is ambiguous. Thus, it was not possible to determine from the report the number of patients who had their only seizure within 6 months of injury.
Russell (1968) conducted a conditional cohort study of the prevalence of epilepsy after penetrating head injury in 185 patients injured in WWII. The men were followed for 10–20 years after injury. Of the 185, 77 (41.6%) had posttraumatic grand mal epilepsy, and 40 (21.6%) were still having seizures 10 years or more after injury. The study was limited in that it did not include a control population and the cohort was not described in terms of age, sex, and nationality. It also is not clear who may have had only one seizure in the first 6 months after injury.
Weiss et al. (1983) studied 1,221 head-injured Vietnam veterans enrolled in the Vietnam Head Injury Study (VHIS). Although the focus of the study was on prognostic factors in the occurrence of epilepsy, they reported that 31% of the cohort had seizures more than a week after injury. A followup study of this cohort was reported by Salazar et al. (1985) and by Rish et al. (1983).
The four studies of closed TBI and seizure risk in civilians come from the Rochester Epidemiology Project (see Chapter 5). Annegers et al. (1980) reported the risk of unprovoked seizures in a cohort of 2,747 patients (1,132 children and 1,615 adults) in Olmsted County, Minnesota, who had sustained TBI in 1935–1974. An additional 4,541 patients who sustained TBI in 1975–1984 were added later (Annegers et al., 1998). As part of the Rochester Epidemiology Project, medical records containing physician diagnoses of TBI were linked to later medical records documenting unprovoked seizures in the study interval and compared with records of those who did not sustain TBI. The authors found the overall risk of unprovoked seizures after TBI to be 3.6 times (95% CI, 2.7–4.8) that in the noninjured population. That risk, also known as the standardized incidence ratio (SIR), was highest among those with severe TBI (SIR, 17.0; 95% CI, 12.3–23.6), followed by those with moderate TBI (SIR, 2.9; 95% CI, 1.9–4.1) and mild TBI resulting in loss of consciousness (LOC) or posttraumatic amnesia (PTA; SIR, 1.5; 95% CI, 1.0–2.2). The overall unprovoked-seizure risk was found to be highest in the first year after injury (SIR, 12.7) and to fall (to 4.4) 1–4 years after injury and fall further (to 1.4) 5 years or more after injury. That pattern of seizure risk over time was also found in each TBI-severity group. Although the risk of unprovoked seizures after mild TBI was increased at all times after injury, it was significantly different from the risk in the uninjured population only during in the period 1–4 years after injury.
A limitation of both Annegers et al. studies (1980, 1998) is that the authors included children in the study and in the risk estimates. It is not possible with the available data to calculate rates for adults only, and reported rates may be misleading inasmuch as seizures occur more frequently in children than adults. Therefore, the generalizability to the veteran population is unclear.
Two additional published analyses of the Rochester Epidemiology Project included children but presented the post-TBI incidence of seizures in adults separately. Annegers and colleagues (1995) reported that the age-adjusted incidence of post-TBI seizures in adults ranged from 2.0/100,000 person-years in 25- to 34-year-olds to 14.0/100,000 person-years in those over 74 years old. In addition, the age-adjusted incidence of post-TBI seizures was higher in males at all ages (8.6/100,000 person-years in males and 4.8/100,000 person-years in females). Singer (2001) used data from the Rochester Epidemiology Project to compare the incidence of post-TBI seizures with the incidence of idiopathic epilepsy previously determined for Olmsted County.