Epidemiology of Posttraumatic Epilepsy: A Critical Review
Address correspondence and reprint requests to Dr. L. Frey at University of Colorado Health Science Center, B150, 4200 E 9th Ave., Denver, CO 80262, U.S.A. Email: email@example.com
Summary: Problem: Traumatic brain injury (TBI) is a major cause of epilepsy. We need to understand its frequency and its contribution to the total spectrum of the convulsive disorders.
Methods: A review of selected articles dealing with epilepsy after brain trauma was undertaken.
Results: The number of epidemiologic studies of posttraumatic seizures has increased substantially over the past 40–50 years, offering steadily increasing knowledge of the frequency, natural history, and risk factors of this well-recognized complication of TBI. In general, the incidence of posttraumatic seizures varies with the time period after injury and population age range under study, as well as the spectrum of severity of the inciting injuries, and has been reported to be anywhere from 4 to 53%. As high as 86% of patients with one seizure after TBI will have a second in the next 2 years. Longer-term remission rates of 25–40% have been reported. Significant risk factors for the development of seizures in the first week after injury include acute intracerebral hematoma (especially subdural hematoma), younger age, increased injury severity, and chronic alcoholism. Significant risk factors for the development of seizures >1 week after TBI include seizures within the first week, acute intra-cerebral hematoma (especially subdural hematoma), brain contusion, increased injury severity, and age >65 years at the time of injury.
Conclusions: Epilepsy is a frequent consequence of brain injury in both civilian and military populations. We understand some factors associated with its development, but there remain many unanswered questions.
“Through trauma, the brain may be injured by contusion, laceration, compression, and it is well known that these insults may result in epilepsy after a ‘silent period of strange ripening.’ That period lasts for months or years, but these insults produce epilepsy in the case of one individual and not in the case of another… . Our attention should therefore be directed toward the discovery of this mysterious difference” (1).
The average incidence of traumatic brain injury (TBI) in the United States is estimated to be between 180 and 220 per 100,000 population per year (2). With recent improvements in prehospital care and emergency diagnosis and care of TBI, mortality rates have decreased from 50%∼30 years ago to ∼30% currently. However, TBI survivors still carry a tremendous burden of disability as a result of their injuries. Of mild head injuries, 10% are thought to result in permanent disability, as well as 66% of moderate head injuries and 100% of severe head injuries (3). The occurrence of seizures after head injury is a recognized complication of TBI and has recently been shown to worsen functional outcome significantly after TBI (4). Much work has been done in hopes of preventing the development of, or, at least, minimizing the impact of posttraumatic epilepsy. Understanding the risk factors and natural course of the disease is one step toward controlling this costly sequela of TBI.
The number of published epidemiologic studies on posttraumatic epilepsy has increased substantially in the past 40–50 years. Despite differences in study features, many of the same conclusions are drawn by different groups of authors. In general, epidemiologic studies on topics like posttraumatic epilepsy offer two basic categories of information: information about the frequency and natural history of the disease within a certain population, and information about the associated risk factors. Before discussing these any further, however, it is more than worthwhile to formulate working definitions of both posttraumatic epilepsy and traumatic head injury.
Throughout the body of literature on the subject, the terms posttraumatic epilepsy and posttraumatic seizures (PTSs) are both used to describe the same entity: seizures occurring after head trauma that are thought to be causally related to the trauma itself. Most of the articles published to date define posttraumatic epilepsy as one or more unprovoked seizures after head trauma. Only recently has the diagnosis of epilepsy (of any kind) been reserved for patients who have had two or more unprovoked seizures. Distinct advantages and disadvantages exist to adopting this newer definition. Defining epilepsy as two or more unprovoked seizures is more in keeping with the current usage of the term, but it also defines a pattern of susceptibility to seizures and excludes those patients who would ever have had only one seizure. Although the exact percentage of head-injury patients in whom this ultimately would be true is not known, we do know that, in one population-based study, 86% of patients with one late posttraumatic seizure had a second seizure within 2 years (5).
At least two major methodologic disadvantages of waiting for a second unprovoked seizure exist, however (6). The first disadvantage lies in the fact that some patients, regardless of how long they are followed up, will have their first seizure at the end of their follow-up period. Because the risk of a first seizure continues to be elevated for >10 years after a severe head injury (7), establishing a recurrent pattern of events in these patients may take an unreasonably long time. The second disadvantage is the increasingly predominant practice of starting antiepileptic drugs (AEDs) after just one seizure, a practice that may change the natural history of posttraumatic seizures. Two recent studies have attempted to minimize the impact of AED use on their incidence data. Angeleri et al. (8) dropped patients from their study if they were started on AEDs. The study protocol used by Vespa et al. (9) provided for uniform use of prophylactic phenytoin (PHT) in their patients, while ensuring that the serum levels for the drug were not statistically different between the patients who seized and the patients who did not seize.
The term posttraumatic seizure (PTS) will be used throughout the rest of this review, even if the original authors used posttraumatic epilepsy, in an attempt to minimize any confusion that would result from using both terms. If it is relevant to the discussion, studies whose patients were included only after two or more seizures will be identified as such.
An association between new-onset seizures and an antecedent head trauma has been found with various degrees of certainty. Only recently have efforts been made to search for, identify, and then eliminate other factors that would independently increase the risk of seizures in any population, such as a history of epilepsy, a history of previous head injury, previous brain damage of any type, severe alcoholism (8), fluid/electrolyte imbalance, or other causes of acute symptomatic seizures, like hypoxia or known brain ischemia (10).
Posttraumatic seizures are usually divided into three categories: “immediate” seizures, early seizures, and late seizures. Early seizures are those occurring while the patient is “still suffering from the direct effects of the head injury” (6), a period commonly defined as 1 week after head injury. Approximately 90% of seizures occurring within the first 4 weeks after head injury will happen during this first week (11). In general, seizures that happen at or minutes after impact—“immediate,”“contact,” or “concussive” seizures—are not included in studies of early PTSs. The exact pathophysiology behind immediate seizures and their exact clinical significance therefore remains unclear. Although these seizures are often thought not to increase susceptibility to later, unprovoked, seizures, data presented by Kollevold (12) suggest otherwise. In his study, both immediate and early seizures were associated with an increased risk of late seizures. No statistically significant difference was found between the two as risk factors for late PTSs. Late PTSs are usually defined as seizures occurring >1 week after injury.
Traumatic head injury
According to Jennett (11), “what matters in head injury is damage to the brain, either on impact or subsequently due to processes initiated by that impact.” The best means of accurately assessing whether damage to the brain has actually occurred, however, is controversial. At this time, no unequivocal clinical markers of brain damage are used consistently in studies of PTSs. In most of the studies done to date, head trauma to the degree that a patient needed some level of medical attention was taken as presumptive evidence of additional underlying brain injury. However, the level of medical attention required was often variable and incompletely defined, making it difficult ultimately to compare the results from different studies. For example, some study participants were identified simply because they sought medical care after head injury (6). Other populations were drawn from a pool of patients requiring either hospital admission (8,10,11,13), or, more specifically, admission to an intensive care unit (ICU) (9) after head trauma. Still other populations were defined by their eventual need for rehabilitation services (4).
Statistical comparisons within these groups also are difficult because of variability in the type of injury sustained. Many studies were done at a time when hospital admission criteria were very different and patients were admitted who, today, may be treated in an emergency department and then discharged home. Current hospital-based series, therefore, are biased in favor of more serious head injuries. Advances in medical technology also have improved survival after head injury, especially after more severe injury, thus further increasing the proportion of severely injured patients in the hospital. In addition, patients in series of wartime injuries (13–18) presumably had a higher percentage of missile wounds compared with those in civilian populations (4–11,19). These missile wounds, with their increased likelihood of dural penetration, are more likely to cause severe brain injury.
Specific clinical features have often been used to stratify head injury severity. In some cases, this was as simple as dividing patients into groups based on length of their posttraumatic amnesia (<24 h or >24 h) (11). Annegers et al. (6) developed a three-tier classification for head-injury severity: mild, moderate, and severe. Mild injuries were those in which no skull fracture was found and the period of posttraumatic amnesia or loss of consciousness was ≤30 min. The moderately injured group of patients included those with skull fractures or other injuries with >30 min of posttraumatic amnesia or loss of consciousness who did not otherwise meet the criteria for a severe injury. Severe injuries were those with a documented brain contusion, an intracerebral hematoma (ICH), or >24 h of posttraumatic amnesia or loss of consciousness. A later group of authors, however, thought that loss of consciousness or duration of posttraumatic amnesia may not be important independent risk factors for PTSs. To support their argument, they point to the fact that loss of consciousness is not necessarily always associated with cortical (and, thus, potentially epileptogenic) injuries (4). A universally accepted classification for head-injury severity has yet to be developed, although a number of alternate strategies have been used. Hahn et al. (10), for example, proposed a severity scale for pediatric injuries based on the Glasgow Coma Scale (GCS) score or, in children younger than 3 years, on the Children's Coma Score (CCS). For these patients, mild injury meant a GCS or CCS ≥13. Moderate injuries had a GCS or CCS between 9 and 12. Severe injuries had a GCS or CCS of ≤8.
FREQUENCY OF POSTTRAUMATIC SEIZURES
In general, the incidence of PTSs correlates with the severity of the inciting injury (Tables 1 and 2). This correlation between the incidence of PTSs and injury severity explains most of the variability in the overall incidence of PTSs, a value that ranges from a low of ∼4% to a high of 53%. Compared with series done in the civil sector, a consistently higher proportion of severe head injuries was seen in series of wartime injuries. This disproportionate severity of wartime injuries is in large part due to a higher proportion of injuries that involve dural penetration and widespread brain damage in military patients, as mentioned earlier. For example, 40.9% of the veterans involved in the Vietnam Head Injury Study (15) had head injuries with damage to multiple lobes of the brain, whereas only 7.1% of the civilian patients studied by Annegers et al. (6) qualified for a diagnosis of severe brain injury. It only makes sense, then, that these series of wartime injuries, with their higher proportion of severe head injuries, would show greater PTS frequencies than would civilian series (Tables 3 and 4). However, even among the civilian series, the populations of patients who one would expect to have had more serious injuries, such as those requiring hospitalization or an ICU admission or those with prominent posttraumatic neurologic deficits requiring rehabilitation, should, and do, have higher seizure frequencies.
Table 1. Frequency of posttraumatic seizures (PTSs) stratified by injury severity in civilian populations
|Adults or mixed series|
| ||PTA <24 h||PTA >24 h|
|(7)||1.5 (SIR) ||2.9 (SIR) ||17.0 (SIR) |
|(19)|| 2.8% (Early)|| 1.7% (Early)|| 21.6% (Early)|
Table 2. Frequency of posttraumatic seizures (PTSs) stratified by injury severity from select military series
| ||Dura mater intact||Dura penetrated|
|World War I (29)||26.1%||47% |
|World War II (29)||—||43.4%|
|Korea (29)||23.8%||42% |
| ||Scalp or skull||Single lobar||Multilobar injury|
| ||only||injury||or worse|
Table 3. The frequencies of posttraumatic seizures (PTSs) in civilian populations
|Adults or mixed series|
|(7)||10+ yr||4.4% || 2.6%|| 2.1%||Population based|
|(n = 4,541)|| || || || || |
|(19)||7+ days||–|| 4.1%||–||Hospital admissions|
|(n = 702)|| || || || || |
|(11)||1–4 yr||–|| 4.6%||5% ||Hospital admissions|
|(n = 1,000)|| || || || || |
|(4)||5+ yr||–||16.9%||24.5%||Required intensive|
|(n = 490)|| || || || || rehabilitation|
|(20)||4–12 yr||7.03%|| 6.5%|| 1.3%||Pediatric hospital|
|(n = 4,465)|| || || || || admissions|
|(10)||7+ mo||9.8% || 9.6%||–||Pediatric hospital|
|(n = 937)|| || || || || admissions|
Table 4. The frequencies of posttraumatic seizures (PTSs) in military series
|World War I (29)||7–20 yr||34% ||–||–||Gunshot wounds|
|(n = 317)|| || || || || |
|World War II (29)||5 yr||43% ||–||–||Gunshot wounds|
|(n = 820)|| || || || || |
|World War II (29)||7–8 yr||28% ||–||–||Unselected|
|(n = 739)|| || || || || |
|Korean War (14)||8–11 yr||30% ||10% within first month||–||Unselected|
|(n = 356)|| || || || || |
|Vietnam War (15)||4+ yr||33.4%||4.4%||27.9%||Required neurosurgical|
|(n = 1,030)|| || || || || attention|
|Vietnam War (16)||4+ yr||35% ||3.9%||31% ||Required neurosurgical|
|(n = 1,221)|| || || || || attention|
|Vietnam War (18)||15+ yr||53% ||–||–||Penetrating missile|
|(n = 421)|| || || || || injuries|
The incidence of early seizures (usually within 1 week of injury) ranges from 2.1 to 16.9% and, in general, is correlated with the distribution of head-injury severity within the specific group being studied. Of the patients in one series, 10% were in status epilepticus, a presentation more common in children (11).
Depending on the series, the incidence of late seizures ranges from 1.9% to >30%. Like the incidence of early PTSs, the variability in this finding is likely due to variability within the patient populations being studied, especially with respect to injury severity. In general, most late PTSs occur during the first year after injury (7,11), although they can also occur for many years afterward. Exactly how long the increased risk of seizures persists after head injury is unclear. Investigators for the Vietnam Head Injury Study concluded that, in their population, “…posttraumatic seizure incidence does not reach that of the general uninjured population until well after 15 years after injury” (17). In an initial study of the civilian population in Olmsted County, Minnesota, PTS risk overall did not differ significantly from baseline risk after 5 years (6). However, with extended follow-up of the same population of head-injured patients, the authors concluded that, “…even unprovoked seizures occurring more than 10 years after a severe traumatic brain injury can be attributed in large part to the injury” (7).
The exclusion of other causes of seizures is an important factor in the accurate determination of the risk of seizures strictly attributable to a head injury. Looking back over the data from the Vietnam Head Injury Study, investigators thought that “neither subsequent head injury, other encephalopathy, nor alcohol abuse played an important role” in the occurrence of late PTSs, particularly in those patients whose seizures began >10 years after their index injury (18). More often, however, investigators formally incorporated these factors into their study exclusion criteria. Other important exclusions included a history of seizures before head injury (4–7,11), known prior head injury or brain damage (6–8), fatal injuries (7,8), evidence of hypoxia, and fluid or electrolyte imbalances (10), as mentioned earlier.
THE PEDIATRIC EXPERIENCE
The overall incidence of PTSs in children ranges from 0.2 to 9.8%. Overall, early seizures were more common than late ones in the pediatric population: 9.3% of children with head injuries in one series had early PTSs, whereas only 0.2% had late PTSs (9). Early seizures in children tended to occur earlier within the first week after injury than they did in the adult population. In the study by Hahn et al. (10), 94% of their patients with PTSs seized during the first 24 h after injury.
Younger children are at proportionally increased risk of both early and late seizures after head injury; 30.8% of children younger than 7 years in one civilian series had early PTSs, compared with 20% of children aged 8–16 years and 8.4% of children 16 years or older (p < 0.0001) (4), findings similar to those of Jennett (11) and Hendrick and Harris (20). The correlation between age at injury and risk of PTSs is less clear in the very youngest age group. In one series, early seizures were even more common in patients younger than 5 years at time of injury (11). However, two other studies found no significant difference in the risk of seizures in the very young (up to 3–5 years) (10,19).
Younger children are more likely to have status epilepticus as the presenting feature of their PTSs. In one study, status was more than twice as common in children younger than 5 years than in all other age groups combined (p < 0.001) (11).
Higher incidences of PTSs are seen in more severely injured children (Table 1). In general, this is true regardless of the age of the child. Uniquely, however, in one study, 17% of children younger than 5 years with “trivial” injuries (no loss of consciousness, no posttraumatic amnesia, no depressed skull fracture, and no ICH) had an early seizure, compared with only 2% of all patients older than 5 years (adults and children) (11), findings similar to those of Kollevold (21). Epilepsy “rarely followed a trivial injury except in children under 5” (11), a phenomenon that is of unclear clinical significance.
NATURAL HISTORY OF POSTTRAUMATIC SEIZURES
The lifetime total number of seizures in patients with PTSs is not associated with any identifiable variables such as age or severity of injury, and often varies widely even within generally homogeneous populations. For example, 39% of patients in the Korean conflict veteran series had a total of between one and three seizures during a 10-year period of follow-up. Of the same group, however, 38% had >30 seizures (15).
Remission rates among patients with PTSs range from 25 to 40%, with higher overall remission rates reported in studies done after the development of effective AEDs. One early study found that seizure remission was less likely in patients whose seizures began later after injury, especially if the latency to seizure onset was >4 years (11). However, the burden of the other evidence suggests that no significant relation exists between the latency to first seizure and seizure duration or persistence (15,18), although patients with frequent seizures in the first year will often continue to have frequent seizures and have a smaller chance of seizure remission (18).
Most patients who will have a second unprovoked late PTS do so during the first 2 years after their first late PTS. Haltiner et al. reported that up to 86% of TBI survivors with a first PTS will also have a second within the following 2 years (5). A certain percentage of PTS patients remain refractory to AED therapy. For example, in the treatment arms of various anticonvulsant prophylactic trials, a pooled estimate of 13.3% seized despite aggressive treatment regimens (22). Very little is specifically known about the chance of remission of PTSs that begin in childhood.
RISK FACTORS FOR POSTTRAUMATIC SEIZURES
The risk factors that have been found to be significant for the development of early PTSs are listed in Table 5. The most consistent risk factor is the presence of an intracerebral blood collection, which confers a ≤30% increase in the risk of early PTSs, regardless of the patient's age or other features of their injury (11,19). This increased risk is especially true for subdural hematomas in children (10). In this same age group, intraparenchymal and epidural hematomas do not confer the same risk of early seizure development, presumably because of the lesser degree of direct cortical irritation from blood products (10).
Table 5. Independent risk factors for posttraumatic seizures
|Risk factors for early posttraumatic seizures||Risk factors for late posttraumatic seizures|| |
|Acute intracerebral hematoma||Early posttraumatic seizures|| |
|Acute subdural hematoma, in children||Acute intracerebral hematoma, especially subdural hematoma (in all patients)|| |
|Younger age||Brain contusion|| |
|Increased injury severity, including loss of consciousness or posttraumatic amnesia lasting >30 min||Increased injury severity, including loss of consciousness or posttraumatic amnesia lasting >24 h|| |
|Chronic alcoholism||Age older than 65 yr at time of injury|| |
The second most predominant risk factor for the occurrence of early PTSs is higher injury severity, although there is considerable lack of consistency in the use of different markers to define subgroups of injury severity (Table 6). This is important because many of these markers of injury likely represent differing aspects of actual brain damage. For example, patients who respond only to pain at presentation after head injury are assumed to have sustained a more severe injury than have those patients who are less encephalopathic (can follow commands). Patients who respond only to pain have a statistically significant increase in the risk of early PTSs. However, if the severity of injury is measured by the extent of actual brain involvement in the same group of patients (i.e., single vs. multilobar involvement), the significantly increased risk of early PTSs no longer exists (although those patients remain at a significantly higher risk for late seizures) (15).
Table 6. Commonly used clinical markers of head-injury severity
|Extent of brain damage: multilobar vs.||Level of alertness|
| single-lobe involvement|| on presentation|
|Brain-volume loss on CT scan||Glasgow Coma Score|
|Length of posttraumatic amnesia||Depth of wound|
|Length of loss of consciousness||Size of cranial defect|
In many studies, early seizures occurred 50–100% more frequently in children than they did in adults with comparable injuries (7,21). In a population study of head injury in the Bronx, this conclusion held at all levels of injury severity (W.A. Hauser, personal communication, 2002). In a population study of inhabitants of Rochester, Minnesota, however, it was true for only severe and mild injuries (6). Younger age is not an independent risk factor for the occurrence of early PTSs if the incidence is adjusted for the increased risk of seizures that exists for this population, regardless of antecedent head injury (7).
Other risk factors that likely influence the occurrence of early PTSs include diffuse cerebral edema (in children), intracranial metal-fragment retention, residual focal neurologic deficits, and depressed or linear skull fractures (adults only).
The risk factors that have been found to be significant for the development of late PTSs are listed in Table 5. In this case, the presence of early PTSs was the most consistently significant risk factor for the development of late PTSs, although, in one study (7), the occurrence of early seizures was not an independent risk factor in multivariate analysis. If it is significant, the presence of early PTSs will increase the risk of late seizures regardless of how many seizures actually occurred during the first week after injury (11). The relation between early PTSs and the development of late PTSs varies based on age. Jennett (11) found that the risk of late PTSs is not significantly increased in children younger than 16 years who only have focal early seizures. In another study, no children with early PTSs were at increased risk of developing late seizures, regardless of their early seizure type (6).
ICH also is a risk factor for the occurrence of late PTSs and may confer a ≤10-fold increase in risk (7). Subdural hematomas are likely responsible for most of this increased risk in both children and adults. The presence of a brain contusion was as strong of a predictor of late seizure occurrence as was the presence of a subdural hematoma.
Markers of increased injury severity have been shown to increase significantly the risk of late PTSs, although this relation is inconsistent. When severe brain injury is defined as multilobar injury (or worse) or loss of consciousness or posttraumatic amnesia lasting >24 h, it is a significant risk factor for the occurrence of late PTSs. However, when injury severity is measured by brain-volume loss on computed tomography (CT) scan or by GCS, the significance of the relation disappears, although it is likely that these parameters will prove useful once they are used in a more standardized fashion (especially a uniform period between injury and scanning/GCS calculation).
Age older than 65 years at the time of injury is a significant risk factor for the development of late PTSs. This relation is especially important because elderly survivors of TBI, even without seizures, recover more slowly and are more likely to be permanently disabled as a result of their injuries than are younger head injury survivors (D.A. Arciniegas, personal communication, 2001).
Premorbid chronic alcoholism likely increases the risk of the development of late PTSs. In one study (21), a higher percentage of patients with chronic alcohol use had late PTSs than did comparable patients who did not drink. In addition, the same study suggested that chronic alcohol use may decrease the chance of seizure remission, although the data were not necessarily tested for significance. Conversely, patients with seizures after head injury have been shown to be at higher risk for both drug and alcohol abuse (23). The relation between alcohol use and seizures (of any type) was studied more specifically by Ng et al. (24), who concluded that “heavy alcohol use is an independent, dose-related risk factor for seizures.”
Other risk factors for the development of late PTSs include metal-fragment retention, skull fracture, residual cortical neurologic deficits, a single CT lesion in the temporal or frontal regions, and persistent focal abnormalities on EEG >1 month after injury.
Is genetic propensity for seizures a risk factor for posttraumatic seizures?
The role of genetic susceptibility in posttraumatic epilepsy is far from being clearly understood. A wide range of variability is found in individuals' responses to similar injuries. Some patients seize frequently after a head injury, whereas others may seize once or not at all, despite an almost identical injury severity. Empirically, Caveness (15) thought that much of this variability was likely due to “constitutional” or genetic factors regulating response to cerebral injury. Since then, multiple authors have attempted more formally to characterize the influence of an individual's genes by including the presence or absence of a family history of epilepsy in their risk-factor analyses. The majority of these studies have found that a family history of epilepsy is not a significant risk factor for the development of seizures after head injury. For example, Jennett (11) found that a family history was more common in patients with late PTSs, but the difference was significant only in patients younger than 16 years. In the Vietnam Head Injury Study cohort, family history was not a significant risk factor for either early or late PTSs (18). Schaumann et al. (25) found no increased risk of seizures among the parents, siblings, or offspring of patients with PTSs. Ottman et al. (26), in their study of clinical markers of genetic susceptibility to epilepsy, also found that the risk of seizures among relatives of patients with posttraumatic seizures was no higher than that of the general population, suggesting that, ultimately, the prominent variability in individual response to head injury is not genetically determined.
Using risk factors to predict the occurrence of posttraumatic seizures
Attempts have been made to use risk factors, in varying combinations, to develop a mathematical method for predicting likelihood of seizures during various periods after head injury (16,17,27). Although the predictions generated from these formulas do seem to match the observed occurrences of PTSs in selected series, they have not yet found widespread clinical use.
The occurrence of PTSs is a medically and functionally costly complication of TBI. The incidence of PTSs ranges from 4.4 to 53%, depending on the population being studied. Risk factors can be identified for the occurrence of both early and late PTSs. The presence of an acute ICH (especially a subdural hematoma), younger age, and higher injury severity are all risk factors for seizures occurring early after TBI. The presence of early PTSs, an acute ICH, a brain contusion, age older than 65 years at the time of injury, and higher injury severity all increase the risk of seizures occurring later after TBI.