Introduction

Top Abstract Introduction Methods Results Discussion References

Electroencephalography (EEG) has been readily available to clinicians throughout the United Kingdom for more than 30 years. It has gained a reputation as a diagnostic test with a range of indications, but its real uses and limitations are not widely appreciated and its usefulness in clinical practice has never been properly evaluated.¹

What can EEG do?
A single interictal EEG often helps to classify epilepsy ^{2 3} and can provide support for a diagnosis of epilepsy when the clinical features are highly suggestive. However, EEG is rarely, if ever, the sole determinant of this diagnosis.⁴ EEG can help to predict the risk of recurrence after a first seizure⁵ and the risk of relapse after drug withdrawal.⁶ Uncommon indications include the differentiation between functional psychoses (presenting with altered consciousness) and organic confusional states, and demonstration of characteristic periodic discharges in rare encephalopathies (subacute sclerosing panencephalitis, sporadic or acquired Creutzfeldt-Jakob disease). A single procedure cannot diagnose or exclude epilepsy, assess the severity of epilepsy or the response to treatment,⁷ diagnose or exclude the presence of a brain tumour, or differentiate dementia from pseudodementia.⁴

Is EEG being used effectively?
There is unrestricted access to EEG in most medium sized district general hospitals throughout the United Kingdom. Most requests for EEG come from non-specialists,⁴ and only 60 out of 250 neurophysiology departments in the United Kingdom are staffed by adequately trained neurophysiologists (personal communication). Therefore there is considerable potential for unnecessary requests and misinterpretation of the results.

	Methods

Top Abstract Introduction Methods Results Discussion References

Retrospective audit
We reviewed the case notes and request forms of 368 patients who underwent EEG in one district general hospital between August 1994 and January 1995. In the four patients who had more than one procedure we included only the first record.

Analyses
We analysed the data to establish if there were differences in requesting practice among users of the service. We also determined the quality of information provided by single interictal EEG in various scenarios. We assessed the role of EEG in establishing the diagnosis. This was categorised as classification of epilepsy, diagnosis of epilepsy, diagnosis of another specific condition, and as justifiable or inappropriate. We considered that requests were inappropriate when a diagnosis had been made on clinical grounds and the subsequent ordering of an EEG could not have produced useful information or when there had been an unsatisfactory attempt to achieve a clinical diagnosis, usually failure to obtain an eyewitness account of the patient's attacks. We assessed the usefulness of the procedure, defined as influencing management, (classification or diagnosis of epilepsy, diagnosis of another specific condition, and when a correctly interpreted EEG influenced decision making), justifiable, or inappropriate.

Intervention
We presented the results of the retrospective audit to the clinicians at two district general hospitals. At these meetings we presented guidelines and circulated them to all doctors using the service. Under the heading "indications" we included several clinical categories. Potential findings and their impact on management decisions were explainedfor example, in someone with a confident diagnosis of epilepsy which was difficult to classify on clinical grounds an EEG revealing either a focal or generalised epileptiform abnormality would assist classification and choice of treatment. Similarly we categorised "not indications" and discussed misconceptions underlying common reasons for requestsfor example, "blackouts, exclude epilepsy," where a normal reading does not exclude epilepsy and overinterpretation of non-specific abnormalities carries a risk of misdiagnosis.

Prospective audit
Referring clinicians had agreed to try to reduce the proportion of inappropriate requests to between 35% and 40%. The prospective audit entailed the review of notes and request forms from 241 patients undergoing EEG between March and August 1996. No patients had more than one EEG. No patients underwent EEG in both the retrospective and prospective components of the audit. We used ² tests to assess association between the group (retrospective, prospective) and categorical variables such as requesting clinician, reason for request, results, and usefulness.

	Results

Top Abstract Introduction Methods Results Discussion References

Retrospective audit
Reasons for requesting EEG varied between clinicians. While neurologists concentrated on patients with epilepsy and seizures (62%), other doctors used the procedure as a diagnostic toolfor example, 134/166 (81%) patients with "funny turns" or aggressive outbursts were referred for "exclusion of epilepsy." In these 166 patients the results were normal in 107 (65%), non-specifically abnormal in 42 (25%), and showed epileptiform discharges in only three (2%). Three children who presented with inattentiveness had generalised spike and wave, indicating a diagnosis of childhood absence epilepsy.

Prospective audit
There was a significant change in the requesting practice after intervention (²=33.85, df=5, P<0.0001), mainly because of a relative decrease in requests by physicians (table 2). Similarly there was a significant change in the nature of requests (²=21.90, df=6, P=0.0013) mainly because of a reduction in the number of requests in patients with funny turns or aggressive outbursts (table 2). There was a concomitant change in the distribution of EEG results (²=11.12, df=3, P=0.01), where the relative proportion of non-specifically abnormal results fell considerably (table 3). Table 3 also shows the change in usefulness of EEG (²= 26.99, df=2, P<0.0001).

	Discussion

Top Abstract Introduction Methods Results Discussion References

This retrospective audit of use of EEG in a district general hospital showed clear differences in requesting practice between clinicians. A large proportion of these requests were considered to be inappropriate because of the prevalent misconception that the EEG is a useful diagnostic tool in various clinical settings, notably in patients with "funny turns." A secondary finding was that a quarter of these procedures showed non-specific abnormalities. While this study did not specifically examine the issue, evidence exists that overinterpretation of these minor abnormalities contributes to erroneous diagnosis of epilepsy. ^{11 14 15}

The EEG department at the hospital where this study was carried out is staffed by a full time EEG technician, access to the test is unrestricted, and records are reported by a local general practitioner. Before and including the period of the retrospective audit, the number of procedures performed (about 800 a year) was consistent that seen in a survey in the Thames region carried out by the Association of British Clinical Neurophysiologists.⁴ This service, and its use, is typical of that seen in district general hospitals throughout the United Kingdom and the results of this original work are likely to be generalisable.

Intervention
The intervention involved an educative, non-confrontational approach with time taken to demonstrate to clinicians the limitations and pitfalls of the interictal EEG and that in patients with "funny turns" they had usually achieved a diagnosis on clinical grounds alone without the aid of EEG. That the prospective audit was successful reflects the quality of communication and cooperation between investigating team and the users of the service.

Conclusions
We have shown that by using an educative approach sustainable change in practice can be achieved, which has benefits for patients, clinicians, and service provision. Through greater understanding of its limitations clinicians have adopted a more selective EEG requesting policy. Doctors may have a lower risk of making, and patients may have lower risk of receiving, a misdiagnosis of epilepsy. The reduction in the number of unnecessary procedures releases technical capacity which can be used in the conduct of other investigations. Systematic replication of this work, possibly on a regional basis, would yield savings which would permit development of accessible local neurophysiology services.

	Acknowledgments

Julie Jones and Clare Jowett in the audit departments of Wrexham Maelor and Royal Shrewsbury Hospitals facilitated the meetings and distribution of guidelines. We specially thank the users of the service for listening and changing their practice

Contributors: DS had the original idea for the audit, helped with data collection and analysis, and is guarantor for the study. RB did most of the data collection. RP reported on all the EEGs. The paper was jointly written by DS, RB, RP, and BT.

	Footnotes

Funding: None.

Competing interests: DS presented the data at an educational meeting sponsored by GlaxoWellcome, for which he received payment.

	References

Top Abstract Introduction Methods Results Discussion References

1.	Fowle AJ, Binnie CD. Uses and abuses of the EEG in epilepsy. Epilepsia 2000; 41(suppl 3): 10-18[Medline].
2.	Commission on classification and terminology of an international league against epilepsy. Proposal for classification of epilepsies and epileptic syndromes. Epilepsia 1989; 30: 389-399[Medline].
3.	King MA, Newton MR, Jackson GD, Fitt GJ, Mitchell LA, Silvapulle MJ, et al. Epileptology of the first seizure presentation; a clinical, electroencephalographic and magnetic resonance imaging study of 300 consecutive patients. Lancet 1998; 352: 1007-1011[CrossRef][Medline].
4.	Binnie CD, Prior PF. Electroencephalography. J Neurol Neurosurg Psychiatry 1994; 57: 1308-1319[Abstract/Free Full Text].
5.	Berg AT, Shinnar S. The risk of recurrence following a first unprovoked seizure. Neurology 1991; 41: 965-972[Abstract/Free Full Text].
6.	MRC Antiepileptic Drug Withdrawal Study Group. Prognostic index for recurrence of seizures after remission of epilepsy. BMJ 1993; 306: 1374-1378.
7.	Binnie CD. Electroencephalography. In: Laidlaw J, Richens A, Chadwick D, eds. A textbook of epilepsy. 4th ed. Edinburgh: Churchill Livingstone, 1992:277-278.
8.	Ajmone-Marsan C, Zivin LS. Factors related to the occurrence of typical paroxysmal abnormalities in the EEG records of epileptic patients. Epilepsia 1970; 11: 361-381[Medline].
9.	Bridgers SL. Epileptiform abnormalities discovered on electroencephalographic screening of psychiatric in-patients. Arch Neurol 1987; 44: 312-316[Abstract/Free Full Text].
10.	Gregory RP, Oates T, Merry RTG. Electroencephalogram epileptiform abnormalities in candidates for aircrew training. Electroencephalogr Clin Neurophysiol 1993; 86: 75-77[CrossRef][Medline].
11.	Riley TL. Normal variants in EEG that are mistaken as epileptic patterns. In: Gross M, ed. Pseudoepilepsy. Lexington KY: Heath, 1983:25-27.
12.	Nicolaides P, Appleton RE, Beirne EM. EEG requests in paediatrics: an audit. Arch Dis Child 1995; 72: 522-523[Abstract/Free Full Text].
13.	Binnie CD. EEG audit: increasing cost efficiency of investigations in epilepsy. Electroencephalogr Clin Neurophysiol 1990; 76: 29P[CrossRef].
14.	Gibbs J, Appleton RE. False diagnosis of epilepsy in children. Seizure 1992; 1: 15-18[CrossRef][Medline].
15.	Smith D, Dafalla B, Chadwick DW. The misdiagnosis of epilepsy and the management of refractory epilepsy in a specialist clinic. Q J Med 1999; 15-23.

(Accepted 14 February 2001)