Seizures and adverse events during routine scalp electroencephalography: A clinical and EEG analysis of 1000 records

Clinical Neurophysiology 118 (2007) 22–30 www.elsevier.com/locate/clinph

Seizures and adverse events during routine scalp electroencephalography: A clinical and EEG analysis of 1000 records Heather Angus-Leppan

*

Department of Clinical Neurosciences, Royal Free Hospital, University College London, Pond Street, London NW3 2QG, UK Accepted 27 August 2006 Available online 6 October 2006

Abstract Objective: To quantify the incidence of seizures and adverse events during standard electroencephalography (EEG). Methods: A retrospective random sample of 1000 of a total of 3391 reports of standard scalp EEG recordings during 2002 at Kings College Hospital were studied, and adverse events during standard EEG were recorded. Photic induced seizures and epileptiform activity were compared with the resting, hyperventilation and sleep EEG. Results: Adverse events occurred in 131 records (13.1%), including seizures in 60 records (45 electro-clinical and 15 non-epileptic seizures). The overall incidence of electro-clinical seizures was not statistically different during the resting EEG (2.8%), sleep EEG (2%), hyperventilation EEG (2.1%) and during photic stimulation EEG (1.4%). There was a higher frequency of electro-clinical seizures during hyperventilation and sleep in those with a diagnosis of idiopathic generalised epilepsy (31.5%) and during photic stimulation in photosensitive patients (31%). The incidence of electro-clinical seizures was significantly less during activation procedures in focal epilepsies (2.6%). Activation techniques made a unique diagnostic contribution when routine resting EEG was normal or equivocal in 11% of cases. Conclusions: Adverse events occurred in 13.1% of records, and most were minor. Sixty of the adverse events were seizures. Those generated during the EEG were brief and safety precautions operated successfully. In those without a prior diagnosis, the chance of seizures is the same during both resting and activation EEG. In those patients with generalised epilepsy or photosensitivity, activation procedures have a higher rate of seizure induction. Significance: This study has implications for informed consent for EEG. Ó 2006 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights reserved. Keywords: Adverse events; Seizures; Electroencephalography; Scalp EEG; Photic stimulation; Hyperventilation; Sleep EEG; Activation procedures

1. Introduction Electroencephalography (EEG) is a readily available functional test of cerebral activity used in the diagnosis, assessment and classification of epilepsy in adults and children. Standard scalp EEG is often carried out with activation techniques (hyperventilation, photic stimulation and sleep) to increase the diagnostic yield. For example, sleep studies increase the overall diagnostic yield of EEG from 50% to 80% (Marsan and Zivin, 1970).

*

Tel.: +44 2082165472; fax: +44 2074726829. E-mail address: [email protected]

There are studies on the risks of invasive EEG (Simon et al., 2003) and long term monitoring (Sanders et al., 1996), but little information on the risk of standard EEG. There are case reports of complications such as skin reactions (Day, 1968) and hyperventilation induced stroke in patients with sickle cell anaemia (Fatunde et al., 2000), but no published large clinical series or audits of complications of standard EEG were found. In the United Kingdom a seizure during an EEG prevents driving for a period determined by the driving regulatory body (DVLA, 2006). The British Society for Clinical Neurophysiology was unsuccessful in gauging the risk of seizures during photic stimulation using current protocols (Dr. Robin Kennett, personal communication). After examining the issue, a working party of the British

1388-2457/$32.00 Ó 2006 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.clinph.2006.08.014

H. Angus-Leppan / Clinical Neurophysiology 118 (2007) 22–30

Society for Clinical Neurophysiology recommended that all patients should be consented for EEG, in discussion with the referring doctor and with an information leaflet provided in advance. This warns specifically that photic stimulation may ‘‘very rarely cause a seizure in people that are sensitive’’ (British Society for Clinical Neurophysiology, 2003). It does not warn about seizures occurring at other times during the EEG. This study sought to quantify the incidence of seizures and adverse events during standard EEG recordings in a large series and to examine when these seizures occurred: whether during routine (non-activated) EEG, or during sleep, during hyperventilation or during intermittent photic stimulation (IPS). This information is important in informing subjects about potential risks of standard resting and activated EEG. This study has been reported in abstract form (Angus-Leppan, 2004). 2. Methods This was a retrospective study on computerised reports of standard EEG recordings done at Kings College Hospital for 2002. Reports comprised a factual report by an accredited trained technician, and a conclusion by a consultant epileptologist (neurophysiologist or neurologist with epilepsy training). 2.1. Terminology Adverse events were defined as an incident or occurrence from which potential or actual ‘‘harm resulted to a person receiving health care’’ (Stevens, 1986) occurring at or immediately around the time of the EEG. It does not imply that the EEG caused the event. Sometimes the event had benefits as well as potentially causing harm. For example a seizure during an EEG may clarify the diagnosis, classification and the correct treatment, as well as preventing driving for a year in the United Kingdom. Activation techniques are defined as methods used to elicit or accentuate abnormal cerebral activity (Kaufmann and Watson, 1949). Activation techniques used in this study were sleep, IPS and hyperventilation. The term standard EEG is used to refer to both routine EEG and combined routine and sleep EEG. Routine EEG refers to resting scalp EEG with or without hyperventilation and/or photic stimulation. Routine EEG without any activation procedure is referred to as a routine resting or non-activation EEG. Sleep EEG refers to scalp EEG recorded during spontaneous, sleep deprived or induced sleep. Seizures were classified as electro-clinical seizures, clinical seizures without EEG change, possible seizures and non-epileptic seizures. Possible seizures were events judged to have a moderate to high likelihood of being an epileptic seizure on combined clinical and electrical grounds, without conclusive evidence. Non-epileptic seizure was defined as a clinical event without EEG evidence of a seizure,

23

presumed to have a psychological or psychiatric mechanism. It was not used as a blanket term for all non-epileptic events, which were classified elsewhere. Sometimes there was strong positive evidence against an epileptic mechanism – for example, apparent loss of consciousness with preserved alpha rhythm. Seizures were subdivided according to whether they occurred during routine recording, sleep, hyperventilation, photic stimulation, or unspecified time. The term ‘‘epileptiform activity’’ (EA) was used to denote ‘‘spiky waveforms in epileptic subjects due to abnormal, excessive, hypersynchronous cerebral neuronal activity’’, (Binnie and Holder, 1999). Spiky waveforms are subjectively assessed as pointed and distinguished from background activity (Chatrian et al., 1983). EA consists of paroxysmal waveforms of cerebral origin with a high correlation with a seizure disorder (Westmoreland, 1985) subdivided into spikes (transients less than 80 ms in duration), sharp waves (transients 80–120 ms in duration), and spike and wave complexes (spike followed by slow wave of more than 120 ms) (Binnie and Holder, 1999) and less common types (slow spike and wave, atypical spike and wave discharges, paroxysmal rhythmic fast activity and hypsarrhythmia (Binnie and Holder, 1999). 2.2. Sample Demographic data on sex and type of EEG were recorded for all 3391 scalp EEG (1763 males and 1673 females) for 2002. There were 2380 (70%) routine EEG studies, and 1011 (30%) routine and sleep EEG studies. One thousand randomly selected records of routine EEGs were analysed in detail. Randomisation was done using computer generated random numbers for a discrete uniform distribution. This sub-group of reports was on 522 males, 477 females (one sex not recorded) ranging in age from two days to 101 years (mean 31.3 years, SD 20.9). Eighty-one were emergency (bedside) recordings. Within the 1000 records, repeat studies were done in five patients (a total of 10 records for these five patients). These were analysed within the total group. 2.3. Techniques Digital EEG recordings were carried out using the Maudsley electrode placement system, with one recording channel for concurrent electrocardiograph (ECG) recordings. In 855 of the 1000 reports, one or more activation procedures were recorded, after resting recordings of approximately 20 min. Sleep studies were considered satisfactory if recorded during sleep to at least Stages I or II (Hughes, 1982; Billiard, 1982). They were completed in 342 studies and incomplete in 10. Sleep studies, of variable duration, were usually performed using medication – in children over the age of two years, alimemazine tartrate and in adults, quinalbarbitone. In a few patients, sleep was spontaneous

24

H. Angus-Leppan / Clinical Neurophysiology 118 (2007) 22–30

rather than due to medication. Sleep studies were analysed along the same lines as the routine resting or non-activation EEG. Hyperventilation, completed in 580 records, was carried out for 3 min with the subject sitting up, breathing deeply and regularly at a rate of approximately 20–30 respirations each minute (Cooper et al., 1974), and adequacy of hyperventilation was noted in reports. It was routinely omitted in those over 50 years, with history of stroke or possible stroke and in those unable to co-operate because of illness or youth. IPS, using standard techniques (Topalkara et al., 1998; Binnie and Holder, 1999; Kasteleijn-Nolst Trenite et al., 1999), was complete in 732 recordings, and incomplete in 14 recordings. It was omitted in 254 recordings, including 81 bedside recordings, six patients who refused and five because of concern about inducing a seizure and loss of driving licence. Responses were classified as normal (following response, driving response, no change, no further information added), ‘‘incidental’’ epileptiform activity or electro-clinical seizure, or photo-paroxysmal responses (PPR). PPR in the EEG is defined as the occurrence of spikes or spike and wave responses in response to intermittent light stimulation. Responses were graded as Type 1 (spikes without occipital alpha rhythm), Type 2 (parietooccipital spikes with biphasic slow waves), Type 3 (parieto-occipital spikes spreading frontally) or Type 4 (generalised spike and wave and polyspike and wave outlasting the stimulus) (Waltz et al., 1992; Kasteleijn-Nolst Trenite et al., 1999). As the only sub-type with a high correlation with epilepsy, the term photosensitivity refers to Type 4 PPR, with the epileptogenic response outlasting the stimulus (Kasteleijn-Nolst Trenite et al., 1999). 2.4. Adverse events Adverse events reported in this study were those recorded by the accredited EEG technician or reporting consultant doctor or both. Events were noted independent of whether the EEG was causal. 2.5. Conclusion of the EEG reports Conclusions, reached in 602 records, were epilepsy, possible epilepsy and other diagnoses. Epilepsy was classified on the basis of seizure, epilepsy or syndromic type, dependent on the degree of diagnostic certainty (International League against epilepsy, 1989; Cenusa, 2002). Analysis of medication was not possible because these details were incomplete. There were no significant cardiac arrhythmias or conduction disturbances recorded in the ECG channel.

pathologically slow, EA, electro-clinical seizures and other seizures. Comparisons were made between the frequency of seizures and epileptiform activity on routine records, and each of the activation procedures using contingency tables and the v2 statistic (Statgraphics Version 7, STSC, 1996). 3. Results 3.1. Resting routine EEG It was normal in 652 reports. Abnormalities were borderline (four patients), in the background rhythm with or without EA, EA and electro-clinical seizures. 3.1.1. Epileptiform activity This was present in 282 records: sharp activity in 133 (no patients with seizures); mixed epileptiform activity in 72 (13 with seizures); spike wave activity in 44 (10 with seizures); spikes in 31 (10 with seizures); electrodecremental response in 2 (two with seizures). 3.1.2. Electro-clinical seizures Seizure types were convulsive status epilepticus (three), non-convulsive status epilepticus (NCSE) (three), absence (three), myoclonus (four), and others (15). Nine seizures occurred only during routine EEG recordings. Four occurred once during resting and once during activation procedures, two occurred three times during recording. In seven cases seizures were frequent (myoclonus, salaam spasms, focal motor seizures, absence), and six were continuous. Reports concluded that six of the 28 resting EEG with electro-clinical seizures represented primary epilepsy, seven were focal (one NCSE), eight were symptomatic generalised epilepsy (SGE) and seven were other diagnoses. The conclusions recorded in the reports are summarised below in Table 1. 3.1.3. Adverse events All 131 adverse events in the 1000 records are summarised in Table 2. All seizures, whether epileptic or non-epileptic, were considered as adverse events because of their potential harm (including driving implications). It was appreciated that concurrently they might have positive diagnostic effects. Apart from patients in status epilepticus, all recorded seizures terminated rapidly. 3.2. EEG during activation procedures

2.6. Analysis of data

3.2.1. Sleep EEG 342/352 sleep EEGs were completed. 172 were normal.

The results of each section of the EEG were analysed separately. The routine resting or non-activation EEG records were summarised as normal, borderline,

3.2.2. Epileptiform activity and electro-clinical seizures There was EA in 152 sleep records (44%), consisting of sharp waves (82), spikes (23), spike wave (40), a combination

H. Angus-Leppan / Clinical Neurophysiology 118 (2007) 22–30 Table 1 Conclusions recorded in 1000 random standard EEG reports Diagnosis

n

%

Normal Borderline Epilepsy – Possible – Classification uncertain – Generalised – Focal* – Status** Photosensitivity Non-specific slowing Other diagnoses

486 28 74 38 37 143 8 16 112 58

48.6 2.8 7.4 3.8 3.7 14.3 0.8 1.6*** 11.2 5.8

*

Including 42 with temporal lobe epilepsy, 15 with childhood benign partial epilepsies. ** Two with electrical status epilepticus of sleep. *** 1.6% of total records, 2.2% of those tested with IPS.

of EA (seven). In a further seven (2%), patients had electroclinical seizures (two were electrical status epilepticus of sleep). In 10 patients there was pathological slowing and one neonate had a flat EEG. 3.2.3. Hyperventilation EEG There were 119/580 abnormal hyperventilation records: including 39 borderline or asymmetric slowing. Epileptiform activity was sharp waves in 29, spike wave activity in 15, spikes in 12, mixed epileptiform activity in four.

Table 2 Number of records with adverse events Adverse events

n

No. of records

%

Reactions to barbiturate (nausea) Psychological or psychiatric reactions* Seizure – epileptic** – Routine/no activation – Sleep – Hyperventilation – By photic stimulation – During photic stimulation – Unspecified/other – Non-epileptic – Routine/no activation – Sleep – Hyperventilation – Photic stimulation – Unspecified

2 24 45 28 7 12 6 4 18 15 2 0 0 2 11

1000 1000 1000 1000 352 580 732 732 1000 1000 1000 352 580 732 1000

0.2 2.4 4.5 2.8 1.4 0.9 0.8 0.5 1.8 1.5 0.2 0 0 0.3 1.1

Unclassified motor and sensory events – total*** – Routine/no activation – Sleep**** – Hyperventilation – Photic stimulation – Unspecified

45

1000

4.5

0 3 5 11 26

1000 352 580 732 1000

0 0.9 0.3 1.5 2.8

*

Including 17 patients upset, nervous, depressed, and anxious. In some of these patients, more than one seizure occurred during the EEG. *** Including jerking without EEG change (16 records), twitching (2 records), tremor (3 records), dizziness (3 records), headache (2 records). **** Possible sleep apnoea in 2. **

25

Seizures occurred in 12 patients (nine mixed EA and three spike-wave activity). Seven patients had symptoms during hyperventilation that were not epilepsy: syncope and pre-syncope in five patients, anxiety in one patient, distress in one patient. No patients had tetany. 3.2.4. Photic stimulation EEG Photic stimulation was normal or non-contributory in 402/732 records, following responses were seen in 247, driving rhythms in 39 patients, and borderline abnormalities in two. Epileptiform activity – with photosensitivity and photoparoxysmal responses. Type 4 photosensitivity was found in 16 patients (2.2% of those tested). In three patients photo-paroxysmal responses of Types 2 and 3 were found. Nine of the 16 patients with definite photosensitivity were females. The age range of those reported was 3–31 years, with a mean of 17 years, younger than the overall mean of 31 years for the total of 1000 records. Only one of six patients in the series of 1000 had positive findings on extended photic testing, despite the clinical suspicion. In those with definite photosensitivity, 11/16 had EA in other parts of the EEG (five during resting EEG and hyperventilation, two during all sections of the EEG, two during resting EEG, one during hyperventilation). Epileptiform activity – without photosensitivity. EA without photosensitivity occurred in nine patients – spike wave activity in five patients, sharp wave activity in two, spike activity in two. Electro-clinical seizures. In seven photosensitive patients there was no specific comment on the clinical state during IPS. Five had seizures. Of those with seizures, four had myoclonic jerks and one dizziness with distress. In one patient, myoclonus occurred with IPS and during the routine (non-activation) record. No patients had convulsions. Diagnoses reached in photosensitive patients are summarised in the Table 3. In 14 of the patients with definite photosensitivity, there were no seizures during other sections of the EEG, in one patient there was a seizure during hyperventilation and in one during routine recording. One of three patients with a photo-paroxysmal response (Types 2–3) had a myoclonic seizure and the photoTable 3 Diagnoses in patients with photosensitivity and number with electroclinical seizures Classification or syndrome Idiopathic generalised epilepsy (IGE) Juvenile myoclonic epilepsy Juvenile absence epilepsy Huntington’s disease/cortical myoclonus Uncertain (focal epilepsy) Not specified Total

n

% of photosensitive

n with seizures

6 2 1 1 2 4

37 13 6 6 13 25

2 0 0 1 0 2

16

100

5

26

H. Angus-Leppan / Clinical Neurophysiology 118 (2007) 22–30

Table 4 Patients with ‘‘coincidental’’ seizures during photic stimulation Seizure type

When seizures

Epileptiform activity

Conclusion

Staring Hand wringing Myoclonus Shaking, restless

IPS IPS, sleep IPS, routine IPS, routine

IPS, routine, sleep (spike) IPS, sleep (spike wave) All (spike wave) IPS, routine, sleep

Frontal seizures IGE BME Focal epilepsy

IGE, Idiopathic generalised epilepsy; BME, benign myoclonic epilepsy of childhood; IPS, intermittent photic stimulation.

paroxysmal response took on a pathological significance. In this record, the sleep and routine (non-activation) records showed EA (spikes). There were four patients who had electro-clinical seizures during photic stimulation considered coincidental to IPS (Table 4). 3.2.5. Comparisons between epileptiform activity during routine resting EEG and during activation procedures Epileptiform activity was significantly more frequent during the non-activated routine EEG recordings (30.8%) than during the activation procedures (24.8%) (v2 = 7.91, p = 0.0049). Epileptiform activity detected only with activation procedures. In 31 of 855 patients having activation procedures (3.6%) epileptiform activity was seen with one or more activation procedures but not during the resting routine EEG recording. The sleep EEG showed epileptiform activity in all 31, and the yield of sleep EEG in this regard was significantly higher (v2 = 97.1 p = 0.0000), compared with hyperventilation (5/580, 0.08%) and photic stimulation (0/732, 0%). Abnormal resting routine EEG and normal activation procedures. In eight records the resting EEG was abnormal with normal activation procedures. One of these EEGs showed epileptiform activity (spikes). The rest showed slow activity. 3.2.6. Comparisons between electro-clinical seizures in routine resting EEGs and activation procedures Clinical seizures with electrical changes occurred in 2.8% of routine resting EEG records (0.6% of these were multiple), 2% of sleep EEG, 2.1% of hyperventilation records and in 1.4% of photic stimulation EEG. There was no statistical difference in the incidence of clinical seizures occurring during resting EEG (28/1000) and during all activation procedures (29/855) (v2 = 0.053, p = 0.81) Clinical seizures during activation procedures in different clinical groups. Seizures during activation procedures were significantly more frequent in photosensitive patients with photic stimulation (31%) and in generalised epilepsies with hyperventilation or sleep (40%) than in focal epilepsies (2.6%) (v2 = 35.4, p = 0.0000) (Table 5). There were no clinical seizures during hyperventilation in any of the 117 patients with focal epilepsies in this series. There was no significant difference in the frequency of clinical seizures during photic stimulation in photosensitive patients (31%), hyperventilation in generalised

epilepsies and sleep in generalised epilepsies (v2 = 3.1, p = 0.21). Electro-clinical seizures and epileptiform activity occurring only during activation procedures. In eight records electro-clinical seizures occurred during activation procedures but not at other times during the EEG (five during IPS, two during hyperventilation, one during sleep). In seven of these, EA was present in other sections of the EEG. The frequency of seizures that occurred only during the three activation procedures were not statistically different – IPS (5/732), Hyperventilation (2/580) and Sleep (1/342) (v2 = 1.097, p = 0.58) Clinical seizures both during the resting EEG and during activation procedures are shown in Table 6. Twelve of the 855 records (1.4%) with activation procedures had clinical seizures both during at least one of the activation procedures and during the resting, non-activated, routine EEG recording. All also had EA during at least one activation procedure and during the resting, non-activated, routine EEG. Clinical seizures only during routine resting EEG. Of the 855 cases with one or more activation procedures six had clinical seizures during the routine EEG and not during the activation procedure. In four the seizures were focal and two were generalised. A v2 test showed no significant difference (v2 = 0.29, p = 0.59) between the frequency of clinical seizures occurring only during routine resting EEG (6/855) and only during activation procedures (8/855). 4. Discussion Adverse events occurred in 13.1% of records, and 60 of these were seizures (45 records were one or more electroTable 5 Clinical seizures during activation procedures in generalised and focal epilepsies Seizures

n

No. of seizures

Incidence (%)

Type of activation

Focal Generalised total IGE JME SGE

117 22 16 3 3

3 9 6 0 3

2.6 40 37.5 0 100

Sleep = 1, Photic = 2 Sleep = 3, HV = 6 Sleep = 1, HV = 5 Sleep=2, HV=2

n, number having activation procedures; IGE, idiopathic generalised epilepsy; JME, juvenile myoclonic epilepsy; SGE, symptomatic generalised epilepsy.

H. Angus-Leppan / Clinical Neurophysiology 118 (2007) 22–30

27

Table 6 Clinical seizures during resting EEGs and during activation procedures Sex

Age (years)

Resting EEG

Sleep

HV

IPS

Diagnosis

M F M M M M F M M M M M

9 29 1.5 31 0.5 1 17 36 2 3 8 2

Brief clonic Absence Myoclonus Clonic Spasm Vacant Jerk Stupor Myoclonus Focal motor Motor Focal motor

Not done Not done Myoclonus Clonic Spasm EA Not done Stupor Myoclonus Focal motor Not done Focal motor

Brief clonic Absence Not done Not done Not done Not done Not done Not done EA EA Motor Not done

Normal Incomplete Incomplete Normal EA Vacant Jerk/PS Not done Myoclonus Focal motor Normal Not done

SGE Unclassified Myoclonus SGE SGE Unclassified IGE NCS BME Partial Unclassified Unclassified

M, male; F, female; HV, hyperventilation; IPS, intermittent photic stimulation; PS, photosensitive; EA, epileptiform activity; BME, benign myoclonic epilepsy of childhood; NCS, non-convulsive status epilepticus; SGE, symptomatic generalised epilepsy.

clinical seizures, and 15 were non-epileptic seizures). Eight were status epilepticus, present before the EEG recording had commenced, and not causally linked to the EEG. All other seizures were brief and self-terminating. Electro-clinical seizures occurred with similar frequency during resting EEG, and activation procedures. Of those seizures triggered by activation, there was no significant difference in the incidence between photic stimulation, hyperventilation and sleep. The chance of a seizure during activation was significantly higher for photic stimulation in patients with definite photosensitivity (5/16, 31%) and for hyperventilation and sleep in those with a diagnosis of generalised epilepsy having activation procedures (9/22, 37.5%). In this last group, 19 were idiopathic epilepsies (IGE and JME); and six of these (31.5%) had seizures triggered by activation (five by hyperventilation and one by sleep). In contrast, only 2.6% of those with focal epilepsy had seizures during activation (one during sleep and two during photic stimulation). EA was present in about half of the sleep records, and this was significantly more than during photic stimulation or hyperventilation. In 31 of 855 (3.6%), EA occurred during activation but not during the resting EEG; all 31 showed EA during sleep. Four hundred and eighty-six records were normal. Of the 300 records consistent with epilepsy, 74 were considered possible, 143 focal, 37 generalised, 38 were of uncertain classification and eight had status epilepticus.

In a study of a general population at a county hospital in the United States, the overall rate of abnormalities (56% of all 1326 EEG studied) was similar to the rate of 51% seen in this study (Hughes, 1960). The rate of photosensitivity in the Hughes study was also similar (3% of all records) to that of the current study (2.2% of those tested). This suggests that the current study looked at a population similar enough to a general population to draw conclusions that apply to a general population referred to an EEG department rather than only to a tertiary EEG referral centre. Data was not available in many cases on points which may have influenced results of activation procedures in particular – for example, medications, sleep, time of last meal and referral diagnosis. The first three factors can play an important role in the effect of activation procedures, and the reproducibility of repeat EEGs (Topalkara et al., 1998). No follow up was carried out in the current study. In particular, the impact of the EEG results, its predictive value and the follow up of seizures and final diagnosis would have added to the utility of this study. The recordings during different EEG sections vary in duration, usually with sleep > resting > hyperventilation > IPS in duration. In this study, sections of the EEG were compared independent of their duration as this best reflects clinical practice. However, on a ‘‘yield per minute’’ basis, the highest yield would be for hyperventilation, then IPS, then resting EEG and the lowest yield would be during the sleep study. 4.2. Adverse events

4.1. Methodology It is possible that the random sample of 1000 of the total 3391 records for the year 2002 was unrepresentative. However age and sex were compared in the two groups and found similar, making a significant difference less likely. The referral base at the hospital studied includes a large number of tertiary referrals and it would not be surprising if the rate of abnormalities was higher in this group than in departments with more primary and secondary referrals.

All negative events occurring at the time of procedure were considered in this study as adverse events of EEG recording. Causality cannot be assumed for all of them. In two patients sleep apnoea was reported, but not caused by the EEG. Eight were patients who had status epilepticus and were too unwell for activation procedures; the resting EEG did not cause these seizures, as they were in status before commencement of the recording. Apart from these, all other patients having seizures in this series had a rapid

28

H. Angus-Leppan / Clinical Neurophysiology 118 (2007) 22–30

and full recovery. There were no tonic-clonic seizures triggered by photic stimulation, a potential complication previously described (Bickford et al., 1952). Observed seizures were important in making a definite diagnosis so that they are adverse events only in one sense, contributing to correct and potentially life-saving treatment in some patients. Other adverse events were minor. No published series of adverse events during non-invasive EEG were found for comparison. 4.3. Routine EEG The incidence of epileptiform activity during routine EEG was higher than during activation procedures. Activation procedures highlighted abnormalities not found in other sections of the recordings in 31/855 patients (3.6%). The risk of having a clinical seizure is similar in all phases of the EEG recordings. Interpretation of this is complex, as some patients did not have activation procedures. In some, it was impossible to perform activation because of the clinical state (drowsiness, status or coma in patients precluding hyperventilation and photic stimulation). Hyperventilation was omitted for safety reasons in patients over the age of 50 years. In others, photic stimulation was not done because of driving implications, patient refusal or discomfort. The impact of this missing data is uncertain. It is possible that the groups having activation procedures were different from those who did not. Traditionally, in undiagnosed patients, the selection is towards performing activation procedures in those most likely to have epilepsy. This would skew the results towards more abnormalities during activation procedures, but this did not occur. The study does not allow further analysis of the reason for the relatively low yield from activation procedures. Activation procedures provided unique information in those with generalised epilepsy and photosensitive patients. Activation procedures added less information in focal epilepsy than primary epilepsy, a finding well established in the literature dating back to Melsen in 1959, and corroborated by others since (Melsen, 1959; Binnie and Holder, 1999). As the overall rate of definite primary epilepsy in this study was only 37/1000 (3.7%), and of those only 22 had activation procedures, the incidence of abnormality in the present study is likely to be lower than in cohorts with a higher incidence of primary generalised epilepsy. 4.4. Sleep EEG Sleep EEG is reported to increase the yield in those later shown to have epilepsy from 50% to 80% on average (Billiard, 1982; Binnie and Holder, 1999). The incidence of epileptiform activity during sleep was approximately 50% higher than in other circumstances in this series. There may be some selection bias – a combined routine and sleep EEG was planned in advance in

one-third of patients, targeting referrals suggesting a high chance of epilepsy. The incidence of clinical seizures was similar during sleep to other periods of the EEG. Part of this may be due to difficulties in observing subtle clinical seizures during sleep. 4.5. Hyperventilation EEG Hyperventilation was reported as unpleasant or causing dizziness, and in one case syncope, in 7/580 (1.2%) of patients studied. In 2/580 (0.3%) of patients who had clinical seizures during hyperventilation, they did not occur at other times – this was 2/45 (4.4%) of the total yield of records with seizures. In 5 patients, epileptiform activity was seen during hyperventilation but not in the resting EEG, where it added unique information. 4.6. Photic stimulation EEG Photic stimulation was helpful in at least 15% of abnormal cases in a series of 1324 patients having IPS as well as resting EEG (Hughes, 1960). An additional 6% of abnormal cases with bilateral abnormalities showed a significant unilateral depression of photic responses and provided evidence for the side maximally involved. He argued that inclusion of photic stimulation in routine EEG, requiring a relatively short amount of time, increased of the yield of EEG valuably. In the current study, photic stimulation contributed to the therapeutic yield of the EEG significantly in 54/514 abnormal records (10.5 %), a number of records showing photosensitivity, seizures and/or epileptiform activity. This figure is slightly lower than in Hughes’ study. One possible explanation is that some of the patients who did not have IPS may have shown significant abnormality if tested. In support of this explanation, it was difficult to predict which patients were photosensitive: only one of six patients selected for prolonged photic stimulation showed an abnormality. In the current study, the ictal manifestations in 5/16 patients during photosensitivity were subtle, some objective and some subjective. Comments were made on the patient’s clinical state during IPS in eight reports, and it is possible that some mild ictal manifestations were missed. Seizures were recognised as such by clinical observations including questions about symptoms and simple cognitive testing by the EEG Technician, who must also observe for patient safety, observe the EEG and instruct the patient. There are not the facilities during standard EEG recordings to test for transient cognitive impairments and subtle clinical seizures (Binnie and Marston, 1992). Self reporting is unreliable, as many patients are unaware of the photic triggers for their seizures (Kasteleijn-Nolst Trenite, 1998). It has previously been shown that most patients with epileptiform discharges induced by IPS showed subtle ictal signs but were unaware of them (Kasteleijn-Nolst Trenite et al., 1987).

H. Angus-Leppan / Clinical Neurophysiology 118 (2007) 22–30

4.7. Information for patients Photosensitivity as defined in this study (Type 4) has a 90% correlation with clinical seizures (Reilly and Peters, 1973) and therefore provides definite information. Only 25% of subjects with photosensitivity lose it, and, if present, it allows the patient to avoid triggers (Harding and Jeavons, 1994). In some countries such as the United Kingdom, photic induced seizures during an EEG preclude driving for one year. This was the reason for omitting photic stimulation in five patients in the current study. This could reduces the diagnostic yield of the EEG and patients may miss the benefit of specific advice about avoidance of photic triggers – advice which can result in seizure avoidance in 40% of patients with photosensitivity, and seizure reduction in 70% of patients (Jeavons and Harding, 1975; Harding and Jeavons, 1994). It would be interesting to perform a prospective study on EEG adverse events and the use of activation studies, to see if BSCN guidelines have changed the practice of EEG. The occurrence of EA in the EEG without recognisable clinical concomitance would not necessarily preclude driving. However, transient cognitive impairment can occur with brief discharges, and on occasions this has been shown to interfere with function (Binnie and Marston, 1992; Binnie, 1996, 2001). The distinction between EA, transient cognitive impairment and clinical seizures is not clear-cut. This study does not support the practice of warning unselected patients specifically about activation procedures, and in particular photic stimulation. The overall incidence of clinical seizures and complications was the same for all phases of the EEG. There are particular groups at high risk of seizures triggered by activation procedures. In this study, patients with known photosensitivity had a seizure rate during IPS of 31%. The literature suggests that this rate is an underestimation because subtle clinical events during IPS may be missed (Kasteleijn-Nolst Trenite, 1998). IPS may be an important way of checking the effectiveness of medication and the ongoing risk of seizures in photosensitive patients, but special caution is needed in this group. This involves vigilance on the part of the EEG Technician to ensure that IPS is terminated when a generalised discharge occurs, to reduce the chance of a convulsion. The second safety feature is the use of a modified protocol (Topalkara et al., 1998) starting with lower risk frequencies in those with known photosensitivity. In this study, safety measures were successful in avoiding convulsions in photosensitive patients, and this provides important reassurance for patients. Patients with generalised epilepsies having hyperventilation or sleep activation had a 40% incidence of seizures during these activation procedures. The figure was 31.5% for the primary generalised epilepsies, very similar to the risk of photic stimulation in photosensitive patients in this series.

29

There is no rationale for warning undiagnosed patients specifically about the risk of clinical seizures during activation procedures as opposed to the resting EEG. Of the 45 electro-clinical seizures in 1000 random records eight occurred only during activation procedures, but there was no significant difference in the incidence between photic stimulation 10/732), hyperventilation (12/580) and sleep (7/342). There was a much higher, and similar, risk of electroclinical seizures in photosensitive patients during photic stimulation (31%) and in idiopathic generalised epilepsies during hyperventilation and sleep (31.5%) compared with the overall population and with focal epilepsies where the incidence with activation procedures was only 2.6%. Patients with these diagnoses should be specifically warned about this risk. The incidence of seizures with photic stimulation was 31% in the 2.2% of patients with photosensitivity, but this may be an underestimate, as subtle clinical signs could be missed. Activation procedures added unique diagnostic information on epileptiform activity in a total of 3.6% records, not available from the resting EEG. Eight of the total electro-clinical seizures (18%) occurred only in activation records. Photosensitivity was seen in 16/732 (2.2%). Activation records made a unique diagnostic contribution, when resting EEG was normal or equivocal, in 55/514 (11%). This study suggests that information given to patients about the risk of EEG should incorporate the following information: - there is little chance of EEG triggering a full-blown convulsion - there is about a one-third chance of having a seizure during activation procedures in primary epilepsy or photosensitivity - activation techniques add unique information in 11% of EEG recordings. Competing interests None declared. Acknowledgements My thanks to Professor Colin Binnie, Dr Gonzalo Alarcon, and Dr Roberto Guiloff for advice. References Angus-Leppan H. The risk of activation procedures during standard electroencephalography (EEG). A study of 1000 records. J Neurol Neurosurg Psychiatry 2004;75:3. Bickford RG, Sem-Jacobsen CM, White PT, Daly D. Some observations on the mechanism of photic and photo-Metrazol activation. Electroencephal Clin Neurophysiol 1952;4: 275–82.

30

H. Angus-Leppan / Clinical Neurophysiology 118 (2007) 22–30

Billiard M. Epilepsies and the sleep wake cycle. Sleep and epilepsy. New York: Academic Press; 1982, p. 269-86. Binnie CD. Cognitive effects of subclinical EEG discharges. Neurophysiol Clin 1996;26:138–42. Binnie CD. Cognitive performance, subtle seizures, and the EEG. Epilepsia 2001;42(Suppl. 1):16–8. Binnie CD, Holder DS. Electroencephalography. In: Vinken PJ, Bruyn GW, editors. The Epilepsies Part I. Amsterdam: Elsevier; 1999. p. 283–317. Binnie CD, Marston D. Cognitive correlates of interictal discharges. Epilepsia 1992;33(Suppl. 6):S11–7. British Society for Clinical Neurophysiology. EEG information leaflet. British Society for Clinical Neurophysiology. 2003. Available from: www.bscn.org.uk. Cenusa, M. Bedeutung der durch Photische Stimulation ausgelosten epileptiformen Potentiale bei Anfallspatienten. 1-65. Medizinischen Fakultat der Universitat Zurich. 2002. Thesis/Dissertation. Chatrian GE, Bergamini M, Dondey DW, Klass M, Buchthal L, Petersen I. A glossary of terms most commonly used by clinical electroencephalographers. In: IFSECN, editor. Recommendations for the practice of clinical neurophysiology. Amsterdam: Elsevier; 1983. p. 11–27. Cooper R, Osselton JW, Shaw JC. EEG technology. London: Butterworths; 1974. Day JL. EEG skin damage. Med Res Eng 1968;7:10–1. DVLA. At a glance guide to the current medical standards of fitness to drive – A guide for medical practitioners. 2006. Available from: www.dvla.gov.uk. Fatunde OJ, Sodeinde O, Familusi JB. Hyperventilation-precipitated cerebrovascular accident in a patient with sickle cell anaemia. Afr J Med Med Sci 2000;29:227–8. Harding GFA, Jeavons PM. Photosensitive epilepsy. London: MacKeith Press; 1994. Hughes JR. Usefulness of photic stimulation in routine clinical electroencephalography. Neurology 1960;10:777–82. Hughes JR. EEG in clinical practice. London: Butterworths; 1982. International League against epilepsy. Proposal for revised classification of epilepsies and epileptic syndromes. Commission on classification and terminology. Epilepsia 1989;30:389–99.

Jeavons PM, Harding GFA. Photosensitive epilepsy. A review of the literature and a study of 460 patients. London: Heinemann; 1975. Kasteleijn-Nolst Trenite DG. Reflex seizures induced by intermittent light stimulation. Adv Neurol 1998;75:99–121. Kasteleijn-Nolst Trenite DG, Binnie CD, Meinardi H. Photosensitive patients: symptoms and signs during intermittent photic stimulation and their relation to seizures in daily life. J Neurol Neurosurg Psychiatry 1987;50:1546–9. Kasteleijn-Nolst Trenite DG, Binnie CD, Harding GF, Wilkins A. Photic stimulation: standardization of screening methods. Epilepsia 1999;40(Suppl. 4):75–9. Kaufmann IC, Watson CW. Brief review of methods used to elicit or accentuate abnormalities in the electroencephalogram. Electroencephalogr Clin Neurophysiol 1949;1:237–40. Marsan CA, Zivin LS. Factors related to the occurrence of typical paroxysmal abnormalities in the EEG records of epileptic patients. Epilepsia 1970;11:361–81. Melsen S. The value of photic stimulation in the diagnosis of epilepsy. J Nervous Mental Dis 1959;128:508–19. Reilly EL, Peters JF. Relationship of some varieties of electroencephalographic photosensitivity to clinical convulsive disorders. Neurology 1973;23:1050–7. Sanders PT, Cysyk BJ, Bare MA. Safety in long-term EEG/video monitoring. J Neurosci Nurs 1996;28:305–13. Simon SL, Telfeian A, Duhaime AC. Complications of invasive monitoring used in intractable pediatric epilepsy. Pediatr Neurosurg 2003;38:47–52. Stevens EA. Phase IV studies and post marketing surveillance. In: Glenny H, Nelmes P, editors. Handbook of clinical drug research. Oxford: Blackwell; 1986. p. 127–50. Topalkara K, Alarcon G, Binnie CD. Effects of flash frequency and repetition of intermittent photic stimulation on photoparoxysmal responses. Seizure 1998;7:249–55. Waltz S, Christen HJ, Doose H. The different patterns of the photoparoxysmal response – a genetic study. Electroencephalogr Clin Neurophysiol 1992;83:138–45. Westmoreland BF. The electroencephalogram in patients with epilepsy. Neurol Clin 1985;3:599–613.