Abstracts

Most instruments ask patients about their quality of life (QOL) in the last four weeks or less. On the other hand, clinicians are interested in the effect of epilepsy therapies over months and years, not weeks. Researchers often report QOL results as the difference between two measures, each looking only at a few weeks of the patient[apos]s life. The clinical meaningfulness of such data is uncertain in epilepsy, which is characterized by fluctuating seizure frequency, severity, seizure clustering, and medication side effects. The alternative, asking frequent 4-week questions, is impractical and burdensome. It is argued that recall bias could invalidate results from questionnaires spanning longer periods of time. However this has not been formally assessed. Based on previous work, we now examine whether commonly used QOL epilepsy tools with extended time frames of 2 and 4 months reflect the patients[apos] QOL during that time.
Patients with epilepsy were seen at the London Health Sciences Centre. Two cohorts of 40 patients each completed valid, monthly questionnaires asking about their QOL in the last 4 weeks, for 2 months (cohort 1) and for 4 months (cohort 2). During the last month patients also answered a set of the same questionnaires, but with an extended time frame, asking them to report QOL over the preceding 2 months (cohort 1) or 4 months (cohort 2). The questionnaires used were QOLIE-89, QOLIE-31, SF-36, and the Liverpool battery including Drug Toxicity, Impact of Epilepsy, and Seizure Severity. For each cohort, we used Pearson correlations to assess the association between a) the scores from the typical monthly questionnaires and the extended time frame questionnaires (2 and 4 months), and b) the scores from each typical monthly questionnaire and the last monthly questionnaire. The hypothesis that the first set of correlations was higher than the second set was tested using a 1-tailed Type I error of 5%, as described by Meng, et al (Psychological Bulletin 1992;111:172-5).
The mean age was 33.7 (SD=13.1) in cohort 1, and 34.4. (SD=12.6) in cohort 2. Thirty eight percent and 33% were females, respectively in cohorts 1 and 2. Twenty eight of 36 correlations (78%) were higher for the extended time frame questionnaires than for the monthly questionnaires. In 13 (35%) this difference was statistically significant, favouring the extended time frame questionnaires. None of the correlations that were higher for the monthly questionnaires were statistically significant. Correlations varied somewhat among different QOL instruments, suggesting that results of extended time frame questions may vary depending on the domain of QOL being explored.
The results suggest that questionnaires asking patients to rate their QOL over the preceding 4 months QOL (rather than the typical 4 weeks) may adequately reflect QOL during this time. Pending confirmation, this could have important implications for the assessment and interpretation of QOL in randomized trials of epilepsy interventions.
[Supported by: LHSC internal research fund]