- Top of page
- PATIENTS AND METHODS
Summary: Purpose: To characterize seizures after intracerebral hemorrhage (ICH), evaluating the risk of occurrence and relapse, predisposing factors, and prognostic significance, and to assess the utility of antiepileptic drug (AED) therapy as used in clinical practice.
Methods: The study sample consisted of 761 patients with spontaneous, nonaneurysmal, supratentorial ICH. Seizures were classified as immediate (within 24 h of ICH) and early (within 30 days of ICH). Baseline variables and clinical events were compared in the seizure and nonseizure group by using a multivariate regression model of failure time data.
Results: Fifty-seven patients had one or more seizures. The 30-day actuarial risk of a post-ICH seizure was 8.1%. Lobar location and small volume of ICH were independent predictors of immediate seizures. Early seizures were associated with lobar location and neurologic complications, mainly rebleeding. In patients with lobar ICH, the risk of early seizures was reduced by prophylactic AED therapy. Among seizure patients, history of alcohol abuse increased the risk of status epilepticus. Immediate and early seizures were not independent predictors of in-hospital mortality.
Conclusions: Patients with ICH are exposed to a substantial risk of seizures; however, short-term mortality was not affected, and the risk of epilepsy was lower than previously thought. The likelihood of immediate seizures is influenced by factors that are inherent characteristics of ICH, whereas the chance of developing early seizures is influenced not only by certain characteristics of ICH, but also by unpredictable events. A brief period of therapy soon after ICH onset may reduce the risk of early seizures in patients with lobar hemorrhage.
Seizures as a clinical feature of intracerebral hemorrhage (ICH) have not been fully investigated. Little is known about the frequency, temporal distribution, and characteristics of seizures, and even less about factors predisposing to seizures and their prognostic significance for short-term mortality and risk of epilepsy. Major aspects have often been ignored, including the fact that delayed post-ICH seizures may have different predisposing factors from onset seizures, that the number of patients at risk for seizure varies in time, and that many predisposing factors may act synergistically in time to cause seizures.
In this study the occurrence of seizures in patients with computed tomography (CT)-proven supratentorial nontraumatic nonaneurysmal ICH was analyzed by using multivariate analyses to determine the risk of developing initial and recurrent seizures, to identify predisposing factors for onset and delayed seizures, to evaluate the impact of seizures on outcome, and to assess the value of prophylactic antiepileptic drug (AED) therapy as used in clinical practice.
PATIENTS AND METHODS
- Top of page
- PATIENTS AND METHODS
The study sample consisted of 761 consecutive patients with nontraumatic, nonaneurysmal ICH. The sample was part of a prospective, observational study that collected short-term and follow-up data on patients with ICH admitted to neurologic, neurosurgical, and intensive care departments of our university hospital between January 1979 and December 1996. Of the 761 patients, 675 (88.7%) were admitted directly to our hospital, 60 (7.9%) reached our hospital through emergency services of other hospitals, nine (1.2%) were transferred from other hospitals (including four patients from the psychiatric hospital), and 17 were transferred from other wards of our hospital, because ICH occurred while they were hospitalized for other reasons. ICH cases were identified mainly on the basis of admission diagnosis, because patients with suspected stroke underwent neurologic evaluation and CT brain scan on arrival in the emergency department, and on the basis of discharge diagnosis. The diagnosis of ICH was based on CT scan or autopsy in all cases. Patients with histories of epileptic seizures and patients with primary intraventricular ICH, infratentorial ICH, or ICH due to brain tumors, vascular malformations, brain surgery, or infections were excluded from the study.
Seizures were classified according to the recommendations of the International League Against Epilepsy (1). Status epilepticus (SE) was defined as more or less continuous behavioral seizure activity or repetitive seizures without full recovery of neurologic function between seizures, occurring over a period of ≥30 min (2). Because onset seizures and more delayed seizures may have different risk factors, we analyzed these categories separately, classifying seizures as immediate (occurring within 24 h of onset of ICH) and early (occurring within 30 days of onset). A single episode of SE, and clusters or repetitive seizures with <24 h between events were regarded as a single seizure episode. Diagnosis was based on direct observation of seizures by medical staff at the time of hospitalization or on history provided by the clinician in charge of the patients or was determined from reliable description by ambulance personnel when seizures occurred during transportation, by patients, family members, or eye witnessess when seizures occurred before medical attention, or by medical staff when patients were transferred from other hospitals or other wards of our hospital (patients who had ICH while hospitalized for other reasons). All suspected cases of seizures were reviewed by two epileptologists. Decerebrate seizures, transient amnesia, and isolated changes in level of consciousness were not classified as seizures.
The following baseline characteristics were considered: sex, age, diabetes (previous diagnosis of diabetes and/or past or present use of antidiabetic agents or need of antidiabetic treatment on discharge), alcohol abuse (consumption of >400 ml/week pure ethanol), Glasgow Coma Scale (GCS) on admission, and features of ICH (location, cortical involvement, volume, and presence/absence of ventricular spread, midline shift, and early hydrocephalus). ICH was classified as deep (putamen, caudate nucleus, internal capsule, and thalamus), deep with lobar extension, and lobar. The volume of ICH was calculated by using the method suggested by Kothari et al. (3). For the purposes of this study, the following clinical events, occurring during hospitalization, were considered: surgical evacuation of the hematoma, neurologic deterioration due to CT scan–proven rebleeding, brain ischemia, hydrocephalus, and brain edema. Seizures occurring soon after clinical events were classified as associated with these events. The prophylactic administration of AEDs to patients who had not had a seizure was based on the judgment of the responsible clinician.
Patients with post-ICH seizures and who survived the immediate event were followed up by clinical examinations and structured telephone interviews. All records regarding to subsequent hospitalization and death were reviewed, and relevant events occurring during follow-up, including changes in AED therapy, were recorded. In patients with seizures during follow-up, CT scan was performed after the seizure had occurred.
For the purposes of the analysis, continuous and ordinal variables were transformed into dichomotous variables based on clinically meaningful subdivisions as follows: age (older vs. younger than 65 years), hematoma volume (≤18 vs. >18 ml), and admission blood glucose (≤160 vs. <160 mg/dl). Univariate logistic regression analysis and the Cox regression model were used to determine risk factors for immediate and early seizures. Independent predictors of immediate seizures were assessed by using stepwise logistic regression analysis. To assess predisposing factors of early seizures, baseline characteristics were treated as fixed covariates, and variables that developed in time (clinical events) were treated as time-dependent covariates; analysis was performed by using the stepwise Cox regression model. The influence of prophylactic AED therapy on the occurrence of early seizures was assessed by the same model. Because the decision to give AEDs was influenced by many factors, the main one being the location of the hematoma, the analysis was stratified according to ICH location, and all confounders were included in the multivariate analysis. The impact of seizures on outcome was assessed separately for immediate and early seizures by using the Cox proportional hazard model, which included traditionally accepted determinants of prognosis, such as age, volume of hemorrhage, GCS on admission, and cerebral complications. The Kaplan–Meier life table method was used to determine cumulative probabilities of developing seizures and recurrent seizures. The log-rank test was used in the comparison of survival experiences. When applicable, the odds ratio (OR) and 95% confidence interval (95% CI) were indicated.
- Top of page
- PATIENTS AND METHODS
Of the 761 patients (mean age, 65.6 ± 12.4 years), 300 (39.4%) had deep ICH; 139 (18.3%), deep ICH with lobar extension; and 322 (42.3%), lobar ICH. ICH was without known precipitating causes other than arterial hypertension in 505 patients; 48 patients had cerebral amyloid angiopathy, and 44 had hemorrhagic diatheses resulting from coagulopathy or anticoagulant therapy. Etiology was unknown in 164.
Forty-six patients had strokes before the index ICH. The incidence of immediate seizures was not significantly different in patients with (4.3%) and those without (4.2%) prior stroke (p = 0.69), nor was the incidence of early seizures (2.3 vs. 3.5%; p = 0.99). The results did not change when these patients were excluded.
Ninety excluded patients had infratentorial ICH, but none of them had seizures.
Seizures at onset of ICH
Thirty-two (4.2%) patients had their initial seizure at the onset of the hemorrhage (n = 14; 43.8%) or within 24 h (n = 18; 56.3%); 18 of them had a single seizure, and 14 had repetitive seizures. Twenty (62.5%) patients had simple partial seizures (nine with secondary generalization), nine (28.1%) had generalized seizures, one had simple partial SE, and two had generalized SE. Patients with immediate seizures more often had lobar ICH and cortical involvement than did patients without seizures. The volume of hematoma was smaller and intraventricular spread less frequent in patients with than in those without seizures. Clinical events (in the 18 patients who had seizures after ICH onset) were observed in two patients: one had seizures after surgical evacuation of hematoma, and the other had seizures after rebleeding, but the frequency of these events was not significantly different from that observed in patients without seizures (Table 1). Multivariate analysis showed that lobar location of ICH was the most powerful predictor of immediate seizures (OR, 4.05; 95% CI, 3.45–4.65; p < 0.0000). The volume of ICH had some predictive value because seizures were less frequent in patients with larger hematomas (OR, 0.60; 95% CI, 0.22–0.98; p = 0.0092; Table 1).
Table 1. Characteristics of patients with and without seizures within 24 hours of intracerebral hemorrhage
| ||No seizures (n = 729)||Seizures (n = 32)||Univariate OR (95% CI)||Multivariate OR (95% CI)|
|Baseline characteristics|| || || || |
| Sex, male||458 (62.8)||19 (59.4)||0.93 (0.57–1.29)|| |
| Age >65 yr||490 (67.2)||21 (65.6)||0.96 (0.59–1.34)|| |
| Alcohol abuse||131 (18.2)||6 (18.8)||1.01 (0.56–1.47)|| |
| Diabetes||94 (12.9)||1 (3.1)||0.46 (0.00–1.47)|| |
| Blood glucose >160 mg/dl||178 (24.5)||4 (12.5)||0.66 (0.13–1.19)|| |
| Lobar location||293 (40.2)||29 (90.6)||3.77 (3.17–4.36)||4.05 (3.45–4.65)|
| Cortical involvement||241 (33.1)||25 (78.1)||2.69 (2.26–3.11)|| |
| Hematoma volume >18 ml||380 (52.1)||11 (34.4)||0.69 (0.32–1.06)||0.60 (0.22–0.98)|
| Intraventricular spread||311 (42.7)||7 (21.9)||0.61 (0.19–0.85)|| |
| Midline shift||394 (54.0)||13 (40.6)||0.76 (0.40–1.47)|| |
| Early hydrocephalus||88 (12.1)||4 (12.5)||1.02 (0.48–1.55)|| |
|Clinical events|| || || || |
| Surgical treatmenta||59 (8.1)||1 (5.6)||0.82 (0.00–1.83)|| |
| Neurologic complicationsa||12 (1.6)||1 (5.6)||0.99 (0.98–1.00)|| |
Of the 32 patients with immediate seizures, only one patient had recurrent seizures in the next 29 days. This patient had recurrence after rebleeding, 8 days after ICH onset.
Early post-ICH seizures
Among patients without immediate seizures, 650 survived the first day after ICH, and 25 (3.8%) had one or more seizures in the following 29 days. Sixteen patients had a single seizure, and nine had repetitive seizures. Seven (28%) patients had simple partial seizures, 13 (52%) had generalized seizures, three had simple partial SE, and two had generalized SE. Among baseline characteristics, only lobar location was significantly more frequent in patients with than in those without seizures. Clinical events preceding seizures were observed in 15 (72.2%) patients and included surgical evacuation of the hematoma in four patients, rebleeding in five, brain ischemia in one, and hydrocephalus or brain edema in five. The rate of surgery was not significantly different among patients with and without seizures, whereas the frequency of neurologic complications was significantly higher in patients with seizures than without seizures (Table 2).
Table 2. Characteristics of patients with and without seizures within 30 days of intracerebral hemorrhage
| ||No seizures (n = 625)||Seizures (n = 25)||Univariate OR (95% CI)||Multivariate OR (95% CI)|
|Baseline characteristics|| || || || |
| Sex, male||389 (62.2)||18 (72.0)||1.24 (0.80–1.93)|| |
| Age >65 yr||411 (65.8)||20 (80.0)||1.48 (0.89–2.36)|| |
| Alcohol abuse||116 (18.6)||4 (16.0)||0.90 (0.53–1.54)|| |
| Diabetes||80 (12.8)||6 (24.0)||1.44 (0.91–2.28)|| |
| Blood glucose >160 mg/dl||130 (20.8)||9 (36.0)||1.55 (1.03–2.34)|| |
| Lobar location||247 (39.5)||21 (84.0)||2.75 (1.61–4.69)||2.80 (1.63–4.82)|
| Cortical involvement||209 (33.4)||13 (52.0)||1.47 (0.99–2.17)|| |
| Hematoma volume >18 ml||292 (46.8)||16 (64.0)||1.51 (1.00–2.27)|| |
| Intraventricular spread||232 (37.1)||10 (40.0)||1.13 (0.76–1.69)|| |
| Midline shift||303 (48.5)||16 (64.0)||1.43 (0.95–2.15)|| |
| Early hydrocephalus||72 (11.5)||3 (12.0)||1.05 (0.57–1.91)|| |
| Prophylactic AED therapy||410 (65.6)||13 (52.0)||0.79 (0.53–1.17)||0.58 (0.39–0.87)|
|Clinical events|| || || || |
| Surgical treatment||100 (16.0)||4 (16.0)||0.98 (0.57–1.67)|| |
| Neurologic complications||87 (13.9)||11 (44.0)||2.22 (1.49–3.29)||2.30 (1.46–3.31)|
Results of multivariate analysis of the impact of fixed and time-dependent covariates on the occurrence of early seizures after ICH are shown in Table 2. The risk of early seizures was increased by lobar location of ICH (OR, 2.80; 95% CI, 1.63–4.82; p = 0.0002), and by neurologic complications (OR, 2.20; 95% CI, 1.46–3.31; p = 0.0002), and was reduced by prophylactic AED therapy (OR, 0.58; 95% CI, 0.39–0.87; p = 0.0087).
SE as a first manifestation of seizure occurred in eight patients, constituting 1.1% of the entire cohort, and 14% of seizure patients. Comparison of seizure patients with and without SE showed that SE occurred exclusively in patients with lobar ICH and was more frequent in diabetics (37.5 vs. 8.2%), alcohol abusers (37.5 vs. 14.3%), and patients with elevated blood glucose on admission (50 vs. 18.4%). However, multivariate analysis indicated alcohol abuse as the only independent predictor of SE (sex- and age-adjusted OR, 3.37; 95% CI, 1.18–9.65; p = 0.023).
Seizures and outcome
Sixteen (50%) patients with immediate seizures and 16 (64%) patients with early seizures died during hospitalization. Immediate seizures (OR, 1.08; 95% CI, 0.84–1.40, p = 0.53) and early seizures (OR, 1.01; 95% CI, 0.77–1.33; p = 0.94) were not independent predictors of in-hospital mortality. Similar results were found when immediate and early seizures were analyzed together and when 30-day mortality was analyzed. The survival experience among seizure patients with and without SE was not significantly different (log rank test, p = 0.89).
Risk of seizure and recurrence
The cumulative actuarial risks of experiencing a seizure or a recurrent seizure are reported in Table 3. The probability of post-ICH seizure was 7.2% (95% CI, 6.2–8.2) within 5 days and 8.1% (95% CI, 7.8–10.3) within 30 days. The crude incidence rates of seizures among different services was 8.1% for neurology, 7.5% for neurosurgery, and 5.1% for intensive care. Among services, actuarial analysis showed no significant differences in 30-day cumulative incidence of seizures (log rank test, p = 0.64).
Table 3. Cumulative actuarial risks (95% confidence interval) of experiencing seizures and recurrent seizures after intracerebral hemorrhage
|Time after ICH (days)||Immediate and early seizures||Time after ICH (yr)||Recurrent seizures|
|1||4.4 (3.7–5.2)||1||5.3 (1.6–8.9)|
|5||7.2 (6.2–8.2)||2||5.3 (1.6–8.9)|
|10||7.6 (6.6–8.6)||3||19.3 (9.6–29.0)|
|15||8.0 (7.1–9.1)||4||27.0 (15.6–38.4)|
|30||8.1 (7.8–10.3)||5||27.0 (15.6–38.4)|
The 26 surviving patients with immediate or early seizures were followed up for a mean period of 59 months. Relapse was observed in six patients (three with initial immediate seizures and three with early seizures). Recurrent seizures were associated with clinical events in five patients, including brain infarct, hematoma enlargement (one patient each), and sudden suspension of AEDs in three patients. The actuarial risk of relapse after an initial seizure was 5.3% (95% CI, 1.6–8.9) in the first year and 27% (95% CI, 15.6–38.4) within 5 years. There were too few cases of relapse to allow meaningful analysis of risk factors.
Antiepileptic drug therapy
Of the 650 patients without immediate seizures who survived the first day, 423 (65.1%) received phenobarbital (PB) soon after ICH onset, and 227 (34.9%) were never treated or received AEDs after seizure manifestation. The influence of prophylactic AED treatment on the occurrence of early seizures was evaluated in subgroups of patients divided according to the location of ICH (deep ICH, lobar ICH, and ICH with any lobar involvement). Comparison of risk factors and personal attributes among groups showed that (a) treated patients with deep ICH were more often younger than 65 years (46.6 vs. 23.6%; p < 0.001), and more often had a larger hematoma (61.1 vs. 24.8%; p < 0.001) and midline shift (54.3 vs. 34.2%; p < 0.001) than did untreated patients; (b) among patients with lobar ICH, only age was significantly different between treated and untreated patients: treated patients were more often younger than 65 years than were untreated patients (34.7 vs. 12.1%; p < 0.001), and the same was found when patients with any lobar involvement were analyzed (36.9 vs. 19.8%; p = 0.004). Prophylactic AED treatment did not seem to modify the risk of early seizures in patients with deep ICH and deep ICH with lobar extension, but risk was significantly reduced by treatment in patients with lobar ICH (OR, 0.62; 95% CI, 0.40–0.96; p = 0.033; Table 4).
Table 4. Impact of prophylactic AED therapy on occurrence of early seizures in relation to hematoma location
| ||Untreated||Treated|| || |
|ICH location||No.||With seizures||No.||With seizures||Crude risk (95% CI)||Adjusted risk (95% CI)|
|Deep (n = 382)||161||3 (1.9)||221||1 (0.5)||0.51 (0.16–1.58)||0.35 (0.13–1.10)|
|Lobar (n = 268)||66||9 (13.6)||202||12 (5.9)||0.59 (0.38–0.92)||0.62 (0.40–0.96)|
|Any lobar involvement (n = 362)||91||11 (12.1)||271||13 (4.8)||0.64 (0.43–0.95)||0.57 (0.38–0.86)|
|Any location (n = 650)||227||12 (5.3)||423||13 (3.1)||0.79 (0.53–1.17)||0.62 (0.41–0.94)|
- Top of page
- PATIENTS AND METHODS
Our finding that 4.2% of patients with supratentorial parenchymal ICH had immediate seizures is less than the percentage (10–17%) reported in some previous studies (4–7), but lower percentages of immediate seizures (from 1.4 to 3.5%) also have been reported (8–10). Referral bias and different selection criteria may explain this variability.
Actuarial analysis has rarely been used to determine cumulative probabilities of developing seizures. Despite the different types of stroke considered, many authors agree that the risk of experiencing seizures is higher in the first days (5,11,12). In our cohort, the cumulative actuarial risk of experiencing seizures was 7.2% in the first 5 days and 8.1% within 30 days, thus confirming these observations.
Irrespective of the latency of seizures, 60% of patients had a single seizure as initial seizure manifestation, and the others had more than one closely spaced seizure, in line with previous observations (4,13). SE as the initial manifestation of seizure occurred in 1.1% of the entire cohort and 14% of seizure patients. This compares with 8.5–14% reported for seizure patients in previous studies (5,8,14,15).
Several demographic, clinical, and imaging features have been evaluated as possible predictors of seizures. Our study clearly demonstrated that predictors are different, depending on the latency of seizures. Immediate seizures were exclusively correlated with characteristics of ICH (lobar location and small volume). Lobar location has been widely recognized as the most potent predictor of immediate seizures (4–6,8,12,13,16–18). Some authors have reported an association between seizures and large lesions; however, these results are based on analysis of patients with ischemic and hemorrhagic stroke (13,19–21). Authors who studied only patients with ICH have not found this relation (4) or have observed that patients with seizures had smaller hemorrhages than patients without seizures (5,6), as in the present study. Our finding that patients with small hematomas were more prone to immediate seizures than were patients with large hematomas may be explained by various hypotheses. However, in our cohort, we found that lobar hematomas were more often small (54%) than deep (41%; p < 0.001) and that small lobar hematomas more often involved the cortex (85%) than did large lobar hematomas (68%; p < 0.001). This may be related to the etiology of ICH and vessel anatomy.
Lobar location of hemorrhage continues to be an independent predictor of early seizures, indicating that patients with lobar ICH do have a significant continuing predisposition for seizures, but neurologic complications also contribute significantly to the occurrence of early seizures. The occurrence of SE seems to be influenced by nonlesional factors (alcohol abuse). The relations between alcohol and seizures are complex; however, both alcohol abuse and alcohol withdrawal may precipitate seizures (22). In our cohort, alcohol abuse seemed not to favor seizures but only their clustering. Clustering of seizures is a known feature of alcohol withdrawal–related seizures (22).
No systematic analysis of clinical events associated with seizures has been reported, and only occasional reference to the question may be found in the literature. Our findings suggest that with increasing latency of seizures after ICH, an increasing proportion of seizure manifestations are associated with events that are in themselves potentially epileptogenic, or at least favor seizures. With regard to early seizures, surgical evacuation of the hematoma does not seem to favor them, whereas rebleeding, brain ischemia, and other neurologic complications were significantly more frequent in patients with than in those without seizures. Rebleeding as a prognostic factor for early seizures has occasionally been reported in studies of patients with ICH, whereas this event has been found to be a major predictor of seizures in patients with subarachnoid hemorrhage (23).
It is not clear whether seizures per se worsen stroke prognosis. Assessment of outcome has been attempted in some studies, but all types of strokes were often analyzed together without discrimination between immediate and early seizures (9,13,20,21,24). Some authors (21,24) but not others (9,10,13,20) suggested that the occurrence of seizures after stroke may worsen clinical outcome. As far as ICH is concerned, our results suggest that immediate seizures do not affect mortality, which is in line with the fact that immediate seizures manifest principally in lobar ICH without intraventricular spread, and are smaller, which means a better prognosis. Nor were early seizures an independent predictor of mortality, because many of these seizures were associated with neurologic complications of negative prognostic significance. Consistent with previous studies (10,14,15) dealing specifically with poststroke SE, we found no significant difference in mortality rate between seizure patients with and without SE.
Some authors have suggested that immediate or early seizures are associated with a significant risk of recurrence (25); others have observed that early seizures do not predict late seizures (5), and others have found that patients with early seizures are less likely to develop recurrent seizures than are patients with late seizures (6,8,16,26). The reported crude estimates of recurrence rates have varied from 28 to 93%, depending on whether immediate, early, or late seizures were considered and whether repetitive seizures or clusters of seizures were considered as a single seizure episode or as recurrent seizures (8,12,16,25,26). In our study, the crude rate of relapse was 11% in a mean of 59 months of follow-up, and the cumulative actuarial risk of relapse was 27% within 5 years.
The utility of prophylactic AED therapy in the prevention of seizures in patients with ICH remains unclear (4,5). Because our study was observational and not a randomized trial, the findings regarding AED treatment must be interpreted as an evaluation of treatment as used in clinical practice. Our findings suggest that if AED therapy is begun immediately after onset of ICH and is maintained through the acute and subacute phases, it may be effective in preventing early seizures in patients at higher risk (i.e., those with lobar ICH).
In conclusion, although patients with ICH are at substantial risk for seizures, the prognosis in terms of short-term mortality and subsequent development of epilepsy seems better than previously thought. The likelihood of immediate seizures is influenced by predisposing factors that are inherent characteristics of ICH, whereas the chance of developing early seizures is influenced not only by certain characteristics of ICH, but also to a large extent by unpredictable events that may themselves induce seizures, or at least favor them. When begun soon after ICH onset, prophylactic AED therapy may reduce the risk of early seizures in patients with lobar hemorrhage.