), and a second pending patent to Drs. Miller, Smyth, Rothman, and D’Ambrisio, “Methods and devices for brain cooling for treatment and prevention of acquired epilepsy” (#US2010/0312318A1). He also received research support from CURE and UCB Pharma. Dr. Weisenberg is funded by NIH/NINDS grant # U01 NS 05398. Dr. Smyth served on the scientific advisory board for Novartis and receives research support from the CURE/DOD Special Grant Program in the prevention of epilepsy after traumatic brain injury. We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.