Abstract
Objective: To compare the health-related quality of life (HRQL) of a nonsurgical sample of adults with epilepsy with that of age- and gender-equivalent norms, and to analyze the relative importance of seizure frequency, time since last seizure, gender, and comorbidity on HRQL in the epilepsy sample.
Methods: Data were obtained from 139 adults with epilepsy from three US centers and published norms on the Medical Outcomes Study Short-Form 36 (SF-36). Patients were classified according to number of seizures over the prior 4 weeks (zero, one to five, six or more). Bivariate and multivariate modeling was used.
Results: HRQL scores for seizure-free patients were similar to the general population. Significant differences between seizure frequency groups were found for seven domains and the physical and mental component summary scales of the SF-36 (p < 0.001). No differences were found in bodily pain. The largest differences were in physical role and social functioning, and general health (p < 0.001). In the multivariate model, seizure frequency was a significant inverse predictor of HRQL across all domains (p < 0.01 to 0.001). Men reported poorer physical function than women (p < 0.05), and patients with a comorbid condition had poorer HRQL in the areas of pain (p < 0.05) and general health perception (p < 0.01). Time since last seizure was not related uniquely to HRQL.
Conclusions: Seizure-free adults can have HRQL levels comparable with those of the general population. As seizure frequency increases, patients report more impaired HRQL, regardless of time since last seizure, gender, and comorbid status. Potential for difficulties in HRQL should be considered in clinical assessment and in evaluating treatment outcomes.
There is evidence to suggest that adults with epilepsy experience significant impairment in their health-related quality of life (HRQL).1-14 The psychological and social domains appear to be particularly problematic, with patients reporting high levels of anxiety and depression, poor self-esteem, and problems with social interaction and involvement.1-8 According to Wagner et al.,9 adults with epilepsy have significantly poorer HRQL than the general population across multiple mental and physical domains, including mental health, emotional role functioning, social functioning, vitality, physical role functioning, and general health perceptions. The extent to which these effects vary across patients with different seizure frequencies is not known. Perhaps those with more frequent seizures are experiencing severe impairment in HRQL whereas those with well-controlled epilepsy are maintaining a quality of life similar to that of the general population.
The results of several studies3,10-13 indicate patients who are seizure free for 1 year fare better in terms of HRQL than those with persistent seizures during this time period, with the most remarkable differences appearing in mental health, physical functioning, social activity, and general health perception. Patients averaging more than one seizure per month during a 1-year period report poorer HRQL than patients averaging less than one seizure per month.3,12 Mental health, social functioning, general health perceptions, and epilepsy-specific concerns (seizure worry, medication effects, health discouragement, and work, driving, and social function) appear to be most sensitive to the effects of ongoing seizure activity.3,12,13
These results suggest there are demarcations in seizure frequency that can be linked to differences in HRQL, and that certain domains of HRQL are more sensitive to the impact of seizure frequency over 1 year. However, the extent to which the HRQL of seizure-free patients is comparable with that of the general population is not known. Furthermore, treatment outcomes are often evaluated over shorter time intervals; 4-week assessments are not uncommon. These observation periods call for a seizure frequency classification system that is HRQL sensitive over a short period of time and robust to the presence of the potentially confounding effects of time since last seizure,14 comorbid conditions, and gender.15-18
The purpose of this study was to examine the relationship between HRQL and seizure frequency during the previous 4 weeks in a nonsurgical sample of adults with epilepsy. Specifically, the intent was to compare the HRQL of three frequency groups (seizure free, one to five seizures, and six or more seizures during the past 4 weeks) with that of age- and gender-equivalent norms, and to test the relationship between seizure frequency and HRQL while concomitantly considering the effects of time since last seizure, comorbidity, and gender. The extent to which HRQL varied by time since last seizure in the seizure-free group was also examined.
Methods.
Data gathered during a previous cross-sectional study of self-administration versus telephone administration of the Quality of Life in Epilepsy Inventory-89 (QOLIE-89) were used in this study. The methods and results of the administration comparison are reported elsewhere.19 Unless otherwise indicated, data used in the current study were gathered through self-administration.
Sample.
Patients were recruited through three epilepsy centers in the United States (Boston, Chapel Hill, and Richmond). Convenience sampling was used, with eligible patients identified by clinic personnel through patient records and clinic visits. Criteria for inclusion were as follows: 18 years of age or older, diagnosis of epilepsy for a minimum of 1 year, prescribed antiepileptic drug (AED) at the time of recruitment, and seizure free for the previous 24 hours or return to normal functioning at the time the questionnaires were completed. Patients were excluded if they could not complete the instruments for reasons of blindness, hearing impairment, or other physical disability precluding participation, or if they could not speak or read English sufficiently to complete the questionnaires. In addition, patients with psychiatric or neurologic disorders that would impair judgment or impact quality of life beyond the effects caused by epilepsy (including but not limited to dementia, mental retardation, stroke, head injury, cerebral palsy, tremor, schizophrenia, and current clinical depression) or who had undergone any form of epilepsy surgery were excluded. Patients who were prescribed any of the following medications for reasons other than AED therapy were also excluded from the study: anxiolytics, sedatives, hypnotics, antidepressants, antipsychotics, narcotics, and tranquilizers (exclusive of medications prescribed on an as-needed basis, such as sleeping medication).
Measures.
Health-related quality of life.
Medical Outcomes Study Short Form-36 (SF-36). The SF-36, which serves as the generic core of the QOLIE-89, was used to evaluate HRQL. This 36-item measure is made up of eight subscales, each evaluating a different domain of HRQL: physical functioning, physical role, bodily pain, general health, vitality, social functioning, emotional role, and mental health. Subscale scores are calculated according to standard procedures, yielding score values of 0 to 100; higher scores indicate better quality of life.20 The eight subscales can be aggregated into the two composite indices (Z scores): physical component summary (PCS) score and mental component summary (MCS) score according to standard procedures.21 These summary indices are often used as primary HRQL end points in clinical trials, for conceptual simplicity, or to avoid the necessarily stringent probability levels that are applied when performing independent tests on multiple end points.
The SF-36 has been shown to be a reliable and valid generic measure of HRQL, and has been used as an outcome indicator in numerous clinical trials and across various diagnostic groups, including adults with epilepsy.20-22 The internal consistency reliability levels in this study ranged from 0.74 (emotional role) to 0.93 (physical functioning). The alpha level for the PCS score was 0.91, and was 0.87 for the MCS score. These values suggest minimal measurement error associated with HRQL evaluation.
Normative SF-36 values. Normative values for the general population were derived from Ware et al.20,21 These published data were estimated from responses to the 1990 National Survey of Functional Health Status using respondents drawn from the 1989 and 1990 General Social Survey of noninstitutionalized adults in the United States.20,21 The general population values are based on a sample size of 2,474, with age- and gender-stratified groups ranging in size from 71 (18 to 24-year-old men) to 413 people (women, age 65 years and older).20,21 For the purposes of this study, general population norms were adjusted for the age and gender distribution of the sample by assigning each patient a normative value corresponding to the mean value from his or her age- and gender-equivalent group.
Sociodemographic and clinical variables.
Patients were asked the number of seizures they had experienced during the past month and, during the telephone interview, the length of time since their last seizure. Based on their response to the seizure frequency question, patients were categorized into one of three groups: seizure free, one to five seizures, and six or more seizures during the past 4 weeks. Patients were also queried about the presence of asthma, cancer, diabetes, heart disease, hypertension, and kidney disease, with space provided for listing other comorbid conditions.
Analytical approach.
SF-36 subscale scores and the PCS and MCS scores for the participants in this study were compared across seizure frequency group and with norms for the general population using multivariate analysis of variance (MANOVA). Two analyses were performed: one for the eight subscales and one for the PCS and MCS composite scores. In each case, significant overall test statistics (p < 0.05) were followed by post hoc analyses to determine which subscale or composite scores were showing group differences, and which specific groups were significantly different from one another.
MANOVA procedures were also used to evaluate the relative impact of seizure frequency, time since last seizure (continuous variable), presence of at least one comorbid condition, and gender. The eight subscale scores comprised the first model; the PCS and MCS were represented in the second model. Interaction effects were included in the models. If these effects were not significant, they were dropped and the models were rerun with the main effects only. The extent to which HRQL varied with time since last seizure in the seizure-free group was examined using Pearson’s product–moment correlation coefficient.
Results.
Sample.
A total of 139 patients completed the SF-36. Demographic and clinical characteristics of the sample are provided in table 1. Patients were taking a variety of AEDs, including carbamazepine, divalproex sodium, gabapentin, lamotrigine, phenobarbital, phenytoin, and primidone.
Sample sociodemographic and clinical characteristics
Distribution of the sample by seizure frequency was as follows: seizure free (n = 49), one to five seizures (n = 46), and six or more seizures (n = 40). Median time since last seizure for seizure-free patients was 365 days (mean, 800 ± 1,128 days; mode, 730 days). Median time for those in the one-to-five seizure category was 6.5 days (mean, 17 ± 26 days; mode, 4 days), and for those in the six-or-more seizure frequency category the median time since last seizure was 1.5 days (mean, 3.3 ± 5.3 days; mode, 1 day). More than one-quarter of the sample had one (n = 29, 21%) or more (n = 10, 7%) comorbid conditions.
HRQL: SF-36 profile.
The SF-36 profile for patients with epilepsy by seizure frequency group, together with the profile of age- and gender-equivalent adults from the general population, are provided in table 2. The overall test statistic was statistically significant (p < 0.001) for the eight subscales, indicating there was a relationship between group membership (norm, seizure frequency) and HRQL. Post hoc analyses showed group differences in all of the domains of HRQL measured by the SF-36, with the exception of bodily pain. Physical, social, and emotional role functioning and general health subscales were particularly sensitive. Mean scores of the seizure-free group were similar to those of the general population across all domains, except for vitality, which was significantly higher in the seizure-free group than in the general population. Patients experiencing seizures during the previous 4 weeks (one to five or six or more) reported significantly poorer HRQL than the seizure-free patients and general population norms in physical role functioning, general health, social functioning, and emotional role functioning. Patients experiencing six or more seizures reported significantly poorer HRQL than those with one to five seizures among six domains: physical functioning, general health, mental health, and physical, social, and emotional role functioning.
HRQL scores for the general population and adults with epilepsy (by seizure frequency)
The overall test statistic was also significant for the two component scores: PCS and MCS. Post hoc analyses indicated both scales were sensitive to group differences (see table 2). PCS and MCS scores for the seizure-free patients were similar to those of the general population. MCS scores for patients experiencing one to five seizures during the previous month were also similar to the general population. Patients reporting six or more seizures during the previous 4 weeks had significantly poorer mental and physical HRQL than any of the other groups.
Seizure frequency, time since last seizure, gender, and comorbid conditions.
MANOVA for the eight SF-36 subscales indicated significant overall effects for seizure frequency (F = 2.65, p < 0.0001), gender (F = 2.24, p < 0.05), and the presence of a comorbid condition (F = 2.08, p < 0.05) on HRQL. Time since last seizure was not significant (F = 0.49, p = 0.86). Post hoc analyses indicated seizure frequency was a significant predictor of HRQL across all of the SF-36 subscales (p < 0.01 to 0.001). Main effects for comorbidity and gender varied by domain: The presence of a comorbid condition was associated with significantly poorer HRQL in the areas of pain (F = 5.73, p < 0.05) and general health perception (F = 9.60, p < 0.01). Gender differences were found in physical functioning, with women reporting significantly better HRQL in this domain than men (F = 4.64, p < 0.05).
In the component score (PCS and MCS) analyses, seizure frequency (F = 6.00, p < 0.0001) and the presence of a comorbid condition (F = 2.08, p < 0.05) were significant predictors of HRQL, whereas time since last seizure (F = 0.05, p = 0.95) and gender (F = 2.86, p = 0.06) were not. Post hoc analyses indicated seizure frequency was a predictor of both the physical (F = 8.05, p < 0.001) and the mental (F = 7.48, p < 0.001) components of HRQL. The presence of at least one comorbid condition was associated with poorer physical HRQL (F = 8.05, p < 0.01), with no effect on mental HRQL.
Bivariate correlations between time since last seizure and HRQL in the seizure-free group were near zero and were not statistically significant. Coefficients were −0.05 for the PCS and 0.02 for the MCS, and ranged from 0.01 (general health) to −0.19 (emotional role) for the eight subscales.
Discussion.
The results of this study support the premise that adults with epilepsy, as a group, have poorer HRQL than the general population. However the data also indicate that adults with epilepsy who are seizure free can have HRQL levels comparable with those of the general population. It is important to note that this study sample excluded patients with comorbid psychiatric disorders, including depression; thus, the HRQL levels in this sample may be higher than the epilepsy population as whole. Nonetheless, patients with persistent seizure activity reported severe impairment across multiple domains of HRQL. The results of this study suggest that analyses of HRQL data that do not take into account seizure frequency may be overestimating the HRQL impairment of seizure-free patients and underestimating the impairment of those experiencing one or more seizures per month.
Seizure frequency is used as a clinical assessment tool, and reduction in seizure frequency is considered a key outcome measure in clinical practice and in trials of antiepileptic therapy.1 The meaning of seizure reductions to the patients themselves has yet to be defined. Devinsky et al.13 pointed to the need for additional research into the classification of seizure frequency to reflect variations in multiple domains of HRQL and to gain insight into the meaning patients assign to seizure frequency. Recently, the Commission on Outcome Measurement in Epilepsy recommended that additional research be undertaken to determine the changes in seizure frequency and severity important to the social domain of HRQL.1 Results of the current study indicate the difference in the perception of HRQL between patients experiencing one to five seizures and those who experience six or more seizures during a 1-month period can be substantial, even when the potentially confounding effects of time since last seizure, gender, and comorbid conditions are controlled. The sensitivity of social function to seizure activity observed in this study was consistent with studies evaluating seizure frequency over 12 months.10,13,22 However, the results of this study also indicate seizure frequency has a significant impact on multiple domains of HRQL, with a particularly strong effect on physical role functioning and general health perceptions.
Although intergroup data cannot be equated with intragroup change over time, these findings suggest that reducing seizure frequency by only a few seizures could have a dramatic impact on HRQL. Prospective studies of HRQL among patients experiencing a change in seizure frequency are required before this relationship can be established. Additional research on the patients’ perspective of clinically meaningful change in seizure frequency and the tradeoff between reduction in seizure frequency and side effects of medication would provide important information on patient motivation to adhere or seek a change in the treatment regimen.
These results also suggest that additional study of the sensitivity of the PCS and MCS scores to change is needed before these aggregate scores are used as primary HRQL end points in a clinical trial. The PCS score was as sensitive to seizure group differences as the subscale scores. However, the MCS did not detect differences between population norms and patients with one to five seizures per month, despite the fact that subscale score differences were substantial between these two groups. Perhaps individual subscale scores are more appropriate than aggregate scores for evaluating the mental health dimensions of HRQL in this population. Condition-specific measures should also be more sensitive to the HRQL effects of changes in seizure frequency, including worries about seizure occurrence, medication effects, and work, driving, or social concerns. The QOLIE-89 is an example of a measure that integrates the generic SF-36 and the specific issues faced by those with epilepsy.7
One could argue that the evaluation of seizure frequency and time since last seizure by patient recall was a limitation of the study. A prospective study with direct observation or monitors to verify that a seizure occurred might have provided a more accurate estimate of these variables. However, clinical assessment is generally based on patient self-report. The fact that seizure frequency was clearly associated with HRQL suggests that frequency is more than a count of occurrences. Rather, it represents a series of events and perceptions, including seizure severity, duration, sequelae, setting, and the perception of social reaction, all of which can affect quality of life adversely. The factors contributing to the perception of seizure severity and HRQL in persons with epilepsy is an area in need of additional research.
Acknowledgments
Acknowledgment
The authors thank Robert Beach, MD, PhD, and Sue Lannon and Gaye Harris of the Department of Neurology, University of North Carolina Hospitals, Chapel Hill, NC; John Pellock, MD, Kathy O’Hara, and Sally Elliott of the Comprehensive Epilepsy Clinic, Medical College of Virginia, Richmond, VA; and Steven Schachter, MD, and Tracey Reardon of the Comprehensive Epilepsy Center, Beth Israel Hospital, Boston, MA, for data collection. They also thank Chris Thompson for computer programming and Dr. Michael Halpern of MEDTAP International Inc. for his comments on an earlier version of this paper.
Footnotes
-
Funded by Astra Pharmaceuticals, Inc.
- Received August 14, 1998.
- Accepted February 13, 1999.
References
Letters: Rapid online correspondence
REQUIREMENTS
You must ensure that your Disclosures have been updated within the previous six months. Please go to our Submission Site to add or update your Disclosure information.
Your co-authors must send a completed Publishing Agreement Form to Neurology Staff (not necessary for the lead/corresponding author as the form below will suffice) before you upload your comment.
If you are responding to a comment that was written about an article you originally authored:
You (and co-authors) do not need to fill out forms or check disclosures as author forms are still valid
and apply to letter.
Submission specifications:
- Submissions must be < 200 words with < 5 references. Reference 1 must be the article on which you are commenting.
- Submissions should not have more than 5 authors. (Exception: original author replies can include all original authors of the article)
- Submit only on articles published within 6 months of issue date.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- Submitted comments are subject to editing and editor review prior to posting.
