Article Text
Abstract
Introduction Little is known about the quality of medical record data used to evaluate febrile seizures despite its influence on vaccine recommendations.
Methods Parent report from phone interviews and mailed questionnaires was compared to abstracted medical records of 110 children with febrile seizures between ages 3 and 60 months. Concordance between parent report and medical records for characteristics and predisposing factors of febrile seizures was assessed by per cent total agreement and κ statistic. Sensitivity of medical record and parent report was determined setting parent report and medical record as the standard, respectively (Tisnado, 2006).
Results Per cent total agreement between medical record and parent report was good for many variables studied (62.5% had >70% agreement). However, κ was low for all measures (100% had κ <0.33) For seizure characteristics, parent report was more sensitive than medical record (69.9% (95% CI 50.0 to 88.9) and 31.9% (95% CI 18.2 to 45.7), respectively) while both were sensitive for fever (91.0% (95% CI 85.4 to 96.6) vs 91.9%(95% CI 86.6 to 97.3), respectively). Neither was sensitive for predisposing factors (62.1% (95% CI 41.5 to 81.8)) and 64.1% (95% CI 44.0 to 83.3), respectively).
Conclusion These data demonstrate the limitations of using only medical records or parent report to assess febrile seizures. Collection of information from both sources is required to most accurately portray the spectrum of predisposing factors and seizure characteristics of childhood febrile seizures.