SEARCH

SEARCH BY CITATION

Keywords:

  • Seizure severity;
  • Epilepsy;
  • Functional impairment

Abstract

  1. Top of page
  2. Abstract
  3. A HISTORY OF METHODS OF SEIZURE MEASUREMENT
  4. ADVANTAGES AND DISADVANTAGES OF SCALES
  5. COMPONENTS OF SEIZURE SEVERITY SCALES
  6. OVERALL LIMITATIONS
  7. FUTURE DEVELOPMENTS
  8. REFERENCES

Summary: Quantitative assessment of seizure severity has been approached using a variety of systems. This review describes currently available methods and possible new approaches to seizure assessment for clinical trials. A review of the literature on methods of seizure assessments resulted in tabulation of the seizure rating scales known as VA, Chalfont-National Hospital, Liverpool, Hague, and the Occupational Hazard Scale. Seizures have been evaluated by simply counting all events, counting events by type, by clinician ratings, patient ratings, and combinations. Each of the scales has advantages and disadvantages. Most scales share core components: seizure frequency, seizure type, seizure duration, postictal events, postictal duration, automatisms, seizure clusters, known patterns, warnings, tongue biting, incontinence, injuries, and functional impairment. This review revealed a partial consensus about aspects of seizures that are important markers for severity. However, usefulness of the existing scales is limited by lack of data on responsiveness. New approaches are needed to assess changes in seizure severity as a result of an intervention in a clinical trial.

Epilepsy therapies have the potential to alter both seizure frequency and severity. Seizure type and seizure frequency have been the outcome assessment variables for most evaluations of therapy [e.g., antiepileptic drugs (AEDs), surgical intervention, vagal nerve stimulator] because these end points are readily measured. Investigators recognize that change in seizure severity would also be an important outcome to assess. It is not clear whether the currently available tools (seizure severity scales) are adequate measures of seizure severity. This report reviews the available methods of seizure-assessment use as outcome measures of new therapy.

A HISTORY OF METHODS OF SEIZURE MEASUREMENT

  1. Top of page
  2. Abstract
  3. A HISTORY OF METHODS OF SEIZURE MEASUREMENT
  4. ADVANTAGES AND DISADVANTAGES OF SCALES
  5. COMPONENTS OF SEIZURE SEVERITY SCALES
  6. OVERALL LIMITATIONS
  7. FUTURE DEVELOPMENTS
  8. REFERENCES

Seizure frequency: counting seizures

The basic unit of measurement in epilepsy is the frequency of seizures. Events have been reported in ordinal scales as the number of seizures, or in cardinal scales as few, many, fewer, or more seizures. Additionally, data have been transformed into percentage change from a baseline frequency or a rate based on frequency per specified time period (e.g., number per month). The standard end point in clinical trials has been the proportion of patients whose seizure frequency decreased by ≥50% from baseline. Other interpretations include the number of patients whose seizure frequency or rate decreased or reached a specified range (e.g., ≥50%, 100%), as well as the number of months with no seizures (1).

In early studies of AEDs, the end point was a simple summation of all types of seizures, with no regard for seizure type. Measuring only seizure frequency disregards the obvious differences in clinical manifestations among types of seizures. Although the International Classification of Seizures was not adopted until 1981 (2), there have been discussions about the clinical and electroencephalographic features of typical events and an appreciation of the differences among seizure types for many years.

Seizure frequency: assessment by seizure severity

In 1977, a new rating system was developed for the VA Cooperative Study to compare AED monotherapy for newly treated patients (3). The system assessed the frequency of events in partial seizures (secondarily generalized tonic–clonic, complex partial, simple partial) as well as by severity. The VA Seizure Frequency and Severity Scale (VA Scale) was completed by the clinician based on specific questions about seizure components performed by examination and interview (3). Seeking a simplification of the VA scale, the Chalfont Seizure Severity Scale was described in 1990 (4). It was updated and renamed the national Hospital Seizure Severity Scale (NHS3) (5). These scales were used for assessment of institutionalized patients. Janz (6) described an Occupational Hazard Scale that ascribed levels of hazard depending on the frequency and severity of seizures. General statements about frequency and duration as well as impairments caused by typical events define the six levels.

The next approach to seizure severity was a departure from the traditional, physician-based clinical assessment, turning instead toward patient-based assessment of severity (7). The Liverpool seizure severity questionnaire was developed with support from Burroughs-Wellcome for use in clinical trials of lamotrigine (LTG). It was designed to evaluate both the severity of ictal events and perceptions of seizures from the perspective of the patient. This concept was extended to the assessment of seizure severity in children by asking parents to complete a questionnaire (8). The parent-based Hague Seizure Severity Scale (HASS) was based largely on the Liverpool scale, with adaptation for parental responses (8).

Seizure severity versus health-related quality of life

Some drift occurred in attempts to combine objective, physiologic, and clinical aspects of seizures with subjective perceptions by patients and parents. The recent development of quality-of-life measures using subjective, patient-based opinion has changed the approach to seizure severity assessment, although the two types of scales measure different issues (9). Several reports blur these distinctions in discussions of outcome assessment using rating scales by comparing these two types of approaches with a focus on a patient-based quality of life (10,11). Unfortunately, this is not an appropriate comparison because objective and subjective approaches arise from different perspectives; neither can be considered equivalent, nor can one be considered optimal. This article reviews these approaches in depth, defining advantages and disadvantages, limitations, and opportunities.

Seizure severity versus functional status

Scales that assess the overall impact of epilepsy or seizures may focus on the frequency of events, on the type of seizure as an index of severity, or on the effect of epilepsy on daily function. In other medical disorders, assessment of functional status is the key to severity because it is a measure of the impact of the disorder on the patient's ability to perform daily activities. None of these definitions encompasses quality of life, which is a different construct. Functional status measures may be used to determine the level of handicap or disability. Inclusion of a variety of problems in the definition of degree of disability makes it far broader than a measure of seizure severity. Although this is an important categorization, a measure of disability is very different from a measure of seizure severity related to change in therapy.

Seizure severity scale responsiveness

A related issue is the design of a seizure severity scale so that it can detect change as a result of treatment (responsiveness). If a scale is not constructed with a balance of items appropriate to most types of seizures, the result could be that many patients have very low scores because the items do not reflect their events. An example is the NHS3 (5) that includes a series of questions about relatively infrequent aspects of seizures (generalization, falls, injury, incontinence) (NHS3 items 2 to 5, Table 2). If the patient does not experience those events, only two items (items 7 and 8) remain for estimation of a seizure severity score. This aspect of scale development could create a “floor effect” in which many scores are very low. Scale structure may also make it impossible to determine whether changes have occurred in seizure type, a common experience wherein a new drug diminishes seizure severity by preventing secondary generalization [more complex partial seizures (CPSs)] or preventing spread to altered consciousness [more simple partial seizures (SPSs)]. Although current scales evaluate seizure severity from a variety of approaches, few data exist demonstrating the sensitivity of any scale to treatment effect.

Table 2. National Hospital Seizure Severity Scale (NHS3)
1. Record the name of the seizure types that occur under headings “type 1, 2, 3 …” Since the last visit: Type 1Type 2Type 3
  1. From O'Donoghue et al. (5), with permission.

2. Does the patient have aYes444
2. generalized convulsion during this type of seizure?No000
3. How often has the patient fallen to the ground in this type of seizure?Nearly always or always Often Occasionally Never4 3 2 04 3 2 04 3 2 0
4. Has this type of seizure caused any of the following? (score only the worst)Burns, scalds, deep cuts, fractures Bitten tongue, severe headaches Milder injuries, mild headache No injuries4 3 2 04 3 2 04 3 2 0
5. How often has the patient been incontinent of urine in this type of seizure?Nearly always or always Often Occasionally Never4 3 2 04 3 2 04 3 2 0
6. If the seizure causes loss of consciousness, is there a warning long enough for the patient to protect him/herself? (no loss of consciousness or seizures only while asleep scores 0)Never Somtimes Nearly always or always2 1 02 1 02 1 0
7. How long is it until the patient is really back to normal after the seizure?Less than 1 min Between 1 and 10 min Between 10 min and 1 h Between 1 and 3 h More than 3 h0 1 2 3 40 1 2 3 40 1 2 3 4
8. Do the following events occur in this type of seizure?Seriously disruptive automatisms (e.g., shouting, wandering, undressing) Mild automatisms or focal jerking None4     2 04     2 04     2 0
Add 1 point to each column  1  1  1 
     
Total score for each seizure typea
1) Define how many different types of seizures occur (e.g., aura, complex partial, generalized convulsion …). Call these types 1–3 arbitrarily
2) Apply questions 2–8 to each seizure type separately. As the NHS3 indicates current seizure severity, define the time frame (e.g., 1–3 months or time since the last clinic visit). Use clinical judgment whether each factor occurs in the seizure type (i.e., the physician decides if there is a convulsion after questioning the patient). Allow the patient to judge the frequency of each event. Then tick the box opposite the response options. The number in the box is the score for that question
Note: Q.3 Only actual falls are recorded (i.e., if the seizures could cause falls but have not because they all occurred while in bed, then the score is 0)
Q.7 Refers to the time until the patient feels fully functional
Note the specific scoring instructions for Q 4 and 6.
a The column totals give the seizure severity score

Requirements for a seizure severity scale

The ideal seizure severity scale would be brief and easy to use by patients and physicians. It would have excellent psychometric properties of internal and external validity and stability. Most important, the scale would be sensitive to change after a change in treatment (e.g., drug, device, surgical). Scales that are shown to have significantly different scores between patients with mild and severe epilepsy are helpful. The key to usefulness in a clinical trial is the ability to demonstrate differences between baseline and follow-up scores. Following is a review of the properties of existing scales.

ADVANTAGES AND DISADVANTAGES OF SCALES

  1. Top of page
  2. Abstract
  3. A HISTORY OF METHODS OF SEIZURE MEASUREMENT
  4. ADVANTAGES AND DISADVANTAGES OF SCALES
  5. COMPONENTS OF SEIZURE SEVERITY SCALES
  6. OVERALL LIMITATIONS
  7. FUTURE DEVELOPMENTS
  8. REFERENCES

VA seizure frequency and severity scale:

The VA scale (3) does not meet the current standard for scale development. Attribution of points for the scoring system was based on estimates of the four developers that were accepted as reasonable by 15 reviewers (VA study investigators) (12). Item generation based on clinical experience evidently included most pertinent issues because later scales added few topics. The scale was used in two long-term clinical trials to assess more than 1,100 patients for as long as 5 years of follow-up and was developed for a study of newly treated patients who had few seizures. The VA scale has been criticized for what appears to be a complex scoring system, although scores for multiple seizures can be calculated. Use of the scale for frequent seizures can be complex. For instance, one question allows only a yes/no response to a question on whether seizure(s) were precipitated by unusual remediable factors such as sleep deprivation or alcohol use. In patients with frequent seizures, some seizures may have been precipitated by these factors, and others, not.

The construction of the VA scale is based on the frequency of the three types of partial-onset seizures (simple, complex, secondarily generalized) (Table 1). Scores for each type of seizure are generated for an interval. After calculating a score based on seizure type and frequency, the score is reduced by reviewing a series of modifying circumstances. If a helpful warning preceded a seizure, avoidable precipitant (e.g., lack of sleep), low drug levels, or nocturnal pattern, the score would be reduced for each factor. SPS and CPS scores were further modified if they did not significantly affect functioning. Modifications are reductions of 30–80% of the points assigned for frequency. Scores for SPSs, CPSs, and secondarily generalized seizures were combined to create a total seizure score for the interval assessed in the clinical trial. In this design, severity is defined as physiologic factors that reflect the spread of the seizure or affect seizure threshold.

Table 1. VA Seizure type and frequency rating (as revised for VA-2)
  1. From Cramer et al. (3,9), with permission.

Instructions: Record all seizures reported by patient since last visit. Use sections A, B, and C as necessary for each type of seizure reported.
Section ASection B
GENERALIZED TONIC–CLONIC SEIZURESCOMPLEX PARTIAL SEIZURES
Complete this section if patient has either primary or secondarily generalized tonic–clonic seizures.Complete this section if patient had complex partial seizures (with altered consciousness) that did not generalize.
1. Total number of seizures since last visit (or interval)  Total number of seizures since start drug  2. Select the seizure frequency category that applies to the patient from the following list and enter on score line. a. Three or more seizures/12 months = 20/each seizure (After 1 year, review past 12 months only for multiple seizures.) b. Two seizures/first 3 months =  50 c. Two seizures/6 months =  45 d. One seizure/first 3 months =  40 e. One seizure/6 months =  40 f. Two seizures/6–12 months =  30 g. One seizure/6–12 months =  20 h. Two seizures/12–24 months =  20 i. One seizure/12–24 months =  10      Score  3. Was seizure(s) modified by an aura (allowing avoidance of harm)? Y  N  If yes, change score to 80% of score in item 2; if no, keep score the same. Score  4. Was seizure(s) precipitated by unusual, remediable factors (e.g., lack of sleep, alcohol, illness)? Y  N  If yes, change score to 50% of score in item 3; if no, keep score the same. Score 5. a. If subtherapeutic drug levels, change score to 30% of score in item 4; if not, keep the score the same.     Score b. If low therapeutic drug levels, change score to 80% of score in item 4; if not, keep score the same.     Score 6. Are seizures modified by known cyclic or diurnal pattern (e.g., nocturnal or early a.m.) Y  N  If yes, change score to 60% of score in item 5; if no, keeps score the same.     Score1. Total number of seizures since last visit  2. Did seizures occur as a cluster (greater than two seizures within 24 hours? Y  N  a. If yes, how many seizures occurred? . b. If three or more seizures/cluster, count only half the number of seizures in that cluster. 3. Select the seizure frequency category that applies to the patient from the following list and enter on score line. a. Equal to or greater than 4/month (+10 for each additional seizure/mo over 4) = 50 (+10 each) b. Equal to 3/month =   40 c. Equal to 2/month =   30 d. Equal to 1/month =   20 e. Less than 1/mo & greater than 1/3 mo = 15 f. Less than or equal to 1/3 months = 10     Score  4. Was seizure(s) modified by an aura (allowing avoidance of harm)? Y  N . If yes, change score to 80% of score in item 3; if no, keep score the same.     Score  5. Was seizure(s) precipitated by unusual, remediable factors (e.g., lack of sleep, alcohol, illness)? Y  N . If yes, change score to 50% of score in item 4; if no, keep score the same.     Score 6. a. If subtherapeutic drug levels, change score to 30% of score in item 5; if not, keep score same. b. If low therapeutic drug levels, change score to 80% of score in item 5; if not, keep score same.    Score 7. Are seizures modified by known cyclic and diurnal patterns (e.g., all nocturnal or early a.m.)? Y  N  If yes, change score to 60% of score in item 6; if no, keep score the same. Score 8. Did seizure(s) cause loss of consciousness (include psychic, cognitive, and autonomic disturbances) or significant interference with functioning? Y  N . If no, change score to 50% of score in item 7; if yes, keep score same.     Score
  Generalized Tonic–Clonic:  COMPLEX PARTIAL SEIZURES:
 Section A Final Score   Section B Final Score 
  
Section C SIMPLE PARTIAL SEIZURES Complete this section if patient had simple partial seizures (without altered consciousness). 1. Total number of seizure since last visit  2. Did seizures occur as a cluster (greater than two seizures within 24 hours)? Y  N . a. If yes, how many seizures occurred? b. If three or more seizures/cluster, count only half the number of seizures in that cluster (e.g., six seizures/cluster = three seizures/week) = seizures/week 3. Select the seizure frequency category that applies to the patient from the following list and enter on score line. a. Equal to or greater than 7/month (+3 for each additional seizure/mo over 7) =  33 (+3 each) b. Equal to 6/month =  30 c. Equal to 5/month =  28 d. Equal to 4/month =  25 e. Equal to 3/month =  23 f. Equal to 2/month =  20 g. Equal to 1/month =  15 h. Less than 1/mo & greater than 1 per 3 mo =  13 i. Less than or equal 1 per 3 mo  10     Score4. Was seizure(s) precipitated by unusual, remediable factors (e.g., lack of sleep, alcohol, illness? Y  N . If yes, change score to 50% of score in item 3; if no, keep score the same.     Score 5. a. If subtherapeutic drug levels, change score to 30% of score in item 4; if not, keep score same. b. If low therapeutic drug levels, change score to 80% of score in item 4; if not, keep score same.     Score  6. Are seizures modified by known cyclic and diurnal patterns (e.g., all nocturnal or early a.m.)? Y  N . If yes, change score to 60% of score in item 5; if no, keep score the same.     Score  7. Did seizure(s) cause significant interference with functioning? Y  N . If no, change score to 50% of score in item 6; if yes, keep score same.     Score    SIMPLE PARTIAL SEIZURES:   Section C Final Score    SEIZURE TYPE, FREQUENCY, AND SEVERITY RATING (Total score is the sum of sections A, B, and C): Score A + Score B + Score C  = Total Score 
  

A Dutch group undertook an independent investigation of the validity of the combined VA seizure, systemic, and neurotoxicity scales (Composite Score) (3), renamed the Index of Impairment for this study (13). For interrater validation, two clinicians using the complete questionnaire interviewed patients (r = 0.90). The suitability of scoring was evaluated by clinicians who interviewed patients using questionnaires without scoring instructions that designated the severity of the ratings. Good correlation was found between scores and the length of time to the next appointment as an index of epilepsy severity (r = −0.66). Data were not reported for the seizure scale alone. Unfortunately, this study did not use adequate methods to test whether the scores were appropriate. The ultimate test of an instrument is sensitivity to differences between groups. In this case, the VA scale did well after revision. Finding no significant differences among carbamazepine (CBZ), phenobarbital (PB), phenytoin (PHT), and primidone (PRM) indicated lack of sensitivity of the scale (12). However, the developers noted that the frequency of CPSs was set far too high for this group of newly treated patients who had few seizures. [Lack of difference also could have reflected the general equivalence among these drugs.] During the second VA trial, the scoring was revised to reflect lower seizure frequency. Scale scores were significantly lower (i.e., better) for patients with CPSs taking CBZ than for patients taking valproate (VPA; 2.0 vs. 6.2; p = 0.04) (14). The scale score outcome was similar to the seizure count and monthly seizure rate outcomes, all of which showed significantly better results for CBZ, probably because the scale was based largely on seizure frequency. The combination of seizure frequency and severity in this scale may make it useful for an overall assessment, but not useful for an evaluation of changes in seizure severity only.

Chalfont–National Hospital seizure severity scale (NHS3)

The Chalfont scale was developed as a measure of clinical events (4). The questionnaire was administered by a clinician who interviewed the patient and by people who witnessed typical seizures. Items were created from patient interviews, and the anchors for a 5-point response were derived from patients and clinicians. Although the Chalfont scale showed some change in an open-label tiagabine (TGB) trial, it was not responsive to change in a later double-blind, placebo-controlled TGB trial. Based on data from the later trial, the authors realized that a 35-point change would have been needed to show a significant difference between groups. The scale was then revised to create the NHS3 (5)(Table 2). The new version starts with listing as many as three types of seizures (e.g., SPS, CPS, or “generalized convulsion”). This is similar to the VA scale except that the types are not predetermined, and all types are scored using the same system. The scale consists of seven questions that are posed for each type of seizure.

The first item asks if a generalized convulsion occurs with this type of seizure. One can immediately note that this structure allows reporting of what is essentially the same type of seizure more than once. For example, if the seizure often starts with a warning before altered consciousness, but rarely proceeds to tonic–clonic movements, those events are listed as CPSs. However, these same events may be listed again as secondarily generalized seizures. Both may receive 4 points because they are described as separate types. Another question asking if the patient falls to the ground also adds points to generalized tonic–clonic seizures (GTCSs). The item about other injuries and headaches includes burns, scalds, deep cuts, and fractures as major; tongue biting and severe headaches as less severe; as well as milder injuries and mild headaches. There is no option for individuals who may have all these elements. Only the most severe will determine the point attribution. Therefore if seizures no longer include severe postictal headaches, they will not be scored with reduced points if an injury occurred with the event. The lack of a frequency factor in the scoring could allow a high score for a rare injury. Even patients with very severe epilepsy will rarely have burns or cuts. These injuries may be more related to circumstances during a given seizure than to the severity of seizures overall. The inclusion of headaches is not explained or justified as part of the revision. Urinary incontinence is scored separately. The score is not increased if a warning is long enough for the patient to seek protection (similar to the VA scale). If there was no altered consciousness or seizures occurred only during sleep, the score is zero. The duration of time until the patient is “back to normal” after the seizure could be problematic because it is difficult for an observer to know when the patient resumes full physical and neurologic status. The final item asks about selected events: seriously disruptive automatisms (e.g., shouting, wandering, and undressing) are rated as severe; mild automatisms or focal jerking are given a moderate score, but no consideration is given for other automatisms that might be socially embarrassing. Inclusion of focal jerking as mild might be disputed. The questionnaire concludes with the addition of 1 point to the total score for each seizure type, as a modification to avoid assigning a score of zero severity to any seizure. Another problem with the scoring system is that each item is rated on a scale of 1 to 4 points. For example, falling “often” is scored 3, whereas an “occasional” fall is scored 2. Designation that one option is 33% worse than the other is arbitrary. Therefore changes as a result of drug therapy might be over- or underrated. This scale has been included in at least one clinical trial in which no change was seen. It would be interesting to gather data from any clinical trials in which the scale was used, even if it was not responsive to change. Such data would help the developers understand whether selected items were useful.

Another approach to overall assessment of the impact of epilepsy was the development of the Subjective Handicap of Epilepsy scale by O'Donoghue et al. (15). The scale is more like a quality-of-life scale than a seizure or epilepsy severity scale. It is a patient-based questionnaire with six domains: Work and activity; Social and personal life; Feelings about oneself; Physical; Life satisfaction; and Change. A number of questions relate to seizure severity and the impact of epilepsy on employment, driving, and other daily activities. It is not included in this review because it is not truly a seizure severity scale.

Occupational hazard scale

Janz (6) considered the taxonomy of seizure classification less important to the patient than the possible social impairment caused by a seizure (e.g., maintaining posture instead of falling reduces the risk of accidents). They used the scale to evaluate the suitability of patients for selected occupations based on the severity of their seizures. The seven-item scale (Table 3) gives examples of types of seizures and treatment status. Seizure-free patients who qualify for driver's licenses are least affected, and those patients whose seizures are frequent, include falling, or impair activity, are most impaired. Although the scale describes only a small number of possible examples, it can serve as a guide for grading occupational suitability or changes in patient status over time if seizure severity changes.

Table 3. Occupational Hazard Scale
Type of seizures and treatment statusEpilepsy
  1. From Janz (6), with permission.

1. Seizure free (allowed to drive motor vehicle)1
2a. Seizures only during sleep ≤1/mo2
2b. Unilateral jerks without loss of consciousness ≤1/wk2
3a. Seizures only during sleep ≤1/wk3
3b. Very brief seizures impairing or interrupting ongoing activity ≤1/six mo3
4a. Brief seizures impairing or interrupting ongoing activity ≤1/six mo4
4b. Unilateral jerks without loss of consciousness4
4c. Seizures with falling lasting <5 min including reorientation ≤1/6 mo4
5a. Seizures during which person does not behave according to the demands of the situation ≤1/mo5
5b. Seizures with falling lasting <15 min including reorientation ≤1/mo5
6a. Seizures impairing or interrupting ongoing activity6
6b. Seizures during which person does not behave according to the demands of the situation ≤1/wk6
6c. Seizures with falling lasting <15 min ≤1/wk6
Occupational hazard rating 

Liverpool seizure severity scale

The Liverpool group (7) developed a scale to assess patient-perceived seizure severity. The 19 items in the questionnaire (Table 4) are answered by the patient without interpretation by the health care provider to represent seizure activity during the previous 4 weeks. A drawback to this method is that patients often are the least knowledgeable about what their seizures are like compared with reliable witnesses who can describe the seizure and postictal state. The scale is divided into two sections, one reviewing perceptions of control and the other reviewing ictal and postictal effects. The scale items serve as descriptions of the patient's perception of the impact of seizures on daily function without assigning points for each item. This format does not provide information about the exact number of seizures nor does this scale describe specific seizure classification, which would have to be documented by an epileptologist. Baker et al. (7) reported the test–retest reliability and validity of the scale (by comparing patients with different seizure types). Test–retest reliability (Pearson correlation coefficient) was 0.8 for the ictal/postictal scale and 0.79 for the percept scale, confirming consistency over time. Cronbach alpha values determining internal consistencies were 0.85 and 0.69, respectively. Pearson correlations between patients and their relatives or carers were 0.64 to 0.77 for four items drawn from the scale. The scale does not differentiate perception of control by seizure type but does reflect seizure severity effects. A revised version tested in the United States showed similar data: Alpha = 0.77 total, 0.80 ictal, 0.64 percept; test–retest = 0.75, 0.74, and 0.66, respectively (16). Baker et al. (11) reported on a revision of the scale, with four additional items (noted in Table 4) that were considered enhancements of validity and reliability.

Table 4. Liverpool Seizure Severity Scale Items
 Major seizuresMinor seizures
  • Adapted from Baker (7,11) with permission.

  • a

     Four items were added during 1998 revision.

  • b

     Item 9 moved from Percept to Ictal/Postictal subscale during 1998 revision.

Perception items  
 a1. How often have your attacks occurred at a particular time of day or night? 1 = always; 2 = usually; 3 = sometimes; 4 = never—my attacks occur at any time  
 2. When your attacks have happened, how often have you been able to tell when you will have them? 1 = always; 2 = usually; 3 = sometimes; 4 = never  
 b 3. How often have you been able to fight off your attacks? 1 = always; 2 = usually; 3 = sometimes; 4 = never  
 b 4. How often have you had an aura or warning with your attacks? 1 = always; 2 = usually; 3 = sometimes; 4 = never  
 b 5. In the last year, how much control have you had over your attacks? 1 = very good control; 2 = fairly good control; 3 = little control; 4 = no control  
 a6. When you have had attacks, how often have they occurred together in clusters, with quite long periods between each cluster? 1 = always; 2 = usually; 3 = sometimes; 4 = never  
 b 7. How often did your attacks occur when you were asleep? 1 = always; 2 = usually; 3 = sometimes; 4 = never  
 b 8. How many of the things you want to do have your attacks stopped you doing? 1 = all of them; 2 = a lot of them; 3 = a few of them; 4 = none of them  
Ictal/postictal items  
 b9. Overall, how severe have your attacks been in the past year? 1 = very severe; 2 = severe; 3 = mild; 4 = very mild  
 b10. In the last year, have you blanked out or lost consciousness during attacks? If yes, generally for how long? 1 = <1 min; 2 = 1–2 min; 3 = 2–5 min; 4 = >5 min; 5 = No, I have not blanked out or lost consciousness  
 b11. When you recovered from your attacks, how confused did you feel? 1 = very confused; 2 = fairly confused; 3 = slightly confused; 4 = not feel confused at all  
 b12. When you recovered from your attacks, did you feel confused? If yes, for how long? 1 = <1 min; 2 = 1–5 min; 3 = 6–60 min; 4 = >1 h; 5 = not feel confused at all  
 b13. When you had your attacks, how often did you fall to the ground? 1 = always; 2 = usually; 3 = sometimes; 4 = never  
 b14. When you recovered from your attacks, how often did you have a headache? 1 = always; 2 = usually; 3 = sometimes; 4 = never  
 a15. When you recovered from your attacks, how often did you feel sleepy? 1 = always; 2 = usually; 3 = sometimes; 4 = never  
 b16. When you recovered from your attacks, how often did you find that you had wet yourself? 1 = always; 2 = usually; 3 = sometimes; 4 = never  
 b17. When you recovered from your attacks, how often did you find that you had bitten your tongue? 1 = always; 2 = usually; 3 = sometimes; 4 = never  
 b18. When you recovered from your attacks, how often did you find that you had injured yourself (other than biting your tongue)? 1 = always; 2 = usually; 3 = sometimes; 4 = never  
 b19. When you had your attacks, how quickly could you usually return to what you were doing? 1 = <1 min; 2 = 1–5 min; 3 = 6–60 min; 4 = >1 h  
 a20. How often did you smack your lips, fidget, or behave in an unusual way during attacks? 1 = always; 2 = usually; 3 = sometimes; 4 = never  

Although the Liverpool scale was reported to show diminished seizure severity in a trial of LTG versus placebo, only the ictal subscale showed a statistically significant difference in one trial (17). However, the very small difference (19.5 vs. 20.6 points) cannot be considered clinically significant. The Percept Subscale has not been sensitive to change in three studies (16–18). Schering-Plough supported a detailed study of the Liverpool Seizure Severity and Quality of Life scales in which the patient-perception subscale neither met criteria for internal consistency nor was sensitive to differences among treatments (18). Glaxo-Wellcome commissioned Rapp et al. (16) to review the Liverpool Seizure Severity Scale in hopes of finding some positive data that might demonstrate effectiveness for LTG. When the scale was revised for use in the United States, eight of the 12 ictal items differentiated between generalized and partial seizure types, whereas only one of eight percept items did so (16). Their data demonstrated that the Percept Subscale added little to the Ictal Subscale because it did not function as designed.

One of the problems with the structure of the scale is the delineation of two types of seizures as “major” and “minor” seizures. The developers have reported scores only for major seizures, but it is not clear that patients have identified the same seizure type at follow-up as they designated at baseline. Another problem may be that some of the questions, particularly in the Percept Subscale, refer to trait (stable characteristic) rather than state (variable characteristic). These items are unlikely to change during a short clinical trial. As with the Liverpool scale, inclusion of items that occur infrequently (tongue biting, urinary incontinence) decreased the likelihood that the scores would change with therapy. Also similar to the Liverpool scale, an arbitrary 4-point scale is used, often with the points ascribed to “always,”“usually,”“sometimes,” or “never,” respectively. Note also that if only one third of patients have decreased seizure frequency during the short-term clinical trial, changes in seizure severity for responders will be diluted by lack of change among nonresponders. Very large sample size is needed to overcome the lack of significance or small mean changes.

The developers have recently discarded the Percept Subscale. The remaining Ictal Subscale is similar to the VA and NHS3 scales in its approach to the objective assessment of seizures. Although the Liverpool scale was widely adopted for use in European clinical trials during the past few years, no data from these studies have been published, suggesting a lack of significant findings. Whether the problem is related to the subjective nature of the questions asked of patients, or the lack of effectiveness of the drugs tested is not known. It would be interesting to gather the data from clinical trials in which the scale was not responsive to learn which items were useful.

Hague seizure severity scale

Carpay et al. (8) initiated development of a seizure severity scale for children as part of a doctoral dissertation project. The 13-item scale is based on parental perceptions of the child's seizures (Table 5). It was based largely on the Liverpool scale in content and in design as a self-administered questionnaire, to be completed by parents, not patients. Testing three groups of children with only one seizure type showed significant differences among the groups with minor, intermediate, and major seizures. The correlations between parents' and neurologists' scores were reasonable (r = 0.45; p < 0.0001) but not significant after stratification for seizure type (r = 0.10–0.26). In a conundrum resulting from the multiple analyses, the most frequent seizures had the lowest severity scales (i.e., frequent absence seizures). A visual analogue scale of severity used concomitantly with the Hague Scale was not correlated with seizure frequency (r = 0.28). No data are available about the sensitivity of the Hague Scale in a clinical trial setting.

Table 5. Hague Seizure Severity Scale
  • *

    Questions to be completed by a parent related to the child's seizures in the past three months.

  • Adapted from Carpay et al. (8), with permission.

1. How often do you notice a decrease of consciousness in your child during a seizure? 1 = always; 2 = usually; 3 = sometimes; 4 = never
2. For how long a time does such a decrease of consciousness last? (From the time of onset to time of normal consciousness) 1 = very long; 2 = long; 3 = short; 4 = very short
3. How severe have the seizures been overall? 1 = very severe; 2 = severe; 3 = mild; 4 = very mild
4. Were there any muscle jerks or cramps in the arms or legs during an attack? 1 = always; 2 = usually; 3 = sometimes; 4 = never
5. How long did the jerks or cramps last during an attack? 1 = very long; 2 = long; 3 = short; 4 = very short; 5 = not applicable, there are no jerks or cramps
6. How noticeable were the seizure symptoms? 1 = very noticeable, obvious to everyone; 2 = fairly noticeable, most people will notice an attack; 3 = not very noticeable, most people will not notice; 4 = not at all noticeable, you have to be very alert to notice an attack
7. During or after an attack, how often was your child confused? 1 = always; 2 = usually; 3 = sometimes; 4 = never
8. During an attack, how often did your child wet him/herself? 1 = always; 2 = usually; 3 = sometimes; 4 = never
9. During an attack, how often did your child bite his/her tongue? 1 = always; 2 = usually; 3 = sometimes; 4 = never
10. How often did your child become injured during an attack (other than biting the tongue)? 1 = always; 2 = usually; 3 = sometimes; 4 = never
11. Once the attack was over, was your child sleepy (include sleepiness caused by the use of a rescue medication such as diazepam.) 1 = always; 2 = usually; 3 = sometimes; 4 = never
12. After an attack, did your child complain of sickness, headache, and/or pain in the muscles? 1 = always; 2 = usually; 3 = sometimes; 4 = never or unknown, my child would not be able to complain about that
13. After an attack, how long a time did it take until your child could resume normal activity? 1 = very long; 2 = long; 3 = short; 4 = very short or immediately after an attack

COMPONENTS OF SEIZURE SEVERITY SCALES

  1. Top of page
  2. Abstract
  3. A HISTORY OF METHODS OF SEIZURE MEASUREMENT
  4. ADVANTAGES AND DISADVANTAGES OF SCALES
  5. COMPONENTS OF SEIZURE SEVERITY SCALES
  6. OVERALL LIMITATIONS
  7. FUTURE DEVELOPMENTS
  8. REFERENCES

Although the developers of each new instrument used different approaches to assess seizure severity, some basic components of seizures are assessed in all the scales. The approach to data collection is by interview of the adult patient and an observer (VA, NHS3, Hazard), the patient alone (Liverpool), or the parent of a child with seizures (Hague). The inclusion of an observer is standard clinical practice because it is often difficult for patients to describe their seizures. Reasons include poor memory and lack of awareness of the relationship of events (e.g., olfactory aura as a sensory component of a seizure). Patients can recall associated precipitating factors that they relate to seizure occurrence (e.g., missed medication, illness). Rather than attempting to document all clinical details of seizures, as can be done during inpatient closed-circuit television monitoring, an outpatient self-report system such as a seizure severity scale should be based on information collected from multiple sources (19).

Table 6 lists the types of items used in at least one of the scales, with much similarity among the approaches to seizure severity issues. In the following sections, the basic components of seizure severity are described, as used in each scale. The issue of whether each component might be sensitive to change as a result of treatment is addressed. The VA and Hague scales include frequency as part of the rating. Most of the scales allow patients to include different types of seizures, although a specific seizure type (e.g., CPS) may not be identifiable. One problem that most scales share is that scores are given separately for different seizure types. Without designation of seizure type, a patient who was previously having GTCS, CPS, and SPS monthly, who had a positive response to treatment with fewer seizures of each seizure type would still be represented, and the severity score might not change. Prolonged seizure duration often represents functional disability, and also probably is a good marker for treatment effect. Unfortunately, it may be difficult to measure accurately. Postictal dysfunction may be one of the most disabling features of seizures. Ictal and postictal phases are a continuum for the patient and observer who are not interested in the electroclinical definitions that separate these phases of a seizure.

Table 6. Aspects of seizure severity included in scales
ItemVANHS3HazardsLiverpoolHague
Seizure frequencyPoints based on frequencyLevel of frequency
Seizure type3 specific types: SPS, CPS, GTC3 nonspecific categories, generalized convulsion itemBy description2 nonspecific typesNonspecific
Seizure durationBy descriptionAltered function, confusionAltered consciousness
Postictal eventsSpecific levels: headacheBy descriptionConfusion, headache, sedationImpaired consciousness
Postictal durationTime until normalBy description3 items: Altered consciousness, confusion, altered functionImpaired consciousness, confusion, sleepiness, other problems
AutomatismsNo = reduce scoreSpecific levels
Seizure clustersYes = Score halfFrequency
Known patternYes = Reduce scoreBy descriptionFrequency
WarningYes = Reduce scoreFrequencyFrequency
Tongue biting, incontinenceFrequencyFrequencyFrequency
InjurySpecific levelsBy descriptionFrequencyFrequency
Impaired functionNo = Reduce scoreDurationBy descriptionDurationDuration
Other itemIf low drug levels or compliance = Reduce scoreAbility to abort seizures, frequencyFrequency, duration of jerks, cramps
Other itemIf preventable = Reduce score

Automatisms can be highly disabling because they are visible to all observers, who recognize that the patient is not acting normally. Severity of automatisms is an aspect of functional impairment. Some automatisms with obvious physical impairment can be deterrents to patients' ability to drive, work, or be alone in a public place. Patients who know that their seizures usually occur in clusters may be able to drive and work on days when no clusters occur. Some patients have seizures only at night or in the early morning during awakening. This is an advantage over the usually sporadic occurrence of seizures because these patients know that once they are fully awake and alert, they can drive and work without fear of a seizure during the day. Women who have catamenial epilepsy have a cyclic occurrence of seizures during the premenstrual phase. Patterns are helpful only if they are consistent, with all seizures occurring into the expected pattern. A consistent type of warning of an impending seizure is a valuable component if it allows the patient to avoid danger or embarrassment. In addition, it may have some relevance to treatment effect, as a treatment that slows seizure spread could theoretically allow increased warning time before loss of awareness. Tongue biting and urinary incontinence are often considered the hallmarks of a GTCS, although urinary incontinence can occur with CPSs. Many types of seizures involve loss of muscle tone, tonic posturing, or altered consciousness that lead to bodily harm. Typical injuries are broken bones, spinal fractures, burns, lacerations, muscle strain, etc. These seizures add to the medical burden of epilepsy because of the need for emergency care. All types of seizures have the potential to cause impairment of function. Even if consciousness is not altered during a simple sensory seizure, a patient may be distracted by the event. Some patients report an ability to abort seizures by using a variety of tactics, such as mental concentration or physical muscle tightening or biofeedback techniques. Many patients know that a specific precipitating factor can provoke a seizure. Common examples are lack of sleep, high fever, or use of alcoholic beverages. Erratic or inconsistent compliance with the prescribed drug regimen leads to low or subtherapeutic drug levels.

OVERALL LIMITATIONS

  1. Top of page
  2. Abstract
  3. A HISTORY OF METHODS OF SEIZURE MEASUREMENT
  4. ADVANTAGES AND DISADVANTAGES OF SCALES
  5. COMPONENTS OF SEIZURE SEVERITY SCALES
  6. OVERALL LIMITATIONS
  7. FUTURE DEVELOPMENTS
  8. REFERENCES

The similarities among existing scales suggest that there is a partial consensus about selected aspects of seizures that are important markers for severity. Nonetheless, developers have moved too quickly into use of the scales for clinical trials before adequate pilot testing. All the scales have limitations that fall into two general categories: responsiveness and appropriateness.

Responsiveness

Data on this aspect of scale validity are scanty. Several developers presented evidence of consistency and reliability. However, these tests are inadequate to determine suitability for determining treatment effect. As noted earlier, the unvalidated VA scale did demonstrate significant differences between CBZ and VPA of the same magnitude as other analyses of seizure frequency. The Liverpool scale showed differences between LTG and placebo in the first use of the scale. However, no differences between groups have been found in other clinical trials using that scale.

Reporting

The lack of information about a scale is accentuated by lack of reporting of negative data. Thus future investigators do not have adequate information to judge the limitations of a scale. Unfortunately, journal reviewers do not request results of seizure severity scales either because they are secondary outcomes, or because the use of the scale is not described in the Methods section. Experience with the scale is needed to develop an estimate of the difference expected between groups. This information is needed to establish the power of a clinical trial to detect significant change (using the sample size for the primary outcome).

Appropriateness

Seizure severity scales have developed along the lines of objective and subjective assessments. The VA, NHS3, and Hazard scales use objective ratings based not merely on the clinician's perspective, but also on the specific questions posed. The subjective section of the Liverpool scale (percept subscale) has been abandoned because of its poor psychometric performance (e.g., reliability). The Hague scale, a modification of the Liverpool scale, is further removed because it assesses the subjective views of the parents about their child's seizures. All the scales suffer from a limited number of questions that may or may not cover seizure severity for an individual. All the scales also suffer from an arbitrary scoring system. The VA scale hypothesized that a specific total score (50 points) would parallel a clinically appropriate need to change therapy. Other scores are open-ended with no clinical definition for the point scores (e.g., 50 points is unsatisfactory). Much work is needed to define how much change is clinically important.

Instrument use

Another issue of appropriateness is how the scale should be used in a clinical trial. Frequency of administration, consistency among interviewers, interpretation of marginal situations, and other problems complicate the use and scoring of scales. For example, the Liverpool scale allows patients to designate seizure types as major and minor. In follow-up, patients frequently have confused or altered the labels, making it impossible to compare baseline with follow-up responses. Instructions for administration of the questionnaires are essential for clinician, patient, and parent/observer. Inadequate instructions will lead to drift that can greatly affect the data.

Combining of elements

Most seizure severity scales, as they now exist, may represent a mixture of disability elements and elements that may change with treatment effect. Also, as can be seen from earlier examples, scale elements may or may not be relevant to given individuals. Even treatments that reduce seizure spread could cause some scales to have higher point values and others to have lower ones. Some items may not relate to treatment effect at all. It is no wonder that it has been difficult to use these severity scales in their entirety to demonstrate treatment effects.

A final caveat is the need to perform adequate pilot tests to determine the amount of change in scale scores that could be expected with a change in therapy. Use of a seizure severity scale as a secondary end point should be based on a reasonable expectation of statistically significant change. If the sample size will be inadequate to find differences between treatment groups, the scale should not be listed as a secondary end point. Analyses will be limited to exploration of the data for use in later trials.

FUTURE DEVELOPMENTS

  1. Top of page
  2. Abstract
  3. A HISTORY OF METHODS OF SEIZURE MEASUREMENT
  4. ADVANTAGES AND DISADVANTAGES OF SCALES
  5. COMPONENTS OF SEIZURE SEVERITY SCALES
  6. OVERALL LIMITATIONS
  7. FUTURE DEVELOPMENTS
  8. REFERENCES

The next step in assessment of seizure severity could be an entirely new approach. Among several possibilities is an anatomic–clinical correlation of electrical events. Knowledge of the origin or specific components of seizures, based on intracranial EEG, allows definition of the anatomy of each seizure based on the international classification of seizures. Thus, it is possible to determine the spread of an electrical event across the brain, recruiting additional areas as expressed in the clinical seizure. The “area under the curve,” amount of brain, or segments of brain involvement could be calculated as an assessment of seizure severity. If a new drug affected the spread of a seizure, the effect would be reflected in reduced severity based on this type of assessment. If the drug aborted the clinical expression of a seizure, this would also be reflected in the brain area involved. One reason that this type of assessment has not been practical is the lack of specificity of drug action on the brain (e.g., no drug can be considered optimal for temporal- vs. extratemporal-onset seizures). When a superior drug is developed, it will visibly affect seizure severity in terms of spread through the brain, as well as reduce (or stop) seizures) for many patients.

Another approach to severity is to develop a scale to assess the severity of epilepsy instead of seizures. Such a scale might parallel the Karnofsky Performance Scale (20) used in neurology or the Global Assessment of Functioning (GAF) scale (21) used in psychiatry. These scales evaluate the overall level of function of the individual. The GAF is a 100-point scale that incorporates psychological symptoms and occupational and social functioning in an overall psychosocial rating. The scores range from the sickest to the healthiest patient with definitions of severity by scores (e.g., 70 points or more indicates that the patient is functioning at an acceptable level). We could develop an Epilepsy Severity Scale that would attempt to define the difference between a disability, a handicap, and normal function for people with epilepsy. A 100-point scale could be developed with appropriate anchors to define the stages of improvement from total incapacity to normal function.

New ideas are needed to develop a novel seizure or epilepsy severity scale for the new millennium.

REFERENCES

  1. Top of page
  2. Abstract
  3. A HISTORY OF METHODS OF SEIZURE MEASUREMENT
  4. ADVANTAGES AND DISADVANTAGES OF SCALES
  5. COMPONENTS OF SEIZURE SEVERITY SCALES
  6. OVERALL LIMITATIONS
  7. FUTURE DEVELOPMENTS
  8. REFERENCES
  • 1
    Cramer JA. Seizure measurement in clinical trials. J Epilepsy 1998;11:15.DOI: 10.1016/s0896-6974(97)00104-7
  • 2
    Commission on Classification and Terminology of the International League Against Epilepsy. Proposal for revised clinical and electroencephalographic classification of epileptic seizures. Epilepsia 1981;22:489501.
  • 3
    Cramer JA, Smith DB, Mattson RH, et al., and the VA Epilepsy Cooperative Study Group. A method of quantification for the evaluation of antiepileptic drug therapy. Neurology 1983;33(suppl 1):26–37.
  • 4
    Duncan JS & Sander JWAS. The Chalfont Seizure Severity Scale. J Neurol Neurosurg Psychiatry 1991;54:8736.
  • 5
    O'Donoghue NF, Duncan JS & Sander JWAS. The National Hospital Seizure Severity Scale: a further development of the Chalfont Seizure Severity Scale. Epilepsia 1996;37:56371.
  • 6
    Janz D. How does one assess the severity of epilepsy? In: TrimbleMR, ed. Chronic epilepsy: its prognosis and management. New York: John Wiley, 1989:2136.
  • 7
    Baker GA, Smith DF & Dewey M, et al. The development of a seizure severity scale as an outcome measure in epilepsy. Epilepsy Res 1991;8:24551.
  • 8
    Carpay JA, Vermeulen J & Stroink J, et al. Seizure severity in children with epilepsy: a parent-completed scale compared with clinical attacks. Epilepsia 1997;38:34652.
  • 9
    Cramer JA. A clinimetric approach to assessing quality of life in epilepsy. Epilepsia 1993;34(suppl 4):S813.
  • 10
    Carpay HA & Arts WFM. Outcome assessment in epilepsy: available rating scales for adults and methodological issues pertaining to the development of scales for childhood epilepsy. Epilepsy Res 1996;24:12736.DOI: 10.1016/0920-1211(96)00013-7
  • 11
    Baker GA, Smith DS & Jacoby A, et al. Liverpool Seizure Severity Scale revisited. Seizure 1998;7:2015.
  • 12
    Mattson RH, Cramer JA & Collins JF, et al. Comparison of carbamazepine, phenobarbital, phenytoin, and primidone in partial and secondary generalized tonic-clonic seizures. N Engl J Med 1985;313:4551.
  • 13
    Wijsman DJP, Hekster YA, Renier WO & Meinardi H. Clinimetrics and epilepsy care. Pharmaceutisch Weekblad 1991;13:1828.
  • 14
    Mattson RH, Cramer JA & Collins JF, et al. A comparison of valproate with carbamazepine for the treatment of partial seizures and secondarily generalized tonic-clonic seizures in adults. N Engl J Med 1992;327:76571.
  • 15
    O'Donoghue MF, Duncan JS & Sander JWAS. The Subjective handicap of epilepsy: a new approach to measuring treatment outcome. Brain 1998;121:31743.DOI: 10.1093/brain/121.2.317
  • 16
    Rapp S, Shumaker S & Smith T, et al. Adaptation and evaluation of the Liverpool Seizure Severity Scale and Liverpool Quality of Life battery for American Epilepsy patients. QOL Res 1998;7:35363 and 467–77.
  • 17
    Smith D, Baker GB & Davies G, et al. Outcomes of add-on treatment with lamotrigine in partial epilepsy. Epilepsia 1993;34:31222.
  • 18
    Wagner AK, Keller SD & Kosinski M, et al. Advances in methods for assessing the impact of antiepileptic drug therapy on patients' health-related quality of life. QOL Res 1995;4:11534.
  • 19
    Cramer JA, Mattson RH. Quantitative approaches to seizure severity. In: Meinardi H, Cramer J, et al. eds. Quantitative assessment in epilepsy care. New York: Plenum Press, 1993:55–71.
  • 20
    Yates JW, Chalmer B & Mckegney FP. Evaluation of patients with advanced cancer using the Karnofsky Performance Status. Cancer 1980;45:22204.
  • 21
    Endicott J, Spitzer RL & Fleiss JL, et al. The global assessment scale: a procedure for measuring overall severity of psychiatric disturbance. Arch Gen Psychiatry 1976;33:76671.