Utility of the electroencephalogram

While the electroencephalogram (EEG) has been available to psychiatrists for over 30 years, its usefulness in psychiatry remains unclear. One study shows that the yield appears low in psychiatry, particularly for epileptic disorder which is fundamentally a clinical diagnosis (Stone & Moran, 2003). However, this contrasts with Fenton & Standage's (1993) finding that 92% of EEGs were useful in a psychiatric series.

We compared the requests from psychiatrists for EEGs with the corresponding report in 186 tests (patient group aged 16 years and above, trial over a 28-month period, target population 924 000). This information is held electronically, but we also inspected the original written request in a quarter of cases.

Clear abnormalities suggesting epilepsy or cerebral dysfunction were found in 15% of the study cohort (9% of <65 years old, 39% of ≥65 years old). We defined a test as being useful if it was either clearly abnormal or clearly normal and was likely to add diagnostic weighting in the context of the information on the request form; this usefulness was found in 37% of tests (32% in <65 years old, 55% in ≥65 years old). The apparent usefulness was reduced if suspected cases of epilepsy were excluded, which happened in 19% of tests (16% in <65 years old, 35% in ≥65 years old).

In terms of abnormal positive results for epilepsy, there were no tests supporting unsuspected epilepsy; however, 7 out of 96 in the younger group and 2 out of 26 in the older group did support suspected cases of epilepsy. For cerebral dysfunction, there were 5 out of 45 suspected and 2 out of 103 unsuspected instances in the younger group, and 2 out of 9 suspected and 11 out of 29 unsuspected in the older group. The division between suspected and unsuspected cases was dependent on the quality of the referral, which was often limited.

Our findings suggest that the EEG gives useful diagnostic information in a little over a third of cases. However, in practice the effect is likely to be reduced by such factors as the primacy of a clear clinical diagnosis in suspected epilepsy, the nature of the EEG report being usually suggestive rather than indicative, and the superiority of other investigations (e.g. neuroimaging) in certain situations. The EEG test remains important in the differential diagnosis of both possible cerebral dysfunction (encephalopathy) and seizures, as well as the monitoring of epilepsy. In order to keep the rate of uninformative tests to a minimum, clinicians should carefully describe the presenting signs and symptoms, considering whether these are consistent with epilepsy and whether other investigations are preferable. This information should be included in the EEG referral to improve the utility of the subsequent report.

Declaration of interest

L.P. was awarded Laughlin prize for outstanding performance in old format MRCPsych exam, Autumn 2007. He is also involved in writing a multiple choice questions' book for the new format MRCPsych.