Epilepsia. 33(Suppl. 4):S15-S25, 1992 Raven Press, Ltd., New York Q lnlernalional League Againsl Epilepsy

Status Epilepticus in Children, Adults, and the Elderly *?Robert J. DeLorenzo, *Alan R. Towne, *$John M. Pellock, and IDaijin KO Departments of *Neurology, tPharmacology and Toxicology, $Pediatrics and §Bioslatistics, Medical College of Virginia, Virginia Commonwealth University, Richmond, Virginia, U.S.A.

Summary: Status epilepticus (SE)is a major neurological and medical emergency associated with a high morbidity and mortality. Retrospective and prospective studies from the Medical College of Virginia Epilepsy Research Center have been utilized to investigate several clinical and epidemiological factors associated with SE.Univariate and multivariate logistic regression analysis of predictive indicatorsin patients with SE indicated that seizure duration, certain specific etiologies, and age were predictors of mortality. Sex, race, and certain other etiologies were not found to be factors that Significantly affect mortality. These findings demonstrate that specific indicators are important in predicting mortality in

Status epilepticus (SE)is a major neurological and medical emergency requiring prompt treatment to prevent significant brain damage and possible mortality (Oxbury and Whitty, 1971; Celesia, 1983; DeLorenzo et al, 1987; DeLorenzo, 1990ab). Acute management of SE has been reviewed (Delgado-Escueta et al., 1982, 1983; Leppik, 1987; DeLorenzo, 1990a,b; Treiman, 1990),and standard protocols for treatment have been recommended. Despite improved treatment techniques, SE is still assoCiated with a significant morbidity and mortality that can only be diminished by developing a better understanding of the pathophysiology of this condition and more effective treatments. SE is a difficult condition to study because of its clinical presentation. Although individuals with epilepsy comprise a large group at risk for developing SE, SE most frequently occurs in the context of an acute neurological or systemic disease (Hauser, 1990). Until the present study, no prospective population-based studies of SE have been published. The large series of clinical studies previously published on SE (Table 1)

Address comspondence and reprint requests to Dr. Robert J. DeLorenzo at the Department of Neurology, Box 599, MCV Station, Richmond,VA 23298, U.S.A.

S15

SE. Preliminary prospective epidemiological data from Richmond, Virginia suggested that more than 250,000 cases of SE may occur annually in the United States with a mortality greater than 55,OOO individuals. Studies comparing community and university hospitals in Richmond indicate that the mortality and clinical variables associated with SE in these two hospital populations are essentially identical. Further investigations employing prospective populationbased studies are essential in evaluating the epidemiology and mortality of SE. Key Words: Status epilepticus-Cause

of death-Child-Adult-Aged-Incidence.

are based primarily on large hospital populations in major tertiary care medical centers. Thus, the actual incidence and clinical presentation of SE may not be reflected accurately by these populations. Addressing the need for improving existing data bases for studying SE,the Medical College of Virginia (MCV)Epilepsy Research Center has organized a relational data base to study SE. This article summarizes some of the data from this research effort and presents both retrospective and prospective population-based data from the community of Richmond, Virginia. DEFINITION OF SE

SE has been defined as “a condition characterized by an epileptic seizure that is sufficiently prolonged or repeated at sufficiently brief intervals so as to produce an unvarying and enduring epileptic condition” in the Dictionary of Epilepsy, published by the World Health Organization (Gastaut, 1973). This initial definition is rather imprecise and allows an overly wide latitude for the inclusion of numerous casesthat would be excluded by a more precise definition of SE. More recently, investigators have defined SE according to the duration ofthe seizures. The current, widely accepted definition of SE in the International Classification of Epileptic Seizures, developed under the leadership of Gastaut

S16

R. J. DELORENZO ET AL. TABLE 1. Mortality of status epilepticus in several clinical series Author

Year

Browne Binswanger Lorenz Gowen Clark and R o u t Turner Hunter Janz Dreyfus-Brisac and Monod Brett Aicardi and Chevrie Rowan and Scott Oxbury and Whitty Heintel Bladin et al. Aminoff and Simon Hauser Barry and Hauser Cclaia et al. Dunn Maytal et al. DeLorenm et al. Towne et al.

1873 1886 1890 1901 1904

I907 1959 1963 I964 I966 I970 1970 1971 1972 1917 1980 1983 1983 1988 I988 1989 1987 1992

Mortality (%)

32 50

45 18 33 20 23 7 53 18 11

21 45 25 30

16 6 32 25 8 3 18 25

( 1 969, 1983), describes SE as “any seizure lasting for

a duration of at least 30 minutes or intermittent seizures lasting for 30 minutes or longer from which the patient does not regain consciousness.” The MCV/Virginia Commonwealth University (VCU) epilepsy research project has utilized this international description of SE to maintain a data base consistent with other studies in the recent literature. Further subclassification of SE based on seizure type can also be provided, but the duration of the seizures represents the main criterion for the definition.

MORTALITY IN SE SE has been recognized as having a significant morbidity and mortality. Mortality rates as high as 50% have been reported, but more recent studies have yielded mortality estimates of 8-3296 (Hauser, 1983, 1990). In addition to mortality, other common sequelae of SE include permanent neurological deficits, intellectual dysfunction, and continuing recurrent seizures (Celesia, 1983). Table 1 summarizes the findings of major studies, regarding mortality in patients with SE. Prior to the 196Os, the mortality in SE approached 305W in many studies. More recent studies have indicated a lower mortality. However, no study in adults has shown a mortality of less than 7-10%. In addition, the studies of Maytal et al. (1989) indicate that children may have a significantly lower mortality than previously observed. Many explanations account for the wide variability in mortality in patients with SE reEpilepsia. Vol. 33, Suppl. 4, 1992

ported in these large series (Hauser, 1990). Some of the variability is related to the distribution of underlying etiologies of disease as well as to the duration of patient follow-up. However, all series with SE show a significant mortality. Even a mortality of 8-10% in adults is a significant medical problem that needs critical attention. The morbidity and mortality from SE has been suggested to be related to three major factors (Leppik, 1990): (i) damage to the central nervous system (CNS) caused by the initial insult causing SE; (ii) systemic stress from r e p t e d convulsions; and, (iii) direct injury to the brain from severe electrical discharges within the CNS. Neuropathological studies in animals by Meldrum and others have indicated that extensive electrical activity within the central nervous system can cause irreversible neuronal damage (Meldrum and Brierley, 1973; Meldrum et al., 1974; Meldrum, 1982). These collective studies suggest that seizures that last for greater than 60 min may produce significant neurological injury that may not be reversible. Meldrum ( 1982) and Simon (I 985) have provided data indicating that seizures exceeding 60 min in duration can produce neuronal damage. These observations support initial findings in human studies from this research center that duration of seizures during SE is directly related to mortality (DeLorenzo et al., 1987; Towne et al., 1987). Thus, seizure duration may be an important predictive indicator of mortality in SE. Several other clinical parameters have also been associated with mortality in SE. Aicardi and Chevrie (1983) have suggested that prognosis depends on the interval of time allowed to elapse between the onset of SE and start of effective treatment. In their study of 98 patients, Aminoff and Simon ( 1 980) provide data to suggest that seizureduration and initiation of treatment are both related to outcome in SE. It has also been observed that adverse outcomes following SE may be attributable to the underlying neurological or systemic cause of the seizures (Janz, 196 1). The relationship between the cause of SE and its resultant mortality has been hrther evaluated by recent studies (Terrence et al., 1981; Hauser, 1990). The role of an identifiable C N S process in association with SE has also been associated with increased mortality (Leppik, 1985, 1987). In the retrospective data analyzed in this report,several clinical variables are evaluated as predictive indicators for outcome in SE. MATERIALS AND METHODS Retrospective data base Adult and pediatric cases of SE at the MCV/VCU Medical Center and McGuire Veterans Medical Center

SE IN CHILDREN, ADULTS, AND THE ELDERLY

were collected from 1982 to 1989. A total of 546 patients were identified, including 375 adults and 171 children. Patients were identified by using the MCV computerized discharge data base for ICDM-9-CM (International Classification of Diseases, 9th revision, Clinical Modifications; Codes 345-345.9, 780.3, and 779.0). Following the identification of the appropriate patient charts, each chart was reviewed to determine if the case met the definition of SE in the International Classification of Epileptic Seizures that has been described above. Based on the number of SE patients from this initial chart review, it became apparent that there was still a significant undercounting of patients that actually had SE in the MCV/VCU Medical Center during this time period. To expand our search for SE patients, every emergency room department record of a seizure presentation with subsequent admission to the hospital was reviewed. In the period from 1982 to 1989, a p proximately 10% of these available records contained information that was consistent with the diagnosis of SE. Less than 4% of the reviewed charts did not contain information on seizure duration. In addition, EEG reports during the same time period were screened to identify cases of SE electrographically. By reviewing emergency room records and EEG reports, additional cases of SE were added to the retrospective data base. These sources were especially useful in identifying patients not associated with the neurology service but with the general medical and surgical services in the hospital. To further identify patients on other services, consultation records of the neurology department in the same time period were reviewed. Utilizing this technique, we identified a significant proportion of the SE patients presenting to the medical center from 1982 to 1989. However, based on our more recent prospective data described, it was clear that the total number was still an underestimation of total patients presenting with SE in the institution. Nevertheless, by combining the data from MCV’s computer-coded ICD-9-CM records, the emergency room case data, EEG reports and consult service information, we obtained a significant number of cases of SE in the MCV/VCU and McGuire Veterans Medical Center hospital complex population consisting of 1,872 beds. Charts of both adult and pediatric cases were reviewed by research personnel. Data were manually and computationally checked for completion as well as consistency. Any discrepancies were corrected by refemng to the original patient charts. Patients in whom the seizure duration could not be clearly documented were excluded from the study and represented approximately 2-3% of the patient population reviewed. Using standard etiological definitions, etiologies of SE were

S17

determined as the immediate precipitating cause of SE or as related etiologies when no acute cause was determined. Previous medical conditions that were not acute and not directly the cause of SE were not included in the determination of SE. Some cases of SE were caused by more than one precipitating etiology. Children represented cases presenting from l month to 16 (yearsof age. The adult population represented patients older than 16 years. Statistical methodology included both univariate and multivariate logistic regression analyses (Santner and Duffey, 1989). X 2 or Fisher’s exact test for dichotomous variables and a Student’s t test for continuous variables were used as univariate tests. Multivariate logistic regression analysis was employed to examine the simultaneous association between several factors and survival. Predictive indicators of outcome were evaluated using multivariate regression analysis. Data were analyzed using SAS version 5 . RESULTS Mortality in SE Mortality in the retrospective study for children and adults is presented in Fig. 1. This data represent a direct comparison of the mortality among adults and children with SE who were in the same hospital population and treated by the same neurological staff and standard treatment procedures. Pediatric and adult neurology services are both in the same department at MCV. The results demonstrate that, in the period from 1982 to 1989, children with SE had a significantly lower mortality (2.3%) than adults with SE (25%). These findings confirm the observation of Maytal et al. (1989). This study utilizes for the first time a combined adult and pediatric data base studied in the same institution over the same time period. Age distribution and mortality in SE The age distribution of SE cases is presented in Fig. 2. The number of patients presenting at various age groups from birth to greater than 80 years of age demonstrate a bimodal distribution. The number of cases of SE had a slight peak at 16 years of age (n = 375). corrgent, congentialcauses; Drugs, drug overdose; CVA. cerebrovascular accidents; Hem, hemxhages; Infec, systemic infections; CNS Inf., central nervous system infections; Metab, metabolism; DAED, decrease in antieplleptic drugs; ETOH, alcohol withdrawal; Idiopath, idiopathic.

30

20

10 0

Etiology Epilepsiu Vd.33. Sup&. 4, 1992

(2.5%), and tumors, although rarely presenting with

SE IN CHILDREN, ADULTS, AND THE ELDERLY 100

................._..

h

s21

....................................... ..............................................

----'

FIG. 8 Survival curves for prolonged (sdidline) and non-prolonged (dotted line) seizure duration.

Thedataere preserlted 89 percent survival based on a 3-y follow-up period.

0

5

10

15

20

30

25

Days

excluding recurrent cases of SE (Fig. 8). The survival curves for prolonged and non-prolonged SE differed significantly. The prolonged seizure group had a mortality rate of 34.8%,whereas the nonprolonged SE group had a 3.7% mortality rate within 1 month. A generalized Wilcoxon test showed that the mortality distribution function of prolonged SE was significantly different from the survival distribution of nonprolonged SE (p < 0.001). This analysis indicates that statistical comparison of the two survival curves demonstrates a very significant difference between the prolonged and nonprolonged seizure groups. Using univariate analysis, the data indicate that seizure duration significantly correlates with mortality. Multiregression analysis of predictive indicators in SE Utilizing a retrospective data base fiom 1982 to 1986 and univariate multiregression analysis, several clinical variables were found to be significantly correlated with mortality (Towne et al., 1993). Seizure duration of less than 1 h was assOciated with a 2.7% mortality and more than 1 h with a 32% mortality (Table 2). The odds ratio, using univariate multiregression analysis, was 17.1930 and was highly significant with a value ofp = O.OO0 1. Thus, by univanate analysis, seizure duration was highly correlated with outcome (Table 2). Certain specific etiologies were also significantlycorrelated with mortality. Table 2 indicates that alcohol withdrawal and lowered antiepileptic drugs showed a statistically significant association with a better outcome. Anoxia was significantly associated with a negative outcome with a 60% mortality. Age and race also showed a statistically significant correlation with mortality in univariate analysis. Other etiologies and sex showed no correlation with mortality. Other etiologies, although assoCiated with mortality, did not show a statistically significant correlation.

The utilization of univariate analysis described above indicated several variables had significant correlations with mortality. Seizure duration, specific etiologies, age, and race were shown by univariate statistical analysis to be significantly associated with mortality. The reTABLE 2 . Univariate analysis of risk factors for mortality Factor

Mortality (5%)

Odds ratio"

Seizure duration >I h vs.

32.0

17.1930'

c1 h

Etiology Drugs ETOH DAED Trauma Anoxia Hem lnfec Metab Tumor CVA Idiopathic Age (years) 16-30 30-60

60Race Black vs. Non-black Sex Male

vs. Female

2.7 16.7 8.3 8.8 20.0 60.0 38.5 30.8 31.0 36.4 26.3 19.4

0.65 17 0.2614' 0.2529' 0.8 158 6.6585 2.1527 1.4950 1.5660 1.5660 I .9064 0.8 105

12.5 15.2 32.3

(96 10 yrs)'

19.4

0.5436

'

1.4538'

30.7 22.2

1.1136

24.1

Drugs,drug overdose; ETOH, alcohol Withdrawal;DAED, decreasc in antiepileptic drugs; Hem, hemorrhages;Infec, infections; Metab, metabolism; CVA, cerebrovascular accidents. Estimated odds ratio in the univariate logistic models. Results of testing odds ratio = 1. It is based on corrtsponding asymptotic distribution theory. Significant factor (p = I h vs.

Status epilepticus in children, adults, and the elderly.

Status epilepticus (SE) is a major neurological and medical emergency associated with a high morbidity and mortality. Retrospective and prospective st...
866KB Sizes 0 Downloads 0 Views