Abstracts

Rationale: Misinterpreted EEGs are a commonly recognized problem. They can lead to unnecessary treatment and adverse outcomes. The causes of misinterpretation are likely complex. One component of this may be insufficient training during neurology residency.Methods: As part of the EEG training in the Boston VA Healthcare System, neurology residents participated in an automated computer-based EEG instruction program. The residents reviewed and interpreted EEGs that were graded previously by consensus of three epileptologists. The program compared the residents’ interpretations with the attendings’ consensus, and provided feedback. The program allowed for review of residents’ recognition and interpretation of EEG findings. This program contained 30 EEGs representing a range of findings. Twelve of the 30 were classified as normal and 18 were abnormal. Of the 30 EEGs, 7 had generalized slowing, 12 had focal slowing, 9 had epileptiform discharges, and 5 had seizures. Two normal EEGs had non-epileptic seizures (NES).Results: In total, 742 EEG interpretations of the 30 EEGs were completed by 34 senior neurology residents (18 PGY-3 and 16 PGY-4). Residents took, on average, 1 hour for each EEG interpretation. Residents identified generalized slowing with 63% sensitivity and 85% specificity, focal slowing with 49% sensitivity and 79% specificity, and focal epileptiform discharges with 58% sensitivity and 86% specificity. Importantly, of the 128 EEG interpretations with events (seizure or NES) resident sensitivity for seizures was 65% and specificity was 67%. Sensitivity dropped to 48% when only considering very focal seizures (n=69). Overall, the residents correctly classified abnormal EEGs as abnormal with 80% sensitivity but only 45% specificity.Conclusions: To our knowledge, these findings are the first to objectively examine the ability of neurology residents to review and interpret specific EEG findings. Since these results represent the evaluations of a large cohort of senior-level residents from two separate adult neurology programs, we expect them to be generalizable to other residencies. The EEGs were selected to represent an array of findings, but may not represent the distribution of findings seen in general practice. Nevertheless, the ability to identify these patterns should be considered necessary for any physician interpreting EEG. Although this cohort of residents identified abnormal EEGs with a high sensitivity, they tended to overcall normal EEGs, resulting in low specificity. This is consistent with the known problem of overinterpretation. Additionally, residents were provided with ample time for interpretation, and we would expect reduced performance if common clinical time restrictions were in place. Although their training period was limited, they did show some learning trends. The first steps to improvement are identifying problems and creating outcome measures. Standardized EEG interpretation education with quantified performance could be a major contributor to improvement.