J Neurosurg 110:319326, 2009

Prognostic factors in the persistence of posttraumatic epilepsy after penetrating head injuries sustained in war
Clinical article
Behzad eftekhar, M.d., M.P.h.,1 MohaMMad ali Sahraian, M.d., 3 BanafSheh nouraliShahi, M.d.,1 ali khaji, M.d.,1 zahra VahaBi, M.d.,1 MohaMMad GhodSi, M.d., M.P.h.,1 haSSan araGhizadeh, M.d., 2 MohaMMad reza SorouSh, M.d., 2 SiMa karBalaei eSMaeili, M.d., 2 and Mehdi MaSouMi, M.d. 2
1 2

Sina Trauma Research Center, and 3Department of Neurology, Sina Hospital, Tehran University; and Medical and Engineering Research Center, Janbazan Foundation, Tehran, Iran

Object. The goal of this paper was to investigate the long-term outcome and the possible prognostic factors that might have influenced the persistence of posttraumatic epilepsy after penetrating head injuries sustained during the IraqIran war (19801988). Methods. In this retrospective study, the authors evaluated 189 patients who sustained penetrating head injury and suffered posttraumatic epilepsy during the IraqIran war (mean 18.6 4.7 years after injury). The probabilities of persistent seizures (seizure occurrence in the past 2 years) in different periods after injury were estimated using the Kaplan-Meier method. The possible prognostic factors (patients and injury characteristics, clinical findings, and seizure characteristics) were studied using log-rank and Cox regression analysis. Results. The probability of persistent seizures was 86.4% after 16 years and 74.7% after 21 years. In patients with < 3 pieces of shrapnel or no sphincter disturbances during seizure attacks, the probability of being seizure free after these 16 and 21 years was significantly higher. Conclusions. Early seizures, prophylactic antiepileptics drugs, and surgical intervention did not significantly affect long-term outcome in regard to persistence of seizures. (DOI: 10.3171/2008.4.17519)

key WordS penetratingwarheadinjury posttraumaticepilepsy prognosis

most significant risk factor for PTE is the severity of the injury; the risk is highest ( 50%) after penetrating head injuries.12 Although the long-term course of the majority of idiopathic epilepsy disorders is not clear, it is interesting that epilepsy due to military brain injury tends to become less frequent or to disappear after 2030 years.5,11 In this study, based on the IraqIran war (19801988) experience, we investigated the aforementioned concept and the possible prognostic factors that might have influenced the persistence of the PTE after penetrating head injuries sustained during war.
he

We used a comprehensive list of war veterans with PTE. These patients were reviewed in 1994 by a neurology team in the Medical and Engineering Research CenAbbreviations used in this paper: EEG = electroencephalography; PTE = posttraumatic epilepsy.

Methods

ter, Janbazan (Veterans) Foundation. Also included were a small group of civilians who sustained battle-related head injuries The records of the patients including the history, clinical examination, and paramedical studies such as CT scanning and EEG reports (where applicable) were reviewed. Those patients who had penetrating head injuries were selected and interviewed via telephone. On some occasions the family members were interviewed as well. Of the patients who consented to a telephone interview, those with a first-degree family history of epilepsy, a history of epilepsy prior to injury, or a history of being exposed to potential confounding factors (such as chemical warfare agents or brain tumors) were excluded from the study. Information on age at the time and at the time of injury; sex; educational level; occupation; cause of injury; entry and exit points; present locations of any shrapnel or bullets; number of shrapnel pieces (based on preoperative imaging findings); history of brain surgery, brain infection, and unconsciousness after trauma; accompanying injuries, sensory and motor disturbances; degree of disability; date of last seizure; and characteristics of the
319

J. Neurosurg. / Vol 110 / February, 2009

B. Eftekhar et al.
seizures (such as onset, seizure type, frequency, severity, aura, seizure-related injuries, trigger factors, and antiepilepsy medication) were extracted from the previous records (1994 study) and interviews. In addition, those with decreased frequency of seizures or those who became seizure free in the past 2 years or more were identified. We defined persistence of seizures as having any seizure in the past 2 years.12 This amounts to Engel Class I subgroup for outcome of epilepsy.8
Statistical Analysis
Table 1: General characteristics of the 189 patients suffering PTe Characteristic
male sex mean age (yrs) at interview at time of injury cause shrapnel only bullet only both unknown no. of in-driven shrapnel fragments entrance side/region rt lt bilat frontal parietal temporal occipital passed through midline patients w/ bullet/shrapnel remnant (mainly temporal) craniotomy for removal of bullet/shrapnel cranioplasty intracranial infection infection following cranioplasty period of unconsciousness (mean 31.4 44 days) accompanying injury (missile fragment injury to the limbs) seizure timing 1 wk (early epilepsy) <1 yr >1 yr (mean 1.3 up to 12 yrs) seizure type (%) generalized secondary generalized partial generalized seizure component sphincter disturbances during seizure attacks triggering factors for seizures (emotional stress, physical/mental exhaustion, loud noises, certain foods, proximity to TV screens, weather) antiepileptic medication none regular intake irregular intake 1 drug 2 drugs 3 drugs decrease in seizure frequency seizure free in past 2 yrs

No. of Patients (%)*

184 (97.4) 40.39 5.78 19.66 153/189 (81) 29/189 (15.3) 6/189 (3.2) 1/189 (0.5) 2.87 2.64 (91 cases) 88/189 (46.6) 82/189 (43.4) 3/189 (1.6) 64/189 (33.9) 25/189 (13.2) 67/189 (35.4) 33/189 (17.5) 30/189 (15.9) 106/189 (56) 155/188 (82.4) 95/184 (51.6) 46/179 (25.7) 8/171 (4.7) 185/188 (98.4) 110/189 (58.2)

A seizure event was defined as the last seizure in those who were seizure free for > 2 years during the follow-up period. Persistent seizure probabilities were estimated and graphically represented as time-to-event curves by using the Kaplan-Meier method. To consider possible changes through the time, we first compared the categorical variables in patients with and without persistent seizures by using the chi-square test. The Student t-test was used to compare continuous variables between groups. In the next step we performed log-rank analysis of the factors that might affect the persistence of seizures. Then we used Cox regression analysis to determine whether our findings were statistically significant during the period of study and after adjustment for other confounding variables. There were 5 patients in whom the date of last seizure was not known. We did not include them in the Cox regression analysis. Four of them were seizure free for > 2 years. The results were considered statistically significant at a probability level < 0.05. The data were analyzed using Intercooled STATA for Windows, Version 6 (STATA Corp.).

A total of 189 patients with PTE after penetrating injuries sustained during the IraqIran war met our inclusion criteria. There were 5 women and 184 men who ranged in age from 20 to 63 years. Most of the patients were veterans (~93%), but we also included a group of civilians injured as a result of military action. Most of the patients (114 [60.3%]) had some degree of disability that made it impossible to work. Seven patients included in the original list died during the follow-up period. Table 1 shows the general characteristics of the study population. Of those who did not have partial seizures, 62.6% had sphincter disturbances. Table 2 shows the comparison between the patients who had persistent seizures and those who did not. The percentages of early seizures and sphincter disturbances were statistically different between the 2 groups. Table 3 shows the results of log-rank analysis of the factors that might affect the persistence of seizures. Only presence of sphincter disturbances and number of shrapnel pieces (< 3 vs 3) at the initial injury (regardless of whether they were still embedded) were found to be statistically significant. Table 4 shows the frequency of seizures as recorded in the initial study. Comparing patients with and without persistent seizures, there was no statistically significant difference (p = 0.235). The rate of persistent seizures was
320

Results

18/174 (10.3) 130/175 (74.3) 45/175 (25.7) 75.9 20.3 3.7 96.3 113/186 (60.1) 95/189 (50.3)

26/189 (14) 141/186 (75.8) 19/186 (10.2) 82/159 (51.6) 60/159 (37.7) 17/159 (10.7) 147/185 (79.5) 47/189 (24.9)

* Data on some patients are missing. Values are number of patients unless

otherwise noted. Where appropriate, mean values are presented standard deviations.

J. Neurosurg. / Vol 110 / February, 2009

Prognostic factors in the persistence of posttraumatic epilepsy

Table 2: Comparison between patients who had persistent seizures and those who did not Characteristic no. of patients mean age at injury (yrs) mean age at present (yrs) patients w/ presence of bullet/shrapnel mean no. of fragments shrapnel (only) bullet (only) cranioplasty craniotomy for removal of the bullet/shrapnel coma duration of coma (days) history of infection history of infection after cranioplasty anticonvulsant prophylaxis irregular antiepileptic drug use taking >1 antiepileptic drug early seizure mean onset latency postinjury (yrs) generalized seizure partial seizure secondary generalized focal motor sign before attack sphincter disturbance during seizure headache during seizure permanent speech disturbance speech disturbance after seizure motor neurological deficit (present) Seizure Free <2 Yrs 142 19.4 6.3 40.3 5.9 83/142 3.2 2.9 (66 cases) 117/142 20/142 72/138 115/141 139/141 34.3 4.4 (118 cases) 38/136 6/129 46/99 15/124 60/63 17/132 1.23 1.84 (97 cases) 106/140 4/140 30/140 23 96/142 79/142 28/142 100/142 42/131 Seizure Free >2 Yrs 47 20.2 4.9 40.7 5.2 23/47 2.0 1.6 (25 cases) 36/47 9/47 23/46 38/47 46/47 22.6 4.5 (39 cases) 8/43 2/42 19/36 4/43 17/19 1/42 1.55 2.63 (33 cases) 36/47 3/47 8/47 3/47 17/46 29/46 9/46 30/46 14/44 p Value 0.480 0.715 0.255 0.076 0.38 0.404 0.798 0.914 0.737 0.15 0.222 0.976 0.516 0.665 0.869 0.052 0.445 0.903 0.271 0.516 0.090 0.000 0.377 0.982 0.507 0.976

lower in patients whose seizure frequency was < 12 per year than in those with greater seizure frequency , and the difference was marginally significant (p = 0.055). Using log-rank analysis, this factor did not retain its statistical significance (p = 0.112). Overall, 142 patients had persistent seizures during the follow-up period. Persistent seizure probability was 74.7% at the end of follow-up according to Kaplan-Meier analysis (Fig. 1). We used the Cox proportional hazards model to assess factors significantly influencing the persistence of PTE. The patients were observed for an average of 18.7 years (range 126 years). Table 5 shows the results of analysis. Only the presence of sphincter disturbances and 3 pieces of shrapnel had statistically significant hazard ratios for persistence of seizures. They maintained their significance when adjusted for other variables. Table 6 shows persistent seizure probability of patients after 16 and 21 years after injury. The persistent seizure probability of all patients was 86.4% after 16 years and 74.7% after 21 years. In those who had < 3 pieces of shrapnel at the initial injury the probability decreased to 80.7% after 16 years and to 61.6% after 21 years. For
J. Neurosurg. / Vol 110 / February, 2009

those who did not have sphincter problems during the seizures, the probabilities were 75.7% after 16 years and 57.4% after 21 years. Figures 13 show the probabilities diagrammatically.

The persistent seizure probability of the entire cohort of patients was 86.4% after 16 years and 74.7% after 21 years. In those with < 3 pieces of shrapnel or those who did not have sphincter disturbances the probability of being seizure free after these periods were statistically higher. Authors of previous studies have commented on the persistence of seizures after a certain period of time (for example, 4, 8, or 10 years). Attempts to determine the predictive value of a particular finding by correlating it with the persistence of seizures may yield different results, depending on the interval between the injury and the seizure assessment. To consider possible changes over time, we used logrank and Cox regression analyses to determine whether our findings were statistically significant during the pe321

Discussion

B. Eftekhar et al.
Table 3: log-rank analysis of prognostic factors for persistent seizure Variable early seizure sex (male/female) entry point (rt, lt, frontal, parietal, temporal, occipital)* exit point (rt, lt, frontal, parietal, temporal, occipital)* present location of the fragments (rt, lt, frontal, parietal, temporal, occipital)* craniotomy cranioplasty infection coma seizure frequency generalized seizure generalized seizure component secondary generalized focal seizure focal seizure component traversing the midline or the ventricles shrapnel only bullet only presence of shrapnel or bullets prophylactic antiepileptics 1 antiepileptic medication vs more irregular drug usage speech problem before/after seizures motor deficit before/after seizures presence of triggering factors for seizures sphincter disturbances visual/auditory hallucination before seizure shrapnel pieces <3 vs 3 mentioned in a single cell. p Value 0.08 0.98 >0.05 >0.05 >0.05 0.61 0.93 0.10 0.65 0.38 0.85 0.36 0.12 0.06 0.41 0.35 0.70 0.48 0.17 0.56 0.82 0.69 0.64 0.92 0.06 0.0000 >0.05 0.01 Table 4: Comparison of frequency of seizures between patients who had persistent seizures and those who did not Seizure Frequency* 1/day 1/wk but <1/day 1/mo but <1/wk 1/6 mos but <1/mo total Seizure Free <2 Yrs 10 42 38 32 122 Seizure Free >2 Yrs 3 16 18 5 42

Total 13 58 56 37 164

* As reported by the patients on initial examination. The differences were not statistically different (p = 0.235, chi-square test).

not agree to be interviewed did not have a major problem with epilepsy, so our selection may have been biased toward those with more serious problems. Although we cannot rule it out completely, because these patients have a close and almost lifetime relationship with the Janbazan Foundation, this should be minimal. We did not review the CT scans themselves. Our study is based on the CT reports, which seem to be reasonably reliable; however, this makes our information limited to the reports, and we were not able to comment on factors such as brain volume loss as a prognostic factor. Similarly, not all the patients underwent EEG, so we cannot comment on the EEG findings as prognostic factors.
Rate of Persistent Seizures After Penetrating CombatRelated Head Injuries

* Each location was tested separately. To save space the results are As reported by the patients and categorized in Table 4. Includes generalized and secondary generalized seizures. Includes partial and secondary generalized seizures.

riod of study and after adjustment for other confounding variables. We think that the differences between our study and the previous studies may partially be explained based on the differences in the time and duration of follow-up and methods of study. As noted in Results, there were factors that were statistically significant during preliminary analysis but did not show significance in log-rank and Cox regression analyses.
Limitations of the Study

The study is methodologically retrospective, and the second phase of the study was based on telephone interviews. This is partially due to economic issues, physical disability, and geographic distribution of the patients. There may be a chance that those patients who did

There are a few studies available for comparison in which the authors have tried to clarify the long-term outcome of patients with PTE after penetrating battle-relatedhead injuries. 6,12,13,15,16 Walker15 followed up 246 soldiers for 10 years who sustained head injuries in World War II. All but 10 had open wounds. Of the entire living group that could be followed up (212 patients), 46.3% did not suffer seizures of any type for > 2 years. Weiss and Caveness16 reported a 47.2% persistence rate of seizures after the Korean war; 67% of their patients had penetrating injuries and were observed for 811 years after injury. Salazar et al.12 reported persistent seizures in 47% of his patients from the Vietnam War after 15 years. Based on all previous experiences with World Wars I and II and the Korean and Vietnamese wars, Caveness et al.6 have come to the conclusion that within 510 years, 50% of patients with PTE stopped having seizures, with or without therapy. In his series of patients with PTE after penetrating head injury sustained during the IraqIran war, Aarabi et al.2 reported persistent seizures in 126 (80.3%) of 157 patients during 6154 months (mean 39.4 months, median 23 months) of follow-up. The rate of persistent seizures from other major wars has been reported to be ~ 50%; however, the rate of becoming seizure free in reported by Aarabi et al.2 and by us (with a longer duration of follow-up) is much lower
J. Neurosurg. / Vol 110 / February, 2009

322

Prognostic factors in the persistence of posttraumatic epilepsy

less common in patients with residual aphasia and organic mental disorders. In his 10-year follow-up of World War II veterans with PTE, Walker15 reported that ~ 50% of patients with gross neurological deficit had experienced a cessation of seizures for 2 years or more, whereas only 25% of the group without neurological deficit had attained such freedom from seizures. He hypothesized that the neurological deficit may act as a surgical extirpation to decrease seizures. We could not find any statistically significant difference between frequency of speech problems and sensory and motor deficits in patients with persistent seizures and those who were free of seizures. In our study, the probability of persistent seizures after 16 and 21 years was higher in patients who had sphincter problems during their seizures. The statistical difference remained significant (p < 0.001, Cox regression) when adjusted for the type of seizure, irregular usage of antiepileptic medication, and shrapnel pieces. This clinical finding had not been addressed by the previous studies. During the interviews we realized that answers regarding factors like presence of retrograde or anterograde amnesia or their duration were not reliable, so we did not use this information.
Does the Presence of Metal Fragments and Their Number Affect the Persistence of Seizures?

Fig. 1. Kaplan-Meier graph showing the probability of persistent seizures in the entire cohort of patients.

(24.8% of our patients were seizure free after 18.6 4.7 years). The differences in the study populations (for example, possible bias toward patients with persistent seizures in our study, unknown distribution of bullet or shell fragments as the cause in studies by Salazar et al.12 and others, or differences in the distribution of seizure types) and study methods may contribute to the different rates of persistent seizures.
Relationship of Patients Characteristics and Clinical Findings to Persistence of Seizures

In the report by Salazar et al.,13 persistent seizure was

Table 5: Cox regression analysis for the prognostic factors* Factor

In the series reported by Gliddon,9 the presence of retained bone fragments increased the chances of PTE. In other studies the investigators failed to demonstrate a

No. of Obs 169 184 184 183 181 180 87 88 87 88 84 88 159

Hazard Ratio 0.204 0.562 0.647 0.265 0.259 0.264 0.294 0.269 0.238 0.279 0.262 0.272 0.476

p Value 0.117 0.070 0.362 0.000 0.000 0.000 0.007 0.018 0.010 0.029 0.016 0.021 0.123

95% Confidence Interval 0.0281.487 0.3021.049 0.2541.648 0.1370.513 0.1330.504 0.1370.511 0.1210.714 0.0910.796 0.0800.713 0.0890.875 0.0880.782 0.0900.821 0.1861.223

early seizure (1st wk) presence of triggering factors for seizures traversing midline/ventricles sphincter disturbance during seizure sphincter disturbance during seizure (adjusted for generalized tonic-clonic seizure) sphincter disturbance during seizure (adjusted for irregular usage of antiepileptic medication) sphincter disturbance during seizure (adjusted for 3 vs < 3 pieces of shrapnel) patients w/ 3 pieces shrapnel vs < 3 pieces patients w/ 3 pieces shrapnel vs < 3 pieces (adjusted for sphincter disturbances) patients w/ 3 pieces shrapnel vs < 3 pieces (adjusted for presence of shrapnel) patients w/ 3 pieces shrapnel vs < 3 pieces (adjusted for history of infection) patients w/ 3 pieces shrapnel vs < 3 pieces (adjusted for history of craniotomy) frequency of seizures (<12 per year vs more)
* Obs = observations. This column refers to the number of observations without missing values.

J. Neurosurg. / Vol 110 / February, 2009

323

B. Eftekhar et al.
Table 6: Persistent seizure probabilities of patients after 16 and 21 years after injury Persistent Seizure Probability in % (95% confidence interval) Yrs Postinjury 16 21 No. of Patients (184) 86.4 (80.590.6) 74.7 (67.0580.9) w/ Sphincter Problems During Seizures (109 patients) 93.6 (8796.9) 86.8 (78.0592.2) w/o Sphincter Problems During Seizures (74 patients) 75.7 (64.283.9) 57.4 (43.369.2) Patients w/ 3 w/ < 3 Pieces of Shrap- Pieces of Shrapnel nel (52 patients) (39 patients) 80.7 (67.289.1) 61.6 (45.474.3) 94.4 (79.698.6) 84.5 (60.294.6)

significant relationship between retained bone fragments and the occurrence of late epilepsy.2,4,12,17 All of these studies addressed the possible contributing factors for occurrence and not persistence of PTE. In our study, there was no relationship between retained metal fragments and seizure persistence. This supports the hypothesis that the initial injury plays a major role in the persistence of the seizures, and further surgical intervention and removal of fragments do not change the outcome significantly in terms of seizure persistence. Although the number of shell fragments at initial injury (where the numbers were available) was not significantly different between the 2 groups (that is, those with < 3 pieces of shrapnel and those with 3 pieces), this variable proved to contribute significantly to persistence of seizures. In interpreting these results, we should consider that the number of fragments was available in 91 cases. Although the extent of injury seems to be related to the occurrence of PTE,12,13 it does not seem to be the case for persistence of PTE. The number of fragments does not reliably represent the extent of injury, and we did not find a statistically significant relationship between multiple lobes of involvement or a fragment passing through the midline or transventricular trajectory (again weak indices of the extent of injury) and persistence of seizures. Salazar et al.12 found that persistent epilepsy was less common in patients with large brain volume losses. As mentioned above we could not comment on the brain volume loss as one of the reliable factors from this standpoint.
Side, Location, and Type of Injury and Persistence of PTE

rates of seizure persistence regardless of prophylactic antiepileptic use. The medication dosages (for example, phenytoin) given as prophylaxis were not available, so our comments on the prophylactic role of the medications have this shortcoming. It has been mentioned that seizure remission was less likely in patients whose seizures began later after injury, especially if the latency to seizure onset was > 4 years.6 In the studies by Salazar et al.12,13 the authors did not show any statistically significant relationship between time of onset of PTE in the first 5 years after injury and persistence of seizures. Weiss et al.16 concluded that early seizures (onset within the 1st week postinjury) make cessation of seizures more likely after 4 years. We could not find any statistically significant relationship between the latency to seizure onset or early seizures and persistence of the seizures. Salazar et al.12,13 reported that patients with persistent seizures had higher seizure frequency, particularly when the frequency was > 11 seizures during the 1st year after trauma. Based on his series, Walker15 found that if frequency of seizures was < 6 per year with or without medication, the odds were better than 3:2 that the attacks would stop. Based on their findings with Korean war victims Weiss and Caveness16 reported that the persistence rate at 4 and 8 years after injury was statistically higher in those patients with 3 seizures in the 1st year postinjury than those who had less frequent seizures. We classified the frequency of seizures in the initial period postinjury similarly to Weiss and Caveness.16 As is seen in Table 4, the frequency of seizures in patients with persistent seizures (seizure free < 2 years) is higher than in others, but this difference was not statistically different. Combination of groups and classification to 2 groups of < 12 seizures per year and 12 per year shows a marginally significant difference in rate of seizures (p = 0.055). This factor was not significant in log-rank and Cox regression analysis. The frequency of seizures in the initial stage of our study had been documented on an arbitrary basis (once per day, week, month, or 6 months). This has caused studying of the initial frequency of seizures as a possible prognostic factor less satisfying.
J. Neurosurg. / Vol 110 / February, 2009 Relationship of Frequency of Seizures to Their Persistence Early Seizure and Persistence of Seizures

We studied the side of the entrance wound (left or right), location (frontal, parietal, temporal, occipital lobes, and multiple lobe lesions), and type of injury (shell fragment or bullet). None of these factors reached the level of statistical significance in Cox regression analysis. Although a specific location of the penetrating injuries causing specific neurological deficits has been reported to be important in determining PTE3,12 and persistence of seizures,12,13 we did not find any supporting evidence.

Although it remains unknown whether early treatment has an impact on the long-term prognosis of epilepsy, preliminary findings suggest that it does not.14 In this study, no statistical significance was found between
324

Effect of Prophylactic Antiepileptic Medications

Prognostic factors in the persistence of posttraumatic epilepsy

Fig. 2. Kaplan-Meier graph comparing probabilities of persistent seizures in patients with and without sphincter disturbances.

Fig. 3. Kaplan-Meier graph comparing probabilities of persistent seizures in patients with < 3 pieces of shrapnel and those with 3 pieces.

Seizure type is an important predictor of response to therapy. The prognosis for the control of pure partial seizure is not as good as that of secondarily generalized seizures.14 According to Salazar et al.,12,13 persistent seizures were more common among the partial simple and partial complex seizure groups (p < 0.001), which also had longer mean durations of epilepsy and significantly higher frequency of seizures (p < 0.001). Weiss and Caveness16 suggested that early focal-general seizures are more likely to persist than either initial general or focal seizures. We reconstructed tables using the available data in the study by Weiss and Caveness.16 No statistical difference was found when the persistence rate was compared between groups of patients with focal (partial without generalization) and nonfocal (generalized or secondary generalized) seizures. We also combined 2 groups of patients with focal and secondary generalized seizures and compared this group with patients having generalized seizures and found no significant difference. In this study we could not repeat the findings of Salazar et al.12,13 Moreover, we could not provide evidence in favor of Weiss and Caveness.16 In the studies by Salazar et al.12,13 44 patients (20% of cases) had complex partial seizures. We had only 1 case of partial complex seizure. According to the scheme proposed by the International League Against Epilepsy, we classified our patients based on partial and generalized seizures. We also classified them regarding the presence of a generalized component (generalized or partial seizures evolving to secondarily generalized seizures) and partial seizures without generalization. No statistical significance was found between rates of persistence of seizures in the 2 groups. We also did not find any statistically significant difference in the rate of persistent seizures between those with partial seizures evolving to secondarily generalized seizures and the others. The small proportion of patients with partial seizures (3.7%) may be one of the reasons that we did not find any significant differences in terms of persistence
J. Neurosurg. / Vol 110 / February, 2009

Relationship of Type of Seizures to Their Persistence

between types of seizures. However, a subset (62.6%) of the patients with nonpartial seizures who had sphincter disturbances during the seizures showed a higher probability of persistent seizures after 16 and 21 years.
Does the Surgical Intervention and its Type (Radical vs Conservative Techniques) and Postoperative Infection Affect the Long-Term PTE?

We studied the role of craniotomy and the surgical removal of retained bone or metal fragments, cranioplasty, and brain abscess in the persistence of the seizures. None of these factors reached a statistically significant level based on Cox regression analysis. These findings are in agreement with those of Salazar et al.12,13 There are reports in which the authors have argued the need for radical surgery and exploration in penetrating head injuries sustained in war.4,7,10 In our series the majority of the explorations were as complete as possible,1 so we could not comment on the effect of the different surgical techniques on the persistence of the seizures. From a neurosurgical standpoint, we were interested in technical factors like dural grafts and extension of the craniotomy for evacuation of the accompanying lesions (such as hematoma), which may have been potentially influential on the persistence of the seizures, but unfortunately the information in the available records was not reliable in this regard.
Effect of Constitutional Factors Versus Brain Damage Characteristics

Caveness et al.6 concluded that in persistence of seizures, the constitutional factor (probably a multifactorial genetic trait) played the dominant role. This study, in accordance with the work of Salazar et al.,12,13 does not provide evidence in support of this finding, but the negative prognostic effect of > 2 pieces of shrapnel in the persistence of seizures shows a significant nonconstitutional (that is, a lack of genetic cause of the seizures) influence on the persistence of the seizures.
325

B. Eftekhar et al.
We think that examining the time in Cox regression analysis yields a more reliable and realistic result regarding the contributing factors for persistence of seizures. Further studies in this area are warranted.
Disclaimer The authors report no conflict of interest concerning the materials or methods used in this study or the findings specified in this paper. Acknowledgments We express our gratitude to Dr. Mojgan Karbakhsh, Dr. Vafa Rahimi Movaghar, and Dr. S. Saadat from Sina Trauma Research Center for their comments. We also thank Dr. Nevin Colgrave of Royal Hobart Hospital for his assistance with the editing of this paper. References 1. Aarabi B: Surgical outcome in 435 patients who sustained missile head wounds during the Iran-Iraq War. Neurosurgery 27:692695, 1990 2. Aarabi B, Taghipour M, Haghnegahdar A, Farokhi M, Mobley L: Prognostic factors in the occurrence of posttraumatic epilepsy after penetrating head injury suffered during military service. NeurosurgFocus 8(1):E1, 2000 3. Adeloye A, Odeku EL: Epilepsy after missile wounds of the head. JNeurolNeurosurgPsychiatry 34:98103, 1971 4. Brandvold B, Levi L, Feinsod M, George ED: Penetrating craniocerebral injuries in the Israeli involvement in the Lebanese conflict, 1982-1985. Analysis of a less aggressive surgical approach. J Neurosurg 72:1521, 1990 5. Caveness WF: Onset and cessation of fits following craniocerebral trauma. J Neurosurg 20:570583, 1963 6. Caveness WF, Meirowsky AM, Rish BL, Mohr JP, Kistler JP, Dillon JD, et al: The nature of posttraumatic epilepsy. J Neurosurg 50:545553, 1979 7. Chaudhri KA, Choudhury AR, al Moutaery KR, Cybulski GR: Penetrating craniocerebral shrapnel injuries during Op-

Conclusions

8.

9. 10.

11. 12.

13.

14. 15. 16. 17.

eration Desert Storm: early results of a conservative surgical treatment. ActaNeurochir(Wien) 126:120123, 1994 Engel J Jr, Van Ness PC, Rasmussen TB, Ojemann LM: Outcome with respect to epileptic seizures, in Engel J Jr (ed): SurgicalTreatmentoftheEpilepsies,ed2.New York: Raven Press, 1993, pp 609621 Gliddon WO: Gunshot wounds of the head (a review of the after-effects in 500 Canadian pensioners from the Great War, 1914-1918). CanMedAssocJ 49:373377, 1943 Levi L, Borovich B, Guilburd JN, Grushkiewicz I, Lemberger A, Linn S, et al: Wartime neurosurgical experience in Lebanon, 1982-85. I: Penetrating craniocerebral injuries. Isr J MedSci 26:548554, 1990 Ropper AH, Brown RH: AdamsandVictorsPrinciplesof Neurology,ed8. New York: McGraw-Hill, 2005, p 294 Salazar AM, Jabbari B, Vance SC, Grafman J, Amin D, Dillon JD: Epilepsy after penetrating head injury. I. Clinical correlates: a report of the Vietnam Head Injury Study. Neurology 35:14061414, 1985 Salazar AM, Schwab K, Grafman JH: Penetrating injuries in the Vietnam war. Traumatic unconsciousness, epilepsy, and psychosocial outcome. Neurosurg Clin N Am 6:715726, 1995 Sander J, Sillanpaa M: Natural history and prognosis, in Engel J Jr, Pedley TA (eds): Epilepsy:aComprehensiveTextbook. Philadelphia: Lippincott-Raven, 1997, pp 6986 Walker AE: Prognosis in post-traumatic epilepsy. JAMA 164:16361641, 1957 Weiss GH, Caveness WF: Prognostic factors in the persistence of posttraumatic epilepsy. J Neurosurg 37:164169, 1972 Weiss GH, Salazar AM, Vance SC, Grafman JH, Jabbari B: Predicting posttraumatic epilepsy in penetrating head injury. ArchNeurol 43:771773, 1986

Manuscript submitted July 27, 2007. Accepted April 14, 2008. Please include this information when citing this paper: published online October 31, 2008; DOI: 10.3171/2008.4.17519. Address correspondence to: Behzad Eftekhar, M.D., M.P.H., Sina Trauma Research Center, Sina Hospital, Tehran, Iran. email: Beftekhar @yahoo.com or eftekhar@sina.tums.ac.ir.

326

J. Neurosurg. / Vol 110 / February, 2009